diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 0746d3a020370d243bd761f365aecf03b3c83475..0e49465f3dc9c4617186ba73cbafa0699598e986 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": "20260512T120027Z", - "generated_at": "2026-05-12T12:09:53Z", + "snapshot_id": "20260512T180027Z", + "generated_at": "2026-05-12T18:10:15Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1426,7 +1426,7 @@ "cluster_id": "cluster-43824-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #43824.", "status": "open", - "confidence": 0.75, + "confidence": 0.55, "canonical_issue_number": 43824, "canonical_pr_number": 43836, "issue_numbers": [ @@ -1437,8 +1437,7 @@ 43842 ], "evidence_types": [ - "closing_reference", - "shared_issue_target" + "closing_reference" ], "pr_comparisons": [ { @@ -1684,6 +1683,201 @@ } ] }, + { + "cluster_id": "cluster-45861-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 45861, + "issue_numbers": [], + "pr_numbers": [ + 45861, + 45903 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45861, + "right_pr_number": 45903, + "code_similarity": 0.617, + "size_similarity": 0.516, + "file_overlap": 1.0, + "area_overlap": 0.377, + "patch_similarity": 0.546, + "shared_filenames": [ + "src/transformers/models/qwen3_5/modeling_qwen3_5.py", + "src/transformers/models/qwen3_5_moe/modeling_qwen3_5_moe.py", + "src/transformers/models/qwen3_omni_moe/modeling_qwen3_omni_moe.py", + "src/transformers/models/qwen3_vl/modeling_qwen3_vl.py", + "src/transformers/models/qwen3_vl/modular_qwen3_vl.py", + "src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", + "tests/models/qwen3_vl/test_modeling_qwen3_vl.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/qwen3_5/modeling_qwen3_5.py", + "left_ranges": [ + [ + 146, + 154 + ] + ], + "right_ranges": [ + [ + 146, + 154 + ], + [ + 161, + 183 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_5_moe/modeling_qwen3_5_moe.py", + "left_ranges": [ + [ + 142, + 150 + ] + ], + "right_ranges": [ + [ + 142, + 150 + ], + [ + 157, + 179 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_omni_moe/modeling_qwen3_omni_moe.py", + "left_ranges": [ + [ + 1315, + 1323 + ] + ], + "right_ranges": [ + [ + 1315, + 1323 + ], + [ + 1330, + 1352 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl/modeling_qwen3_vl.py", + "left_ranges": [ + [ + 354, + 362 + ] + ], + "right_ranges": [ + [ + 354, + 362 + ], + [ + 369, + 391 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl/modular_qwen3_vl.py", + "left_ranges": [ + [ + 294, + 302 + ] + ], + "right_ranges": [ + [ + 32, + 38 + ], + [ + 270, + 292 + ], + [ + 311, + 319 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", + "left_ranges": [ + [ + 865, + 873 + ] + ], + "right_ranges": [ + [ + 865, + 873 + ], + [ + 880, + 902 + ] + ] + }, + { + "filename": "tests/models/qwen3_vl/test_modeling_qwen3_vl.py", + "left_ranges": [ + [ + 17, + 26 + ], + [ + 29, + 35 + ], + [ + 45, + 78 + ], + [ + 522, + 547 + ] + ], + "right_ranges": [ + [ + 17, + 26 + ], + [ + 29, + 35 + ], + [ + 45, + 79 + ], + [ + 523, + 582 + ] + ] + } + ] + } + ] + }, { "cluster_id": "cluster-44018-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", @@ -2005,7 +2199,7 @@ 43842 ], "target_issue_number": 43824, - "reason": "PRs in cluster-43824-3 are treated as duplicates because they converge on issue #43824 with closing_reference, shared_issue_target evidence." + "reason": "PRs in cluster-43824-3 are treated as duplicates because they converge on issue #43824 with closing_reference evidence." }, { "cluster_id": "cluster-43979-11", @@ -2059,6 +2253,15 @@ ], "target_issue_number": 45647, "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-45861-2", + "canonical_pr_number": 45861, + "duplicate_pr_numbers": [ + 45903 + ], + "target_issue_number": null, + "reason": "PRs in cluster-45861-2 are treated as duplicates because they share soft_similarity evidence." } ], "best_issue": { diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index b7690801d41d279e7f21efca37dd43f17c2eb859..899172dcb2e6228f8444f5d7f3e4777c9908eedb 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": "20260512T120027Z", - "generated_at": "2026-05-12T12:09:53Z", + "snapshot_id": "20260512T180027Z", + "generated_at": "2026-05-12T18:10:15Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -131,20 +131,20 @@ "issue:43425|issue:45070", "issue:43596|issue:45800" ], - "prepared_review_unit_hash": "498516001acde0722b052d5c34e5ceaf9b3946961a72911a391991d81f06a14c", + "prepared_review_unit_hash": "cd0b97f14548d2b625bf2900852fefd72a613f4f9a4704c8c2ae34b614a5ce30", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 63, - "serialized_chars": 45922, + "serialized_chars": 45921, "estimated_input_tokens": 11481, "estimated_eval_tokens": 23218 }, @@ -153,366 +153,318 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous. I only see one clear duplicate pair: the two DeepSpeed ZeRO-3 + BertModel index-out-of-bounds reports. The rest are superficial similarity matches across different models, backends, or docs topics.", - "confidence": 0.91, - "canonical_issue_reason": "issue:43596 is the broadest, earliest report in the clear duplicate pair and best captures the shared DeepSpeed ZeRO-3/BertModel index-0 out-of-bounds failure.", + "summary": "Mostly unrelated issues, with two small duplicate clusters: v5 docs cleanup for removed references, and DeepSpeed ZeRO-3 load-time index-0-out-of-bounds failures. The rest are only superficially similar title matches.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43638 is the clearest representative of the shared DeepSpeed ZeRO-3 loading bug: the exact index-0-out-of-bounds failure appears here in a broad, model-agnostic form, while nearby reports look like variants of the same load path.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best representative issue overall because it matches the strongest duplicate edge and states the common failure mode succinctly.", + "best_issue_reason": "issue:43638 is the strongest issue to keep as the cluster anchor because it states the concrete failure mode clearly and generalizes beyond a single model family.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different models and failure paths: Qwen3_5 config attribute error vs Qwen3.5 streaming API failure." + "reason": "Different models and failure modes: Qwen3.5 config attribute error vs streaming chat/completions issue." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Different backend-specific bugs: AMD GPU load failure vs MPS histogram Int error." + "reason": "Unrelated hardware/runtime failures: AMD GPU loading problem vs MPS histogram-on-int error." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Both are model/device issues, but one is dtype mismatch in preprocessing and the other is AMD GPU runtime failure." + "reason": "Different bugs: image-preprocessor dtype mismatch vs AMD GPU load failure." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Different concrete bug: autocast dtype mismatch vs AMD GPU incompatibility." + "reason": "Autocast dtype mismatch and AMD GPU device/runtime failure are not the same bug." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different models and code paths; one is GPU backend failure, the other is an invalid method call in a MoE model." + "reason": "Different code paths: GPU load failure vs invalid method call in GraniteMoEHybrid." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Qwen-3 NVFP4 runtime issue is unrelated to AMD GPU execution failure." + "reason": "NVFP4 runtime error and AMD GPU load failure are unrelated." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Both are runtime errors, but one is DeepSpeed collective mismatch and the other is ONNX export failure." + "reason": "DeepSpeed collective mismatch vs ONNX export error; different subsystems and symptoms." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Different model/export problem vs Qwen3Moe FP8 failure." + "reason": "ONNX export failure and FineGrainedFP8 model failure are unrelated." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ONNX export error and CUDA assert with device mapping are different failures." + "reason": "Different problems: export-time ONNX issue vs auto device-map cuda assert." }, { "left": "issue:44509", "right": "issue:44512", - "accept": false, - "reason": "Both are v5 docs cleanup items, but they target different removed features and do not look like the same concrete change." + "accept": true, + "reason": "Both are v5 docs cleanup for removed/deprecated references; they can be fixed in one documentation PR." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "ONNX export vs GenerationMixin import regression are unrelated code paths." + "reason": "ONNX export issue vs import-time GenerationMixin regression; different failures." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Generic mask-shape IndexError and Gemma4 PLE device mismatch are different bugs." + "reason": "Mask/tensor shape mismatch and Gemma4 PLE device mismatch are different runtime errors." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Similar IndexError wording, but different model/setup and not enough evidence of the same root cause." + "reason": "Different bugs: ZeRO-3 index-out-of-bounds during load vs generic mask shape mismatch." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Different failure shapes and setups; the zero-size Bert error is not the same as the mask mismatch IndexError." + "reason": "Not the same underlying failure; one is ZeRO-3 loading, the other is a mask indexing mismatch." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Mask/tensor shape mismatch is unrelated to GPT-OSS tensor-parallel issues." + "reason": "Mask shape mismatch is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Different model/training path: mask mismatch vs TRL SFT embedding unpacking." + "reason": "Different symptoms and code paths: mask indexing error vs Qwen3.5/TRl embedding unpacking." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset bug is unrelated to ModernBERT ONNX export." + "reason": "Processor voice_preset bug and ONNX export error are unrelated." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "DeepSpeed collective mismatch on Jamba and ZeRO-3 Bert index error are different distributed failures." + "reason": "Collective mismatch during ZeRO-3 and index-0-out-of-bounds loading are different failures." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same underlying DeepSpeed ZeRO-3/BertModel bug; both report the identical index-0 out-of-bounds failure, with one being a non-pretrained Bert variant." + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds load failure; only the model family differs." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Different problem domains: Liger Kernel calling cross_entropy vs GPT-OSS tensor-parallel not working." + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are not the same bug." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Both concern tokenization/LoRA behavior, but they are different model-specific bugs and not the same change." + "reason": "Detokenization regression and LoRA merge/save mismatch are different issues." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Different failures: NVFP4 runtime error vs LoRA merge/reload output mismatch." + "reason": "NVFP4 runtime error vs LoRA save/reload inconsistency are unrelated." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Both mention shape-related runtime errors, but the FSDP activation-checkpointing issue is a different code path." + "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 involve DeepSpeed, but one is a weight-dimension error during training and the other is a collective mismatch in ZeRO-3." + "reason": "Different DeepSpeed problems: weight shape error vs collective mismatch." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "Bert/ZeRO-3 index error is unrelated to GPT-OSS tensor-parallel failure." + "reason": "ZeRO-3 model-loading index error is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "FSDP activation checkpointing tensor-size mismatch and CustomTrainer loss-signature regression are unrelated." + "reason": "Activation checkpointing/FSDP issue and CustomTrainer signature change are unrelated." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Same generic distributed vibe, but the concrete bug is different: Bert ZeRO-3 init vs GPT-OSS TP." + "reason": "Different failures: ZeRO-3 loading bug vs GPT-OSS TP issue." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "Different model families and failures: dtype mismatch in preprocessing vs invalid method call." + "reason": "Dtype mismatch with autocast/image preprocessing is unrelated to an invalid method call." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Gemma3 vLLM config attribute error and Mistral tokenizer regex crash are unrelated." + "reason": "Missing sliding_window_pattern and tokenizer backend_tokenizer crash are different regressions." }, { "left": "issue:45823", "right": "issue:45874", "accept": false, - "reason": "Both are Gemma4-related, but one is a device mismatch and the other is an expensive reverse embedding lookup." + "reason": "Device mismatch during forward and expensive reverse embedding lookup are different Gemma4 issues." }, { "left": "issue:43828", "right": "issue:45161", "accept": false, - "reason": "Autocast dtype mismatch is not the same as GPT-OSS tensor-parallel failure." + "reason": "Autocast dtype mismatch and tensor-parallel failure are unrelated." }, { "left": "issue:43425", "right": "issue:44292", "accept": false, - "reason": "General torch/version incompatibility and Qwen-3 NVFP4 runtime error are different issues." + "reason": "Torch version incompatibility and Qwen3 NVFP4 runtime error are different issues." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "FSDP checkpointing mismatch and Qwen3 auto device-map CUDA assert are different distributed/runtime problems." + "reason": "FSDP activation-checkpointing mismatch is unrelated to cudaErrorAssert under auto device mapping." }, { "left": "issue:44918", "right": "issue:45161", "accept": false, - "reason": "TRL SFT embedding unpacking and GPT-OSS tensor-parallel failure are different code paths." + "reason": "TRL SFT embedding unpacking and GPT-OSS TP failure are different code paths." }, { "left": "issue:43716", "right": "issue:45685", "accept": false, - "reason": "Mistral image-preprocessor dtype mismatch and MPS histogram Int limitation are unrelated." - }, - { - "left": "issue:43828", - "right": "issue:44292", - "accept": false, - "reason": "Different model/runtime failures: autocast dtype mismatch vs NVFP4 run error." - }, - { - "left": "issue:43541", - "right": "issue:43716", - "accept": false, - "reason": "Mixtral grouped_mm tracing error is unrelated to Mistral preprocessor dtype mismatch." + "reason": "Image-preprocessor dtype mismatch is unrelated to the MPS histogram int limitation." }, { - "left": "issue:43828", + "left": "issue:45507", "right": "issue:45685", "accept": false, - "reason": "Autocast dtype mismatch and MPS histogram unsupported Int are different backend bugs." + "reason": "Invalid method call and MPS histogram int error are not the same bug." }, { "left": "issue:43716", "right": "issue:44292", "accept": false, - "reason": "Different models and failure modes; no shared root cause." + "reason": "Different runtime failures: dtype mismatch vs NVFP4 model error." }, { "left": "issue:44292", "right": "issue:44918", "accept": false, - "reason": "Qwen-3 NVFP4 runtime error is unrelated to Qwen3.5 embedding unpacking." + "reason": "NVFP4 runtime error and Qwen3.5 embedding issue are unrelated." }, { "left": "issue:43039", "right": "issue:44918", "accept": false, - "reason": "Liger Kernel cross-entropy invocation and TRL embedding unpacking are different problems." + "reason": "Liger Kernel cross_entropy issue is unrelated to TRL SFT input-embedding unpacking." }, { "left": "issue:41762", "right": "issue:43638", - "accept": false, - "reason": "Gemma3 ZeRO-3 load failure and Bert ZeRO-3 index error are different model-specific bugs." + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds loading bug; the model differs, but the failure pattern is the same." }, { "left": "issue:43827", - "right": "issue:44512", - "accept": false, - "reason": "Both are docs regressions after v5 removals, but they reference different removed APIs and do not appear to be the same change." + "right": "issue:44509", + "accept": true, + "reason": "Both are stale v5 documentation references that need the same cleanup pass." }, { "left": "issue:43716", "right": "issue:44560", "accept": false, - "reason": "Different multimodal models and failures: image-preprocessor dtype mismatch vs video StopIteration." + "reason": "Different bugs: autocast dtype mismatch vs Qwen3-vl video StopIteration." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." }, { "left": "issue:44918", "right": "issue:45685", "accept": false, - "reason": "TRL SFT trainer embedding issue is unrelated to MPS Int histogram limitation." + "reason": "TRL trainer embedding issue and MPS histogram int error are unrelated." }, { "left": "issue:43039", "right": "issue:43828", "accept": false, - "reason": "Same general dtype/theme, but different concrete bug and model path." + "reason": "Liger Kernel cross_entropy issue is not the same as Phi-tiny-MoE autocast dtype mismatch." }, { "left": "issue:30277", "right": "issue:36331", "accept": false, - "reason": "DeepSpeed collective mismatch and CustomTrainer signature change are unrelated." + "reason": "DeepSpeed collective mismatch and trainer compute_loss signature mismatch are unrelated." }, { "left": "issue:34928", "right": "issue:36010", "accept": false, - "reason": "Activation-checkpointing/FSDP mismatch is unrelated to GenerationMixin import breakage." + "reason": "FSDP activation-checkpointing mismatch and GenerationMixin import regression are different problems." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Cross-entropy call under Liger Kernel is unrelated to mask/tensor shape mismatch." + "reason": "Liger Kernel cross_entropy issue is unrelated to the mask shape mismatch bug." }, { "left": "issue:43901", "right": "issue:44512", "accept": false, - "reason": "Both are docs mismatches, but one is about return_all_scores and the other about the removed run command." + "reason": "return_all_scores docs behavior mismatch is not the same issue as the removed run-command docs reference." }, { "left": "issue:41762", "right": "issue:43596", "accept": false, - "reason": "Gemma3 ZeRO-3 load error is a different bug from the Bert Zero-3 index-out-of-bounds failure." + "reason": "Same general ZeRO-3 theme, but not enough direct evidence to treat these as the same concrete duplicate pair." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Gemma3 sliding_window config error and Qwen3 auto device-map CUDA assert are different issues." - }, - { - "left": "issue:45237", - "right": "issue:45901", - "accept": false, - "reason": "AMD GPU model loading failure is unrelated to table-question-answering crash." - }, - { - "left": "issue:44841", - "right": "issue:45823", - "accept": false, - "reason": "Processor failure for Voxtral and Gemma4 PLE device mismatch are different code paths." - }, - { - "left": "issue:43039", - "right": "issue:44560", - "accept": false, - "reason": "Liger Kernel cross_entropy issue is unrelated to Qwen3-vl video StopIteration." - }, - { - "left": "issue:43866", - "right": "issue:44863", - "accept": false, - "reason": "Checkpoint corruption and NemotronH checkpoint loading failure are not the same underlying bug." - }, - { - "left": "issue:43854", - "right": "issue:43866", - "accept": false, - "reason": "Unable to load GLM-4.7-Flash in tests and Ovis2 checkpoint corruption are different model-specific failures." - }, - { - "left": "issue:41093", - "right": "issue:43638", - "accept": false, - "reason": "Mask/index tensor mismatch and Bert ZeRO-3 zero-size index error are distinct runtime errors." - }, - { - "left": "issue:39290", - "right": "issue:43531", - "accept": false, - "reason": "Gemma3 sliding_window config regression and Qwen3-MoE sliding_window issue are different models and fixes." + "reason": "sliding_window_pattern missing and cudaErrorAssert under device mapping are different failures." }, { "left": "issue:43792", @@ -524,19 +476,19 @@ "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch version incompatibility and PretrainedConfig pydantic regression are different compatibility issues." + "reason": "Torch-version incompatibility and pydantic PretrainedConfig field breakage are unrelated." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "DeepSpeed ZeRO-3 Bert index error is unrelated to torch/transformers version incompatibility." + "reason": "ZeRO-3 load-time index error and torch/transformers incompatibility are different issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The one accepted pair, issue:43596 and issue:43638, is strongly supported by nearly identical DeepSpeed ZeRO-3/Bert index-out-of-bounds titles, and the rest of the rejections are appropriately cautious given the differing models/backends/code paths." + "feedback": "The summary is grounded in the packet and stays conservative. The accepted soft-edge pairs are mergeable as single fixes: the v5 docs references are both documentation cleanup, and the ZeRO-3 cases all share the same load-time index-out-of-bounds failure pattern despite different models. The rejected pairs also read as appropriately cautious rather than overstated." }, "evaluator_used": true, "retried": false, @@ -627,11 +579,11 @@ "issue:43541|issue:45161", "issue:43828|issue:45464", "issue:44291|issue:45070", - "issue:45020|issue:45800", "issue:43366|issue:45464", + "issue:45020|issue:45800", "issue:43531|issue:45070", - "issue:41093|issue:43596", "issue:43541|issue:44589", + "issue:41093|issue:43596", "issue:43054|issue:43994", "issue:45542|issue:45800", "issue:40990|issue:44960", @@ -643,250 +595,250 @@ "issue:44530|issue:45005", "issue:44360|issue:44512" ], - "prepared_review_unit_hash": "b2198e28877a7e87a81a11037377b44dce2d79a5642f23356dd02c684507843a", + "prepared_review_unit_hash": "dccfe0eb4d3044f2e9480206697ec53df0990a70b83154fc75818fe504204e19", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 37847, - "estimated_input_tokens": 9462, - "estimated_eval_tokens": 19180 + "serialized_chars": 37849, + "estimated_input_tokens": 9463, + "estimated_eval_tokens": 19182 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of unrelated transformer bug reports. The only plausible duplicate pair is the two custom-module loading issues around `trust_remote_code` / `from_pretrained` cache or module selection; the rest are too far apart in model, error, or code path to merge safely.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45698 is the broader, more active report of the custom-module misload problem after `save_pretrained`/`from_pretrained`; 45632 reads like a narrower cache-path collision explanation of the same bug.", + "summary": "This cluster is mostly heterogeneous: many unrelated model/runtime regressions, with only a couple of plausible duplicate sub-pairs. The strongest duplicate signals are around Siglip2 output regressions and remote_code/custom-module loading collisions.", + "confidence": 0.84, + "canonical_issue_reason": "There is no single true canonical issue for the whole cluster; it mixes unrelated bugs. If forced to pick an anchor, issue:45698 is the best representative because it is concrete, open, and sits in the most plausible duplicate sub-thread about custom-module/remote_code loading.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45698 is the best representative issue in this set because it is open, has the most discussion, and describes the user-visible failure mode most generally.", + "best_issue_reason": "Issue:45698 is the clearest actionable report in the set and the best anchor for deduping the remote_code/custom-module loading cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45859", "right": "issue:45902", "accept": false, - "reason": "Both involve Qwen3.5 MoE rotary embedding, but one is a CPU-offload compatibility failure and the other is a meta/to_empty materialization memory bug; not the same concrete issue." + "reason": "Same model family, but different concrete failures: CPU offload compatibility vs uninitialized-memory access after meta->to_empty(cuda)." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "DeepSeek detokenization and table-question-answering crashes are unrelated." + "reason": "Different subsystems and symptoms: detokenization bug vs table-question-answering crash." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Both mention Qwen3.5 chat/inference, but the failure modes and entrypoints differ; not one underlying bug." + "reason": "Both mention Qwen3.5/chat, but one is a general crash and the other is a streaming API failure on a different model size; not the same bug." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper loading/running and a PretrainedConfig pydantic breakage are unrelated." + "reason": "Whisper runtime failure and a PretrainedConfig/pydantic regression are unrelated." }, { "left": "issue:44292", "right": "issue:45901", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure is unrelated to table-question-answering crashes." + "reason": "Different model and failure mode; NVFP4 execution error is unrelated to table QA crash." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Mistral3 tied-weight bug and Ovis2 checkpoint corruption are different problems." + "reason": "Tied lm_head weights in Mistral3 and a corrupted Ovis2 checkpoint are not the same underlying issue." }, { "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Gemma3 missing `sliding_window_pattern` in vLLM and Gemma4 device mismatch are distinct bugs." + "reason": "Gemma3 missing config attribute with vLLM vs Gemma4 device-mismatch during forward are different bugs." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Qwen3.5 streaming API failure and an MPS histogram kernel issue are unrelated." + "reason": "Streaming chat API failure and an MPS histogram op limitation are unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Torch Dynamo grouped_mm tracing error and MPS int histogram failure are different code paths." + "reason": "Different MoE issues: torch dynamo grouped_mm tracing failure vs MPS int histogram support." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Both are MoE/dtype-adjacent, but one is a tracing failure and the other is an autocast dtype mismatch; not the same bug." + "reason": "Both are MoE/dtype-related, but one is grouped_mm tracing and the other is autocast dtype mismatch; too different to merge." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "GenerationMixin import failure and Gemma3 config attribute failure are unrelated." + "reason": "ImportError for GenerationMixin and a Gemma3 config regression are unrelated." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Voxtral processor failure and non-template-node compile error do not share a concrete root cause." + "reason": "Processor failure for Voxtral and a template-node compile TypeError do not share the same code path." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are index/shape errors, but they affect different models and failure points; too generic to merge." + "reason": "Both are IndexError-shaped, but one is a mask-size mismatch and the other is ZeRO-3 model-loading out-of-bounds; not the same bug." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers compatibility and Gemma4 PLE device mismatch are not the same issue." + "reason": "Torch/Transformers version incompatibility and Gemma4 device placement mismatch are different regressions." }, { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "PaddleOCR-VL loading in vLLM and GPT-OSS GGUF support are unrelated." + "reason": "PaddleOCR-VL load failure and GGUF gpt-oss support are unrelated." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "StableLm missing `pad_token_idx` and a general PretrainedConfig pydantic regression are different bugs." + "reason": "Missing pad_token_idx in StableLmConfig and a pydantic PretrainedConfig regression are distinct config issues." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Int4 reserved-memory/OOM regression and tied-weights translation-model issues are not the same code-path problem." + "reason": "Int4 CUDA memory/OOM regression and tied-weights translation-model bug are unrelated." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Both involve MoE models, but torch-dynamo grouped_mm tracing and GPT-OSS tensor-parallel behavior are distinct." + "reason": "Both involve MoE/GPT-OSS, but one is a grouped_mm tracing bug and the other is tensor-parallel routing; not the same concrete issue." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE and Qwen3.5 streaming API failures are unrelated." + "reason": "Autocast dtype mismatch and streaming chat API failure are unrelated." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "`init_empty_weights` argument handling and PretrainedConfig pydantic field handling are separate regressions." + "reason": "init_empty_weights/_is_hf_initialized argument mismatch is a different loading-path bug from the pydantic PretrainedConfig regression." }, { - "left": "issue:45020", - "right": "issue:45800", + "left": "issue:43366", + "right": "issue:45464", "accept": false, - "reason": "Remote-code breakage and torch/transformers version incompatibility are different classes of issue." + "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference failure are unrelated." }, { - "left": "issue:43366", - "right": "issue:45464", + "left": "issue:45020", + "right": "issue:45800", "accept": false, - "reason": "GPT-OSS GGUF architecture support and Qwen3.5 streaming inference are unrelated." + "reason": "Remote_code breakage across recent transformers versions and a specific torch/transformers incompatibility are not the same bug." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and a PretrainedConfig pydantic regression do not match." + "reason": "Qwen3-MoE sliding_window issue and the PretrainedConfig pydantic regression are unrelated." }, { - "left": "issue:41093", - "right": "issue:43596", + "left": "issue:43541", + "right": "issue:44589", "accept": false, - "reason": "Both are index errors with ZeRO-3, but they occur in different models and are not clearly the same bug." + "reason": "Grouped_mm tracing failure and Float8 storage lookup failure are different runtime errors." }, { - "left": "issue:43541", - "right": "issue:44589", + "left": "issue:41093", + "right": "issue:43596", "accept": false, - "reason": "Grouped_mm tracing and missing Float8 storage registration are unrelated." + "reason": "Both mention IndexError, but the shape mismatch and ZeRO-3 empty-tensor initialization are different failure modes." }, { "left": "issue:43054", "right": "issue:43994", - "accept": false, - "reason": "Both concern SigLIP2 quality, but one is a text-embedding regression and the other is a general nonsensical-output report; not enough to treat as the same concrete bug." + "accept": true, + "reason": "Both report Siglip2 producing bad/nonsensical text-embedding outputs; they look like the same underlying model behavior bug." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "TensorBoard-without-TensorFlow backend detection and torch/transformers incompatibility are unrelated." + "reason": "Undefined tf backend error and torch/transformers incompatibility are unrelated." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "High perplexity on gpt-oss and a GLM5 issue are different model-specific bugs." + "reason": "High perplexity on gpt-oss and GLM5 are different model-specific issues." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Mistral-3 image/text dtype mismatch and Qwen3.5 embedding unpacking under TRL are unrelated." + "reason": "Mistral-3 image-preprocessor dtype mismatch and Qwen3.5 SFT embedding unpacking failure are unrelated." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP weight-shape failure and Qwen3.5 embedding unpacking under TRL are different problems." + "reason": "FSDP weight-shape runtime error and TRL embedding unpacking failure are different code paths." }, { "left": "issue:45632", "right": "issue:45698", "accept": true, - "reason": "Both report the same underlying custom-module loading problem with `trust_remote_code`/`from_pretrained`, where the wrong module is resolved due to cache/path handling." + "reason": "Both point to remote_code/custom-module loading picking the wrong cached module; they are plausibly the same cache-key collision bug." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Mistral3 tied-weight bug and translation-model tied-weights regressions are not the same concrete failure." + "reason": "Mistral3 tied lm_head weights and translation-model tied-weights regressions are in the same broad area but not the same concrete bug." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Custom-model/Jupyter initialization and flash-attn2 fallback checks are unrelated." + "reason": "Custom-model initialization in notebooks and flash-attn fallback blocking are unrelated." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and tied-weights translation issues are unrelated." + "reason": "PagedAttention linear_attention crash and tied weights in translation models are unrelated." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "DSA indexer ReLU discussion and stale docs mentioning the removed run command are unrelated." + "reason": "DSA indexer ReLU discussion and removed docs run command are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays appropriately conservative. The only accepted soft pair, 45632/45698, is reasonably supported by the titles, and the rest of the rejections avoid overclaiming duplicate status from broad model/task similarity alone." + "feedback": "Grounded overall. The cluster is clearly heterogeneous, and the two accepted soft pairs are phrased conservatively enough to be plausible duplicates rather than asserted matches. The only slight caution is the high confidence, but it is still defensible given the evidence provided." }, "evaluator_used": true, "retried": false, @@ -989,112 +941,112 @@ "issue:43881|issue:43883", "issue:41762|issue:42915" ], - "prepared_review_unit_hash": "9849d21510ef53d2f22ee7e02e9c786e1367aa60ca735f359dcd596d3826a844", + "prepared_review_unit_hash": "203f12cbadc52545c81a3f563fa1af137c5e0812e20038874bf321f118de5a3b", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 32, - "serialized_chars": 36363, - "estimated_input_tokens": 9091, - "estimated_eval_tokens": 18438 + "serialized_chars": 36368, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous; only one clear duplicate pair appears to describe the same regression: non-persistent buffers being corrupted/fillled with junk in Transformers 5.x.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43950 is the clearest, most explicit report of the non-persistent-buffer regression, and it matches the strongest duplicate candidate (issue:44534) almost exactly.", + "summary": "This cluster is mostly a loose collection of unrelated Transformers issues, with only one clear duplicate-style pair around non-persistent buffer corruption. I\u2019d anchor on the 5.x buffer regression issue as the best representative.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43950 is the clearest, most directly stated report of the non-persistent-buffer regression, and it has the strongest same-bug overlap with issue:44534.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43950 is the best representative because it states the bug concretely, has a specific repro/regression framing, and aligns with the only high-confidence duplicate pair.", + "best_issue_reason": "issue:43950 is the best overall issue anchor: specific, technically clear, and it captures the concrete regression that appears duplicated elsewhere in the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Different failures: Qwen3 MoE LoRA incompatibility vs flash-attn2 fallback gating when FA2 is missing." + "reason": "Different problems: Qwen3 MoE fine-tuning failure vs flash-attn2 fallback blocked by unrelated checks." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Both involve quantization/offloading broadly, but they describe different problems and code paths." + "reason": "Quantization/offloading behavior is unrelated to tied-weights translation-model warnings." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is unrelated to tied-weights issues on translation models." + "reason": "NemotronH checkpoint loading failure is a different code path than tied-weight handling." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A typo in cache strategies is unrelated to stale docs mentioning a removed run command." + "reason": "A typo in cache strategies is not the same as stale docs mentioning a removed command." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Different model-specific bugs: Qwen3.5 embedding unpacking vs GraniteMoEHybrid calling an invalid method." + "reason": "Different model and failure mode: SFT embedding unpacking vs an invalid method call in GraniteMoEHybrid." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Different model/runtime failures; detokenization bug is not the same as NVFP4 execution failure." + "reason": "Incorrect detokenization and Qwen-3 NVFP4 loading are distinct issues." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "These describe the same underlying regression: Transformers v5 corrupts non-persistent buffers during loading/from_pretrained." + "reason": "Both describe the same 5.x regression where non-persistent buffers get corrupted/junk values on load." }, { "left": "issue:44292", "right": "issue:45507", "accept": false, - "reason": "Unrelated issues: Qwen-3-8B-NVFP4 runtime error vs GraniteMoEHybrid invalid-method call." + "reason": "No shared underlying bug; one is a model/runtime load failure, the other a method-call issue." }, { "left": "issue:43644", "right": "issue:43950", "accept": false, - "reason": "Both are loading-related regressions, but one is about missing config fields with trust_remote_code, the other about non-persistent buffer corruption." + "reason": "Missing fields from trust_remote_code config is unrelated to non-persistent buffer corruption." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Different SigLIP2 problems: docs/example errors vs poorer text embeddings." + "reason": "Both involve SigLIP2, but one is a docs/example error and the other is a model-quality complaint; not the same concrete bug." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Different Qwen/Phi issues with different failure modes and models; not the same bug." + "reason": "Different models and symptoms: dtype mismatch under autocast vs Qwen3.5 position_ids/SDPA behavior." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Trainer API TypeError and Gemma3 ZeRO-3 loading IndexError are unrelated." + "reason": "CustomTrainer signature mismatch and Gemma3 ZeRO-3 load failure are unrelated." }, { "left": "issue:43866", "right": "issue:44403", "accept": false, - "reason": "Checkpoint corruption is not the same as a general loading-noise issue." + "reason": "Corrupted checkpoint vs noisy loading warnings are not the same bug." }, { "left": "issue:41628", @@ -1106,115 +1058,115 @@ "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "Both mention tokenization/behavior changes across versions, but they are not the same concrete bug." + "reason": "SigLIP2 embedding quality and tokenization changes are too broad and not the same concrete issue." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Different quantization incompatibilities with different error signatures." + "reason": "Different incompatibilities: bitsandbytes Int8Params kwarg mismatch vs missing Float8 storage object." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 is not the same as degraded text embeddings." + "reason": "Both concern SigLIP2 behavior, but the titles point to different symptoms without a clear shared code-path bug." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Bad tokenizer error messaging for Voxtral is unrelated to Molmo tied-weights attribute failure." + "reason": "Bad AutoTokenizer error message for Voxtral and Molmo tied-weights attribute error are unrelated." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Different Qwen3.5 failures: Flash Attention crash from 3D position_ids vs CPU-offload incompatibility in rotary embedding." + "reason": "Flash-Attention packed-sequence misinterpretation and CPU-offload rotary-embedding incompatibility are different failures." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Voxtral AutoTokenizer error handling is unrelated to glm-4v-9b loading failure." + "reason": "Voxtral loading UX problem and glm-4v loading failure do not share the same underlying bug." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Both involve training/memory concerns, but one is gradient checkpointing with LoRA and the other is int4 reserved-memory/OOM behavior." + "reason": "Gradient checkpointing ineffectiveness is not the same issue as int4 quantization memory growth/OOM." }, { "left": "issue:43054", "right": "issue:43493", "accept": false, - "reason": "Both are SigLIP2-related, but one is an embedding quality complaint and the other is a JAX vs HF implementation discrepancy." + "reason": "Potentially related SigLIP2 implementation concerns, but not clearly the same concrete bug from the titles alone." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "General loading noise is not the same issue as failing to load a specific model." + "reason": "General loading noise vs inability to load a specific model are different issues." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Missing fields in AutoConfig with trust_remote_code is not the same concrete bug as broader remote_code breakages." + "reason": "A missing-field config return bug is not the same as broader remote_code breakages." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Detokenization regression and Python 3.13 flash_attn import failure are unrelated." + "reason": "Detokenization bug and Python 3.13 flash_attn import error are unrelated." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Detokenization bug is unrelated to GPT-OSS-20B failing on AMD GPUs." + "reason": "Detokenization and AMD GPU execution failure are distinct problems." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Different MoE-related performance/memory concerns with different scopes and symptoms." + "reason": "Different memory-related training issues; not the same code-path bug." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "General loading-noise complaint does not match a specific model-load failure." + "reason": "Unnecessary loading noise and failure to load a specific model are not the same underlying issue." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Qwen3VL weight_only loading error is unrelated to a generic GLM5 issue." + "reason": "Qwen3VL weight_only load error and GLM5 are unrelated model-loading issues." }, { "left": "issue:42491", "right": "issue:44368", "accept": false, - "reason": "LoRA compatibility across hf4.x/hf5.x is not the same as tie_word_embeddings warnings during ms-swift fine-tuning." + "reason": "LoRA compatibility on hf5.x and tied-word-embeddings warnings are different concerns." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different loading errors for different models; no shared concrete code-path bug is evident." + "reason": "Two separate model-loading failures with no clear shared bug." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Gemma3 ZeRO-3 loading IndexError is unrelated to the Qwen3 MoE LoRA incompatibility." + "reason": "Gemma3 ZeRO-3 load crash and Qwen3 MoE fine-tuning failure are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays mostly conservative. The only accepted soft pair, issue:43950 / issue:44534, is well-supported by the near-identical titles and the same non-persistent-buffer regression framing. The remaining verdicts also read as cautious rather than overclaimed. Confidence is a bit high for a cluster with mostly soft-similarity evidence, but not enough to reject." + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft edge, issue:43950 \u2194 issue:44534, is well-supported by the titles and the regression description. The rest of the pair verdicts appropriately avoid overclaiming shared bugs when the evidence is only superficial similarity." }, "evaluator_used": true, "retried": false, @@ -1308,8 +1260,8 @@ "issue:41553|issue:44991", "issue:43257|issue:45305", "issue:45759|issue:45902", - "issue:36331|issue:41093", "issue:44928|issue:45588", + "issue:36331|issue:41093", "issue:41093|issue:44805", "issue:43872|issue:45538", "issue:44530|issue:44863", @@ -1321,20 +1273,20 @@ "issue:44155|issue:45584", "issue:43366|issue:43531" ], - "prepared_review_unit_hash": "3786fc595fc970c867a105b9999cb14f38d5d67934337e1b06066c9ddad0ada8", + "prepared_review_unit_hash": "6d3efd902d4c17f0f10765e2801cf8b23d04549d0b76ab2f66c758f9a670b971", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 38554, + "serialized_chars": 38556, "estimated_input_tokens": 9639, "estimated_eval_tokens": 19534 }, @@ -1343,78 +1295,78 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a grab-bag of unrelated bug reports. I found a few strong duplicate pairs around docs cleanup, the `_is_hf_initialized` loading regression, a repeated mask-shape IndexError, and a Qwen video-input regression; the rest do not look mergeable.", - "confidence": 0.72, - "canonical_issue_reason": "issue:44291 is the clearest, most general statement of the `_is_hf_initialized` model-loading regression; issue:43872 is a narrower bitsandbytes manifestation of the same bug.", + "summary": "This set is mostly a loose semantic cluster around recent Transformers loading/runtime regressions, but the individual issues are usually about different models, error modes, and code paths. Only a few pairs are true duplicates on the same concrete bug.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45020 is the broadest umbrella candidate: it targets recent-version breakage across multiple remote-code models, so it is the most plausible single issue to anchor a heterogeneous loading-regression cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44291 is the strongest anchor in the set: explicit error, clear root cause, and broad applicability across model-loading paths.", + "best_issue_reason": "issue:45020 is the best overall representative because it is open, broadly scoped, and framed as a cross-model regression rather than a model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "Different bugs: MPS histogram/int error in MoE code vs Gemma4 device placement mismatch." + "reason": "Different failures: MPS histogram/int support in MoE code vs a Gemma4 device-mapping mismatch." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Different failures: Qwen2.5-VL import error vs bitsandbytes `_is_hf_initialized` constructor incompatibility." + "reason": "Import failure for Qwen2.5-VL is unrelated to bitsandbytes' unexpected `_is_hf_initialized` argument error." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Too broad and unrelated: remote_code breakage is not the same as Gemma4 PLE device mismatch." + "reason": "Broad remote-code regression report vs a specific Gemma4 device mismatch; not the same bug." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Both are model-loading adjacent, but not the same underlying bug or code path." + "reason": "Tokenizer mapping edge case in `add-new-model-like` is a different code path from remote-code model breakage." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading failure vs flash-attn fallback gating; unrelated subsystems." + "reason": "Tokenizer loading regression for a specific model is unrelated to the flash-attn fallback gating issue." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "Different runtime path: autocast dtype mismatch vs Gemma4 device_map forward mismatch." + "reason": "Different model and different failure mode: dtype mismatch under autocast vs device placement mismatch." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same docs regression after v5 pipeline removals; both ask to remove stale summarization/translation pipeline references." + "reason": "Both report the same docs regression: old pipeline task names still referenced after v5 removals." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Tokenizer regex crash vs flash-attn fallback selection; not the same issue." + "reason": "Mistral tokenizer regex patch crash is unrelated to flash-attn2 fallback blocking." }, { "left": "issue:43950", "right": "issue:45902", "accept": false, - "reason": "Both involve buffer/materialization problems, but one is a general from_pretrained regression and the other is a specific Qwen3Moe uninitialized-memory report." + "reason": "Both involve buffer/materialization concerns, but one is a general `from_pretrained` regression and the other is a specific Qwen3-MoE uninitialized-memory bug." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer/model-loading failures with only loose theme overlap." + "reason": "Remote-code model breakage and Kimi tokenizer codec handling are different problems." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "lm_head tying bug is unrelated to non-persistent buffer corruption." + "reason": "Tied-weight bug in Mistral3 is not the same as non-persistent buffer corruption." }, { "left": "issue:43742", @@ -1426,145 +1378,145 @@ "left": "issue:43257", "right": "issue:43866", "accept": false, - "reason": "Qwen3 MOE weight-conversion issue vs Ovis2 checkpoint corruption; different bugs." + "reason": "Qwen3 MoE weight conversion under DeepSpeed is unrelated to a corrupted Ovis2 checkpoint." }, { "left": "issue:42915", "right": "issue:45020", "accept": false, - "reason": "FineGrainedFP8/Qwen3Moe failure is not the same as remote_code loading regressions." + "reason": "FineGrainedFP8Config failure is a specific quantization/config issue, not the broader remote-code regression." }, { "left": "issue:43638", "right": "issue:45823", "accept": false, - "reason": "DeepSpeed zero3 index error in BERT training is unrelated to Gemma4 device mismatch." + "reason": "Zero-3 + non-pretrained BERT index error is unrelated to Gemma4 device mismatch." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Both are Whisper/audio-related, but the titles do not clearly point to the same concrete bug." + "reason": "Whisper model runtime failure is too generic to merge confidently with the `_torch_extract_fbank_features()` report." }, { "left": "issue:43940", "right": "issue:45313", "accept": false, - "reason": "Both are ZeRO-3 loading failures for Qwen models, but the affected models and symptoms differ too much to merge confidently." + "reason": "Qwen3-Next missing weights under ZeRO-3 and Qwen3.5 language_model loading are distinct DeepSpeed loading bugs." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same `_is_hf_initialized` TypeError during model loading; 44291 states the general regression, 43872 is a specific bitsandbytes manifestation." + "reason": "Same underlying bug: loading under `init_empty_weights` hits the unexpected `_is_hf_initialized` keyword in bitsandbytes." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "Different causes: FP8 config failure vs `_is_hf_initialized` constructor incompatibility." + "reason": "Different root causes: FP8Config behavior vs bitsandbytes constructor incompatibility." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different model-loading failures for different architectures; no shared concrete bug." + "reason": "GLM-4.7-Flash test loading issue and NemotronH checkpoint loading failure are unrelated model-specific problems." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Bad error message for Voxtral tokenizer loading is not the same as est-roberta tokenizer load failure." + "reason": "Both are tokenizer-related, but they concern different models and different failure mechanisms." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Qwen3 MOE weight conversion under DeepSpeed is unrelated to GAS gradient averaging with loss kwargs." + "reason": "DeepSpeed weight conversion bug and gradient averaging bug are separate training issues." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Composite config unwrapping bug vs uninitialized rotary-embedding memory; different code paths." + "reason": "Composite config unwrapping and uninitialized rotary embedding memory are different code paths." }, { - "left": "issue:36331", - "right": "issue:41093", + "left": "issue:44928", + "right": "issue:45588", "accept": false, - "reason": "Different training API mismatch vs mask/tensor shape IndexError." + "reason": "RLHF gradient explosion from position IDs is unrelated to a flash-attention `s_aux=None` crash." }, { - "left": "issue:44928", - "right": "issue:45588", + "left": "issue:36331", + "right": "issue:41093", "accept": false, - "reason": "RLHF NaNs from SDPA/BF16 fallback is unrelated to flash_attention `s_aux=None` crash." + "reason": "CustomTrainer signature mismatch and mask/index shape mismatch are unrelated." }, { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "Same mask-vs-tensor shape mismatch IndexError; 44805 looks like the same bug with generalized placeholders." + "reason": "Same shape-mismatch IndexError, with the later issue appearing to be the same underlying bug report." }, { "left": "issue:43872", "right": "issue:45538", "accept": false, - "reason": "`_is_hf_initialized` loading regression is unrelated to CLIPTokenizer max length handling." + "reason": "bitsandbytes init incompatibility is unrelated to CLIPTokenizer's `model_max_length` constant." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "PagedAttention group-type crash and NemotronH checkpoint loading are different failures." + "reason": "PagedAttentionCache linear_attention crash and NemotronH loading failure are different subsystems and symptoms." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Tokenizer error-message quality issue vs a concrete tokenizer load failure; not the same bug." + "reason": "Bad Voxtral tokenizer error message and failing to load `cjvt/sleng-bert` are separate tokenizer-loading issues." }, { "left": "issue:44479", "right": "issue:44560", - "accept": true, - "reason": "Both describe the same 5.3.0 Qwen video-input regression/StopIteration family in the video preprocessing path." - }, - { - "left": "issue:24643", - "right": "issue:30064", "accept": false, - "reason": "DeepSpeed training weight-shape error vs image processor segmentation-map handling; unrelated." + "reason": "Video-input regression across Qwen VL variants and a Whisper embedding `StopIteration` are different bugs." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "GGUF architecture support and autocast dtype mismatch are unrelated." + "reason": "GGUF gpt-oss architecture support and Phi-tiny-MoE autocast dtype mismatch are unrelated." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Both are Gemma4, but one is FSDP2 shared_kv_states logic and the other is device_map auto placement." + "reason": "Gemma4 FSDP2 shared_kv_states bug is distinct from Gemma4 device-map mismatch." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "AudioFlamingo3 track leakage and Whisper empty-transcription handling are different bugs." + "reason": "AudioFlamingo3 track leakage and Whisper empty-transcription failure are different audio inference bugs." }, { - "left": "issue:43366", - "right": "issue:43531", + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Trainer API change vs tensor mask shape mismatch; not the same underlying issue." + }, + { + "left": "issue:44991", + "right": "issue:45399", "accept": false, - "reason": "GGUF/gpt-oss support vs Qwen3-MoE sliding-window behavior; no shared concrete code-path problem." + "reason": "Tokenizer loading regression and flash-attn fallback gating do not share the same code path." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays within the packet evidence and the accepted/rejected pairs are conservative. The duplicated docs, `_is_hf_initialized`, mask-shape, and Qwen video-regression pairs are plausibly same-bug matches, while the rejects avoid over-connecting only loosely related issues." + "feedback": "The summary is grounded in the packet and stays conservative. The accepted soft-edge pairs are well supported by near-identical titles and failure modes, and the rejected pairs are reasonably distinguished by model, subsystem, or error type. No overclaim stands out." }, "evaluator_used": true, "retried": false, @@ -1644,8 +1596,8 @@ "issue:42915|issue:43957", "issue:42617|issue:43366", "issue:44560|issue:44918", - "issue:44661|issue:45823", "issue:36010|issue:42915", + "issue:44661|issue:45823", "issue:38175|issue:43994", "issue:43632|issue:45834", "issue:44560|issue:44805", @@ -1670,14 +1622,14 @@ "issue:42491|issue:43931", "issue:39692|issue:43994" ], - "prepared_review_unit_hash": "0076ef3ba59e20efd60015d22ed2c1a596cad61c77a3ea2f04f4f4c13f5f794d", + "prepared_review_unit_hash": "59866b35d4a5d134e9c2bfc54978ddf8574eece4a4c5b8c89959c15ccbfba431", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, @@ -1692,222 +1644,222 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous. 43644/44534 is the clear duplicate pair around the non-persistent buffer regression, and 43645/43646 is the other defensible soft duplicate. The SigLIP2 pair 38175/43994 is too weakly supported and should be rejected.", - "confidence": 0.93, - "canonical_issue_reason": "issue 43644 is the clearest concrete regression report for the non-persistent-buffer corruption bug, and issue 44534 is an exact near-duplicate of that failure mode.", + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports rather than one unified duplicate family. The clearest duplicate subcluster is the non-persistent-buffer corruption reports; there is also a likely related BERT-loading pair, but the rest are broader subsystem/theme overlaps rather than the same bug.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43644 is the cleanest canonical for the non-persistent-buffer regression subcluster: its title states the bug directly, and issue:44534 appears to be a near-rephrasing of the same failure mode.", "canonical_pr_reason": null, - "best_issue_reason": "issue 43644 is the strongest canonical issue because it is specific, well-scoped, and directly duplicated by 44534.", + "best_issue_reason": "issue:43644 is the best overall issue representative because it is a concrete, well-scoped v5 regression with a solid duplicate counterpart; the cluster as a whole is too heterogeneous for a single broader canonical beyond that subcluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Different bugs: non-persistent buffer corruption vs. Kosmos2.5 OCR index error." + "reason": "Different bugs: non-persistent buffer corruption versus Kosmos2.5 OCR index failure." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Different models and different failure modes; the shared attention/runtime theme is too broad." + "reason": "Both are Qwen/attention-related, but the failure modes differ: SDPA/torch.compile versus Flash Attention with 3D position_ids." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Unrelated errors: video StopIteration vs. MPS histogram not implemented for Int." + "reason": "Unrelated model and backend issues: video StopIteration versus an MPS histogram implementation error." }, { "left": "issue:43645", "right": "issue:43646", - "accept": true, - "reason": "Same underlying regression: Transformers 5.0 breaks custom model initialization, with notebook-specific wording in one report." + "accept": false, + "reason": "Both involve custom models, but one is notebook-specific and the other is a broader custom-model initialization regression." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Specific ImportError for one class is not the same bug as generic meta-device loading breakage." + "reason": "Both are loading-related, but the specific failures differ: missing import versus meta-device loading behavior." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "Different code paths and symptoms; no shared underlying bug is evident." + "reason": "GGUF/gpt-oss support is unrelated to the Float8 storage lookup error." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "Both are load failures, but they affect different models and mechanisms." + "reason": "Qwen3Moe FP8 failure and meta-device loading failures are different code-path problems." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "No concrete overlap beyond being transformer-related issues." + "reason": "Running 3d_parallel.py and GGUF gpt-oss support are unrelated issues." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Different failures in different model paths; not enough to treat as the same bug." + "reason": "Both touch Qwen3.5 video/input flows, but the observed failures are different: StopIteration versus TRL embedding unpacking." }, { - "left": "issue:44661", - "right": "issue:45823", + "left": "issue:36010", + "right": "issue:42915", "accept": false, - "reason": "Tokenizer-mapping behavior is unrelated to Gemma4 PLE device mismatch." + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 loading." }, { - "left": "issue:36010", - "right": "issue:42915", + "left": "issue:44661", + "right": "issue:45823", "accept": false, - "reason": "Importing GenerationMixin and Qwen3Moe FP8 loading are unrelated problems." + "reason": "Tokenizer-mapping edge case and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:38175", "right": "issue:43994", "accept": false, - "reason": "Same model family, but only vague wrong-output symptoms are shown; not enough evidence of the same underlying bug." + "reason": "Both are SigLIP2-related, but one is a zero-probabilities bug and the other is an AutoModel/pipeline output issue; not the same underlying defect." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "Different regressions: _is_hf_initialized flag behavior vs. Kosmos2.5 OCR indexing." + "reason": "_is_hf_initialized flag regression is unrelated to Kosmos2.5 long-ocr index errors." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Distinct tensor-path failures with no clear shared root cause." + "reason": "Different failures: video StopIteration versus a mask/tensor shape mismatch." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Similar config-validation wording, but different models and different actual failures." + "reason": "Strict config blocking granite_speech loading and tiny-random glm4v config validation are separate config issues." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "DeepSpeed zero3/Bert indexing is unrelated to add-new-model-like tokenizer mapping." + "reason": "Zero3 training with non-pretrained BertModel and tokenizer-mapping-name handling are unrelated." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Different failures: zero3/BertModel indexing vs. model-mapping logic." + "reason": "Zero3 index error in BertModel is not the same bug as add-new-model-like failing inside TOKENIZER_MAPPING_NAMES." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "No overlap beyond both being transformer bugs." + "reason": "Tokenizer mapping behavior and a mask shape mismatch are different problems." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Both are loading/memory related, but the concrete bug mechanisms differ." + "reason": "device_map=auto offload/dequantized-model loading and disabling mmap on Strix Halo are not the same underlying issue." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Exact duplicate regression: both describe v5 filling non-persistent buffers with junk." + "reason": "Same regression: Transformers v5 corrupts non-persistent buffers during loading; 44534 is a reworded duplicate report." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Docs example/model-processor mismatch and quantization failure are not the same as the wrong-output report." + "reason": "Both mention SigLIP2, but one is a runtime output bug and the other is documentation/example errors plus quantization failure." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Different Qwen3.5 failures with different triggers and symptoms." + "reason": "Different models and different failure modes: GLM5 versus Qwen3.5 flash-attention generation." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Different loading bugs: Qwen3 MoE LoRA compatibility vs. packed-weight unpacking." + "reason": "Qwen3 MoE LoRA compatibility and BitNet packed-weight unpacking are unrelated loading issues." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Both concern Qwen3.5, but the failure modes are different: ZeRO-3 loading vs. save_pretrained key corruption." + "reason": "DeepSpeed ZeRO-3 load failure for language_model and incorrect visual encoder keys during save_pretrained are different regression paths." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Tokenizer regressions affecting different model families and different mechanisms." + "reason": "Tokenizer pre-tokenizer override in LlamaTokenizer is not the same as Granite token-ID misgeneration." }, { "left": "issue:44451", "right": "issue:44488", - "accept": false, - "reason": "Likely related class of load failures, but not enough evidence that they are the same bug." + "accept": true, + "reason": "Both report current/latest Transformers failing to load specific BERT checkpoints, which is plausibly the same underlying loading regression." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Different Qwen3 failures with different symptoms and likely different causes." + "reason": "Qwen3 auto device mapping cudaErrorAssert and Qwen3Moe FP8 failures are separate issues." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint math and SGD argument handling are unrelated." + "reason": "Trainer checkpoint-resume step calculation and SGD args not being used are unrelated trainer/optimizer bugs." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different model-loading failures with no shared concrete mechanism." + "reason": "Unit-test loading failure for GLM-4.7-Flash and Qwen3-VL weight-shape mismatch are different model-loading problems." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Qwen3 MoE LoRA compatibility is unrelated to Gemma4 embedding resizing." + "reason": "Qwen3 MoE LoRA compatibility and Gemma4 resize_token_embeddings affecting embeddings are unrelated." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different models and different loading paths." + "reason": "MoE weights not converted with accelerate+deepspeed and GLM-4.7-Flash unit-test loading are different bugs." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "FP8 load failure and mmap/OOM tuning are different issues." + "reason": "Qwen3Moe FP8 failure and mmap OOM on Strix Halo are unrelated." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "Both are Qwen3-related, but the specific bugs are different." + "reason": "Qwen3 MoE LoRA incompatibility and Qwen3-VL weight-shape mismatch are different model-specific load failures." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Docs/example errors are not the same bug as the runtime wrong-output report." + "reason": "Both are SigLIP2-related, but one is an example/documentation issue and the other is a runtime model-output issue." } ] }, "evaluator_result": { "accept": false, - "feedback": "The canonical 43644/44534 duplicate looks well grounded, and most non-matches are conservatively rejected. However, the accepted 38175/43994 pair overstates the evidence: the packet only shows same model family and vague wrong-output symptoms, not enough to conclude the same underlying bug. Because at least one soft-edge accept is not sufficiently conservative, I would reject this review." + "feedback": "The summary is mostly conservative, but it overstates the evidence in two places: (1) calling issue:43644 the cluster\u2019s \"strongest duplicate match\" is not grounded by the packet, since issue:44451/issue:44488 has higher similarity metrics; and (2) the BERT pair is better described as a likely related loading class rather than a confirmed duplicate bug. The non-persistent-buffer pair looks solid, but the representative/canonical choice should be phrased more cautiously." }, "evaluator_used": true, "retried": true, @@ -2013,14 +1965,14 @@ "issue:44164|issue:44756", "issue:43646|issue:43708" ], - "prepared_review_unit_hash": "cc25c6130cefd8e016ef27c97a517570860e29a04d0132484dd92bce89ab82e1", + "prepared_review_unit_hash": "49b71b5c3e81dcc0f4aead9a7ca0d291e920e2ec72f2c3c09824b7091bad5779", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, @@ -2035,223 +1987,226 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose cluster of mostly unrelated Transformers regressions. A few items are thematically close around custom-model save/load and initialization in Transformers 5.x, but none of the soft pairs look like true duplicates under a strict triage standard.", - "confidence": 0.89, - "canonical_issue_reason": "Issue 43646 is the broadest umbrella for the cluster\u2019s recurring Transformers 5.x custom-model initialization/load regressions, and is the best single issue to anchor a mixed cluster like this.", + "summary": "This cluster is mostly a grab bag of Transformers regressions around custom model loading/init, quantization, attention, and model-specific bugs. I only see one near-duplicate pair: a shared `resize_token_embeddings`/`output_embeddings` regression. The rest are too model- or symptom-specific to merge safely.", + "confidence": 0.53, + "canonical_issue_reason": "Issue 43646 is the broadest umbrella report in the set: a generic Transformers 5.0 custom-model initialization breakage. It best represents the cluster\u2019s main recurring theme compared with the narrower model-specific save/load or embedding reports.", "canonical_pr_reason": null, - "best_issue_reason": "43646 is the most representative issue if you need one tracker for the cluster: it captures the general custom-model init breakage that overlaps with several nearby save/load reports, even though the rest of the cluster is mostly separate bugs.", + "best_issue_reason": "43646 is the most representative and actionable issue because it is the most general regression description and can subsume several narrower custom-model initialization failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both are memory-related, but one is Qwen3 MoE training memory usage and the other is quantization/offload behavior; different bugs." + "reason": "Both mention memory/quantization concerns, but one is Qwen3 MoE training memory usage and the other is quantized offloading behavior; different bugs and fixes." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Different models and different failure modes; one is a specific model load issue, the other is a tokenizer regression for a different model." + "reason": "Both are tokenizer/model-load failures, but they affect different models with different failure modes; no clear shared code-path bug." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Same Transformers 5.x theme, but one is junk non-persistent buffers and the other is notebook custom-model initialization; not the same bug." + "reason": "Related to Transformers 5.0 breakages, but one is junk non-persistent buffers and the other is Jupyter/custom model initialization; not the same underlying issue." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Both are model-serving/chat crashes, but they involve different models and different code paths." + "reason": "Different model families and crash sites: Qwen3.5 chat vs Gemma4Processor in serve; too different to be duplicates." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both mention memory, but one is training efficiency and the other is int4 reserved-memory OOM; separate regressions." + "reason": "Both are memory-related, but one is training memory efficiency and the other is int4 reserved-memory OOM; separate regressions." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Both hit flash_attention code, but the failures differ: packed 3D position_ids misinterpretation vs s_aux=None AttributeError." + "reason": "Both touch flash attention, but one is a packed-sequence/position_ids crash and the other is an `s_aux=None` AttributeError; different code paths." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "LoRA/qwen3_moe compatibility and resize_token_embeddings are unrelated bugs." + "reason": "LoRA checkpoint compatibility vs resize_token_embeddings propagation; unrelated bugs." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "ImportError for GenerationMixin is unrelated to Whisper pipeline return_language regression." + "reason": "ImportError for `GenerationMixin` and Whisper pipeline `return_language` regression are unrelated." }, { "left": "issue:45276", "right": "issue:45335", "accept": false, - "reason": "Both are resize_token_embeddings symptoms, but they affect different model implementations and different embedding targets; not enough to call duplicate." + "reason": "Both concern resize_token_embeddings, but they describe different model-specific embedding surfaces; not enough evidence they are the same bug." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both concern untied lm_head weights, but they are model-specific reports with no evidence of the same underlying code-path." + "reason": "Similar symptom of untied `lm_head`, but they affect different model classes and are likely separate model-specific omissions." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "save/from_pretrained extra_state handling and init_empty_weights _is_hf_initialized are different load/init regressions." + "reason": "Different lifecycle bugs: extra_state save/load handling versus an unexpected `_is_hf_initialized` constructor argument." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Auto device mapping CUDA assert and AudioFlamingo3 track leakage are unrelated inference bugs." + "reason": "Different problems: Qwen3 auto device mapping CUDA assert vs AudioFlamingo3 batched inference token leak." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Both touch tokenizer/vocab edge cases, but they are about different models and different symptoms." + "reason": "The latter is a model load failure that may involve vocab/tokenizer issues, but the former is a specific pad/unk vocab_size mismatch; not the same bug." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "Documentation example errors are not the same as the GLM5 issue." + "reason": "Unrelated: SigLIP2 doc-example errors vs GLM5 model issue." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "Gradient checkpointing with PEFT LoRA and an apex RMSNorm memory leak are unrelated." + "reason": "Gradient checkpointing/PEFT LoRA vs T5 apex RMSNorm memory leak are unrelated." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch 2.10 incompatibility is separate from meta-device model loading failures." + "reason": "Torch version incompatibility and meta-device model loading are different regressions." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Quantization offload behavior and int4 reserved-memory OOM are different bugs." + "reason": "Both involve memory pressure, but one is quantization offloading and the other is int4 reserved-memory growth; different fixes." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Sparse-layer config creation and Gemma4 FSDP2 attention state rebuilding are unrelated." + "reason": "SwitchTransformers sparse-layer construction and Gemma4 FSDP2 shared_kv_states are unrelated model bugs." }, { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Gemma4 FSDP2 attention state and Qwen3.5 multimodal config unwrapping are different code paths." + "reason": "Gemma4 FSDP2 shared_kv_states vs Qwen multimodal config unwrapping; different code paths." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Gemma4 attention/FSDP2 and Qwen3_5Moe rotary embedding offload are distinct issues." + "reason": "Gemma4 FSDP2 bug and Qwen3_5Moe CPU offload incompatibility are unrelated." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and Strix Halo mmap OOM are unrelated." + "reason": "`init_empty_weights` constructor arg issue vs Strix Halo mmap OOM; not the same bug." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "AudioFlamingo3 batched inference leakage and Gemma4 PLE device mismatch are different bugs." + "reason": "AudioFlamingo3 track leakage and Gemma4 device mismatch are separate failures." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both are memory-related, but reserved CUDA memory under int4 and mmap OOM are not the same failure." + "reason": "Both are memory-related, but one is int4 reserved memory and the other is mmap on Strix Halo; different root causes." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "GLM-4.7-Flash loading in tests and Qwen3.5 save_pretrained regression are separate issues." + "reason": "GLM-4.7-Flash load failure in tests and Qwen3.5 save_pretrained regression are unrelated." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Different quantization formats, different models, and different outcomes." + "reason": "Different quantization/load problems on different models (NVFP4 vs MXFP4); not the same concrete bug." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Checkpointing ineffective with PEFT LoRA is not the same as checkpointing being disallowed in eval mode." + "reason": "Both mention gradient checkpointing, but one is ineffective LoRA checkpointing and the other is eval-mode prohibition; separate issues." }, { "left": "issue:44743", "right": "issue:44960", "accept": false, - "reason": "Qwen3 modular recurrent-state reset and GLM5 are unrelated." + "reason": "Qwen recurrent-state reset bug and GLM5 issue are unrelated." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Meta-device loading regression and mmap OOM on Strix Halo are different problems." + "reason": "Meta-device loading and Strix Halo mmap OOM are different initialization/memory issues." }, { "left": "issue:45276", "right": "issue:45292", - "accept": false, - "reason": "Both are resize_token_embeddings reports, but they affect different model internals and look like separate model-specific bugs." + "accept": true, + "reason": "Same underlying `resize_token_embeddings` regression: resized embeddings are not propagated to output embeddings; 45276 is just a Gemma4-specific manifestation with an extra symptom." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Both concern save_pretrained/from_pretrained for custom models, but one is about permissions and the other about loading the wrong module." + "reason": "Both involve save/from_pretrained for custom models, but one is file permissions and the other is wrong module resolution; different bugs." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Gradient checkpointing in eval mode and RLHF NaNs from 3D position_ids are unrelated." + "reason": "Gradient checkpointing in eval mode and Qwen3.5 3D position_ids / SDPA fallback NaNs are unrelated." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Different models and different mechanisms: Gemma4 FSDP2 attention state vs Qwen3_5Moe rotary embedding materialization." + "reason": "Gemma4 FSDP2 shared_kv_states and Qwen3_5Moe rotary embedding uninitialized memory are different model internals." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "save/from_pretrained extra_state handling and Strix Halo mmap OOM are unrelated." + "reason": "Save/load extra_state handling and Strix Halo mmap OOM are unrelated." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom model initialization breakage and Trainer resume max_steps miscalculation are different subsystems and different bugs." + "reason": "Generic custom model initialization breakage and Trainer resume max_steps calculation are separate issues." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted near-duplicate pair, `45276`/`45292`, is reasonably supported by the titles as the same `resize_token_embeddings`/`output_embeddings` regression with a Gemma4-specific manifestation. The other pair verdicts appropriately avoid overstating similarity across model-specific or symptom-specific issues." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -2341,31 +2296,31 @@ "issue:43299|issue:43931", "issue:43994|issue:44079", "issue:44479|issue:44805", - "issue:45005|issue:45902", "issue:39404|issue:44671", + "issue:45005|issue:45902", "issue:40990|issue:42947", - "issue:44671|issue:44991", "issue:44589|issue:45084", + "issue:44671|issue:44991", "issue:43425|issue:44918", "issue:44164|issue:45759", - "issue:43596|issue:44683", "issue:36032|issue:43452", - "issue:43856|issue:44756" + "issue:43856|issue:44756", + "issue:43596|issue:44683" ], - "prepared_review_unit_hash": "057cdc226e3dacd40654861fb8f601c857c29d39903fb1d9a068041fd3484171", + "prepared_review_unit_hash": "884edecc5efa642e2df19c63c60e8b47af34aa55df8e309b431ca5e314b5dc82", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37609, + "serialized_chars": 37611, "estimated_input_tokens": 9403, "estimated_eval_tokens": 19062 }, @@ -2374,198 +2329,216 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly heterogeneous. I only see one clear duplicate-style pair: the two resize_token_embeddings reports. Most other pairs share only broad themes like v5 loading regressions, model-specific dtype issues, or docs drift, which is not enough to merge them as the same bug.", - "confidence": 0.68, - "canonical_issue_reason": "issue:43611 is the broadest representative of the cluster\u2019s dominant theme: a Transformers v5 model-loading regression. Several other reports are narrower, model-specific load failures, so this is the best single issue to anchor the cluster.", + "summary": "This cluster is largely heterogeneous and mostly not duplicate-related: it spans unrelated loading, dtype, quantization, docs, and model-specific regressions. The only clear near-duplicate pair is the resize_token_embeddings bug affecting output/decoder embeddings.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 45292 is the broadest, most general statement of the shared bug in the one true duplicate-like pair: resize_token_embeddings not propagating to output embeddings.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the clearest overall representative of the main failure pattern in this set: v5 breaking model loading across different models/configurations.", + "best_issue_reason": "Issue 45292 is the best representative of the cluster\u2019s only concrete duplicate theme and is phrased more generally than the t5gemma-specific report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Both are v5-era regressions, but one is a video-input bug for Qwen VL models and the other is a Granite Speech config-loading issue; different code paths." + "reason": "Different regressions in different model families: Qwen video-input behavior vs granite_speech config loading." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash-attention crash on sink-less models is unrelated to the Gemma4 FSDP2 shared_kv_states KeyError." + "reason": "Different code paths and failures: flash_attention s_aux=None crash vs Gemma4 FSDP2 shared_kv_states KeyError." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Same underlying resize_token_embeddings bug: the generic output_embeddings report matches the t5gemma decoder.embed_tokens symptom." + "reason": "Same underlying bug: resize_token_embeddings is not updating downstream embeddings; 45335 is the t5gemma-specific manifestation." }, { "left": "issue:43257", "right": "issue:44863", "accept": false, - "reason": "Different load failures: Qwen3 MoE conversion under accelerate/deepspeed vs NemotronH checkpoint loading." + "reason": "Both are checkpoint/loading complaints, but they involve different model implementations and failure modes." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Custom-model initialization in Jupyter notebooks is a different bug from save/from_pretrained handling extra_state." + "reason": "Custom-model notebook initialization is unrelated to save/from_pretrained extra_state handling." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "The first is a tokenizer vocab-size mismatch; the second is a generic ScandiBERT load failure with no clear evidence it shares the same root cause." + "reason": "A tokenizer vocab mismatch is not the same as a model-specific load failure for ScandiBERT." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch and meta-device loading breakage are different symptoms and likely different fixes." + "reason": "Different bugs: autocast dtype mismatch vs meta-device loading breakage." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch 2.10 incompatibility is too broad and unrelated to the specific autocast dtype mismatch bug." + "reason": "Generic torch version incompatibility is unrelated to the Phi MoE autocast dtype issue." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline return_language regression is not the same as a general tokenization-behavior change report." + "reason": "Whisper pipeline return_language regression is unrelated to a general tokenization change report." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Missing Float8 storage and disabling mmap on Strix Halo for OOM are unrelated issues." + "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap OOM issue." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 device mismatch are different failures." + "reason": "Compiled flex_attention failure on torch>=2.9 is unrelated to Gemma4 device placement mismatch." }, { "left": "issue:43931", "right": "issue:44863", "accept": false, - "reason": "Qwen3-VL shape-mismatch loading and NemotronH checkpoint loading are distinct model families and bugs." + "reason": "Qwen3-VL weight-shape mismatch is a different loading problem than NemotronH checkpoint incompatibility." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "An isolated reproduction of another issue is not the same bug as the Qwen3.5 PagedAttentionCache linear_attention crash." + "reason": "An isolated reproduction of an older issue is not the same bug as the Qwen3.5 PagedAttentionCache crash." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both are v5 regressions, but one is base_model_prefix loading behavior and the other is custom-model initialization in notebooks." + "reason": "Both are loading-related, but base_model_prefix breakage and Jupyter custom-model initialization are different defects." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as the Gemma4 PLE device-map forward mismatch." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and non-template-node compile failure are different code-path problems." + "reason": "Qwen3.5 linear_attention cache crash is unrelated to the non-template-node compilation TypeError." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "A torch version incompatibility report is not the same as a Mistral-3 image-preprocessor dtype mismatch." + "reason": "Version incompatibility and image-preprocessor dtype mismatch are different issues." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs issues, but they cover different removed/renamed pipeline APIs and should not be merged as one bug." + "reason": "Both are docs issues, but they cover different removed/deprecated pipeline docs and are not the same change." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper pipeline behavior and gpt-oss MXFP4 quantization fallback are unrelated." + "reason": "Whisper pipeline behavior is unrelated to MXFP4 quantization fallback on gpt-oss-20b." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export / torch_compilable_check failure is unrelated to Perceiver non-default-resolution interpolation." + "reason": "torch.export failure on torch_compilable_check is unrelated to Perceiver non-default-resolution classification failure." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Both involve Qwen3-VL loading, but one is a MoE-model regression and the other is a weight-shape mismatch; not clearly the same concrete bug." + "reason": "Different Qwen3-VL loading problems with different symptoms: model-specific v5 breakage vs weight-shape mismatch." }, { - "left": "issue:45005", - "right": "issue:45902", + "left": "issue:43994", + "right": "issue:44079", "accept": false, - "reason": "Tied-weights translation-model issues and the Qwen3_5Moe rotary-embedding memory bug are unrelated." + "reason": "Incorrect predictions for SigLIP2 are unrelated to ModelOutput key assignment logic." }, { - "left": "issue:44164", - "right": "issue:45759", + "left": "issue:44479", + "right": "issue:44805", "accept": false, - "reason": "Handling extra_state in save/from_pretrained is a different bug from not unwrapping text_config in AutoModelForCausalLM.from_config." + "reason": "Video-input regression for Qwen models is unrelated to the mask/tensor shape IndexError." }, { - "left": "issue:43596", - "right": "issue:44683", + "left": "issue:39404", + "right": "issue:44671", "accept": false, - "reason": "deepspeed zero3/BertModel index error is unrelated to compiled flex_attention failing on torch>=2.9." + "reason": "Whisper pipeline regression and CamemBERT masked-LM prediction issues are distinct." }, { - "left": "issue:36032", - "right": "issue:43452", + "left": "issue:45005", + "right": "issue:45902", "accept": false, - "reason": "T5 tokenizer add_special_tokens conflict and gguf_file failures in from_pretrained are different loading bugs." + "reason": "Tied-weights translation-model loading and Qwen3.5 rotary embedding materialization are different bugs." }, { - "left": "issue:43856", - "right": "issue:44756", + "left": "issue:40990", + "right": "issue:42947", "accept": false, - "reason": "Qwen3 MoE training memory usage and Strix Halo mmap OOM are unrelated performance/memory issues." + "reason": "Perplexity regression on gpt-oss-20b is unrelated to gradient checkpointing with PEFT LoRA." }, { - "left": "issue:40990", - "right": "issue:42947", + "left": "issue:44589", + "right": "issue:45084", "accept": false, - "reason": "High perplexity on a model-eval benchmark is not the same bug as gradient checkpointing being ineffective with PEFT LoRA." + "reason": "Float8 storage lookup failure is unrelated to the compile-time template-node error." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "CamemBERT masked-LM prediction errors and tokenizer-load failure for EMBEDDIA/est-roberta are different model-specific problems." + "reason": "CamemBERT prediction regression and tokenizer-loading failure for est-roberta are different defects." }, { - "left": "issue:44589", - "right": "issue:45084", + "left": "issue:43425", + "right": "issue:44918", "accept": false, - "reason": "These are different failures: one is a missing Float8 storage type, the other is a compilation error for non-template nodes." + "reason": "Torch version incompatibility is unrelated to TRL SFT unpacking of Qwen3.5 embeddings." }, { - "left": "issue:43425", - "right": "issue:44918", + "left": "issue:44164", + "right": "issue:45759", "accept": false, - "reason": "Torch 2.10 incompatibility and Qwen3.5 SFT embedding unpacking are unrelated." + "reason": "extra_state save/load handling is unrelated to composite Qwen config unwrapping." }, { - "left": "issue:44479", - "right": "issue:44805", + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer special-token conflict is unrelated to gguf_file breaking AutoTokenizer/AutoModel loading." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "MoE training memory inefficiency is unrelated to the Strix Halo mmap OOM workaround." + }, + { + "left": "issue:43596", + "right": "issue:44683", "accept": false, - "reason": "A video-input regression and a mask/tensor shape mismatch are not the same underlying bug." + "reason": "DeepSpeed ZeRO-3 init IndexError is unrelated to compiled flex_attention on newer torch." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted pair, `issue:45292` / `issue:45335`, is a plausible duplicate-style resize_token_embeddings bug. The other rejections are appropriately cautious given the differing model families, code paths, or problem types." + "feedback": "The summary is grounded in the packet and stays conservative overall. The one accepted soft pair (45292/45335) is well-supported by the titles, and the rejected pairs do not overstate duplicate evidence." }, "evaluator_used": true, "retried": false, @@ -2642,8 +2615,8 @@ "issue:44164|issue:44479", "issue:43582|issue:43632", "issue:44488|issue:45081", - "issue:44661|issue:44683", "issue:40990|issue:43421", + "issue:44661|issue:44683", "issue:43611|issue:43644", "issue:44488|issue:44987", "issue:44206|issue:44220", @@ -2670,239 +2643,242 @@ "issue:43994|issue:44610", "issue:43645|issue:44534" ], - "prepared_review_unit_hash": "16279dedaae2e64e864f679608b7e9fbeea5056ba36d198d3b711edddfd07941", + "prepared_review_unit_hash": "3be1cac12bda73ff62d503adb74c18009a057c5a32e66b0c1bb83e67e3b38df9", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37696, - "estimated_input_tokens": 9424, - "estimated_eval_tokens": 19104 + "serialized_chars": 37700, + "estimated_input_tokens": 9425, + "estimated_eval_tokens": 19106 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issue set; the only clear duplicate is the Transformers v5/non-persistent buffers regression. No PRs are present.", - "confidence": 0.94, - "canonical_issue_reason": "issue:43644 is the clearest representative of the duplicate pair: it precisely describes the v5 regression where non-persistent buffers are filled with junk, and issue:44534 is the same bug phrased more generally.", + "summary": "This cluster is mostly heterogeneous, with issues spanning unrelated Transformers subsystems. I found one likely duplicate pair around GGUF loading in v5; the rest are distinct bugs and should stay separate.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue for the whole set: the items cover many unrelated regressions across training, tokenizers, audio, vision, quantization, and model loading.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the best canonical anchor because it is specific, well-scoped, and directly matches the near-identical later report.", + "best_issue_reason": "issue:43452 is the strongest canonical issue candidate among the near-duplicates because it describes the broader GGUF loading regression directly, while issue:43482 looks like a specific model instance of the same failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:32090", "right": "issue:39404", "accept": false, - "reason": "Different failures: Trainer GPU broadcast NoneType error vs Whisper pipeline return_language regression." + "reason": "Different failures: GPU broadcast of NoneType in Trainer vs Whisper pipeline return_language behavior." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Both involve tokenizer/model loading, but the concrete bugs differ: tokenizer loading for est-roberta vs Mistral regex patching on a backend_tokenizer attribute." + "reason": "Both involve loading/tokenizer code, but one is a tokenizer load failure for a specific model and the other is a Mistral regex patch/backend_tokenizer error." }, { "left": "issue:44737", "right": "issue:45859", "accept": false, - "reason": "Both are device/offload related, but they affect different models and code paths (XLNet positional encoding vs Qwen3.5 MoE rotary embedding)." + "reason": "Different code paths: XLNet positional encoding device placement vs Qwen3.5-MoE CPU-offload incompatibility." }, { "left": "issue:44164", "right": "issue:44479", "accept": false, - "reason": "Unrelated areas: save/from_pretrained extra_state handling vs a video input regression for Qwen multimodal models." + "reason": "Save/load extra_state handling is unrelated to the Qwen video-input regression." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Different bugs: Apple Silicon caching_allocator warmup TypeError vs `_is_hf_initialized` flag breakage in v5." + "reason": "Apple Silicon allocator warmup TypeError and the _is_hf_initialized regression are separate bugs." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Both mention loading failures, but one is a model-loading problem and the other is a tokenizer regex patch crash." + "reason": "Model-loading failure for cjvt/sleng-bert is unrelated to the Mistral tokenizer backend_tokenizer crash." }, { - "left": "issue:44661", - "right": "issue:44683", + "left": "issue:40990", + "right": "issue:43421", "accept": false, - "reason": "Different mechanisms: TOKENIZER_MAPPING_NAMES handling vs compiled flex_attention on torch >= 2.9." + "reason": "Perplexity regression on gpt-oss is not the same as runtime post-processor updates in TokenizersBackend." }, { - "left": "issue:40990", - "right": "issue:43421", + "left": "issue:44661", + "right": "issue:44683", "accept": false, - "reason": "Perplexity issue vs runtime special-token post-processor update bug; same tokenizer/theme but not the same bug." + "reason": "Tokenizer mapping-name bug and compiled flex_attention failure are unrelated." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are Transformers v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer serialization." + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; different bugs." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Both are model-loading failures, but they target different models and likely different regressions." + "reason": "Both are model-loading reports, but they concern different checkpoints and failure modes." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Both are audio-related feature extraction issues, but the concrete failures differ (`center` arg regression vs `_torch_extract_fbank_features()`)." + "reason": "Same audio area, but one is an unsupported center argument and the other is a _torch_extract_fbank_features issue; not clearly the same bug." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Different multimodal bugs: image preprocessor dtype mismatch vs a template compilation TypeError." + "reason": "Image-preprocessor dtype mismatch is unrelated to the template-node compilation TypeError." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "Missing fields in AutoConfig with trust_remote_code is unrelated to PagedAttentionCache/linear_attention failures." + "reason": "Missing trust_remote_code fields and invalid group type in PagedAttentionCache are different problems." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Different problems: missing `all_tied_weights_keys` on Molmo vs tokenizer loading failure for a specific model." + "reason": "Missing all_tied_weights_keys and loading cjvt/sleng-bert are unrelated." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "TrainerState token counting bug vs Qwen2.5-VL iterable dataset with multiple images; no shared code-path." + "reason": "TrainerState token counter issue and Qwen2.5-VL iterable dataset failure are distinct." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper token decoding ignores skip_special_tokens vs empty-transcription failure after align_special_tokens; related area but distinct symptoms/code paths." + "reason": "Both are Whisper-related, but one is batch_decode skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Close in wording, but 43646 is a broader custom-model initialization break while 43644 is specifically non-persistent buffer serialization; not clearly the same bug." + "reason": "Non-persistent buffer junk and custom model initialization regressions are separate v5 issues." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "Tokenizer add_special_tokens conflict vs FineGrainedFP8Config failure on Qwen3Moe; different components entirely." + "reason": "T5 tokenizer special-token conflict is unrelated to Qwen3Moe FineGrainedFP8Config failure." }, { "left": "issue:43452", "right": "issue:43482", - "accept": false, - "reason": "Both mention GGUF/loading, but one is a generic gguf_file break in AutoTokenizer/AutoModel loading and the other is a specific Qwen2.5-GGUF v5 failure; not clearly the same defect." + "accept": true, + "reason": "Both point to the same GGUF loading regression in transformers v5; 43482 looks like a specific model manifestation of the broader 43452 failure." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Two different model-loading failures for different repositories; no evidence they share the same underlying bug." + "reason": "Different model-loading failures for different checkpoints and different triggers." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Mixtral grouped_mm torch dynamo tracing error is unrelated to the template-node compilation error." + "reason": "Torch dynamo grouped_mm tracing failure is unrelated to the template-node compilation error." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "FineGrainedFP8Config failure on Qwen3MoE vs PagedAttentionCache invalid group type; different layers and symptoms." + "reason": "FineGrainedFP8Config failure and linear_attention group-type crash are separate Qwen issues." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Qwen3.5 output_hidden_states bug vs Qwen2.5-VL rope index scaling bug; same family name only, not the same issue." + "reason": "Output_hidden_states bug and RoPE temporal position scaling bug are different Qwen3.5/Qwen2.5-VL problems." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Different loading failures: dequantized model device_map offload vs Qwen2.5-GGUF v5 loading regression." + "reason": "gpu+cpu offload dequantized-model loading and GGUF loading regression are different code paths." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "Tokenizer method conflict and custom model initialization break are unrelated." + "reason": "Tokenizer method-name conflict and notebook custom-model initialization failure do not match." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Different model families and bugs: Mistral3 lm_head tying vs inconsistent lm_head serialization depending on device." + "reason": "Mistral3 lm_head tying bug and device-dependent lm_head serialization are related in theme but not the same concrete bug." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Both are version-gated loading failures, but for different models and different objects (model vs tokenizer)." + "reason": "Separate versioned loading failures for different models/tokenizers; not the same underlying regression." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Molmo tied-weights attribute error is unrelated to tokenizer loading for EMBEDDIA/est-roberta." + "reason": "Missing all_tied_weights_keys is unrelated to est-roberta tokenizer loading." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio chat-template sampling-rate default issue vs MIMI encoder batch-padding-mask bug; different subsystems." + "reason": "Audio processor chat-template sampling-rate default and MIMI batched-vs-single padding-mask bug are different." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Template compilation TypeError is unrelated to GraniteMoEHybrid calling an invalid method." + "reason": "Template compilation TypeError and GraniteMoEHybrid invalid-method calls are different failures." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "43646 is about custom model initialization; 44534 is specifically non-persistent buffers filled with junk. Different bugs." + "reason": "Custom model init breakage and junk-filled non-persistent buffers are separate v5 regressions." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different vision model issues: SigLIP2 nonsensical outputs vs OmDet-Turbo processor/model input-size mismatch." + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo 640px-vs-224px preprocessing mismatch are unrelated." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Jupyter/custom-model initialization regression vs non-persistent buffer junk-fill regression; not the same bug." + "reason": "Notebook custom-model initialization issue does not match the non-persistent-buffer regression." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: this set is clearly heterogeneous, and the lone GGUF-loading pair (43452/43482) is a reasonable near-duplicate call based on the titles. The other soft-edge rejections are conservative and do not overclaim equivalence from theme-level similarity alone." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -2938,7 +2914,6 @@ "issue:43611", "issue:43632", "issue:43742", - "issue:43818", "issue:43819", "issue:44062", "issue:44079", @@ -2976,8 +2951,8 @@ "issue:44265|issue:44610", "issue:45216|issue:45357", "issue:44451|issue:45081", - "issue:44220|issue:44683", "issue:44987|issue:45081", + "issue:44220|issue:44683", "issue:43404|issue:45127", "issue:43632|issue:44756", "issue:42915|issue:44291", @@ -2986,8 +2961,8 @@ "issue:36032|issue:43742", "issue:43296|issue:43531", "issue:43611|issue:44534", - "issue:40444|issue:45325", "issue:42915|issue:44589", + "issue:40444|issue:45325", "issue:43531|issue:44877", "issue:40990|issue:44811", "issue:36032|issue:41720", @@ -2995,225 +2970,218 @@ "issue:43425|issue:44403", "issue:43632|issue:44534", "issue:41553|issue:43582", - "issue:42175|issue:45542", "issue:44462|issue:45812", - "issue:43421|issue:44568", - "issue:43818|issue:45357" + "issue:42175|issue:45542", + "issue:43421|issue:44568" ], - "prepared_review_unit_hash": "a1ce9f57d70c4b2b64766f257a4bb3d7276b388364048efb89db17a7107ab145", + "prepared_review_unit_hash": "e02cc04075cc0f6db45a989660e71a209e565021defc729cfcc2cf37be2ac786", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36713, - "estimated_input_tokens": 9179, - "estimated_eval_tokens": 18614 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35801, + "estimated_input_tokens": 8951, + "estimated_eval_tokens": 18158 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly distinct transformer/model-specific regressions; only one soft pair looks like the same save_pretrained bug for Qwen3.5. The rest share broad subsystems but not the same concrete failure mode.", - "confidence": 0.44, - "canonical_issue_reason": "issue:45812 is the best cluster representative because it describes a broad, user-visible v4\u2192v5 tokenizer regression affecting all Granite models, which is more general than the narrower model-specific failures here.", + "summary": "This cluster is mostly a loose collection of unrelated Transformers bugs. The only clear duplicate-like pair is the Qwen3.5 `save_pretrained` regression, where the later issue narrows the same incorrect-checkpoint problem to visual encoder keys.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45216 is the broadest, earliest report of the Qwen3.5 `save_pretrained` serialization regression; 45357 appears to be a later, narrower symptom of the same bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is the strongest stand-in for the set: it is broad, regression-oriented, and about core loading/tokenization behavior rather than a one-off model quirk.", + "best_issue_reason": "Issue 45216 is the best representative issue for the one confirmed duplicate pair in this set and has the clearest, most general description of the shared bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Both are multimodal loading/input errors, but one is a Llava Next segfault and the other is a Qwen3OmniMoe video ValueError; different code paths and fixes." + "reason": "Different multimodal model families and different failure modes (segfault on Llava Next loading vs ValueError on Qwen3OmniMoe video processing)." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Unnecessary load noise vs a Mistral regex patch crash; unrelated symptoms and root causes." + "reason": "One is a generic loading-noise complaint; the other is a specific Mistral tokenizer regex crash. Not the same code-path bug." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Different models and failures: MIMI padding-mask batching vs DAC latent/STE mismatch." + "reason": "Both are audio-related, but one is a padding-mask batching mismatch in MIMI and the other is a DAC STE/from_latents mismatch. Different bugs." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "Qwen3Moe FP8 config failure is unrelated to the template-compilation TypeError." + "reason": "Qwen3Moe FP8 failure and a template-node compilation TypeError are unrelated." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "torch.export/torch_compilable_check is a compile/export issue; the other is an OmDet-Turbo image-size mismatch." + "reason": "Torch export failure with `torch_compilable_check` is unrelated to OmDet-Turbo processor image-size mismatch." }, { "left": "issue:45216", "right": "issue:45357", "accept": true, - "reason": "Both report the same Qwen3.5 save_pretrained regression with incorrect saved keys; the later issue is a continuation of the same bug." + "reason": "Same Qwen3.5 `save_pretrained` checkpoint regression; the later issue specifically calls out incorrect visual encoder keys, which is a narrower manifestation of the same bug." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "ScandiBERT loading failure is not the same as the Mistral backend-tokenizer crash." + "reason": "Different models and different failures: ScandiBERT load failure vs Mistral regex patch crash." }, { - "left": "issue:44220", - "right": "issue:44683", + "left": "issue:44987", + "right": "issue:45081", "accept": false, - "reason": "Feature extraction/fbank issue vs compiled flex_attention failure; different components and failure modes." + "reason": "Different affected models and separate loading paths; not the same underlying bug." }, { - "left": "issue:44987", - "right": "issue:45081", + "left": "issue:44220", + "right": "issue:44683", "accept": false, - "reason": "Loading physical-intelligence/fast and the Mistral regex crash are separate loading regressions." + "reason": "Feature-extraction helper issue and compiled flex_attention torch-version incompatibility are unrelated." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Untied lm_head weights in Mistral3 are not the same as LoRA-vocabulary collapse after merge." + "reason": "Weight-tying bug in Mistral3 is unrelated to LoRA merging with extended vocab collapse." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "_is_hf_initialized compatibility break and mmap/OOM on Strix Halo are unrelated." + "reason": "`_is_hf_initialized` regression and mmap OOM on Strix Halo are different problems." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Both mention fp8/initialization issues, but one is a Qwen3Moe config failure and the other is an init_empty_weights argument regression." + "reason": "Both mention loading/training regressions, but one is Qwen3Moe FP8-specific and the other is an `init_empty_weights` argument error." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "AddedToken keyword conflict is unrelated to the physical-intelligence/fast loading failure." + "reason": "Tokenizer added-token argument collision is unrelated to loading `physical-intelligence/fast`." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both involve Qwen VL/video position handling, but one is a broad video-input regression and the other is a specific still-image temporal-position scaling bug." + "reason": "Both touch Qwen VL rope/vision logic, but one is a broad video-input regression across several models and the other is a specific still-image temporal position scaling bug." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "T5Tokenizer add_special_tokens conflict and MobileLLM key error are unrelated loading issues." + "reason": "T5 special-token conflict and MobileLLM key error are different tokenizer/model-loading issues." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure is unrelated to the Qwen3-MoE sliding_window issue." + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding-window issue are unrelated." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "base_model_prefix loading break and junk non-persistent buffers are different v5 regressions." + "reason": "Base-model-prefix loading breakage and non-persistent buffer junk-filling are different v5 regressions." }, { - "left": "issue:40444", - "right": "issue:45325", + "left": "issue:42915", + "right": "issue:44589", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is not the same as the Qwen2.5-VL temporal position scaling bug." + "reason": "Qwen3Moe FP8 config failure and Float8 storage TypeError are distinct errors." }, { - "left": "issue:42915", - "right": "issue:44589", + "left": "issue:40444", + "right": "issue:45325", "accept": false, - "reason": "FineGrainedFP8Config failure is unrelated to missing Float8 storage support." + "reason": "Different Qwen2.5-VL bugs: iterable-dataset multi-image finetuning failure vs rope-index temporal-position scaling." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE and strict config loading for granite_speech are separate issues." + "reason": "Qwen3-MoE sliding-window issue and Granite speech strict-config loading are unrelated." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on GPT-OSS and Whisper skip_special_tokens behavior are unrelated." + "reason": "Perplexity on GPT-OSS and Whisper batch_decode special-token handling are not the same bug." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer load conflict and Qwen3 auto-device-map cuda assert are different bugs." + "reason": "Tokenizer load conflict and Qwen3 auto device-mapping cuda assert are unrelated." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared_kv_states KeyError are unrelated." + "reason": "ModelOutput key assignment bug and Gemma4 FSDP2 shared_kv_states KeyError are different issues." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch version incompatibility does not match the generic load-noise issue." + "reason": "Torch version compatibility and noisy loading messages are unrelated." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "_is_hf_initialized break and non-persistent buffer corruption are distinct regressions." + "reason": "Different v5 regressions with different symptoms and code paths." }, { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "Voxtral tokenizer error-message bug and AppleSilicon warmup TypeError are unrelated." + "reason": "Bad AutoTokenizer error message for Voxtral and Apple Silicon warmup TypeError are unrelated." }, { - "left": "issue:42175", - "right": "issue:45542", + "left": "issue:44462", + "right": "issue:45812", "accept": false, - "reason": "Both touch TensorFlow optionality, but one is missing TF in the torch extra and the other is a backend/runtime detection error; not the same concrete bug." + "reason": "Both involve AutoTokenizer regressions, but one is about ignoring repository `tokenizer.json` and the other is a Granite-specific wrong-token-IDs regression; too broad to treat as the same bug without stronger evidence." }, { - "left": "issue:44462", - "right": "issue:45812", + "left": "issue:42175", + "right": "issue:45542", "accept": false, - "reason": "Both are tokenizer regressions, but one is repo tokenizer.json being ignored while the other is Granite token IDs being wrong; not enough evidence they share one root cause." + "reason": "Both concern TensorFlow/backend packaging, but one is missing tf in the torch extra while the other is a tensorboard-without-TF runtime backend error." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Runtime special-token post-processor updates and mdeberta BOS/EOS insertion are related in area but not the same concrete bug." - }, - { - "left": "issue:43818", - "right": "issue:45357", - "accept": false, - "reason": "Video-LLaVA weight/temporal-attention defect is unrelated to Qwen3.5 save_pretrained key serialization." + "reason": "Runtime special-token post-processor updates and `add_special_tokens=True` BOS/EOS behavior are distinct tokenizer bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative about the soft pairs. The only accepted pair, issue:45216 vs issue:45357, is plausibly the same Qwen3.5 save_pretrained regression. The rest are appropriately rejected as distinct bugs, and the representative issue choice is supported by the broader tokenizer-regression title rather than overclaiming cluster-wide evidence." + "feedback": "Grounded and conservative overall. The Qwen3.5 save_pretrained pair is the only plausible duplicate-like match from the packet, and the other soft-edge rejections stay appropriately cautious." }, "evaluator_used": true, "retried": false, @@ -3254,6 +3222,7 @@ "issue:43701", "issue:43749", "issue:43792", + "issue:43818", "issue:43856", "issue:43957", "issue:43994", @@ -3263,7 +3232,6 @@ "issue:44387", "issue:44403", "issue:44466", - "issue:44479", "issue:44534", "issue:44589", "issue:44617", @@ -3275,21 +3243,22 @@ "issue:45072", "issue:45081", "issue:45325", + "issue:45357", "issue:45381", "issue:45479", "issue:45491", "issue:45538", "issue:45800", - "issue:45823", "issue:45834" ], "soft_pairs": [ "issue:43317|issue:43856", + "issue:43818|issue:45357", "issue:33666|issue:40444", - "issue:44683|issue:45020", "issue:43792|issue:44589", "issue:44829|issue:45479", "issue:44403|issue:44991", + "issue:44683|issue:45020", "issue:43482|issue:45538", "issue:35707|issue:40444", "issue:44079|issue:44534", @@ -3313,223 +3282,234 @@ "issue:43792|issue:43957", "issue:44387|issue:44617", "issue:43296|issue:45800", - "issue:43701|issue:45325", - "issue:44479|issue:45823" + "issue:43701|issue:45325" ], - "prepared_review_unit_hash": "f84019fe7d4d71801f94fd80981fa6122005e6aa4663d29cc19780a9f0424aca", + "prepared_review_unit_hash": "0e0781c0851a795e07924fcd1f7d7639a26c4a88457fb8de10d46bb964442d9a", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36957, - "estimated_input_tokens": 9240, - "estimated_eval_tokens": 18736 + "serialized_chars": 36997, + "estimated_input_tokens": 9250, + "estimated_eval_tokens": 18756 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a collection of unrelated issue pairs. Only 43644 and 44534 look like a true duplicate; the other soft pairs share broad model families or subsystems but describe different bugs.", - "confidence": 0.91, - "canonical_issue_reason": "43644 is the best canonical issue because 44534 is essentially the same v5 non-persistent-buffer regression, and 43644 is the earlier, clearer report.", + "summary": "This cluster is mostly a set of distinct model-specific Transformer regressions, with one clear duplicate pair around the non-persistent buffer junk bug. Most soft links are only broad similarity (same model family, same subsystem, or same Transformers v5 regression theme) and should not be merged.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43644 is the earlier, clearer report of the non-persistent-buffer regression, and issue:44534 appears to be the same bug with nearly identical wording. It is the best canonical issue for that duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": "43644 is the safest representative of the only clear duplicate pair in the set; the remaining issues are not close enough to collapse into one canonical bug.", + "best_issue_reason": "issue:43644 is the strongest canonical issue here because it is specific, concrete, and the earliest of the near-identical reports; it best represents the underlying bug among the provided items.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Both mention model loading/training on GPUs, but they are different bugs in different workflows and model families." + "reason": "Different failures: dequantized model loading on CPU/GPU offload vs Qwen3 MoE training memory usage." }, { - "left": "issue:33666", - "right": "issue:40444", + "left": "issue:43818", + "right": "issue:45357", "accept": false, - "reason": "Same Qwen VL area, but one is multi-GPU training and the other is iterable-dataset fine-tuning with multiple images; not the same bug." + "reason": "Both are vision-model related, but they describe different bugs on different models: Video-LLaVA tower weights/attention vs Qwen3.5 save_pretrained key serialization." }, { - "left": "issue:44683", - "right": "issue:45020", + "left": "issue:33666", + "right": "issue:40444", "accept": false, - "reason": "One is a torch>=2.9 flex_attention regression; the other is a broader remote_code loading breakage." + "reason": "Different Qwen2-VL training problems: multi-GPU training vs iterable dataset with multiple images per prompt." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Different failure modes: model-specific runtime failure versus missing Float8 storage object." + "reason": "Unrelated symptoms: Whisper model runtime failure vs float8 storage lookup error." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "Both affect training loss, but one is flash_attention_3 regression and the other is a problem_type/num_labels loss bug." + "reason": "Both affect training, but one is a flash_attention_3 regression and the other is a classification-loss bug for num_labels=1." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "One is generic loading noise; the other is a specific tokenizer loading failure for a single model." + "reason": "Noise on model load vs tokenizer loading failure for a specific model; not the same code-path bug." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "Different issues: compiled flex_attention on torch >=2.9 vs remote_code model breakage." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "Different components: GGUF loading under v5 versus CLIPTokenizer max_length behavior." + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer model_max_length behavior are unrelated." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Different generation/training paths; no evidence they are the same underlying bug." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to multi-image finetuning failures." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "Both touch v5 internals, but ModelOutput key assignment is unrelated to non-persistent buffer initialization." + "reason": "ModelOutput key assignment and non-persistent buffer junk are different regressions." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Meta init argument handling and Float8 storage lookup are distinct loading bugs." + "reason": "init_empty_weights/_is_hf_initialized argument handling is unrelated to the float8 storage object error." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Different errors and code paths: GGUF loading versus missing pad_token_id on Llama4Config." + "reason": "Different models and symptoms: GGUF loading vs missing pad_token_id on Llama4Config." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Both are batching-related model bugs, but one is padding-mask mismatch and the other is NaNs from sliding-window all-padding cases." + "reason": "Both are batching/shape-related, but MIMI padding-mask output mismatch is not the same as Gemma3 sliding-window NaNs." }, { "left": "issue:43749", "right": "issue:44079", "accept": false, - "reason": "FSDP CPU RAM-efficient loading is unrelated to ModelOutput key assignment." + "reason": "FSDP CPU RAM efficient loading and ModelOutput key assignment are separate bugs." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Non-persistent buffer junk is unrelated to ModelOutput key assignment." + "reason": "Non-persistent buffer junk regression is unrelated to ModelOutput key assignment." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen2.5-VL, but one is multi-image finetuning and the other is video position-id regression." + "reason": "Both involve Qwen2.5-VL, but one is multi-image finetuning and the other is video vision_position_ids." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Auto device-mapping CUDA assert and Mistral tokenizer regex crash are unrelated." + "reason": "Auto device-map CUDA assert on Qwen3 is unrelated to Mistral tokenizer regex patch failure." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Sliding-window MoE bug and init_empty_weights argument mismatch are different issues." + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument handling are different bugs." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export with torch_compilable_check is a different problem from dtype mismatch in inference." + "reason": "torch.export with torch_compilable_check is unrelated to dtype mismatch in bfloat16 inference." }, { "left": "issue:43632", "right": "issue:45072", "accept": false, - "reason": "_is_hf_initialized flag handling is unrelated to bfloat16 dtype mismatches." + "reason": "_is_hf_initialized flag regression and dtype mismatch are different failures." }, { "left": "issue:45325", "right": "issue:45381", "accept": false, - "reason": "Related Qwen2.5-VL position-id area, but one is still-image temporal scaling and the other is video vision_position_ids; too different to merge confidently." + "reason": "Closely related Qwen2.5-VL position-id topics, but one is still-image temporal scaling and the other is video input vision_position_ids; not clearly the same bug." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are Transformers v5 regressions, but one is base_model_prefix loading and the other is custom model initialization." + "reason": "Both are broad Transformers v5 breakages, but they describe different aspects of model loading/initialization and are too general to merge confidently." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "Different models and symptoms: SigLIP2 nonsense outputs versus Perceiver resize/interpolation failure." + "reason": "Different vision-model failures: SigLIP2 nonsensical outputs vs Perceiver non-default resolution classification failure." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Package/backend install issue and PaddleOCR-VL loading issue are unrelated." + "reason": "Packaging/backend availability issue vs PaddleOCR-VL loading failure in vLLM." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Tying lm_head weights and serialization inconsistency are related conceptually but not the same concrete bug." + "reason": "Tied lm_head weights causing bad generation and inconsistent serialization of lm_head.weight are related but not the same concrete bug." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Qwen2.5-VL rope-index scaling bug is unrelated to Kosmos2.5 OCR long-input index errors." + "reason": "Different models and failures: Qwen2.5-VL rope indexing vs Kosmos2.5 OCR index error." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen2.5-GGUF loading failure are different issues." + "reason": "Qwen3Moe FP8 config failure is unrelated to Qwen2.5-GGUF loading on Transformers v5." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "Model-specific can't-run report and meta-device loading regression are not the same bug." + "reason": "Whisper runtime failure and torch.device('meta') model loading regression are different problems." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Int4 memory regression and a model OOM report are not the same underlying issue." + "reason": "Int4 quantization memory regression is unrelated to Sam3Video OOM." }, { "left": "issue:43296", "right": "issue:45800", "accept": false, - "reason": "PaddleOCR-VL load failure and torch/transformers incompatibility are separate problems." + "reason": "PaddleOCR-VL loading in vLLM and torch 2.4.1/transformers 5.8 incompatibility are not the same bug." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL position-id scaling." + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope_index position-id bug are unrelated." }, { - "left": "issue:44479", - "right": "issue:45823", + "left": "issue:43611", + "right": "issue:43646", "accept": false, - "reason": "Video input regression in Qwen VL models and Gemma4 device-mismatch forward failure are different bugs." + "reason": "Both mention Transformers v5 load/init regressions, but the titles point to different concrete breakages and not one shared code-path." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Near-identical reports of the same regression: Transformers v5 fills non-persistent buffers with junk." } ] }, @@ -3571,9 +3551,7 @@ "issue:43526", "issue:43550", "issue:43572", - "issue:43577", "issue:43606", - "issue:43708", "issue:43720", "issue:43746", "issue:43749", @@ -3586,6 +3564,7 @@ "issue:44265", "issue:44291", "issue:44466", + "issue:44479", "issue:44488", "issue:44534", "issue:44610", @@ -3603,10 +3582,11 @@ "issue:45823" ], "soft_pairs": [ + "issue:44479|issue:45823", "issue:43881|issue:44488", - "issue:43606|issue:44265", - "issue:43262|issue:43746", "issue:44291|issue:44877", + "issue:43262|issue:43746", + "issue:43606|issue:44265", "issue:40990|issue:45584", "issue:34634|issue:35707", "issue:43873|issue:45701", @@ -3627,242 +3607,235 @@ "issue:43454|issue:43720", "issue:43881|issue:44991", "issue:41720|issue:43322", - "issue:43296|issue:44991", + "issue:43296|issue:44877", "issue:41720|issue:44977", "issue:40990|issue:45701", "issue:44008|issue:45823", "issue:45632|issue:45684", "issue:43404|issue:45357", "issue:44186|issue:45072", - "issue:42617|issue:42915", - "issue:43577|issue:43708" + "issue:42617|issue:42915" ], - "prepared_review_unit_hash": "58d7c7b0fcb38283598158b59ea64b779ac4fbcc2856ca922722b0923f2a2fa3", + "prepared_review_unit_hash": "c56c0f7a47573ef5e6c54dfa445deb998b25ed4f241d6eb3ae85b457dbfbaf7d", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 33, - "serialized_chars": 37822, - "estimated_input_tokens": 9456, - "estimated_eval_tokens": 19168 + "serialized_chars": 37164, + "estimated_input_tokens": 9291, + "estimated_eval_tokens": 18838 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely heterogeneous: the soft pairs share broad HF/transformers themes, but they do not look like the same underlying bug. I would not merge any of the proposed issue pairs, and there are no PRs here to consider.", - "confidence": 0.89, - "canonical_issue_reason": "issue:43873 is the closest representative because it is the broadest and most discussed report around offloading/quantization, which is the nearest common thread among the few somewhat-related items.", + "summary": "The set is highly heterogeneous. The proposed soft edges mostly share broad subsystem words like offloading, device mapping, serialization, or multimodal models, but they do not appear to describe the same underlying bug or change. I would reject all candidate merges.", + "confidence": 0.96, + "canonical_issue_reason": "If a representative issue must be chosen, issue 43873 is the broadest and most cluster-central because it captures the recurring offloading/quantization theme; however, the cluster is too diverse for a true canonical duplicate anchor.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best single issue to anchor the cluster: it is general, active, and sits in the same loading/offload problem space as a few neighbors, even though the overall set is not a true duplicate cluster.", + "best_issue_reason": "Issue 43873 is the best overall representative among these mostly unrelated reports because it is broad, open, and closest to the repeated offloading/device-placement subset, though it is not a strong duplicate canonical.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Different models and failure modes: video-input regression in Qwen VL family vs Gemma4 PLE device mismatch." + }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "Different models and likely different loading failures; same broad symptom of load failure is not enough." + "reason": "Unrelated loading failures for different models; one is glm-4v-9b, the other a tokenizer load issue for a BERT model." }, { - "left": "issue:43606", - "right": "issue:44265", + "left": "issue:44291", + "right": "issue:44877", "accept": false, - "reason": "CPU offload device mismatch vs torch.export/torch_compilable_check export failure are different code paths." + "reason": "Different root causes: unexpected `_is_hf_initialized` argument during init-empty-weights vs strict config rejection for granite_speech." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and PEFT checkpoint loading are unrelated bugs." + "reason": "Audio chat-template sampling-rate default bug is unrelated to GraniteSpeech PEFT local checkpoint loading." }, { - "left": "issue:44291", - "right": "issue:44877", + "left": "issue:43606", + "right": "issue:44265", "accept": false, - "reason": "Both are loading-related, but one is an init_empty_weights signature issue and the other is strict config validation." + "reason": "CPU offload device mismatch for Bark-small is unrelated to torch.export failing on torch_compilable_check." }, { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "Benchmark perplexity on one model vs Whisper empty-transcription generation bug on another model." + "reason": "High perplexity on gpt-oss is a modeling/eval issue, not the Whisper empty-transcription generation bug." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "Bark voice_preset handling and progressive generation with inputs_embeds/past_key_values are unrelated." + "reason": "BarkProcessor voice_preset and progressive generation with inputs_embeds/past_key_values are separate code paths." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "Offloading/quantization behavior and tokenizer changes are different problems." + "reason": "Offloading/quantization failure is unrelated to tokenizer behavior changing across transformer versions." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "Tokenizer padding/truncation crash is unrelated to export-time torch_compilable_check failures." + "reason": "Tokenizer padding/truncation crash is unrelated to torch.export and torch_compilable_check." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Both mention offload/quantization, but one is dequantized model loading with device_map and the other is general offloading behavior; not the same bug." + "reason": "Both mention offloading, but one is a specific dequantized-model load failure and the other is a broader quantization/offloading complaint; not the same bug." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "Progressive generation cache handling and Qwen3 device-map CUDA assert are different failure modes." + "reason": "Progressive generation with inputs_embeds is unrelated to Qwen3 auto device-map CUDA assertion." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Device mismatch during CPU offload and DAC latent/STE mismatch are unrelated." + "reason": "Bark CPU offload device mismatch and DAC.from_latents STE mismatch are different model-specific issues." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "Weight tying bug in AyaVision and torch.compile SDPA failure in Bamba are different issues." + "reason": "lm_head weight tying bug is unrelated to Bamba torch.compile SDPA failure." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Both involve device placement, but the models and root causes differ." + "reason": "Both involve device placement, but the errors and models differ materially: CUDA assert on Qwen3 vs Gemma4 PLE forward mismatch." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "Padding-mask-dependent encoder outputs and processor/model input-size mismatch are unrelated." + "reason": "MIMI padding-mask batching bug is unrelated to OmDet-Turbo processor/model input-size mismatch." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "LM-head tying bug and bfloat16 dtype mismatch are different code paths." + "reason": "Weight-tying/generation bug is unrelated to dtype mismatches in SwitchTransformers and TimmWrapperModel." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "Missing padding-mask support and Perceiver resolution handling are not the same bug." + "reason": "Different subsystems and causes: missing padding-mask support vs Perceiver non-default-resolution failure." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "Documentation example mismatch/quantization failure and multi-image IterableDataset finetuning failure are different." + "reason": "Documentation/example errors for SigLIP2 are unrelated to Qwen2.5-VL finetuning with multi-image iterable datasets." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx in config and text_config unwrapping in composite multimodal configs are different config bugs." + "reason": "Missing `pad_token_idx` in StableLmConfig is unrelated to `from_config` not unwrapping `text_config` for Qwen multimodal configs." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BEiT image label reduction and missing all_tied_weights_keys on Molmo are unrelated." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Molmo missing `all_tied_weights_keys`." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP CPU RAM efficient loading breakage and non-persistent buffer junk in v5 are different loader/state issues." + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are distinct v5 regressions." }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "Both touch serialization, but one is lm_head weight serialization and the other is Qwen3.5 visual encoder key handling." + "reason": "Both concern serialization, but one is tied lm_head.weight behavior and the other is incorrect visual encoder keys for Qwen3.5." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "Weight tying in AyaVision and unpacking packed weights during accelerate loading are different defects." + "reason": "lm_head tying bug is unrelated to BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "Model load failure for glm-4v-9b and tokenizer load failure for est-roberta are unrelated." + "reason": "Model loading failure for glm-4v-9b is unrelated to tokenizer loading failure for EMBEDDIA/est-roberta." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "CUDA assert on auto device mapping and segmentation fault while loading Llava Next are different failures." + "reason": "Different models and failures: Qwen3 CUDA assert vs Llava Next segmentation fault." }, { "left": "issue:43296", - "right": "issue:44991", + "right": "issue:44877", "accept": false, - "reason": "PaddleOCR-VL load failure in vLLM and tokenizer loading failure for a different model are not duplicates." + "reason": "PaddleOCR-VL vLLM load failure is unrelated to granite_speech strict-config loading rejection." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Same family name, but device-map CUDA assert and flash-attention generation regression are distinct bugs." - }, - { - "left": "issue:40990", - "right": "issue:45701", - "accept": false, - "reason": "High perplexity on a benchmark and tokenization version regression are too different to be the same issue." + "reason": "Both are Qwen-family reports, but one is an auto device-map CUDA assert and the other is a flash-attention generation regression." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma 3n variable collision and Gemma4 device mismatch are unrelated." + "reason": "Gemma 3n variable-name collision and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Trust-remote-code cache path collision and read-only permission propagation are different filesystem issues." + "reason": "Cache-path collision for trust_remote_code is unrelated to permission propagation in save_pretrained/register_for_auto_class." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Weight tying in Mistral3 and incorrect visual encoder keys in Qwen3.5 are different model-specific bugs." + "reason": "Mistral3 lm_head tying bug is unrelated to Qwen3.5 visual encoder key serialization regression." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "Tokenizer crash on NER/padding versus dtype mismatch in inference are unrelated." + "reason": "Tokenizer crash on NER/padding is unrelated to bfloat16 dtype mismatches in model inference." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "3d_parallel runtime failure and FineGrainedFP8Config failure on Qwen3Moe are different problems." - }, - { - "left": "issue:43577", - "right": "issue:43708", - "accept": false, - "reason": "Model/qformer dtype preservation and Trainer resume max_steps computation are unrelated." + "reason": "3d_parallel.py execution problem and Qwen3Moe FineGrainedFP8Config failure are different issues." } ] }, @@ -3886,8 +3859,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-397", "nodes": [ - "issue:33357", - "issue:35707", "issue:36010", "issue:36331", "issue:41720", @@ -3905,7 +3876,9 @@ "issue:43540", "issue:43550", "issue:43572", + "issue:43577", "issue:43606", + "issue:43708", "issue:43720", "issue:43782", "issue:43873", @@ -3936,11 +3909,12 @@ "issue:45812" ], "soft_pairs": [ + "issue:43577|issue:43708", "issue:43873|issue:44756", "issue:44387|issue:45704", "issue:44423|issue:45406", - "issue:36331|issue:42617", "issue:44877|issue:44991", + "issue:36331|issue:42617", "issue:43550|issue:43720", "issue:43450|issue:45701", "issue:36331|issue:42915", @@ -3948,12 +3922,12 @@ "issue:44186|issue:44610", "issue:44479|issue:44918", "issue:43550|issue:43606", - "issue:43316|issue:44913", "issue:41720|issue:42617", "issue:42915|issue:43317", + "issue:43316|issue:44913", "issue:43295|issue:45701", - "issue:45081|issue:45812", "issue:44671|issue:44743", + "issue:45081|issue:45812", "issue:43526|issue:43881", "issue:41720|issue:45161", "issue:43531|issue:43572", @@ -3964,231 +3938,223 @@ "issue:42915|issue:43322", "issue:43388|issue:45059", "issue:45406|issue:45464", - "issue:44492|issue:44509", "issue:43421|issue:45701", - "issue:33357|issue:35707", - "issue:43606|issue:43720" + "issue:44492|issue:44509" ], - "prepared_review_unit_hash": "8c4a37cb9b9e562348f77123807afec9d5b77ec84dde22739e56fb4b36e457dd", + "prepared_review_unit_hash": "0c8ff351c40c42985369f6ad679f03dab3a5f727c6d49cb8c9492ab9df2caf20", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37216, - "estimated_input_tokens": 9304, - "estimated_eval_tokens": 18864 + "soft_pair_count": 31, + "serialized_chars": 37022, + "estimated_input_tokens": 9256, + "estimated_eval_tokens": 18768 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: it mixes unrelated bugs across generation, quantization/offloading, multimodal processors, tokenizers, configs, and docs. The soft pairs are mostly broad topical similarities, not the same underlying defect, so no soft edges should be accepted.", - "confidence": 0.98, + "summary": "These issues are mostly unrelated Transformers regressions spread across trainers, tokenizers, multimodal processors, quantization, offloading, and config loading. The soft pairs only share broad subsystems or similar symptoms, not the same concrete underlying bug, so all should be rejected. No single canonical issue clearly represents the whole set.", + "confidence": 0.91, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Different bugs: one is model dtype initialization on Blip2 load, the other is Trainer checkpoint resume step \u0917\u0923ulation." + }, { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Both mention memory/offloading, but one is a quantization offload regression and the other is an mmap/OOM issue on Strix Halo; different failure modes and fixes." + "reason": "Both mention offloading/memory, but one is quantization offload behavior and the other is mmap-related OOM on Strix Halo; different code paths." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "Both concern memory usage under quantization, but one is CUDA reserved memory growth in Transformers 5.x and the other is T5 silently using apex RMSNorm; not the same bug." + "reason": "One is int4 quantization causing higher CUDA reserved memory; the other is a T5/apex RMSNorm memory leak. Not the same defect." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both crash in serving paths, but one is continuous-batching with multimodal models and the other is a Gemma4Processor attribute regression; different code paths." + "reason": "Both are serve/multimodal crashes, but they involve different models and different attribute errors in different processor paths." }, { - "left": "issue:36331", - "right": "issue:42617", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "Both are runtime errors, but one is Trainer compute_loss signature drift and the other is failure to run 3d_parallel.py; unrelated issues." + "reason": "Config strictness for granite_speech and tokenizer loading failure for est-roberta are unrelated failures." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:36331", + "right": "issue:42617", "accept": false, - "reason": "Both are loading/config-related, but one is strict config blocking granite_speech and the other is tokenizer loading for est-roberta; not duplicate bugs." + "reason": "Trainer compute_loss signature regression is unrelated to running 3d_parallel.py." }, { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "Both involve model loading/SDPA or accelerate, but one is a torch.compile+SDPA failure and the other is unpacked packed-weights handling in accelerate loading." + "reason": "Different failures: SDPA with torch.compile in Bamba versus packed-weight handling during accelerate loading for BitNet." }, { "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "Video processor shape regression vs tokenizer behavior changes across versions; different subsystems and symptoms." + "reason": "Video batch-shape bug is unrelated to a tokenization-version regression." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "Trainer compute_loss API mismatch is unrelated to Qwen3Moe FineGrainedFP8Config failure." + "reason": "CustomTrainer compute_loss kwargs issue is unrelated to Qwen3MoE + FineGrainedFP8Config loading." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both are vision processor size problems, but one is OmDet-Turbo processor/model size mismatch and the other is Perceiver interpolation not respecting non-default resolution." + "reason": "Both are image-size/preprocessing issues, but they affect different models and different mismatch logic." }, { "left": "issue:44186", "right": "issue:44610", "accept": false, - "reason": "Both affect image processing, but one is LayoutLMv2 tokenizer NER padding/truncation and the other is OmDet-Turbo input resolution mismatch." + "reason": "LayoutLMv2Tokenizer NER/padding crash is a tokenizer bug, not the OmDet-Turbo processor image-size bug." }, { "left": "issue:44479", "right": "issue:44918", "accept": false, - "reason": "Both mention Qwen3.5 and regressions, but one is video input handling and the other is unpacking input embeddings with TRL SFT trainer." + "reason": "A video-input regression in Qwen VL models is not the same as Qwen3.5 embedding unpacking with TRL SFT." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Both are hardware/runtime failures, but one is SDPA with torch.compile and the other is CPU offload device mismatch in bark-small." - }, - { - "left": "issue:43316", - "right": "issue:44913", - "accept": false, - "reason": "Both are config/API consistency issues, but one is Gemma3TextConfig divergence and the other is GPTNeoX rotary_pct not persisting on reload." + "reason": "Both are loading/execution failures, but one is SDPA/compile in Bamba and the other is CPU-offload device mismatch in bark-small." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert vs inability to run 3d_parallel.py; not the same bug or code path." + "reason": "Qwen3 auto device mapping cuda assert is unrelated to failing to run 3d_parallel.py." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "Both involve loading/quantization, but one is Qwen3Moe with FineGrainedFP8Config and the other is device_map auto loading a dequantized model on cpu/gpu offload." + "reason": "Qwen3MoE FP8 config failure and dequantized-model GPU+CPU offload failure are different loading/offload bugs." }, { - "left": "issue:43295", - "right": "issue:45701", + "left": "issue:43316", + "right": "issue:44913", "accept": false, - "reason": "Processor/tokenizer regression in v4.57.5 vs tokenization changes across versions; one is a specific API break, the other is a broader version-dependent tokenization report." + "reason": "API discrepancy in Gemma3TextConfig and GPTNeoX rotary_pct reload behavior are separate configuration issues." }, { - "left": "issue:45081", - "right": "issue:45812", + "left": "issue:43295", + "right": "issue:45701", "accept": false, - "reason": "Mistral regex patch crash in tokenizer internals vs Granite models producing wrong token IDs; unrelated tokenizer bugs." + "reason": "Processor/tokenizer API regression and generic tokenization-version change are not the same concrete bug." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked LM prediction regression and Qwen3.5 recurrent-state cache behavior are different model-specific issues." + "reason": "CamemBERT masked-LM prediction regression is unrelated to Qwen3.5 recurrent state reset with cache." + }, + { + "left": "issue:45081", + "right": "issue:45812", + "accept": false, + "reason": "Mistral regex patch crash in tokenizer loading is a different failure from Granite token-ID regression." }, { "left": "issue:43526", "right": "issue:43881", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug vs glm-4v-9b loading failure; different components and failure signatures." + "reason": "BEiT label reduction bug and glm-4v-9b loading failure do not share the same root cause." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Both mention MoE/model parallelism, but one is a CUDA assert with auto device mapping and the other is only-TP not working for GPT-OSS MoE; distinct problems." + "reason": "Both involve Qwen/GPT-OSS MoE models, but one is an A800 auto-device-map assert and the other is TP-only support not working." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Sliding_window behavior in Qwen3-MoE is unrelated to StableLmConfig missing pad_token_idx after 5.0 update." + "reason": "Sliding-window behavior in Qwen3-MoE and missing pad_token_idx in StableLmConfig are unrelated." }, { "left": "issue:43421", "right": "issue:45812", "accept": false, - "reason": "Runtime special-token/post-processor update issue vs Granite token IDs regression; both tokenizer-related but different bugs." + "reason": "Special-token post-processor update behavior and Granite token ID regression are different tokenizer issues." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Video processing ValueError in Qwen3OmniMoe vs Qwen3VL weight_only loading error; different models and failure points." + "reason": "Video-input ValueError in Qwen3OmniMoe is unrelated to Qwen3VL weight_only loading error." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "ImportError for GenerationMixin from transformers.generation is unrelated to 3d_parallel.py execution failure." + "reason": "GenerationMixin import error and 3d_parallel.py execution failure are distinct issues." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "LayoutLMv2 tokenizer padding/truncation crash vs MLukeTokenizer AttributeError on tasks; both tokenizer bugs but not the same defect." + "reason": "Both are tokenizer-related, but one is LayoutLMv2 NER/batching crash and the other is MLukeTokenizer task AttributeError." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure vs Llava Next segmentation fault; different models and code paths." + "reason": "Qwen3MoE FP8 loading failure and Llava Next segmentation fault are different model-loading bugs." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric gathering label truncation bug vs SAM3 text/bbox behavior; both input/label handling but unrelated." + "reason": "Metric-gathering label truncation and SAM3 text/bbox odd behavior are unrelated." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve vs Qwen3.5 streaming inference failure; different processor/server bugs." - }, - { - "left": "issue:44492", - "right": "issue:44509", - "accept": false, - "reason": "A typo in cache strategies is unrelated to removed pipeline task docs still being referenced." + "reason": "Both hit serve/chat multimodal paths, but they are different models with different failure modes and code paths." }, { "left": "issue:43421", "right": "issue:45701", "accept": false, - "reason": "TokenizersBackend post-processor update bug vs version-dependent tokenization change report; related area, but not the same issue." - }, - { - "left": "issue:33357", - "right": "issue:35707", - "accept": false, - "reason": "MacOS bus error with pretrained CLIP model vs generation with inputs_embeds/past_key_values; entirely different failures." + "reason": "Runtime special-token/post-processor updates are not the same as a version-induced tokenization change." }, { - "left": "issue:43606", - "right": "issue:43720", + "left": "issue:44492", + "right": "issue:44509", "accept": false, - "reason": "Both touch loading/offload, but one is CPU offload device mismatch and the other is accelerate loading with packed weights not unpacked; not duplicate." + "reason": "A typo in cache strategies is unrelated to stale docs mentioning removed pipeline tasks." } ] }, @@ -4213,6 +4179,7 @@ "cluster_id": "cluster-10000-397", "nodes": [ "issue:30064", + "issue:33357", "issue:34567", "issue:34634", "issue:35707", @@ -4221,17 +4188,16 @@ "issue:42886", "issue:43381", "issue:43421", - "issue:43525", "issue:43531", "issue:43550", "issue:43606", "issue:43653", + "issue:43720", "issue:43746", "issue:43782", "issue:43844", "issue:43873", "issue:43927", - "issue:43994", "issue:44062", "issue:44077", "issue:44186", @@ -4242,7 +4208,6 @@ "issue:44589", "issue:44610", "issue:44625", - "issue:44671", "issue:44779", "issue:44843", "issue:44898", @@ -4262,6 +4227,8 @@ "issue:45859" ], "soft_pairs": [ + "issue:33357|issue:35707", + "issue:43606|issue:43720", "issue:41720|issue:43746", "issue:43550|issue:43653", "issue:30064|issue:34634", @@ -4282,238 +4249,229 @@ "issue:45759|issue:45859", "issue:44987|issue:45042", "issue:43531|issue:44589", + "issue:44918|issue:44936", "issue:43421|issue:43653", "issue:44062|issue:44451", - "issue:44918|issue:44936", "issue:45588|issue:45759", - "issue:39692|issue:44610", "issue:43381|issue:43844", + "issue:39692|issue:44610", "issue:42886|issue:44843", - "issue:43421|issue:44186", - "issue:43994|issue:44671", - "issue:45704|issue:45859", - "issue:43525|issue:43531" + "issue:43421|issue:44186" ], - "prepared_review_unit_hash": "0195a73f2668597e1c869fe4d9a38e78dba9f3e4ab5fff8feee1f84eb3d2f46a", + "prepared_review_unit_hash": "9e976e45e04cad27c62deb18763837515b7bdc8cdf504b16455521de0df7246a", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37167, - "estimated_input_tokens": 9292, - "estimated_eval_tokens": 18840 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36329, + "estimated_input_tokens": 9083, + "estimated_eval_tokens": 18422 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of unrelated bug reports across different models and subsystems. I only accept the two pairs that appear to describe the same concrete tokenizer/processor bug; the rest are similar at a high level but not the same underlying issue.", - "confidence": 0.84, - "canonical_issue_reason": null, + "summary": "The cluster is mostly a loose set of unrelated bug reports that only share broad subsystem overlap (tokenizers, model loading, multimodal processors, offload, training). There is one clear duplicate pair: the Qwen2_5_VLProcessor.apply_chat_template batched-input crash.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44514 is the best canonical issue because it is the earlier report of the exact same batched-input crash in Qwen2_5_VLProcessor.apply_chat_template, with essentially identical reproduction conditions to issue:44545.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is the strongest standalone report: it describes a broad, user-visible tokenizer regression with a clear v4\u2192v5 break and impact across all Granite models.", + "best_issue_reason": "issue:44514 is the strongest representative issue in this cluster since it is a clean, specific bug report with a direct duplicate and a well-scoped failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "Different failures: MacOS bus error with a CLIP model vs progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Both are CI/offload-related, but they describe different code paths and failures (device mismatch vs packed-weight unpacking during loading)." + }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Different problems and code paths: Qwen3 auto device-mapping CUDA assert vs GraniteSpeech PEFT local checkpoint loading." + "reason": "Different models and symptoms: Qwen3 auto device map CUDA assert vs GraniteSpeech local checkpoint/PEFT loading." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "Unrelated: Bamba torch.compile/SDPA failure vs BigBirdTokenizer special-token registration bug." + "reason": "Bamba torch.compile/SDPA failure is unrelated to BigBirdTokenizer mask-token decode behavior." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Different processors and symptoms: void segmentation maps vs BarkProcessor voice_preset handling." + "reason": "Different processors and bugs: void segmentation map handling vs Bark voice_preset not working." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but on different model families and likely different root causes (Deepseek vs Granite)." + "reason": "Both concern tokenizer regressions, but they affect different model families and appear to stem from different bugs." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization/offloading issue vs a T5 apex RMSNorm memory leak; not the same bug." + "reason": "Offloading with quantization is not the same as T5 silently using apex.FusedRMSNorm and leaking memory." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Same processor and same crash path: Qwen2_5_VLProcessor.apply_chat_template on batched input with padding=False." + "reason": "Same underlying bug: identical Qwen2_5_VLProcessor.apply_chat_template crash on batched input when padding=False." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "Both involve labels/classification, but one is Qwen3.5 config propagation and the other is a broader single-label zero-loss bug." + "reason": "Different configuration bugs: Qwen3.5 num_labels propagation vs degenerate zero loss for single-label classification." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Different areas: Bark CPU-offload device mismatch vs tokenizer special-token registration." + "reason": "Device mismatch during CPU offload is unrelated to BigBirdTokenizer special-token registration and decode output." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Both are Qwen-related, but one is sliding_window behavior in MoE models and the other is config unwrapping for composite multimodal models." + "reason": "Qwen3-MoE sliding_window behavior is not the same as missing text_config unwrapping in composite Qwen 3.5/3.6 configs." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Both touch tokenizer internals, but one is runtime special-token post-processor updates and the other is save/load of custom token IDs causing IndexError." + "reason": "Runtime post-processor update for special tokens is a different tokenizer feature from DiaConfig losing custom token IDs on save/load." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "Different models and failures: BarkProcessor voice_preset vs Qwen3 auto device mapping CUDA assert." + "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping failures are unrelated." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "TrainerState token counters vs progressive generation with inputs_embeds/past_key_values are unrelated." + "reason": "TrainerState token counting and progressive generation with cache/embeds are unrelated training/inference bugs." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "Different Qwen3.5 issues: warning about tie_word_embeddings vs incorrect visual encoder keys on save_pretrained." + "reason": "A warning about tie_word_embeddings is not the same as incorrect visual-encoder keys being saved in Qwen3.5." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen VL, but the concrete bugs differ: weight_only from_pretrained error vs wrong video vision_position_ids." + "reason": "Weight-only loading error for Qwen3VL is unrelated to incorrect vision_position_ids in qwen2.5-vl video input." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Perceiver image size handling vs dtype mismatches in SwitchTransformers/TimmWrapperModel; not the same change." + "reason": "Different failures: Perceiver non-default resolution handling vs dtype mismatches in SwitchTransformers/TimmWrapperModel bfloat16 inference." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Separate tokenizer-loading regressions for different models; same broad theme, but not the same underlying bug." + "reason": "Both are tokenizer loading regressions, but they target different models and likely different causes." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "Unrelated: patchtsmixer post_init policy vs Qwen3.5 num_labels propagation." + "reason": "patchtsmixer post_init policy is unrelated to Qwen3.5 num_labels propagation." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Different Qwen issues: config unwrapping for composite multimodal models vs CPU-offload compatibility in rotary embedding." + "reason": "Missing text_config unwrapping is not the same as CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding.forward." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "Separate regressions: model loading failure for one repo vs PIL backend image processors incorrectly requiring torchvision." + "reason": "Model loading regression for physical-intelligence/fast is unrelated to PIL backend processors requiring torchvision." }, { "left": "issue:43531", "right": "issue:44589", "accept": false, - "reason": "No shared code path: Qwen3-MoE sliding_window vs a Float8 storage lookup error." + "reason": "Sliding-window issue for Qwen3-MoE is unrelated to missing Float8 storage type resolution." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "trl SFT input-embedding unpacking and trainer.evaluate() after trainer.train() are different trainer workflows." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Both are tokenizer-related, but one is runtime post-processor syncing and the other is a model-specific special-token registration bug." + "reason": "Special-tokens runtime update behavior is a different bug than BigBirdTokenizer mask-token registration." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "TypeError in AddedToken construction vs model loading failure for ScandiBERT; unrelated." - }, - { - "left": "issue:44918", - "right": "issue:44936", - "accept": false, - "reason": "Different trainer failures: SFT unpacking of Qwen3.5 embeddings vs evaluate() after train()." + "reason": "AddedToken constructor TypeError is unrelated to loading ScandiBERT." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Flash-attention s_aux handling vs Qwen config unwrapping; different layers of the stack and different failures." + "reason": "Flash attention sink-less-model crash is unrelated to Qwen 3.5/3.6 config unwrapping." }, { - "left": "issue:39692", - "right": "issue:44610", + "left": "issue:43381", + "right": "issue:43844", "accept": false, - "reason": "Both are vision/model-input mismatches, but for different models and different failure modes (docs/quantization vs resolution mismatch)." + "reason": "Gradient checkpointing in eval mode and exploding gradients under ZeRO-3 are different training issues." }, { - "left": "issue:43381", - "right": "issue:43844", + "left": "issue:39692", + "right": "issue:44610", "accept": false, - "reason": "Gradient checkpointing eval-mode restriction vs ZeRO-3 gradient explosion are unrelated training issues." + "reason": "Documentation example errors for SigLIP2 are unrelated to OmDet-Turbo processor output-size mismatch." }, { "left": "issue:42886", "right": "issue:44843", - "accept": true, - "reason": "Same offline-cache tokenizer failure: both point to HF_HUB_OFFLINE breaking AutoTokenizer/model_info access during loading." + "accept": false, + "reason": "Both are HF_HUB_OFFLINE-related, but one is cache loading for tokenizers and the other is an unconditional model_info call in mistral regex patching." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Different tokenizer problems: dynamic special-token updates vs LayoutLMv2 batched NER padding/truncation crash." - }, - { - "left": "issue:43994", - "right": "issue:44671", - "accept": false, - "reason": "Different models and symptoms: SigLIP2 nonsensical outputs vs CamemBERT masked LM regression." - }, - { - "left": "issue:45704", - "right": "issue:45859", - "accept": false, - "reason": "Apex RMSNorm memory leak in T5 vs Qwen rotary embedding CPU-offload incompatibility." - }, - { - "left": "issue:43525", - "right": "issue:43531", - "accept": false, - "reason": "Llama4Config missing pad_token_id is unrelated to Qwen3-MoE sliding_window behavior." + "reason": "Runtime special-token/post-processor updates and LayoutLMv2 NER padding/truncation crashes are different tokenizer bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative on the soft-edge pairs. The two accepted pairs are plausibly the same concrete bugfix candidate, while the rejected pairs are appropriately treated as separate issues despite superficial similarity." + "feedback": "The summary is grounded in the packet and stays conservative. The one accepted duplicate pair, issue:44514 vs issue:44545, is well supported by the identical titles and the metadata showing references; the rest of the soft-edge verdicts are appropriately marked false without overclaiming. Minor phrasing like \"essentially identical reproduction conditions\" is a bit stronger than the visible metadata alone, but not enough to undermine the overall assessment." }, "evaluator_used": true, "retried": false, @@ -4543,6 +4501,7 @@ "issue:43322", "issue:43335", "issue:43421", + "issue:43525", "issue:43531", "issue:43540", "issue:43550", @@ -4556,21 +4515,20 @@ "issue:43720", "issue:43746", "issue:43756", - "issue:43782", "issue:43818", "issue:43819", "issue:43873", + "issue:43994", "issue:44008", "issue:44062", "issue:44186", "issue:44220", "issue:44488", "issue:44661", + "issue:44671", "issue:44811", "issue:44843", - "issue:44877", "issue:44898", - "issue:44977", "issue:45072", "issue:45081", "issue:45183", @@ -4580,22 +4538,26 @@ "issue:45685", "issue:45704", "issue:45834", + "issue:45859", "issue:45907" ], "soft_pairs": [ + "issue:43994|issue:44671", + "issue:45704|issue:45859", + "issue:43525|issue:43531", "issue:45081|issue:45356", "issue:41720|issue:45704", "issue:45356|issue:45907", "issue:39692|issue:43873", "issue:43746|issue:43873", - "issue:43596|issue:44220", "issue:40444|issue:43746", + "issue:43596|issue:44220", "issue:43746|issue:45381", "issue:43540|issue:44008", "issue:43720|issue:43819", - "issue:44220|issue:44661", "issue:43618|issue:43697", "issue:45183|issue:45479", + "issue:44220|issue:44661", "issue:43653|issue:43720", "issue:30064|issue:36010", "issue:43335|issue:43756", @@ -4610,225 +4572,222 @@ "issue:43577|issue:43818", "issue:43421|issue:44811", "issue:43531|issue:43643", - "issue:43653|issue:43819", - "issue:43643|issue:44877", - "issue:43540|issue:45381", - "issue:43782|issue:44977" + "issue:43653|issue:43819" ], - "prepared_review_unit_hash": "31822a23be0d88c26735e2ec6895ed3b040fd206a673b530c25da6c4931aad2e", + "prepared_review_unit_hash": "de36c963ba6b79a9602c92128260190d83c0597bf27bad482689aca04e70663e", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36862, - "estimated_input_tokens": 9216, - "estimated_eval_tokens": 18688 + "serialized_chars": 37471, + "estimated_input_tokens": 9368, + "estimated_eval_tokens": 18992 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely heterogeneous: the soft-paired issues span unrelated bugs in tokenizers, multimodal processors, model loading, quantization/offloading, and framework regressions. None of the soft edges look like true duplicates, so I would keep them separate rather than collapsing the cluster.", - "confidence": 0.86, - "canonical_issue_reason": "No single issue clearly serves as a duplicate target for the whole cluster; the items are too diverse. If forced to pick an anchor, issue #43873 is the most triage-worthy representative because it has the strongest cross-references and a broadly impactful infrastructure theme (offloading + quantization), but it is not a duplicate hub for the rest of the set.", + "summary": "The set is highly heterogeneous: it spans unrelated bugs in tokenizers, multimodal processors, quantization/offload, model configs, and CI/documentation. None of the soft pairs look like the same underlying issue, so there is no duplicate cluster to collapse.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue: the items do not share one concrete code path or failure mode, so treating any one as the cluster representative would be misleading.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #43873 is the best single issue for manual follow-up because it is open, has the most inbound references, and concerns a broadly relevant runtime behavior regression. It is still only a representative anchor, not a merge target for the other issues.", + "best_issue_reason": "No clear best issue for the cluster as a whole, because the issues are unrelated rather than variants of one bug. If forced, a broad/high-activity item could be picked, but it would not be representative.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different subsystems: model/pipeline output correctness vs tokenizer-mapping/model addition logic." + }, + { + "left": "issue:45704", + "right": "issue:45859", + "accept": false, + "reason": "Unrelated model/runtime bugs: T5 Apex RMSNorm memory leak vs Qwen3 MoE CPU offload incompatibility." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Different failures: missing pad_token_id on Llama4Config vs sliding_window handling in Qwen3-MoE." + }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Both mention Mistral-related tokenizer behavior, but one is a crash in _patch_mistral_regex while the other is a broader Kimi-K2.5 regression about codec handling and warnings. Different failure modes and likely different fixes." + "reason": "Both involve tokenizer regex handling, but they are different concrete bugs with different triggers and symptoms." }, { "left": "issue:41720", "right": "issue:45704", "accept": false, - "reason": "Qwen3 CUDA assert on device mapping is unrelated to T5 silently using apex.FusedRMSNorm and leaking memory. Different model families and different code paths." + "reason": "Different model/code-path issues: Qwen3 device mapping assert vs T5 Apex memory leak." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Tokenizer regression with mistral regex/codec handling is unrelated to list_repo_files failing to forward revision in CI. Different subsystems entirely." + "reason": "Tokenizer regression vs repo file listing API forwarding; no shared underlying bug." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "SigLIP2 doc example errors are documentation/model-usage issues, while #43873 is about offloading with quantization. Not the same underlying bug." + "reason": "Documentation/model example errors are unrelated to quantization offloading behavior." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "GraniteSpeech PEFT local checkpoint loading and quantization/offloading behavior are different code paths and different symptoms." + "reason": "Local checkpoint/PEFT loading failure is a different path from quantization offload problems." }, { - "left": "issue:43596", - "right": "issue:44220", + "left": "issue:40444", + "right": "issue:43746", "accept": false, - "reason": "Deepspeed ZeRO3/BertModel index error is unrelated to _torch_extract_fbank_features. Different model stack and failure site." + "reason": "Multimodal dataset processing failure vs checkpoint loading with PEFT adapters." }, { - "left": "issue:40444", - "right": "issue:43746", + "left": "issue:43596", + "right": "issue:44220", "accept": false, - "reason": "Qwen2.5-VL iterable dataset multi-image finetuning failure is unrelated to GraniteSpeech PEFT checkpoint loading." + "reason": "Deepspeed/BertModel init failure and audio fbank feature extraction are separate bugs." }, { "left": "issue:43746", "right": "issue:45381", "accept": false, - "reason": "Local checkpoint/PEFT loading for GraniteSpeech is unrelated to Qwen2.5-VL video vision_position_ids behavior." + "reason": "GraniteSpeech loading issue is unrelated to Qwen2.5-VL video position-id errors." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Qwen3OmniMoe video processing ValueError and Gemma3n audio_mask AttributeError are distinct multimodal model bugs with different code paths." + "reason": "Different multimodal models and different failures: video input ValueError vs Gemma3n name collision." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "BitNet packed-weight loading under accelerate and DAC.from_latents missing STE are unrelated issues." - }, - { - "left": "issue:44220", - "right": "issue:44661", - "accept": false, - "reason": "Feature extraction failure and add-new-model-like failing inside TOKENIZER_MAPPING_NAMES are different components and different bug types." + "reason": "Packed-weight loading in BitNet is unrelated to DAC.from_latents / STE mismatch." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "CLIPOutput attentions assignment regression and RTDetrV2 output differences are unrelated model implementation bugs." + "reason": "Both are output regressions, but for different models and different concrete code paths." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "Invalid input validation/error messaging across models is not the same as degenerate zero-loss classification behavior." + "reason": "Input validation messaging bug vs classification loss degeneracy; not the same issue." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction failure and tokenizer-mapping model registration failure are unrelated." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug is unrelated to BitNet packed-weight unpacking during loading." + "reason": "BigBird special-token registration and BitNet weight unpacking are different bugs." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Void segmentation-map processing failure is unrelated to the GenerationMixin ImportError." + "reason": "Image segmentation-map processing and GenerationMixin import error are unrelated." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "SwitchTransformers sparse-layer creation at zero layers and Smollm3 RoPE-layer dropping are different configuration/model construction bugs." + "reason": "Different model architectures and failures; no shared concrete defect." }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "processor.tokenizer regression in v4.57.5 is unrelated to Kosmos2.5 OCR long-input indexing errors." + "reason": "Processor/tokenizer regression and Kosmos2.5 OCR index error are different multimodal bugs." }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Mistral-3 image preprocessor/model dtype mismatch is not the same bug as Gemma3n audio mask collision." + "reason": "Image preprocessor dtype mismatch is unrelated to Gemma3n audio-mask variable collision." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "Bamba torch.compile + SDPA failure and dtype mismatch issues in SwitchTransformers/TimmWrapperModel are separate regressions." + "reason": "torch.compile/SDPA failure in Bamba is different from bfloat16 dtype mismatches in other models." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Bamba torch.compile SDPA bug and DAC.from_latents vs forward mismatch are unrelated model behaviors." + "reason": "Different models and failure modes: compile+SDPA vs DAC latent conversion mismatch." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "AddedToken special-argument TypeError is not the same as loading cjvt/sleng-bert failing." + "reason": "AddedToken keyword collision and sleng-bert loading are not the same underlying bug." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "Llava Next segmentation fault and MoE MPS histogram Int error are distinct runtime failures." + "reason": "Llava Next loading segfault and MPS histogram-on-int error are unrelated." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "LayoutLMv2Tokenizer crashes on NER/padding are unrelated to Perceiver image classification with non-default resolution." + "reason": "Tokenizer NER/padding crash vs Perceiver image-resolution failure are different paths." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Offline hub/model_info Mistral regex patch issue and tokenizer crash with fix_mistral_regex are related only superficially; one is offline-mode behavior, the other is a tokenizer object attribute mismatch." + "reason": "Same helper function, but different defects: offline model_info call vs backend_tokenizer attribute crash." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "Blip2 dtype remaining float32 and Video-LLaVA missing temporal attention / shared weights are unrelated model-loading problems." + "reason": "Blip2 dtype loading bug and Video-LLaVA temporal-attention/weight-sharing issue are unrelated." }, { "left": "issue:43421", "right": "issue:44811", "accept": false, - "reason": "TokenizersBackend post-processor update at runtime and Whisper batch_decode skip_special_tokens behavior are different tokenizer APIs and different bugs." + "reason": "Runtime special-token post-processing update bug is different from Whisper batch_decode skip_special_tokens handling." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Qwen3-MoE sliding_window issue and trust_remote_code AutoConfig missing fields are unrelated." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to trust_remote_code config field loss." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug and DAC STE mismatch are unrelated." - }, - { - "left": "issue:43643", - "right": "issue:44877", - "accept": false, - "reason": "AutoConfig missing fields with trust_remote_code and strict config blocking granite_speech config are both config-related, but they are different failure modes and fixes." - }, - { - "left": "issue:43540", - "right": "issue:45381", - "accept": false, - "reason": "Qwen3OmniMoe video ValueError and Qwen2.5-VL vision_position_ids being wrong are distinct multimodal-video bugs." - }, - { - "left": "issue:43782", - "right": "issue:44977", - "accept": false, - "reason": "Qwen3VL weight_only load error and Qwen3.5 flash-attention generation failure are separate model/version/runtime issues." + "reason": "BigBird tokenizer special-token registration and DAC latent conversion are unrelated bugs." } ] }, @@ -4859,19 +4818,17 @@ "issue:39692", "issue:41720", "issue:42175", - "issue:43262", + "issue:43540", "issue:43577", "issue:43582", "issue:43606", + "issue:43643", "issue:43644", - "issue:43650", "issue:43720", "issue:43749", "issue:43782", - "issue:43819", "issue:43825", "issue:43827", - "issue:43906", "issue:43994", "issue:44060", "issue:44186", @@ -4887,34 +4844,39 @@ "issue:44671", "issue:44805", "issue:44821", - "issue:44857", + "issue:44877", "issue:44936", + "issue:44977", "issue:44991", "issue:45020", "issue:45072", "issue:45081", "issue:45084", "issue:45325", + "issue:45381", "issue:45406", "issue:45584", "issue:45701", "issue:45834" ], "soft_pairs": [ + "issue:43643|issue:44877", + "issue:43540|issue:45381", + "issue:43782|issue:44977", "issue:44448|issue:44671", "issue:45584|issue:45834", "issue:43644|issue:43749", - "issue:44485|issue:44509", "issue:43825|issue:43994", + "issue:44485|issue:44509", "issue:44561|issue:45020", "issue:39692|issue:45834", "issue:44821|issue:45084", - "issue:30064|issue:36331", "issue:41720|issue:44991", + "issue:30064|issue:36331", "issue:44186|issue:44442", "issue:43825|issue:43827", - "issue:44479|issue:45834", "issue:43577|issue:43582", + "issue:44479|issue:45834", "issue:44805|issue:44936", "issue:44671|issue:45701", "issue:43606|issue:45072", @@ -4925,214 +4887,214 @@ "issue:42175|issue:45081", "issue:45081|issue:45406", "issue:43749|issue:44291", - "issue:44060|issue:44368", - "issue:43650|issue:43906", - "issue:43582|issue:44857", - "issue:43262|issue:43819" + "issue:44060|issue:44368" ], - "prepared_review_unit_hash": "483b339caea6222e0cb330ffad9dc33c47bde6f67fc250714393ddf200e31701", + "prepared_review_unit_hash": "44075519b3a63e5f0ddd475f221da71d0e3016f8e8185d7a215a705b4ac1a3e9", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35676, - "estimated_input_tokens": 8919, - "estimated_eval_tokens": 18094 + "serialized_chars": 35677, + "estimated_input_tokens": 8920, + "estimated_eval_tokens": 18096 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: most pairs are superficially similar but point to different models, code paths, or even docs vs runtime behavior. I would reject all soft edges as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue: the cluster mixes unrelated regressions across model loading, tokenization, audio, vision, training, and documentation.", + "summary": "Mostly a heterogeneous set of Transformers bug reports. The only near-duplicate pair is the v5 pipeline-removal docs/error-message fallout; the rest are distinct model-, tokenizer-, loading-, or dtype-specific regressions. No PRs were provided.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45020 is the broadest and most representative umbrella report in the set: it covers recent-version regressions affecting `remote_code` models, has active discussion, and can anchor the cluster better than the narrower model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the broadest representative of the remote_code regression theme, but it is still too general to serve as a true duplicate canonical for this cluster.", + "best_issue_reason": "Issue 45020 is the best overall issue to keep as the canonical tracker because it is the most general and central report, with enough discussion/inbound references to serve as an umbrella for related breakages.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config loading, but one is missing fields from `trust_remote_code` config objects while the other is strict config loading for `granite_speech`; different failure modes." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Different models and different video-processing bugs: Qwen3OmniMoe ValueError vs Qwen2.5-VL wrong `vision_position_ids`." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated symptoms: Qwen3VL `weight_only=True` load error vs Qwen3.5 flash-attention generation failure." + }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Different models and failure modes: Pegasus output drift vs CamemBERT masked-LM prediction regression." + "reason": "Different models and outputs: Pegasus v4/v5 output drift vs CamemBERT MLM prediction regression." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Both are input-shape edge cases, but they affect different modalities and code paths (Whisper empty transcription vs Kosmos2.5 long OCR)." + "reason": "Whisper empty-transcription failure and Kosmos2.5 OCR index error are distinct code paths." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Both are loading-related, but one is non-persistent buffer initialization and the other is FSDP CPU RAM loading." + "reason": "Both touch loading, but one is non-persistent buffer initialization junk and the other is FSDP CPU RAM efficient loading; not the same bug." }, { - "left": "issue:44485", - "right": "issue:44509", + "left": "issue:43825", + "right": "issue:43994", "accept": false, - "reason": "One is a GLM-5 RoPE implementation discussion; the other is stale v5 docs about removed pipeline tasks." + "reason": "Both mention pipeline-related behavior, but one is a bad error message and the other is a SigLIP2 model/processor mismatch plus quantization failure." }, { - "left": "issue:43825", - "right": "issue:43994", + "left": "issue:44485", + "right": "issue:44509", "accept": false, - "reason": "One is an error-message/docs issue; the other is a SigLIP2 model/pipeline correctness bug." + "reason": "Different scope: GLM-5 RoPE implementation discussion vs stale docs mentioning removed pipeline tasks." }, { "left": "issue:44561", "right": "issue:45020", "accept": false, - "reason": "Related theme, but 45020 is a broad umbrella for many remote_code breakages; 44561 is a specific helper-removal regression." + "reason": "Related broad theme around `remote_code`, but 44561 is a specific `is_torch_fx_available` removal breakage and 45020 is a wider umbrella report." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 docs/example errors are unrelated to Kosmos2.5 OCR index failures." + "reason": "Different models and problems: SigLIP2 example errors vs Kosmos2.5 long-OCR index error." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "AutoImageProcessor URL loading and template compilation errors are different subsystems and bugs." + "reason": "AutoImageProcessor-from-URL loading and template compilation errors are unrelated." }, { - "left": "issue:30064", - "right": "issue:36331", + "left": "issue:41720", + "right": "issue:44991", "accept": false, - "reason": "Segmentation-map processing and CustomTrainer.compute_loss signature changes are unrelated." + "reason": "Different subsystems: Qwen3 device-map CUDA assert vs tokenizer loading failure for EMBEDDIA/est-roberta." }, { - "left": "issue:41720", - "right": "issue:44991", + "left": "issue:30064", + "right": "issue:36331", "accept": false, - "reason": "Qwen3 device-mapping CUDA assert and tokenizer loading for est-roberta are distinct issues." + "reason": "ImageProcessor void segmentation handling and CustomTrainer `compute_loss` signature mismatch are unrelated." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "LayoutLMv2 NER/padding crashes and FastSpeech2Conformer tokenizer loading are different tokenizer bugs." + "reason": "LayoutLMv2 NER/padding crash and FastSpeech2Conformer tokenizer loading failure are different tokenizer issues." }, { "left": "issue:43825", "right": "issue:43827", - "accept": false, - "reason": "Both relate to v5 pipeline removals, but one is runtime messaging and the other is stale documentation." + "accept": true, + "reason": "Same underlying v5 pipeline-removal fallout: one reports the bad runtime error message, the other the stale docs referencing removed summarization/translation pipeline tasks." }, { - "left": "issue:44479", - "right": "issue:45834", + "left": "issue:43577", + "right": "issue:43582", "accept": false, - "reason": "Video-input regression and long OCR index error are different model-specific failures." + "reason": "Blip2 dtype preservation bug and Apple Silicon `caching_allocator_warmup` TypeError are unrelated." }, { - "left": "issue:43577", - "right": "issue:43582", + "left": "issue:44479", + "right": "issue:45834", "accept": false, - "reason": "Blip2 dtype propagation and Apple Silicon allocator warmup TypeError are unrelated." + "reason": "Different regressions: Qwen2.5/3.5 VL video-input issue vs Kosmos2.5 OCR index error." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "Mask-shape IndexError and trainer evaluate-after-train failure are not the same bug." + "reason": "Mask-shape indexing error and trainer evaluate-after-train failure are separate bugs." }, { "left": "issue:44671", "right": "issue:45701", "accept": false, - "reason": "Masked-LM prediction changes are not the same as general tokenization changes, even if they may be related." + "reason": "CamemBERT masked-LM regression and vague tokenization change report are not the same concrete bug." }, { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Bark CPU-offload device mismatch and dtype mismatches in other models are different loading bugs." + "reason": "CPU offload device-mismatch for Bark and bfloat16 dtype mismatches for SwitchTransformers/TimmWrapper are different issues." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "BarkProcessor voice_preset failure and GenerationMixin import failure are unrelated." + "reason": "BarkProcessor `voice_preset` failure and `GenerationMixin` import error are unrelated." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "OmDet-Turbo image-size mismatch and dtype mismatch regressions are different defects." + "reason": "Processor output size mismatch for OmDet-Turbo vs bfloat16 dtype mismatch in unrelated models." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Qwen3VL weight_only loading error and Qwen2.5-VL rope-index scaling are separate VL issues." + "reason": "Different Qwen VL regressions: `weight_only=True` load error vs still-image temporal position-id scaling bug." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "BitNet packed-weight unpacking and dtype mismatch problems do not share the same code path." + "reason": "BitNet packed-weight unpacking bug and bfloat16 dtype mismatch are different model-loading paths." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Missing tensorflow backend in pip extras is unrelated to the Mistral tokenizer patch crash." + "reason": "TensorFlow backend packaging and Mistral regex patch crash are unrelated." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Both are attribute errors, but they hit different objects and loading paths (tokenizer patch vs Gemma4Processor in serve)." + "reason": "Different tokenizer/processor bugs: Mistral regex patch vs Gemma4Processor missing `_tokenizer`." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "Both concern model loading, but FSDP CPU RAM loading and init_empty_weights `_is_hf_initialized` are different failures." + "reason": "FSDP CPU RAM efficient loading and `init_empty_weights` unexpected `_is_hf_initialized` argument are separate loading regressions." }, { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "Wrong tied-weights warning and the LoRA tie_word_embeddings warning are distinct issues." - }, - { - "left": "issue:43650", - "right": "issue:43906", - "accept": false, - "reason": "A vague 'ADD THE DATA' issue is not the same as a specific reproduction of another bug." - }, - { - "left": "issue:43582", - "right": "issue:44857", - "accept": false, - "reason": "Apple Silicon allocator TypeError and CUDA AMP image-loss crash are unrelated." - }, - { - "left": "issue:43262", - "right": "issue:43819", - "accept": false, - "reason": "Audio chat-template sample-rate defaulting and DAC latent/STE mismatch are different audio bugs." + "reason": "Different Qwen3-Next warning behaviors: incorrect tied-weights mapping vs noisy `tie_word_embeddings=False` warning." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The canonical choice is grounded in the packet: issue 45020 is the broadest remote_code umbrella and has enough activity/inbound references to justify centering the cluster. The summary stays conservative overall, and the lone accepted soft pair (43825/43827) is clearly tied to the same v5 pipeline-removal fallout. No PR-pair mergeability concerns apply because no PRs were provided." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5163,26 +5125,27 @@ "issue:43531", "issue:43550", "issue:43577", + "issue:43582", "issue:43638", "issue:43645", + "issue:43650", "issue:43653", "issue:43696", "issue:43697", "issue:43720", "issue:43742", "issue:43746", - "issue:43749", "issue:43819", "issue:43854", + "issue:43906", "issue:43927", "issue:44479", "issue:44554", "issue:44568", "issue:44589", "issue:44610", - "issue:44756", "issue:44811", - "issue:44849", + "issue:44857", "issue:44912", "issue:44936", "issue:44977", @@ -5195,17 +5158,19 @@ "issue:45488", "issue:45538", "issue:45600", - "issue:45701", "issue:45803" ], "soft_pairs": [ + "issue:43650|issue:43906", + "issue:43582|issue:44857", + "issue:43262|issue:43819", "issue:43526|issue:43697", "issue:43526|issue:44554", "issue:45459|issue:45803", "issue:34634|issue:36331", "issue:43653|issue:44912", - "issue:45084|issue:45538", "issue:43329|issue:45600", + "issue:45084|issue:45538", "issue:43742|issue:45020", "issue:43638|issue:44936", "issue:43257|issue:43329", @@ -5225,219 +5190,216 @@ "issue:43262|issue:44811", "issue:43329|issue:44479", "issue:43531|issue:45042", - "issue:44568|issue:45356", - "issue:43749|issue:44756", - "issue:36010|issue:36331", - "issue:44849|issue:45701" + "issue:44568|issue:45356" ], - "prepared_review_unit_hash": "73b7fb12a06fef296575c00484cdd7792bd096b71b2cc8b86e7442d2f2864c29", + "prepared_review_unit_hash": "023f8ab5e89266279ca36f6eb6c1ae6d291f2274ce140feaed1e0e4c69b04363", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36882, - "estimated_input_tokens": 9221, - "estimated_eval_tokens": 18698 + "serialized_chars": 36932, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very noisy: most items are distinct model-, tokenizer-, processor-, or training-specific regressions rather than true duplicates. I would not merge any soft pairs. The best triage anchor is the broad open issue about recent version regressions affecting remote-code models.", - "confidence": 0.84, - "canonical_issue_reason": "Issue 45020 is the broadest open umbrella candidate here and best matches the overall theme of version-induced loading/runtime regressions, even though the rest of the cluster is heterogeneous.", + "summary": "The candidate pairs are mostly superficial similarity matches rather than true duplicates. They span unrelated bugs across tokenizers, processors, training, quantization, and model-loading paths, so no soft edge is safe to merge.", + "confidence": 0.94, + "canonical_issue_reason": "issue:45020 is the broadest open umbrella for recent Transformers v5 model-loading regressions, but the cluster is too heterogeneous for a true duplicate canonical.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the most suitable representative for triage because it is recent, open, and broad enough to anchor adjacent version-regression reports without incorrectly collapsing unrelated bugs.", + "best_issue_reason": "issue:45020 is the best global anchor because it is broad and current; the other issues are narrower, model-specific failures that do not form one coherent duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "Completely unrelated: a vague data request vs a reproduction of a separate transformers bug." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Different subsystems and failures: Apple Silicon allocator TypeError vs CUDA/AMP loss crash." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate default bug is unrelated to DAC latent decoding/STE behavior." + }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "Different models and failure modes: BeitImageProcessorFast label reduction vs RTDetrV2 output drift." + "reason": "Different models and code paths: BeitImageProcessorFast label reduction vs RTDetrV2 output regression." }, { "left": "issue:43526", "right": "issue:44554", "accept": false, - "reason": "Unrelated subsystems: image processor label handling vs MPS attention correctness." + "reason": "No overlap beyond being correctness bugs; BEiT label reduction and MPS attention issues are unrelated." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both are error-handling bugs, but they affect different code paths and are not the same underlying defect." + "reason": "Both involve exception handling, but in different tokenizer utilities and with different root causes/code paths." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "BarkProcessor voice preset vs Trainer compute_loss signature mismatch are unrelated issues." + "reason": "Voice preset behavior in BarkProcessor is unrelated to CustomTrainer.compute_loss signature breakage." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "BigBirdTokenizer special-token decode bug is unrelated to MXFP4 quantization loading." + "reason": "BigBirdTokenizer special-token registration bug is unrelated to MXFP4 quantization fallback for gpt-oss." }, { - "left": "issue:45084", - "right": "issue:45538", + "left": "issue:43329", + "right": "issue:45600", "accept": false, - "reason": "Different tokenization/compile paths: template-node compilation error vs CLIPTokenizer max length constant." + "reason": "Video multimodal token counting bug and auto_mappings CI break are unrelated issues." }, { - "left": "issue:43329", - "right": "issue:45600", + "left": "issue:45084", + "right": "issue:45538", "accept": false, - "reason": "Video token counting bug is unrelated to stale auto_mappings references." + "reason": "Compiler template-node error is unrelated to CLIPTokenizer's model_max_length constant." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "Both involve loading, but one is a MobileLLM key error and the other is a broad remote_code regression; not the same bug." + "reason": "A single model loading KeyError is not the same as the broad remote_code regression reported in the other issue." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "Zero3 index error with non-pretrained Bert is unrelated to evaluate-after-train behavior." + "reason": "Deepspeed ZeRO3/empty-parameter index error is unrelated to evaluate() failing after train()." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Qwen3 MoE weight conversion and multimodal video token counting are different problems." + "reason": "Qwen3-MoE weight conversion/loading bug is unrelated to undefined variables in multimodal video token counting." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Both are from_pretrained/tokenizer loading complaints, but the file formats and failure modes differ." + "reason": "Both concern loading, but one is gguf_file support and the other is a tokenizer regression for a specific model." }, { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "GraniteSpeech PEFT checkpoint loading is unrelated to Whisper batch_decode special-token handling." + "reason": "PEFT local checkpoint loading and Whisper batch_decode skip_special_tokens are distinct failures." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Weight conversion bug vs plain CUDA OOM are not duplicates." + "reason": "Qwen3-MoE loading bug is unrelated to a GPT-oss CUDA out-of-memory report." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Float8 storage deserialization and PIL-backend torchvision dependency are unrelated." + "reason": "Float8 storage lookup error and PIL backend torchvision dependency regression are unrelated." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "Different models and symptoms: GLM-4.7-Flash load failure vs Qwen3.5 flash-attn generation bug." + "reason": "GLM-4.7-Flash unit-test loading issue is not the same as Qwen3.5 flash-attention generation failure." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Both are vision-model issues, but one is zero probabilities and the other is processor/model input-size mismatch." + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo processor/model size mismatch." }, { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "SigLIP2 docs example errors are unrelated to MXFP4 quantization fallback." + "reason": "SigLIP2 documentation errors and gpt-oss quantization fallback are different problems." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "DAC latent reconstruction mismatch and DiaConfig save/load token ID loss are separate bugs." + "reason": "DAC latent/forward mismatch and DiaConfig token-ID persistence are unrelated." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "GenerationMixin import failure and Jupyter custom-model initialization breakage are distinct API regressions." + "reason": "GenerationMixin import path break and custom-model notebook initialization are different v5 issues." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "AyaVision weight tying and Blip2 dtype propagation are different model-specific bugs." + "reason": "AyaVision lm_head tying bug and BLIP2 dtype propagation bug affect different models and code paths." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "torch.compile/SDPA failure for Bamba is unrelated to MXFP4 quantization loading." + "reason": "torch.compile + SDPA failure in Bamba is unrelated to MXFP4 quantization fallback." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "BitNet packed-weight loading and OmDet-Turbo image size mismatch are unrelated." + "reason": "Packed-weight unpacking during accelerate loading is unrelated to an OmDet-Turbo input-size mismatch." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Tokenizer special-token addition bug and tokenizer.json pre-tokenizer override are different root causes." + "reason": "Both are tokenizer regressions, but they hit different tokenizers and different behaviors; not the same bug." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting is unrelated to Whisper batch_decode behavior." + "reason": "Audio sampling-rate default in apply_chat_template and Whisper batch_decode skip_special_tokens are unrelated." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "Both touch video, but one is undefined-variable token counting and the other is a broader v5.3.0 video regression." + "reason": "Undefined video-token helper variables and a broader v5.3.0 video-input regression are not the same concrete defect." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and PIL torchvision dependency are unrelated." + "reason": "Qwen3-MoE sliding window behavior and PIL backend torchvision dependency are unrelated." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer regressions affecting different models and code paths." - }, - { - "left": "issue:43749", - "right": "issue:44756", - "accept": false, - "reason": "FSDP RAM-efficient loading and Strix Halo mmap OOM are different memory-related issues." - }, - { - "left": "issue:36010", - "right": "issue:36331", - "accept": false, - "reason": "ImportError for GenerationMixin and CustomTrainer.compute_loss kwarg mismatch are unrelated." - }, - { - "left": "issue:44849", - "right": "issue:45701", - "accept": false, - "reason": "Qwen3.5 hidden-states output bug is not the same as a generic tokenization change regression." + "reason": "Different tokenizer regressions: special-token addition vs codec/regex warning behavior in Kimi-K2.5." } ] }, @@ -5463,7 +5425,8 @@ "nodes": [ "issue:33357", "issue:34567", - "issue:38175", + "issue:36010", + "issue:36331", "issue:40990", "issue:41720", "issue:42175", @@ -5480,19 +5443,18 @@ "issue:43643", "issue:43697", "issue:43742", + "issue:43749", "issue:43756", "issue:43824", - "issue:43906", "issue:43927", "issue:43931", "issue:44062", "issue:44291", "issue:44479", - "issue:44488", "issue:44560", - "issue:44737", + "issue:44756", "issue:44821", - "issue:44898", + "issue:44849", "issue:44912", "issue:44938", "issue:44987", @@ -5504,24 +5466,27 @@ "issue:45381", "issue:45405", "issue:45538", - "issue:45588", + "issue:45701", "issue:45759", "issue:45859", "issue:45907" ], "soft_pairs": [ + "issue:43749|issue:44756", + "issue:36010|issue:36331", + "issue:44849|issue:45701", "issue:43643|issue:45759", "issue:43257|issue:43931", - "issue:34567|issue:40990", "issue:43329|issue:43450", + "issue:34567|issue:40990", "issue:43606|issue:45859", "issue:43257|issue:43756", "issue:43697|issue:45357", "issue:43452|issue:43582", "issue:44291|issue:45042", - "issue:44987|issue:45538", "issue:44987|issue:45907", "issue:42915|issue:43742", + "issue:44987|issue:45538", "issue:33357|issue:34567", "issue:44479|issue:45381", "issue:41720|issue:43540", @@ -5530,232 +5495,225 @@ "issue:42915|issue:45538", "issue:43577|issue:45072", "issue:43317|issue:45538", - "issue:42175|issue:43824", "issue:43540|issue:44479", + "issue:42175|issue:43824", "issue:43540|issue:44560", "issue:44062|issue:44991", "issue:42175|issue:44938", - "issue:45137|issue:45405", - "issue:38175|issue:44898", - "issue:42175|issue:44488", - "issue:44737|issue:45588", - "issue:43906|issue:45381" + "issue:45137|issue:45405" ], - "prepared_review_unit_hash": "483b0834708406e2f191ec3001c69e366fc1a92a34677b7831ffd90f880967a8", + "prepared_review_unit_hash": "20d828a995cf07e15b099110ed621c1d311899900d2fdca9910b832d2b9ab313", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36228, - "estimated_input_tokens": 9057, - "estimated_eval_tokens": 18370 + "soft_pair_count": 29, + "serialized_chars": 35776, + "estimated_input_tokens": 8944, + "estimated_eval_tokens": 18144 }, "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 issues. The only noticeable dense subcluster is around Qwen/Qwen-VL video-input regressions, but even there the items are similar rather than clear duplicates.", - "confidence": 0.43, - "canonical_issue_reason": "Issue 44479 is the broadest and most representative item in the main coherent subcluster: it explicitly covers a multi-model Qwen video-input regression in v5.3.0.", + "summary": "Mostly unrelated issue reports clustered by superficial similarity in model loading, offload, and multimodal/video wording. I only see one likely duplicate pair around the Qwen v5.3.0 video-input regression; the rest look like distinct bugs and should not be merged.", + "confidence": 0.94, + "canonical_issue_reason": "Use issue:44479 as the canonical issue: it is the broadest and clearest description of the Qwen v5.3.0 video-input regression, and issue:45381 reads like a specific manifestation of the same problem.", "canonical_pr_reason": null, - "best_issue_reason": "44479 best represents the largest coherent theme in the set (Qwen video regression reports), whereas the other issues are narrower, model-specific, or on unrelated code paths.", + "best_issue_reason": "issue:44479 is the best issue representative because it captures the shared regression at the right level of abstraction while naming the affected model families and version.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Both are loading/memory-related, but one is FSDP CPU RAM efficient loading and the other is mmap OOM on Strix Halo; different mechanisms and fixes." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "An ImportError for GenerationMixin is unrelated to a Trainer.compute_loss signature break." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "One is a Qwen3.5 output_hidden_states bug; the other is a tokenizer/version-dependent tokenization change. Different layers of the stack." + }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Different loading bugs: missing fields with trust_remote_code vs unwrapping text_config for composite multimodal configs." + "reason": "Both concern config/model construction, but missing fields from trust_remote_code and not unwrapping text_config are distinct code paths." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Different models and failures: Qwen3 MoE weight conversion vs Qwen3-VL weight shape mismatch." + "reason": "Different models and failures: Qwen3 MOE weight conversion with accelerate/deepspeed versus Qwen3-VL weight shape mismatch." }, { - "left": "issue:34567", - "right": "issue:40990", + "left": "issue:43329", + "right": "issue:43450", "accept": false, - "reason": "Unrelated areas: Trainer token-count metric not updating vs high perplexity on a causal LM." + "reason": "Both mention video, but one is undefined helpers in multimodal token counting and the other is a batched video processor shape bug." }, { - "left": "issue:43329", - "right": "issue:43450", + "left": "issue:34567", + "right": "issue:40990", "accept": false, - "reason": "Both are video-related, but one is an undefined helper in token counting and the other is a batched processor shape bug." + "reason": "TrainerState token counting and high perplexity on gpt-oss are unrelated problems." }, { "left": "issue:43606", "right": "issue:45859", "accept": false, - "reason": "Both mention offload, but they concern different code paths and models: generic device mismatch vs Qwen3_5Moe rotary embedding CPU-offload behavior." + "reason": "Both involve CPU offload, but they affect different models and different failure points; not the same concrete bug." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Different model families and bugs: Qwen3 MoE conversion vs SmolLM3 RoPE-layer dropping." + "reason": "Qwen3 MOE weight conversion and Smollm3 RoPE layer selection are unrelated." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "Different subsystems: RTDetrV2 output regression vs Qwen3.5 save_pretrained key mismatch." + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained key handling are separate issues." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "Unrelated failure modes: gguf_file loading vs Apple Silicon allocator warmup TypeError." + "reason": "gguf_file loading failures and an Apple Silicon caching_allocator_warmup TypeError do not share a code path." }, { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "Different bugs: init_empty_weights argument handling vs PIL image processors incorrectly requiring torchvision." - }, - { - "left": "issue:44987", - "right": "issue:45538", - "accept": false, - "reason": "No overlap beyond general loading problems; one is a model load failure, the other is tokenizer max-length handling." + "reason": "Unexpected _is_hf_initialized argument and PIL backend requiring torchvision are unrelated." }, { "left": "issue:44987", "right": "issue:45907", "accept": false, - "reason": "Unrelated: model loading for physical-intelligence/fast vs list_repo_files not forwarding revision." + "reason": "Loading physical-intelligence/fast on transformers>=5.1.0 and forwarding revision in list_repo_files are unrelated." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "Different models and code paths: Qwen3Moe FP8 config vs MobileLLM key error on load." + "reason": "Qwen3Moe fine-grained FP8 loading failure and MobileLLM-125M key error are different loading bugs." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Model loading regression and CLIPTokenizer max length handling are unrelated." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "Completely unrelated: MacOS bus error with CLIP vs TrainerState token counting." + "reason": "MacOS bus error with CLIP and TrainerState token counting are not the same bug." }, { "left": "issue:44479", "right": "issue:45381", - "accept": false, - "reason": "Close theme, but not enough to merge: broad Qwen video regression vs a specific vision_position_ids bug in qwen2.5-vl." + "accept": true, + "reason": "Both describe the same v5.3.0 Qwen2.5-VL video-input regression; 45381 looks like a specific symptom of the broader regression in 44479." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Different concrete problems: auto device mapping cuda assert vs Qwen3OmniMoe video processing ValueError." + "reason": "Qwen3 auto device-mapping cudaErrorAssert and Qwen3OmniMoe video processing ValueError are different failures." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "Unrelated: DiaConfig losing custom token IDs vs MXFP4 quantization fallback." + "reason": "DiaConfig save/load token ID loss and MXFP4 quantization fallback are unrelated." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Different topics: Qwen3Moe FP8 configuration vs AutoImageProcessor URL loading." + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor URL loading are different problems." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "Different topics: Qwen3Moe FP8 configuration vs CLIPTokenizer max_length behavior." + "reason": "Qwen3Moe FP8 loading and CLIPTokenizer model_max_length are unrelated." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both are dtype-related, but they affect different model families and likely different implementation paths." + "reason": "BLIP2 dtype mismatch and SwitchTransformers/TimmWrapperModel bf16 mismatch are separate model-specific dtype bugs." }, { "left": "issue:43317", "right": "issue:45538", "accept": false, - "reason": "Unrelated: device_map/offload loading failure vs CLIPTokenizer max_length." + "reason": "device_map auto offload/dequantized model loading is unrelated to CLIPTokenizer max length." }, { - "left": "issue:42175", - "right": "issue:43824", + "left": "issue:43540", + "right": "issue:44479", "accept": false, - "reason": "Unrelated: missing TensorFlow backend in torch install vs missing Qwen2_5_VL export." + "reason": "Different video bugs in different model families; Qwen3OmniMoe ValueError is not the same as the broader Qwen2.5-VL/Qwen3.5 video regression." }, { - "left": "issue:43540", - "right": "issue:44479", + "left": "issue:42175", + "right": "issue:43824", "accept": false, - "reason": "Same broad area, but the titles point to different reports and symptoms; not enough evidence that they are the same underlying bug." + "reason": "Missing TensorFlow backend in a pip extra is unrelated to a missing Qwen2_5_VLForConditionalGeneration import." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both are video-input issues, but they target different models and errors (ValueError vs StopIteration)." + "reason": "Qwen3OmniMoe video processing ValueError and Qwen3-vl-embedding StopIteration are distinct video-input failures." }, { "left": "issue:44062", "right": "issue:44991", "accept": false, - "reason": "Tokenizer constructor error vs tokenizer loading failure for a specific model; not the same bug." + "reason": "AddedToken special-argument duplication and tokenizer loading for EMBEDDIA/est-roberta are different tokenizer bugs." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "Unrelated: TensorFlow backend packaging vs Python 3.14 import/load failure." + "reason": "Missing TensorFlow in an install extra and Python 3.14 import failure are unrelated environment/package issues." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "Completely unrelated: DeepSpeed ZeRO3 deque error vs PEFT version bump." - }, - { - "left": "issue:38175", - "right": "issue:44898", - "accept": false, - "reason": "Different model-specific inference issues with no shared concrete code path." - }, - { - "left": "issue:42175", - "right": "issue:44488", - "accept": false, - "reason": "Unrelated: TensorFlow backend packaging vs a specific model loading failure." - }, - { - "left": "issue:44737", - "right": "issue:45588", - "accept": false, - "reason": "Different code paths: XLNet positional encoding device placement vs flash_attention s_aux handling." - }, - { - "left": "issue:43906", - "right": "issue:45381", - "accept": false, - "reason": "One is a repro of an earlier issue, the other is a specific qwen2.5-vl video-position bug; not enough to call them duplicates." + "reason": "DeepSpeed ZeRO3 deque underflow and MIN_PEFT_VERSION bump are unrelated." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays conservative. The only accepted pair, issue:44479 vs issue:45381, is plausibly the same Qwen 5.3.0 video-input regression at different levels of specificity. The other soft pairs are rejected with reasons that match the issue titles and do not overclaim shared root causes." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5772,20 +5730,19 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-397", "nodes": [ - "issue:24643", "issue:28282", "issue:29127", "issue:34567", "issue:34689", "issue:36331", - "issue:41762", + "issue:38175", + "issue:42175", "issue:42915", "issue:43452", "issue:43493", "issue:43526", "issue:43531", "issue:43540", - "issue:43577", "issue:43582", "issue:43643", "issue:43688", @@ -5795,19 +5752,19 @@ "issue:43749", "issue:43756", "issue:43782", + "issue:43906", "issue:43994", "issue:44062", "issue:44291", "issue:44484", "issue:44488", "issue:44514", - "issue:44561", "issue:44589", "issue:44704", + "issue:44737", "issue:44792", "issue:44821", "issue:44898", - "issue:44936", "issue:44964", "issue:45042", "issue:45081", @@ -5817,18 +5774,23 @@ "issue:45381", "issue:45405", "issue:45479", + "issue:45588", "issue:45600", "issue:45823" ], "soft_pairs": [ + "issue:38175|issue:44898", + "issue:42175|issue:44488", + "issue:44737|issue:45588", "issue:28282|issue:34689", + "issue:43906|issue:45381", "issue:42915|issue:43749", "issue:43742|issue:44291", "issue:43526|issue:45479", - "issue:43756|issue:45325", "issue:28282|issue:29127", - "issue:43540|issue:45823", "issue:43582|issue:43782", + "issue:43540|issue:45823", + "issue:43756|issue:45325", "issue:43493|issue:43994", "issue:34567|issue:36331", "issue:44291|issue:44821", @@ -5844,208 +5806,228 @@ "issue:43688|issue:43749", "issue:44484|issue:44704", "issue:44514|issue:45381", - "issue:43720|issue:44898", - "issue:41762|issue:44936", - "issue:43643|issue:44561", - "issue:43452|issue:43577", - "issue:24643|issue:29127" + "issue:43720|issue:44898" ], - "prepared_review_unit_hash": "513d958d18f627a8f7515326efaa07f9cdf79a0b5d9224e75afb22519357e500", + "prepared_review_unit_hash": "8103e143b8f9d76fb75d711ade97aeafb21ec3aa166b27715116ce68e151b893", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35769, - "estimated_input_tokens": 8943, - "estimated_eval_tokens": 18142 + "serialized_chars": 35777, + "estimated_input_tokens": 8945, + "estimated_eval_tokens": 18146 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a single duplicate family; it spans many unrelated Transformers bugs across different models and subsystems. None of the soft pairs look like the same underlying fix.", - "confidence": 0.93, - "canonical_issue_reason": "No clear canonical issue: the items cover distinct failure modes (model loading, training APIs, processors, multimodal paths, config/serialization regressions) rather than one duplicated bug.", + "summary": "These items are largely unrelated closed Transformers issues rather than a true duplicate cluster. None of the soft pairs look like the same concrete bug/change with enough confidence to merge.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43749 has the strongest signal in this set (highest discussion/inbound references) and is a concrete, broadly impactful failure report, so it is the best anchor issue among otherwise unrelated tickets.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 is the best anchor only as a broad, high-signal loading-regression report with discussion and inbound references, but it does not subsume the rest of the cluster.", + "best_issue_reason": "issue:43749 is the most actionable and representative issue here: it describes a specific broken loading path with substantial discussion and external references.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different models and failure modes: zero probabilities in SigLIP2 vs resolution/interpolation failure in Perceiver." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Backend packaging/dependency issue vs model loading failure for a specific checkpoint; not the same bug." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "Unrelated code paths: XLNet CPU tensor creation vs flash attention auxiliary-sink crash." + }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "Different problems: missing PyTorch import vs a specific Llama 3.2 Vision loading regression." + "reason": "Generic PyTorch import error vs a regression loading Llama 3.2 Vision; different root causes." + }, + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "No clear evidence they target the same bug; one is a reproduction of another issue, the other is a Qwen2.5-VL video position-id problem." }, { "left": "issue:42915", "right": "issue:43749", "accept": false, - "reason": "Different subsystems and failures: Qwen3Moe FP8 config issue vs FSDP CPU RAM-efficient loading." + "reason": "Different subsystems and symptoms: FineGrainedFP8/Qwen3Moe vs FSDP CPU RAM efficient loading." }, { "left": "issue:43742", "right": "issue:44291", "accept": false, - "reason": "Unrelated code paths: MobileLLM key error vs init_empty_weights handling of an unexpected argument." + "reason": "Different loading failures: MobileLLM key error vs init_empty_weights/_is_hf_initialized argument mismatch." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "Different bug shapes: BeitImageProcessorFast label reduction vs sequence-classification loss degeneracy." + "reason": "Label reduction bug in BeitImageProcessorFast is unrelated to sequence-classification loss degeneracy." }, { - "left": "issue:43756", - "right": "issue:45325", + "left": "issue:28282", + "right": "issue:29127", "accept": false, - "reason": "Both touch RoPE-ish logic, but they are different model families and distinct position-id bugs." + "reason": "ImportError for missing PyTorch and a LayoutLMv3 error-message clarity issue are not the same bug." }, { - "left": "issue:28282", - "right": "issue:29127", + "left": "issue:43582", + "right": "issue:43782", "accept": false, - "reason": "No shared underlying defect: PyTorch import missing vs LayoutLMv3 box-content error messaging." + "reason": "Apple Silicon allocator warmup TypeError is unrelated to Qwen3VL weight_only loading error." }, { "left": "issue:43540", "right": "issue:45823", "accept": false, - "reason": "Different model paths and failures: Qwen3OmniMoe video processing vs Gemma4 device placement with device_map=auto." + "reason": "Different models and failure modes: Qwen3OmniMoe video processing vs Gemma4 device mismatch." }, { - "left": "issue:43582", - "right": "issue:43782", + "left": "issue:43756", + "right": "issue:45325", "accept": false, - "reason": "Apple Silicon warmup TypeError and Qwen3VL weight_only loading error are unrelated." + "reason": "Separate RoPE/position-id bugs in different models; not a single underlying change." }, { "left": "issue:43493", "right": "issue:43994", "accept": false, - "reason": "Same model family theme, but not the same concrete bug: implementation discrepancy vs nonsensical inference results." + "reason": "Both mention SigLIP2, but one is a HF-vs-JAX discrepancy while the other is a nonsensical AutoModel/pipeline result; insufficient evidence of the same bug." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState token accounting and CustomTrainer loss-signature compatibility are separate trainer issues." + "reason": "TrainerState token counting and CustomTrainer compute_loss signature mismatch are distinct trainer issues." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "init_empty_weights argument handling is unrelated to AutoImageProcessor URL loading." + "reason": "Init-empty-weights argument handling is unrelated to loading an AutoImageProcessor from URL." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Tokenizer AddedToken duplication and Mistral regex patch backend access are different tokenizer failures." + "reason": "AddedToken constructor conflict and Mistral regex patch crash are different tokenizer-loading problems." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "Missing fields from trust_remote_code config loading is not the same as a Qwen3_5MoeVisionConfig schema omission." + "reason": "Both involve missing config fields, but one is generic trust_remote_code behavior and the other is a specific Qwen config field drop; not clearly the same code path." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both are loading-related, but they affect different APIs and different root causes." + "reason": "gguf_file loading breakage and a separate sleng-bert load failure are not the same issue." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "A failed janus test and Phi-4 multimodal model loading are not the same underlying bug." + "reason": "Different multimodal model failures and different symptoms; no shared bug is evident." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "PIL backend torchvision dependency regression and Mistral tokenizer regex patch crash are unrelated." + "reason": "Torchvision dependency regression in image processors is unrelated to Mistral tokenizer regex patching." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "Key error loading MobileLLM and missing Float8 storage are different serialization/loading problems." + "reason": "MobileLLM load failure and missing Float8 storage object are distinct errors." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and AutoImageProcessor URL loading do not share a code path." + "reason": "Qwen3-MoE sliding window behavior is unrelated to URL loading for AutoImageProcessor." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "PEFT version bump and stale auto_mappings references are separate maintenance/CI issues." + "reason": "PEFT version pinning and stale auto_mappings references are unrelated CI/setup problems." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "RTDetrV2 output mismatch and Qwen3.5 save_pretrained regression are distinct model-specific regressions." + "reason": "RTDetrV2 output drift and Qwen3.5 save_pretrained checkpoint corruption are different regressions." }, { "left": "issue:43688", "right": "issue:43749", "accept": false, - "reason": "Auxiliary-loss normalization and FSDP loading are different functionality with no direct overlap." + "reason": "Auxiliary-loss normalization and FSDP CPU RAM efficient loading affect different code paths." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "Default max_shard_size question and AutoProcessor kwargs propagation are unrelated." + "reason": "save_pretrained shard-size question is unrelated to AutoProcessor cached_file kwargs forwarding." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Batched processor crash with padding=False and Qwen2.5-VL video vision-position-id bug are different issues." + "reason": "Both are Qwen2.5-VL issues, but one is apply_chat_template batching and the other is video position-id computation; not the same bug." }, { - "left": "issue:24643", - "right": "issue:29127", + "left": "issue:44589", + "right": "issue:44704", "accept": false, - "reason": "DeepSpeed weight-dimension runtime error and LayoutLMv3 box-clarity issue are unrelated." + "reason": "Float8 storage lookup failure is unrelated to AutoProcessor kwargs forwarding." }, { - "left": "issue:43643", - "right": "issue:44561", + "left": "issue:44737", + "right": "issue:45588", "accept": false, - "reason": "Both involve trust_remote_code fallout, but the root causes differ: missing config fields vs removal of is_torch_fx_available." + "reason": "Already distinct XLNet CPU placement and flash_attention s_aux handling issues." }, { - "left": "issue:43452", - "right": "issue:43577", + "left": "issue:44821", + "right": "issue:43531", "accept": false, - "reason": "gguf_file loading breakage and BLIP2 dtype not propagating are different model-loading bugs." + "reason": "AutoImageProcessor URL loading and Qwen3-MoE sliding window are unrelated." }, { - "left": "issue:43720", - "right": "issue:44898", + "left": "issue:45042", + "right": "issue:45081", "accept": false, - "reason": "BitNet packed-weight unpacking in accelerate loading is unrelated to Perceiver non-default resolution failures." + "reason": "These touch different subsystems and have different failure signatures." }, { - "left": "issue:41762", - "right": "issue:44936", + "left": "issue:43720", + "right": "issue:44898", "accept": false, - "reason": "Gemma3 ZeRO-3 loading index error and trainer.evaluate after train are different failure modes." + "reason": "BitNet packed-weight loading and Perceiver interpolation are different model-loading failures." } ] }, @@ -6069,7 +6051,9 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-397", "nodes": [ + "issue:24643", "issue:28282", + "issue:29127", "issue:30064", "issue:32090", "issue:33357", @@ -6077,12 +6061,13 @@ "issue:35141", "issue:38175", "issue:41720", + "issue:41762", "issue:43295", - "issue:43329", "issue:43452", "issue:43531", "issue:43550", "issue:43577", + "issue:43643", "issue:43646", "issue:43653", "issue:43688", @@ -6096,9 +6081,8 @@ "issue:44442", "issue:44464", "issue:44466", - "issue:44479", "issue:44514", - "issue:44554", + "issue:44561", "issue:44568", "issue:44625", "issue:44743", @@ -6110,15 +6094,17 @@ "issue:44991", "issue:45005", "issue:45290", - "issue:45325", "issue:45356", "issue:45440", - "issue:45588", "issue:45701", "issue:45704", "issue:45823" ], "soft_pairs": [ + "issue:41762|issue:44936", + "issue:43643|issue:44561", + "issue:43452|issue:43577", + "issue:24643|issue:29127", "issue:44625|issue:44877", "issue:43452|issue:44843", "issue:44466|issue:45005", @@ -6135,8 +6121,8 @@ "issue:43688|issue:45704", "issue:30064|issue:33666", "issue:41720|issue:43906", - "issue:43577|issue:43697", "issue:28282|issue:32090", + "issue:43577|issue:43697", "issue:28282|issue:33357", "issue:43646|issue:43749", "issue:43688|issue:45701", @@ -6145,233 +6131,222 @@ "issue:43653|issue:43927", "issue:43531|issue:43742", "issue:43550|issue:43927", - "issue:44464|issue:44977", - "issue:43329|issue:45325", - "issue:43577|issue:45588", - "issue:43577|issue:44554", - "issue:44479|issue:44625", - "issue:43550|issue:44361" + "issue:44464|issue:44977" ], - "prepared_review_unit_hash": "6c57fcff8896e85394fa6ca8229e3a5f101b724ed6078ee0a77c19218023f5c8", + "prepared_review_unit_hash": "db311ba966acd53570ee842ca48b7057c3964aeee6015a1f3636b7753596174c", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37487, - "estimated_input_tokens": 9372, - "estimated_eval_tokens": 19000 + "soft_pair_count": 31, + "serialized_chars": 37104, + "estimated_input_tokens": 9276, + "estimated_eval_tokens": 18808 }, "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 regressions across tokenization, config loading, multimodal processors, training, and generation. The soft pairs are generally same-subsystem but not the same underlying bug, so none should be merged as duplicates.", - "confidence": 0.72, - "canonical_issue_reason": "issue:44843 is the strongest representative of the tokenizer-loading regression subset here: it has a concrete root cause, a clear failure mode, and is closer to several other tokenizer-related reports than the broader model/training bugs.", - "canonical_pr_reason": null, - "best_issue_reason": "issue:44843 is the most concrete and reusable duplicate anchor among the tokenizer regressions in this set.", - "best_pr_reason": null, + "summary": "No duplicate cluster here: the items span many unrelated bugs across loading, tokenization, training, MoE, and device-map behavior, so every soft edge should be rejected.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is appropriate; the set is heterogeneous and the paired items are not the same underlying defect.", + "canonical_pr_reason": "No pull requests are present in this cluster.", + "best_issue_reason": "No global best issue stands out as a representative duplicate anchor; at most, a few items share broad subsystems, but not the same concrete bug.", + "best_pr_reason": "No pull requests are present in this cluster.", "soft_edge_verdicts": [ + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: Gemma3 ZeRO-3 load-time index error vs trainer.evaluate() after train; no shared code-path bug." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both touch trust_remote_code, but one is missing fields in AutoConfig output and the other is v5 removal of is_torch_fx_available; distinct regressions." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "gguf_file breaks tokenizer/model loading, while the BLIP2 report is about dtype propagation; unrelated bugs." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed weight-shape training error vs LayoutLMv3 box-clarity/error-message issue; different problems." + }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Both are config-related, but one is Qwen3.5 num_labels propagation and the other is strict loading for granite_speech; different code paths and failures." + "reason": "num_labels propagation in Qwen3.5 is unrelated to strict config loading for granite_speech." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "Both involve tokenizer loading, but one is gguf_file handling for AutoTokenizer/AutoModel and the other is an unconditional model_info() call in _patch_mistral_regex." + "reason": "Both involve loading/tokenizer paths, but gguf_file breakage and unconditional model_info() in offline mode are different bugs." }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights, but one is device-dependent lm_head serialization and the other is translation-model weight tying behavior; not the same bug." + "reason": "Both mention tied weights, but one is device-dependent lm_head serialization and the other is translation-model tied-weight handling; not the same fix." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Same API surface, different triggers: batched input with padding=False versus assistant tool-call messages without content." + "reason": "Both crash in apply_chat_template, but with different inputs and causes: batched padding=False vs assistant tool-call messages with no content." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Both are regressions, but one is processor.tokenizer/image passing breakage and the other is Kimi-K2.5 codec/Mistral regex handling." + "reason": "One is processor.tokenizer regression; the other is Kimi-K2.5 codec/warning behavior. Same area, different defect." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "Completely different areas: DAC latent conversion versus Whisper batch_decode skip_special_tokens behavior." + "reason": "DAC from_latents mismatch and Whisper batch_decode skip_special_tokens are unrelated code paths." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Both affect generation, but one is compiled-forward chunking inconsistency and the other is Qwen3.5 recurrent-state reset with cache." + "reason": "Compiled-forward chunked generation inconsistency is distinct from Qwen3-MoE recurrent-state reset with cache." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "Both concern special tokens, but they are different tokenizers and different failures: mask token registration vs BOS/EOS omission." + "reason": "BigBirdTokenizer special-token registration and mdeberta add_special_tokens behavior are separate tokenizer bugs." }, { "left": "issue:44936", "right": "issue:45823", "accept": false, - "reason": "Trainer lifecycle failure versus Gemma4 PLE device mismatch; unrelated code paths." + "reason": "trainer.evaluate() after train is unrelated to Gemma4 PLE device mismatch under device_map=auto." }, { "left": "issue:43688", "right": "issue:45440", "accept": false, - "reason": "Both are MoE-related, but one is auxiliary-loss normalization and the other is DeepSeekV3MoE implementation divergence." + "reason": "Auxiliary-loss normalization in OLMoE/GPT Oss is not the same as DeepseekV3MoE diverging from the remote implementation." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Siglip2 zero probabilities is unrelated to the processor/tokenizer regression in v4.57.5." + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to the processor.tokenizer regression." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "Both are tokenizer loading issues, but they affect different tokenizer classes with different errors and likely different fixes." + "reason": "MLukeTokenizer AttributeError and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are different tokenizer-loading failures." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "BigBird special-token registration and MLukeTokenizer AttributeError are unrelated tokenizer bugs." + "reason": "Special-token registration in BigBirdTokenizer is unrelated to MLukeTokenizer AttributeError." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "MoE auxiliary-loss normalization is unrelated to T5's apex.FusedRMSNorm memory leak." + "reason": "MoE auxiliary-loss normalization and T5's apex FusedRMSNorm memory leak are unrelated." }, { "left": "issue:30064", "right": "issue:33666", "accept": false, - "reason": "Image segmentation map processing and Qwen2-VL multi-GPU training are unrelated." + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:41720", "right": "issue:43906", "accept": false, - "reason": "Different problems and no evidence the reproduction of 38071 is the same cudaErrorAssert root cause." + "reason": "Qwen3 auto device-mapping cudaErrorAssert is not the same as an isolated reproduction of a different upstream issue." }, { - "left": "issue:43577", - "right": "issue:43697", + "left": "issue:28282", + "right": "issue:32090", "accept": false, - "reason": "Both are regression reports, but one is BLIP2 dtype loading and the other is RTDetrV2 output drift; different model families and symptoms." + "reason": "PyTorch-missing import error and NoneType broadcast failure in Trainer are unrelated." }, { - "left": "issue:28282", - "right": "issue:32090", + "left": "issue:43577", + "right": "issue:43697", "accept": false, - "reason": "ImportError for missing PyTorch versus Trainer broadcast NoneType TypeError; unrelated." + "reason": "BLIP2 dtype state is unrelated to RTDetrV2 producing different outputs in v5." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "Different failure modes: missing torch import versus MacOS bus error on a CLIP model." + "reason": "ImportError about missing PyTorch and Mac bus-error on CLIP are different environment/runtime problems." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization breakage in v5 is not the same as FSDP CPU RAM efficient loading being broken." + "reason": "Transformers v5 custom-model initialization break is distinct from FSDP_CPU_RAM_EFFICIENT_LOADING." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "Auxiliary-loss normalization in MoE models is unrelated to a generic tokenization-change report." + "reason": "Auxiliary-loss normalization is unrelated to tokenization changing across versions." }, { "left": "issue:28282", "right": "issue:35141", "accept": false, - "reason": "Missing PyTorch dependency import error is unrelated to post_init reinitializing output embeddings." + "reason": "Missing-PyTorch import error and output-embedding reinitialization after resize_token_embeddings are different bugs." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer-loading regressions, but the concrete failures differ: mistral-regex/offline model_info versus EMBEDDIA est-roberta loading." + "reason": "Offline model_info() regression and tokenizer-loading failure for a specific model are different failure modes." }, { "left": "issue:43653", "right": "issue:43927", "accept": false, - "reason": "Special-token registration in BigBird and DiaConfig custom token ID persistence are different bugs." + "reason": "BigBirdTokenizer special-token registration and DiaConfig losing custom token IDs after save/load are separate state/token bugs." }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key errors are unrelated." + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error are unrelated." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to DiaConfig token ID persistence." + "reason": "torch.compile with SDPA on Bamba-9B-v2 and DiaConfig save/load token-ID loss are unrelated." }, { "left": "issue:44464", "right": "issue:44977", "accept": false, - "reason": "Both affect generation, but the former is compiled-forward chunking while the latter is flash-attention behavior in Qwen3.5." - }, - { - "left": "issue:43329", - "right": "issue:45325", - "accept": false, - "reason": "Both touch Qwen VLM video paths, but one is an undefined-variable bug in token counting and the other is incorrect RoPE position scaling." - }, - { - "left": "issue:43577", - "right": "issue:45588", - "accept": false, - "reason": "Different subsystems: BLIP2 dtype loading versus flash_attention sink-less s_aux handling." - }, - { - "left": "issue:43577", - "right": "issue:44554", - "accept": false, - "reason": "BLIP2 dtype mismatch and an MPS attention correctness issue are unrelated." - }, - { - "left": "issue:44479", - "right": "issue:44625", - "accept": false, - "reason": "A Qwen video-input regression is not the same as Qwen3.5 num_labels propagation." - }, - { - "left": "issue:43550", - "right": "issue:44361", - "accept": false, - "reason": "Bamba torch.compile/SDPA failure and MLukeTokenizer AttributeError do not share the same underlying bug." + "reason": "Chunked generation with compiled forward and Qwen3.5 flash-attention generation issues are different bugs." } ] }, @@ -6405,12 +6380,10 @@ "issue:39692", "issue:41720", "issue:43122", - "issue:43257", "issue:43262", "issue:43278", "issue:43295", - "issue:43299", - "issue:43334", + "issue:43329", "issue:43452", "issue:43493", "issue:43504", @@ -6418,6 +6391,7 @@ "issue:43540", "issue:43550", "issue:43572", + "issue:43577", "issue:43606", "issue:43688", "issue:43720", @@ -6430,29 +6404,34 @@ "issue:44361", "issue:44451", "issue:44464", + "issue:44479", + "issue:44554", "issue:44568", "issue:44610", "issue:44625", "issue:44821", "issue:44849", "issue:44913", - "issue:44964", "issue:44977", - "issue:45083", + "issue:45325", "issue:45357", - "issue:45405", - "issue:45584", + "issue:45588", "issue:45812" ], "soft_pairs": [ + "issue:43577|issue:45588", + "issue:43329|issue:45325", + "issue:43577|issue:44554", + "issue:44479|issue:44625", + "issue:43550|issue:44361", "issue:43122|issue:43906", "issue:32090|issue:34567", "issue:43504|issue:43550", "issue:43504|issue:43720", "issue:41720|issue:44464", "issue:43504|issue:44610", - "issue:44849|issue:45357", "issue:32090|issue:33666", + "issue:44849|issue:45357", "issue:43452|issue:44451", "issue:43278|issue:43688", "issue:44568|issue:45812", @@ -6469,228 +6448,223 @@ "issue:43493|issue:43844", "issue:33666|issue:36331", "issue:39692|issue:43493", - "issue:43278|issue:43844", "issue:43540|issue:44977", - "issue:44964|issue:45405", - "issue:43257|issue:43299", - "issue:43334|issue:43572", - "issue:44625|issue:45584", - "issue:43756|issue:45083" + "issue:43278|issue:43844" ], - "prepared_review_unit_hash": "a67d0375994034054fcfa4d024a9df227c01f30d3c3dc5d1bfee8bd8d2cc1856", + "prepared_review_unit_hash": "3e2064489a5265579473ceba9ba615fb6769f7796b3aaf87ad9d01fdf81db26e", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 31, - "serialized_chars": 37031, - "estimated_input_tokens": 9258, - "estimated_eval_tokens": 18772 + "serialized_chars": 36622, + "estimated_input_tokens": 9156, + "estimated_eval_tokens": 18568 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated Transformer issues, with the strongest common thread being tokenizer/config regression reports around the v4\u2192v5 transition. No pull requests are present.", - "confidence": 0.61, - "canonical_issue_reason": "issue:45812 is the broadest and most representative tokenizer-regression report here: it affects all Granite models, is a silent v4\u2192v5 behavior change, and overlaps the tokenization-focused subset better than the narrower model-specific bugs.", + "summary": "This cluster is heterogeneous: many items are tokenizer/config/training/loading bugs, but the soft pairs are only superficially similar and do not appear to be the same concrete underlying issue. No soft edge should be merged.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45812 is the broadest and most central tokenizer regression in the set, with multiple inbound refs and a clear v4\u2192v5 impact; the other items are mostly unrelated one-off model/training/loading bugs.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is the best global representative because it describes a widespread, user-visible tokenizer regression rather than a single-model edge case or an isolated runtime error.", + "best_issue_reason": "issue:45812 is the strongest representative issue because it describes a widespread, silent tokenizer regression affecting all Granite models and is more cluster-representative than the narrower, model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43577", + "right": "issue:45588", + "accept": false, + "reason": "Different code paths: BLIP2 dtype handling vs flash-attention sinkless-model crash." + }, + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Both involve video/VL models, but one is undefined variables in multimodal token counting and the other is rope-index scaling for still-image temporal positions." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Unrelated bugs: loading dtype mismatch in BLIP2 vs an MPS attention correctness issue." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is a video-input regression and the other is num_labels propagation." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Different failures: torch.compile + SDPA on Bamba vs MLukeTokenizer attribute error on tasks." + }, { "left": "issue:43122", "right": "issue:43906", "accept": false, - "reason": "Both mention tokenization, but one is a version-to-version tokenization difference and the other is an isolated reproduction of a separate issue; not the same concrete bug." + "reason": "One is a tokenizer output-change regression across versions; the other is an isolated repro of a different issue." }, { "left": "issue:32090", "right": "issue:34567", "accept": false, - "reason": "Trainer broadcast NoneType errors and TrainerState token-count updates are unrelated failures." + "reason": "Trainer errors, but different symptoms and code paths: gpu broadcast NoneType vs num_input_tokens_seen not updating." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "Legacy-field model loading and torch.compile SDPA failure are different code paths and different bugs." + "reason": "Legacy-field model loading bug vs compile/SDPA runtime failure on a different model." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Both are loading-related, but one is a legacy config-field issue and the other is packed-weight unpacking during accelerate loading; not the same fix." + "reason": "Different loading problems: legacy preset field vs packed-weight handling during accelerate loading." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "Qwen3 auto device-map CUDA asserts and chunked-generation inconsistencies under compiled forward are distinct failures." + "reason": "A800 auto device mapping cudaErrorAssert is not the same as inconsistent outputs from compiled forward chunked generation." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "A pretrained-load legacy-field bug is unrelated to a processor image-size mismatch." + "reason": "Different issues: pretrained preset load failure vs processor producing the wrong input size for the model." }, { - "left": "issue:44849", - "right": "issue:45357", + "left": "issue:32090", + "right": "issue:33666", "accept": false, - "reason": "Both are Qwen3.5-related, but one is an output_hidden_states bug and the other is incorrect save_pretrained key serialization." + "reason": "Both mention training, but one is a broadcast TypeError and the other is a Qwen2-VL multi-GPU training question/bug." }, { - "left": "issue:32090", - "right": "issue:33666", + "left": "issue:44849", + "right": "issue:45357", "accept": false, - "reason": "Generic trainer broadcast/type errors do not match a Qwen2-VL multi-GPU training issue." + "reason": "Both are Qwen3.5 regressions, but hidden-states behavior and save_pretrained key mapping are distinct bugs." }, { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "gguf_file support regression and a model-specific load failure are not the same underlying bug." + "reason": "Loading failures, but gguf_file support breakage and inability to load a specific model are not the same bug." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Embedding dtype drift in eval and auxiliary-loss normalization are unrelated." + "reason": "Different model behaviors: dtype drift in eval vs incorrect auxiliary-loss normalization." }, { "left": "issue:44568", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but one is about special-token insertion for a specific tokenizer and the other is wrong token IDs for Granite models; not the same bug." + "reason": "Both are tokenizer regressions in v5, but they affect different models and symptoms (special tokens vs wrong token IDs) with no evidence of the same code-path bug." }, { "left": "issue:43742", "right": "issue:44821", "accept": false, - "reason": "Model key errors and AutoImageProcessor-from-URL loading are unrelated." + "reason": "Both are load/processor problems, but one is a MobileLLM key error and the other is loading AutoImageProcessor from URL." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Multi-GPU training for Qwen2-VL and a GenerationMixin import error are unrelated." + "reason": "Multi-GPU training vs GenerationMixin import error; unrelated failures." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation discrepancy and pipeline translation-task messaging are unrelated." + "reason": "SigLIP2 implementation discrepancy is unrelated to the pipeline translation-task support message regression." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch and DiaConfig token-ID persistence are different bugs." + "reason": "Device-mismatch during CPU offload is unrelated to loss of custom token IDs and generation IndexError." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "Packed-weight handling during accelerate loading does not match config token-ID loss on save/load." + "reason": "Packed-weight loading failure is a different bug from DiaConfig losing custom token IDs after save/load." }, { "left": "issue:43756", "right": "issue:44913", "accept": false, - "reason": "RoPE-layer truncation in Smollm3 and GPTNeoX rotary_pct reload behavior are only loosely config-related, not the same bug." + "reason": "RoPE-layer selection bug and GPTNeoX rotary_pct reload bug are both config-related but not the same issue." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and processor.tokenizer/image-input regression are different multimodal processor issues." + "reason": "Audio chat-template sampling-rate default is unrelated to processor.tokenizer access / image-passing regression." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Both are missing config token fields after v5 changes, but they affect different models and different attributes; too broad to merge." + "reason": "Both mention missing pad token fields, but they affect different configs and different attribute names; no shared concrete bug is shown." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer spacing after added tokens is unrelated to a MacOS bus error on a CLIP model." + "reason": "Tokenizer spacing bug and MacOS bus error in a CLIP model are unrelated." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "BitNet packed-weight loading and MLukeTokenizer task AttributeError are unrelated." + "reason": "Packed-weight loading and MLukeTokenizer task attribute errors are separate failures." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 discrepancy and ZeRO-3 gradient growth are unrelated training/implementation issues." + "reason": "SigLIP2 model discrepancy is unrelated to DeepSpeed ZeRO-3 gradient growth." }, { "left": "issue:33666", "right": "issue:36331", "accept": false, - "reason": "Qwen2-VL multi-GPU training and a custom compute_loss signature mismatch are different trainer issues." + "reason": "Different training issues: multi-GPU Qwen2-VL vs CustomTrainer.compute_loss signature mismatch." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "A doc example with multiple errors is not the same as a core SigLIP2 implementation discrepancy." - }, - { - "left": "issue:43278", - "right": "issue:43844", - "accept": false, - "reason": "BF16-to-FP32 eval dtype drift does not match a ZeRO-3 gradient explosion issue." + "reason": "Both are SigLIP2-related, but one is a docs example mismatch/quantization failure and the other is a core implementation discrepancy; not the same bug." }, { "left": "issue:43540", "right": "issue:44977", "accept": false, - "reason": "Qwen3OmniMoe video-processing errors and Qwen3.5 flash-attention generation bugs are distinct." - }, - { - "left": "issue:44964", - "right": "issue:45405", - "accept": false, - "reason": "A model load regression and an unreleased PEFT version bump are unrelated." - }, - { - "left": "issue:43257", - "right": "issue:43299", - "accept": false, - "reason": "Both involve Qwen3 MoE loading, but one is an accelerate+deepspeed weight-conversion bug and the other is a generic Qwen3VL MoE load failure; not clearly the same concrete defect." - }, - { - "left": "issue:43334", - "right": "issue:43572", - "accept": false, - "reason": "Missing pad_token_id in Qwen3-VL and missing pad_token_idx in StableLmConfig are similar class-wise, but they are separate model/config regressions." - }, - { - "left": "issue:44625", - "right": "issue:45584", - "accept": false, - "reason": "Qwen3.5 num_labels propagation and Whisper empty-transcription generation are unrelated." + "reason": "Video processing ValueError in Qwen3OmniMoe and flash-attention generation failure in Qwen3.5 are separate bugs." }, { - "left": "issue:43756", - "right": "issue:45083", + "left": "issue:43278", + "right": "issue:43844", "accept": false, - "reason": "Smollm3 RoPE-layer loss and a Qwen3 Omni MoE feature-length helper bug are unrelated." + "reason": "BF16-to-FP32 dtype drift is unrelated to abnormal gradient growth under ZeRO-3." } ] }, @@ -6714,16 +6688,16 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-397", "nodes": [ - "issue:28218", "issue:29942", "issue:30333", "issue:33357", "issue:33666", - "issue:34567", "issue:39692", "issue:43232", + "issue:43257", "issue:43278", "issue:43295", + "issue:43299", "issue:43334", "issue:43504", "issue:43572", @@ -6732,6 +6706,7 @@ "issue:43646", "issue:43653", "issue:43701", + "issue:43756", "issue:43824", "issue:43844", "issue:43927", @@ -6745,31 +6720,36 @@ "issue:44488", "issue:44493", "issue:44521", - "issue:44554", "issue:44589", + "issue:44625", "issue:44734", "issue:44792", "issue:44843", "issue:44849", "issue:44964", + "issue:45083", "issue:45200", "issue:45229", "issue:45290", "issue:45405", - "issue:45446", "issue:45520", "issue:45584", "issue:45759", "issue:45823" ], "soft_pairs": [ + "issue:44964|issue:45405", + "issue:43257|issue:43299", + "issue:43334|issue:43572", + "issue:44625|issue:45584", + "issue:43756|issue:45083", "issue:44488|issue:44843", "issue:45200|issue:45229", "issue:43334|issue:45759", "issue:43646|issue:44466", "issue:43824|issue:44589", - "issue:45229|issue:45823", "issue:39692|issue:43295", + "issue:45229|issue:45823", "issue:33357|issue:33666", "issue:44589|issue:45520", "issue:43504|issue:43606", @@ -6786,194 +6766,210 @@ "issue:29942|issue:30333", "issue:44521|issue:45290", "issue:43232|issue:44849", - "issue:43844|issue:44964", - "issue:28218|issue:34567", - "issue:44554|issue:45446" + "issue:43844|issue:44964" ], - "prepared_review_unit_hash": "1c1e9a99db59a6d4a45b63ea89e860d872d45c531b9de30b2dfb261082ad129d", + "prepared_review_unit_hash": "948bf06835ded405396d65d1e684828ebfcdc8593dee50168d20d75bda2f82c3", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 34996, - "estimated_input_tokens": 8749, - "estimated_eval_tokens": 17754 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36537, + "estimated_input_tokens": 9135, + "estimated_eval_tokens": 18526 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous cluster of unrelated bug reports across tokenizers, multimodal models, training, and serving. None of the soft-similarity pairs look like the same underlying defect, so no duplicate merges are justified.", - "confidence": 0.96, - "canonical_issue_reason": "No canonical issue stands out: the items are not duplicates and the apparent similarities are only broad subsystem overlap.", + "summary": "This cluster is mostly a loose similarity group of Transformers/model-regression issues, not true duplicates. The strongest unifying thread is v5-era model/config loading breakage, but most pairs are distinct model-specific failures. No PRs are present.", + "confidence": 0.38, + "canonical_issue_reason": "issue:43646 is the broadest and most central regression in the set: it describes Transformers 5.0.0 breaking custom model initialization, which is the closest umbrella symptom to several other v5 loading/init complaints.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good cluster representative because the cluster does not center on one shared bug or code path.", + "best_issue_reason": "issue:43646 is the best representative issue because it is general, clearly stated, and captures the broader initialization regression theme better than the more model-specific tickets.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "One is a model-loading failure for Phi-4 multimodal; the other is a dependency/version bump issue for PEFT. No shared bug." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen3, but one is MOE weight conversion with accelerate/deepspeed and the other is Qwen3VL loading. Different code paths and failures." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Both are missing config fields, but for different models and different fields (`pad_token_id` vs `pad_token_idx`). Not the same bug." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation and Whisper empty-transcription behavior are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:45083", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and qwen3_omni_moe feature-length helper behavior are different issues." + }, { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "Different failures: model loading for a specific tokenizer vs offline model_info access in tokenizer patching." + "reason": "A specific model load failure versus an offline-mode tokenizer regex patch bug; different failure modes." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Both mention Gemma4, but one is a token-id defaulting bug and the other is a multi-GPU CUDA OOM." + "reason": "Same Gemma4 family, but one is a missing default tensor input and the other is a multi-GPU OOM. Not mergeable as one fix." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Both involve Qwen multimodal configs, but one is a missing pad_token_id and the other is text_config unwrapping in from_config." + "reason": "Both involve Qwen multimodal configs, but one is a missing `pad_token_id` attribute and the other is incorrect unwrapping of `text_config`. Different concrete bugs." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization breakage is unrelated to lm_head.weight serialization differences." + "reason": "Custom model initialization regression versus device-dependent `lm_head.weight` serialization inconsistency. Not the same code-path problem." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Importing a Qwen2.5-VL class is unrelated to the Float8 storage lookup error." + "reason": "ImportError for a missing Qwen2.5-VL class versus a Float8 storage lookup error. Unrelated." }, { - "left": "issue:45229", - "right": "issue:45823", + "left": "issue:39692", + "right": "issue:43295", "accept": false, - "reason": "Gemma4 inference OOM and device mismatch in forward are different bugs and code paths." + "reason": "Docs/example errors for SigLIP2 versus a runtime regression in custom model code using `processor.tokenizer` and images. Different artifacts and bugs." }, { - "left": "issue:39692", - "right": "issue:43295", + "left": "issue:45229", + "right": "issue:45823", "accept": false, - "reason": "Docs example errors are not the same as the processor/tokenizer regression in custom model code." + "reason": "Both are Gemma4-related, but one is CUDA OOM in inference and the other is a device-mismatch forward bug. Distinct issues." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "A MacOS CLIP bus error is unrelated to Qwen2-VL multi-GPU training." + "reason": "MacOS CLIP bus error and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Storage-object lookup failure is not the same as the flash_attn import KeyError on Python 3.13." + "reason": "Float8 storage lookup failure and Python 3.13 flash_attn KeyError are different import/runtime problems." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Legacy preset loading and CPU offload device mismatch are different issues." + "reason": "BEiT pretrained-load legacy-field issue versus Bark CPU-offload device mismatch. Different models and failures." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is unrelated to init_empty_weights passing _is_hf_initialized." + "reason": "Missing `pad_token_idx` in StableLmConfig and unexpected `_is_hf_initialized` in `init_empty_weights` are different initialization regressions." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "DiaConfig losing custom token IDs is a different failure from MLukeTokenizer AttributeError." + "reason": "DiaConfig save/load token-ID loss and MLukeTokenizer task AttributeError are separate tokenizer/config bugs." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Continuous-batching multimodal crash and KV-cache continuation indexing bug are separate serve paths." + "reason": "Both are serve-related, but one is a multimodal `.to` crash and the other is KV-cache continuation tensor indexing. Different endpoints and code paths." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "BEiT legacy field loading and BigBirdTokenizer special-token registration are unrelated tokenizer/model issues." + "reason": "A BEiT model-loading legacy-field issue versus a BigBirdTokenizer special-token registration bug. Unrelated." }, { "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "TP OOM for a Qwen2 model is unrelated to the PEFT minimum-version bump." + "reason": "Qwen2-57B-A14B-Auto tensor-parallel OOM and unreleased PEFT version bump are not the same issue." }, { "left": "issue:44464", "right": "issue:45584", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is not the same as Whisper empty-transcription failure." + "reason": "Compiled-forward chunked generation inconsistency and Whisper empty-transcription failure are unrelated." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 bad outputs and UNEXPECTED Key with position id are different symptoms with different likely causes." + "reason": "SigLIP2 pipeline/AutoModel nonsense outputs and unexpected position-id keys across many models do not describe the same concrete bug." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "CPU offload device mismatch is not the same as a stale GraniteSpeech CI test." + "reason": "Bark-small CPU offload device mismatch versus a GraniteSpeech CI test issue. One is runtime behavior, the other is test flakiness." }, { "left": "issue:43278", "right": "issue:43701", "accept": false, - "reason": "Dtype drift between train/eval and resume_from_checkpoint key mismatch are unrelated." + "reason": "Embedding dtype drift and checkpoint resume key mismatch are unrelated training issues." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure is unrelated to the PEFT version requirement issue." + "reason": "Janus image-generation test failure and PEFT version bump are unrelated." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Flash Attention 2 test failures and MLFlow reporting on job failure are unrelated." + "reason": "Flash Attention 2 test failures and MLFlow reporting on failed jobs are entirely different problems." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "All-zero assistant masks and crashes on tool-call assistant messages are both chat-template bugs but not the same concrete failure." + "reason": "Both involve `apply_chat_template`, but one is all-zero assistant masks and the other is a crash with tool-call messages. Different failure modes." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs sync_gpus handling and Qwen3.5 hidden-state behavior are separate bugs." + "reason": "Generation kwarg update after `sync_gpus` and Qwen3.5 hidden-state behavior are unrelated." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading failure are unrelated." - }, - { - "left": "issue:28218", - "right": "issue:34567", - "accept": false, - "reason": "Tokenizer spacing after added tokens is unrelated to TrainerState input-token counting." - }, - { - "left": "issue:44554", - "right": "issue:45446", - "accept": false, - "reason": "MPS attention correctness and a PyTorch version check in flex_attention are different issues." + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading failure are different bugs." } ] }, @@ -6999,6 +6995,7 @@ "nodes": [ "issue:15354", "issue:28218", + "issue:34567", "issue:35141", "issue:37428", "issue:38175", @@ -7020,7 +7017,6 @@ "issue:43957", "issue:43975", "issue:44315", - "issue:44442", "issue:44464", "issue:44466", "issue:44554", @@ -7040,12 +7036,15 @@ "issue:45325", "issue:45356", "issue:45412", + "issue:45446", "issue:45520", "issue:45563", "issue:45701", "issue:45823" ], "soft_pairs": [ + "issue:28218|issue:34567", + "issue:44554|issue:45446", "issue:43582|issue:44977", "issue:45084|issue:45520", "issue:15354|issue:38175", @@ -7074,231 +7073,233 @@ "issue:43295|issue:43873", "issue:38175|issue:45412", "issue:28218|issue:35141", - "issue:42915|issue:43824", "issue:38175|issue:43493", - "issue:44442|issue:45198" + "issue:42915|issue:43824" ], - "prepared_review_unit_hash": "1b9eaa28797c5df897e1b7f033503bc8fda5fb200791e4f347a7ffaf127833e8", + "prepared_review_unit_hash": "a37dff7dd02d0ed5bce829c300ed284069afda4d19db255f083d136cc8eabe13", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36661, - "estimated_input_tokens": 9166, - "estimated_eval_tokens": 18588 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37505, + "estimated_input_tokens": 9377, + "estimated_eval_tokens": 19010 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated Transformers issues; several soft links are only broad topical similarities (tokenization, config fields, processor bugs), but none look like the same underlying fix.", - "confidence": 0.84, - "canonical_issue_reason": "No single true canonical duplicate for the cluster: the items span many unrelated bugs. If one issue must anchor the set, 45701 is the broadest tokenization-regression report.", + "summary": "Most pairs are unrelated false positives across different models/features. I only see two likely duplicates: the missing `pad_token_id` config regressions and the SigLIP2 implementation discrepancy. The rest are separate bugs. I chose the broad tokenization/version-regression report as the best overall issue umbrella.", + "confidence": 0.41, + "canonical_issue_reason": "Issue 45701 is the broadest umbrella in this set: a version-dependent tokenization regression. It is the closest high-level representative of the cluster, though many items are still distinct model-specific bugs.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45701 is the best representative because it is the most general user-facing regression and overlaps the tokenization-focused subset of the cluster.", + "best_issue_reason": "45701 is the most general issue to keep as the cluster representative; it can absorb the tokenizer/version-regression theme better than the narrower model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer spacing after added tokens and TrainerState token counting are unrelated bugs." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness and a flex_attention import/version guard are different code paths." + }, { "left": "issue:43582", "right": "issue:44977", "accept": false, - "reason": "Different failures: Apple Silicon allocator warmup TypeError vs flash-attention generation behavior." + "reason": "Apple Silicon allocator warmup TypeError is unrelated to Qwen3.5 flash-attention generation." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Non-template compilation error and Python 3.13 flash_attn import KeyError are unrelated code paths." + "reason": "Compile/template-node errors and a Python 3.13 flash_attn import KeyError are separate failures." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript export support issue vs SigLIP2 zero-probability output bug." + "reason": "TorchScript GeneratorExp support and SigLIP2 zero probabilities are not the same issue." }, { "left": "issue:44568", "right": "issue:45701", "accept": false, - "reason": "Specific mdeberta added-token BOS/EOS behavior is not the same concrete bug as a broad tokenization-version regression." + "reason": "A specific mdeberta special-tokens bug is too narrow to be the same as the broader tokenization-version regression." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError and Gemma4 mm_token_type_ids defaulting are different model-path bugs." + "reason": "Video-input processing and Gemma4 text-only token type defaults are different problems." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "bfloat16 dtype mismatch and Wav2Vec2 save/tokenization failure do not share a concrete root cause." + "reason": "bfloat16 dtype mismatches and Wav2Vec2 save/tokenization failures are unrelated." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Both are Gemma4, but one is token-type defaults and the other is forward device mismatch." + "reason": "Missing `mm_token_type_ids` defaults and a Gemma4 device-map mismatch are distinct bugs." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "Missing flash-attention import symbol and meta-device loading regressions are separate failures." + "reason": "A missing flash-attention helper import and meta-device loading breakage are different regressions." }, { "left": "issue:44315", "right": "issue:44849", "accept": false, - "reason": "Liger Kernel/model_init application bug and Qwen3.5 hidden-states bug are unrelated." + "reason": "Liger kernel application in `model_init` and a Qwen3.5 hidden-states bug do not share the same root cause." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "MPS attention correctness and tokenizer codec regression are different subsystems." + "reason": "MPS attention correctness and tokenizer codec handling are unrelated." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 implementation discrepancy and resume_from_checkpoint key mismatch are unrelated." + "reason": "SigLIP2 HF/JAX discrepancy and checkpoint resume key mismatch are separate issues." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Processor/tokenizer API regression and batched video shape bug are not the same underlying issue." + "reason": "Processor.tokenizer regression with images and OmDet-Turbo resize/shape mismatch are not the same bug." }, { "left": "issue:43873", "right": "issue:44464", "accept": false, - "reason": "Quantization offloading and compiled-forward chunked generation are separate code paths." + "reason": "Quantization offloading and compiled-forward chunked generation are different failure modes." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "Tied-weight serialization depends on device, while the other is LoRA merge collapse after vocab extension." + "reason": "Both involve tied weights, but one is serialization inconsistency and the other is LoRA-merge collapse; not clearly the same concrete bug." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Perceiver resolution/interpolation failure and RT-DETR memory release are unrelated." + "reason": "Perceiver resolution handling and RT-DETR memory release are unrelated." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "Janus image-generation test failure and stale generate warning do not share a bug." + "reason": "A Janus generation test failure and a stale paged-generate warning are different issues." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Gemma4 text-only fine-tuning token-type handling is unrelated to Qwen2.5-VL rope-index scaling." + "reason": "Defaulting `mm_token_type_ids` and Qwen2.5-VL temporal rope scaling are unrelated." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "StableLmConfig pad_token_idx regression and physical-intelligence/fast loading failure are different issues." + "reason": "Missing `pad_token_idx` in StableLmConfig may be adjacent, but the loading failure is too unspecific to confirm the same bug." }, { "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different breakage points." + "reason": "Two different missing-import symbols; same broad area, but not the same underlying defect." }, { "left": "issue:43844", "right": "issue:44936", "accept": false, - "reason": "ZeRO-3 gradient growth and trainer.evaluate-after-train failure are unrelated training bugs." + "reason": "ZeRO-3 gradient growth and post-train evaluate failure are different training issues." }, { "left": "issue:43572", "right": "issue:43957", "accept": false, - "reason": "Missing pad_token_idx and meta-device loading regressions do not look like the same fix." + "reason": "A config field regression and meta-device loading breakage are not the same problem." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor-tokenizer regression and batched video output shape bug are different modalities and code paths." + "reason": "Processor.tokenizer regression and batched video-shape handling are distinct." }, { "left": "issue:43334", "right": "issue:43525", - "accept": false, - "reason": "Both mention missing pad_token_id, but they are model-specific config issues for different architectures." + "accept": true, + "reason": "Both are the same `pad_token_id` AttributeError class of config regression, just reported on different model configs." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "ImportError for Qwen2.5-VL and Deepseek detokenization are unrelated." + "reason": "Missing model import and incorrect detokenization are unrelated." }, { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Label dropping in gather_for_metrics and BEiT label reduction are similar in theme only, not the same bug." + "reason": "Metrics label dropping and BeitImageProcessorFast label reduction are different code paths." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Processor API regression and quantization offloading bug are unrelated." + "reason": "Processor/tokenizer regression and quantization offloading are unrelated." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 output probabilities and RT-DETR memory cleanup are different problems." + "reason": "SigLIP2 output issues and RT-DETR memory release are different bugs." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer spacing after added tokens and embedding reinitialization on resize are distinct issues." - }, - { - "left": "issue:42915", - "right": "issue:43824", - "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure and Qwen2.5-VL import error are unrelated." + "reason": "Added-token spacing and output-embedding reinitialization are unrelated." }, { "left": "issue:38175", "right": "issue:43493", - "accept": false, - "reason": "Both touch SigLIP2, but one is inference probability correctness and the other is implementation parity with JAX." - }, - { - "left": "issue:44442", - "right": "issue:45198", - "accept": false, - "reason": "FastSpeech2ConformerTokenizer loading and Wav2Vec2 save/tokenization failure are different tokenizer bugs." + "accept": true, + "reason": "Both describe the same SigLIP2 HF implementation mismatch/incorrect outputs compared with the reference implementation." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays conservative. The two accepted soft pairs are well supported by the titles/body themes, and the rejections avoid overclaiming across clearly different bugs. The choice of issue:45701 as the umbrella representative is also reasonable." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -7341,6 +7342,7 @@ "issue:44206", "issue:44351", "issue:44360", + "issue:44442", "issue:44448", "issue:44462", "issue:44493", @@ -7353,20 +7355,20 @@ "issue:44857", "issue:45083", "issue:45127", + "issue:45198", "issue:45200", "issue:45245", "issue:45290", - "issue:45310", "issue:45381", - "issue:45478", "issue:45584", "issue:45588", "issue:45593", "issue:45823" ], "soft_pairs": [ - "issue:44514|issue:44521", + "issue:44442|issue:45198", "issue:43278|issue:43856", + "issue:44514|issue:44521", "issue:43653|issue:44112", "issue:43827|issue:43901", "issue:43450|issue:43575", @@ -7380,254 +7382,250 @@ "issue:44610|issue:44857", "issue:43278|issue:43825", "issue:43335|issue:43550", - "issue:37428|issue:43425", "issue:43606|issue:45083", + "issue:37428|issue:43425", "issue:44857|issue:45593", "issue:44448|issue:44743", "issue:45290|issue:45381", "issue:45200|issue:45588", "issue:29942|issue:32090", "issue:39692|issue:43526", - "issue:37428|issue:44351", "issue:43611|issue:43931", + "issue:37428|issue:44351", "issue:44206|issue:45823", "issue:44462|issue:44561", "issue:44561|issue:44779", "issue:43335|issue:43653", "issue:43819|issue:44493", - "issue:43550|issue:44112", - "issue:45310|issue:45478" + "issue:43550|issue:44112" ], - "prepared_review_unit_hash": "9dacc5c6ab27afcfa8d06b77eea67f8a875deb0a812972c1b71bc562cb67dff4", + "prepared_review_unit_hash": "86a59824798bcf2f445aa873cef843216448d8645270fc80e93d3f6df34eb986", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37251, - "estimated_input_tokens": 9313, - "estimated_eval_tokens": 18882 + "serialized_chars": 37293, + "estimated_input_tokens": 9324, + "estimated_eval_tokens": 18904 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a collection of unrelated issue reports. Only the two Qwen3.5 MoE from_pretrained reports look like the same underlying regression; the rest are similar in theme at most, not true duplicates.", - "confidence": 0.82, - "canonical_issue_reason": "issue:45478 is the clearest canonical anchor for the only true duplicate pair: it reports the same Qwen3.5 MoE from_pretrained failure as issue:45310, with the later report offering the same concrete bug signature.", + "summary": "The cluster is heterogeneous with no clear duplicate pair. Most items are isolated bug reports or docs regressions across different models/APIs, so I would not merge any soft-edge candidates.", + "confidence": 0.31, + "canonical_issue_reason": "issue:44521 is the strongest representative because it has the most discussion/inbound references and centers on a high-traffic API (`apply_chat_template`) with a concrete multimodal bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45478 is the best issue to keep as the anchor here because it is the duplicate pair\u2019s later, clearer report of the same concrete failure mode, making it the strongest representative issue in this set.", + "best_issue_reason": "issue:44521 is the best standalone issue in this set: concrete, user-impacting, and well-documented compared with the more niche or documentation-only reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44514", - "right": "issue:44521", + "left": "issue:44442", + "right": "issue:45198", "accept": false, - "reason": "Both involve apply_chat_template on multimodal inputs, but one is a batched padding=False crash and the other is broken assistant_masks; different failure modes." + "reason": "Different tokenizer families and different failures: AutoTokenizer loading vs save_pretrained/tokenization in Wav2Vec2." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "One is a dtype mismatch between training and eval; the other is a Qwen3 MoE memory-usage issue. Same broad area, not the same bug." + "reason": "One is a dtype regression during evaluate; the other is MoE training memory usage. Not the same bug." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both involve `apply_chat_template`, but one is a batching crash and the other is incorrect assistant masks; likely separate fixes." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Tokenizer special-token registration and a GraniteSpeech CI device-override test are unrelated." + "reason": "Tokenizer special-token registration vs a GraniteSpeech CI device-override failure; unrelated." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs complaints around pipelines, but they target different documentation and behaviors." + "reason": "Both are docs/deprecation complaints, but they target different pipeline behaviors and different APIs." }, { "left": "issue:43450", "right": "issue:43575", "accept": false, - "reason": "Video processor batching shape bug vs model loading OOM; no shared underlying defect." + "reason": "Video processor shape bug and Qwen2-57B TP OOM are unrelated." }, { "left": "issue:43450", "right": "issue:43493", "accept": false, - "reason": "Different vision-side problems: batched video processor shape vs SigLIP2 implementation discrepancy." + "reason": "Batched video shape issue vs SigLIP2 implementation discrepancy; different code paths." }, { "left": "issue:44206", "right": "issue:45584", "accept": false, - "reason": "Audio feature-extractor arg mismatch and Whisper empty-transcription failure are distinct issues." + "reason": "Feature-extractor arg regression vs Whisper empty-transcription failure; not the same underlying defect." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "LoRA/tied-embeddings collapse and a categories-limit runtime error are unrelated." + "reason": "LoRA/tied-embeddings model collapse is unrelated to the 2^24 category limit error." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "DSA indexer activation-function discussion and a v5 trust_remote_code breakage are different code paths." + "reason": "DSA indexer activation issue vs removal of `is_torch_fx_available`; no shared bug." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "DAC latent reconstruction mismatch and a qwen3_omni_moe helper-length issue are not the same bug." + "reason": "DAC latent mismatch vs Qwen3 Omni MoE helper length behavior; different model components." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio apply_chat_template sampling-rate default vs video batch-shape handling; different subsystems and defects." + "reason": "Audio chat-template sampling-rate default vs batched video input shape bug; different subsystems." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "DAC forward/latents mismatch and GraniteSpeech CI test flakiness are unrelated." + "reason": "Different problems entirely: DAC latent reconstruction vs GraniteSpeech CI override." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "Processor output-size mismatch and an AMP/CUDA loss crash are different problems." + "reason": "Processor input-size mismatch vs image loss AMP crash; not the same issue." }, { "left": "issue:43278", "right": "issue:43825", "accept": false, - "reason": "Embedding dtype regression and a pipeline v5 error-message issue are unrelated." + "reason": "Training/eval dtype regression is unrelated to the pipeline translation-error messaging bug." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "SwitchTransformers sparse-layer configuration bug and Bamba torch.compile/SDPA failure are different code paths." + "reason": "SwitchTransformers sparse-layer config bug vs Bamba torch.compile/SDPA failure; different code paths." }, { - "left": "issue:37428", - "right": "issue:43425", + "left": "issue:43606", + "right": "issue:45083", "accept": false, - "reason": "Importing a removed flash-attention helper and Torch 2.10 incompatibility are not the same underlying change." + "reason": "CPU offload device mismatch is unrelated to the Qwen3 Omni helper output-length issue." }, { - "left": "issue:43606", - "right": "issue:45083", + "left": "issue:37428", + "right": "issue:43425", "accept": false, - "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length behavior are unrelated." + "reason": "Import error around a flash-attention helper vs broad Torch 2.10 incompatibility; too general to merge." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "A mixed-precision crash in LwDetrImageLoss is not the same as D-FINE missing auxiliary losses." + "reason": "Loss crash under AMP vs D-FINE auxiliary-loss omission; separate training bugs." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Pegasus v4/v5 output drift and qwen3_5 recurrent-state reset are distinct regressions." + "reason": "Pegasus v4/v5 output drift vs Qwen3 recurrent-state reset; different model behaviors." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Tool-call chat templating crash and Qwen2.5-VL video position-id bug are different multimodal failures." + "reason": "Chat-template tool-call crash vs video vision position-id bug; unrelated multimodal issues." }, { "left": "issue:45200", "right": "issue:45588", "accept": false, - "reason": "Gemma4 token-type defaults and a flash_attention sinkless-model AttributeError are unrelated." + "reason": "Gemma4 mm_token_type_ids defaulting vs flash_attention sinkless-model AttributeError; different failures." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 test failures and Trainer GPU broadcast NoneType errors are not the same bug." + "reason": "Flash Attention test failure vs Trainer `_gpu_broadcast_one` NoneType error; no common bug." }, { "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "SigLIP2 docs/example issues and BeitImageProcessorFast label reduction are different problems." + "reason": "SigLIP2 docs/model mismatch vs BEiT Fast image processor label reduction bug; unrelated." }, { - "left": "issue:37428", - "right": "issue:44351", + "left": "issue:43611", + "right": "issue:43931", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols from different modules." + "reason": "Base-model-prefix loading breakage vs Qwen3-VL weight-shape mismatch; different model-loading problems." }, { - "left": "issue:43611", - "right": "issue:43931", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "General base_model_prefix loading breakage and a specific Qwen3-VL shape mismatch are not the same defect." + "reason": "Both are import errors, but for different missing symbols and different root causes." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "Both mention device-related problems, but one is an audio extractor arg issue and the other is a Gemma4 PLE forward mismatch." + "reason": "Both mention device mismatch, but one is a feature extractor crash and the other is a Gemma4 forward-pass placement issue." }, { "left": "issue:44462", "right": "issue:44561", "accept": false, - "reason": "Tokenizer file selection and removal of is_torch_fx_available are unrelated." + "reason": "Tokenizer repository resolution bug vs removal of `is_torch_fx_available`; unrelated." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "One is a v5 trust_remote_code breakage, the other is a Deepseek tokenizer regression; different failures." + "reason": "Trust-remote-code breakage from API removal vs Deepseek tokenizer regression; different causes." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and BigBirdTokenizer special-token registration are unrelated." + "reason": "Sparse-layer config bug vs BigBird tokenizer special-token registration; no overlap." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "DAC latent mismatch and unexpected position-id keys across many models are not the same issue." + "reason": "DAC latent reconstruction bug vs widespread position-id key warnings; unrelated." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba and GraniteSpeech CI stale device override are different bugs." - }, - { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "These are the same reported regression: Qwen3.5 MoE from_pretrained fails on transformers>=5.4.0, with matching symptom and scope." + "reason": "Bamba SDPA/compile failure vs GraniteSpeech CI stale override test; different failures." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is well-grounded in the packet and stays conservative. The only clearly duplicated pair is the identical Qwen3.5 MoE from_pretrained reports (45310/45478), and the remaining soft-edge calls are reasonably rejected as different bugs or only loosely related themes. No evidence is overstated." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -7651,12 +7649,10 @@ "issue:33357", "issue:34689", "issue:35141", - "issue:37428", "issue:38175", "issue:39692", "issue:40444", "issue:43335", - "issue:43412", "issue:43504", "issue:43526", "issue:43550", @@ -7676,13 +7672,14 @@ "issue:44855", "issue:44857", "issue:44977", - "issue:44987", "issue:45072", "issue:45083", "issue:45198", "issue:45229", + "issue:45310", "issue:45372", "issue:45459", + "issue:45478", "issue:45491", "issue:45520", "issue:45563", @@ -7694,11 +7691,12 @@ "issue:45859" ], "soft_pairs": [ + "issue:45310|issue:45478", "issue:43335|issue:43504", - "issue:44855|issue:45083", "issue:45588|issue:45859", "issue:29127|issue:32090", "issue:43335|issue:45083", + "issue:44855|issue:45083", "issue:39692|issue:44493", "issue:44292|issue:45520", "issue:44448|issue:44779", @@ -7722,252 +7720,54 @@ "issue:29942|issue:35141", "issue:45459|issue:45712", "issue:43550|issue:45083", - "issue:45372|issue:45588", "issue:43873|issue:44493", - "issue:37428|issue:44987", - "issue:32090|issue:33357", - "issue:43412|issue:45588", - "issue:43873|issue:45229" + "issue:45372|issue:45588", + "issue:32090|issue:33357" ], - "prepared_review_unit_hash": "c0f8f6baf347c33257e26ab8f7e9da82984d49f0996a51ff5c491ac4aa0d45c8", + "prepared_review_unit_hash": "459d6727d4e679a87b44677636080abf2e73ea25d105ea020b08ca808ba9a631", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 38155, - "estimated_input_tokens": 9539, - "estimated_eval_tokens": 19334 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36930, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The soft pairs are mostly superficial keyword matches across unrelated bug reports. I don\u2019t see any true duplicate pair here, and the cluster is too heterogeneous to justify a single strong canonical issue.", - "confidence": 0.97, - "canonical_issue_reason": "No strong canonical issue stands out: the items span unrelated models, subsystems, and failure modes, so any single issue choice would be arbitrary.", + "summary": "The Qwen3.5 MoE pair (#45310/#45478) still looks like a true duplicate: same title, same model family, and same from_pretrained failure. The earlier issue #45310 is the better anchor, but not because it is open; it is simply the earlier report and the more suitable representative.", + "confidence": 0.9, + "canonical_issue_reason": "#45310 is the earlier report for the same Qwen3.5 MoE from_pretrained failure, so it is the best issue anchor without relying on issue state.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue stands out; this is a grab-bag of distinct bugs rather than one underlying defect.", + "best_issue_reason": "#45310 best represents the duplicate pair because it is the first report of the same concrete loading bug and has the broader discussion trail.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43335", - "right": "issue:43504", - "accept": false, - "reason": "Different bugs: one is SwitchTransformers sparse-layer construction, the other is BEiT pretrained preset loading/legacy fields." - }, - { - "left": "issue:44855", - "right": "issue:45083", - "accept": false, - "reason": "One is a Python 3.13 import/IndentationError in DebertaV2; the other is qwen3_omni_moe feature-length helper behavior." - }, - { - "left": "issue:45588", - "right": "issue:45859", - "accept": false, - "reason": "Both mention attention/offload, but one is a flash-attention integration AttributeError and the other is a rotary-embedding CPU offload incompatibility." - }, - { - "left": "issue:29127", - "right": "issue:32090", - "accept": false, - "reason": "Unrelated: LayoutLMv3 error-message clarity vs Trainer failing to broadcast NoneType to GPU." - }, - { - "left": "issue:43335", - "right": "issue:45083", - "accept": false, - "reason": "No shared underlying defect; SwitchTransformers config creation and qwen3_omni_moe length computation are different code paths." - }, - { - "left": "issue:39692", - "right": "issue:44493", - "accept": false, - "reason": "Docs/model mismatch and quantization failure are not the same as the generic position-id key warning." - }, - { - "left": "issue:44292", - "right": "issue:45520", - "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure is unrelated to the Python 3.13 flash_attn import KeyError." - }, - { - "left": "issue:44448", - "right": "issue:44779", - "accept": false, - "reason": "Pegasus output drift between versions and Deepseek tokenizer correctness regressions are different tokenizer/generation bugs." - }, - { - "left": "issue:38175", - "right": "issue:44493", - "accept": false, - "reason": "One is zero probabilities in SigLIP2 inference; the other is a broad unexpected position-id key issue." - }, - { - "left": "issue:43335", - "right": "issue:45072", - "accept": false, - "reason": "Different failures: sparse-layer config bug vs dtype mismatches in bfloat16 inference." - }, - { - "left": "issue:43335", - "right": "issue:45593", - "accept": false, - "reason": "SwitchTransformers sparse-layer creation and D-FINE auxiliary-loss behavior are unrelated." - }, - { - "left": "issue:43720", - "right": "issue:44112", - "accept": false, - "reason": "BitNet packed-weight loading and a stale GraniteSpeech CI test do not share the same failure mode." - }, - { - "left": "issue:44977", - "right": "issue:45229", - "accept": false, - "reason": "Qwen3.5 flash-attention generation issues and Gemma4 multi-GPU OOM are different problems." - }, - { - "left": "issue:40444", - "right": "issue:43526", - "accept": false, - "reason": "Multiple-images-per-prompt dataset handling is unrelated to BeitImageProcessorFast reduce_labels returning only one label." - }, - { - "left": "issue:43723", - "right": "issue:44554", - "accept": false, - "reason": "Tokenizer loading in v5 and an MPS attention correctness issue are different subsystems and code paths." - }, - { - "left": "issue:34689", - "right": "issue:35141", - "accept": false, - "reason": "Llama 3.2 Vision model loading breakage is not the same as output embeddings being reinitialized after resizing token embeddings." - }, - { - "left": "issue:43606", - "right": "issue:44857", - "accept": false, - "reason": "CPU offload device mismatch and a float16 AMP crash in LwDetrImageLoss are unrelated runtime errors." - }, - { - "left": "issue:44448", - "right": "issue:44977", - "accept": false, - "reason": "Pegasus v4/v5 output differences and Qwen3.5 flash-attention generation failures are not the same bug." - }, - { - "left": "issue:44857", - "right": "issue:45072", - "accept": false, - "reason": "AMP/CUDA loss crash and bfloat16 inference dtype mismatches affect different components and symptoms." - }, - { - "left": "issue:44568", - "right": "issue:45491", - "accept": false, - "reason": "Tokenizer special-token insertion and Gemma3 NaN embeddings from mixed-length batching are unrelated." - }, - { - "left": "issue:28218", - "right": "issue:32090", - "accept": false, - "reason": "Tokenizer spacing after added tokens is unrelated to Trainer GPU broadcast failures." - }, - { - "left": "issue:43927", - "right": "issue:45698", - "accept": false, - "reason": "Both involve save/load behavior, but one loses custom token IDs while the other loads the wrong custom module; not the same defect." - }, - { - "left": "issue:45198", - "right": "issue:45803", - "accept": false, - "reason": "Wav2Vec2 save_pretrained/tokenization failure is not the same as a bare except swallowing real errors in FuyuBatchFeature." - }, - { - "left": "issue:44977", - "right": "issue:45563", - "accept": false, - "reason": "Flash-attention generation breakage and a stale warning about num_return_sequences are clearly different issues." - }, - { - "left": "issue:43701", - "right": "issue:45083", - "accept": false, - "reason": "Checkpoint key mismatch during resume is unrelated to qwen3_omni_moe helper output behavior." - }, - { - "left": "issue:29942", - "right": "issue:35141", - "accept": false, - "reason": "Flash Attention 2 test failures and token-embedding resize reinitialization do not describe the same bug." - }, - { - "left": "issue:45459", - "right": "issue:45712", - "accept": false, - "reason": "Tokenizer error masking due to protobuf handling is unrelated to leftover dummy classes in dummy_pt_objects.py." - }, - { - "left": "issue:43550", - "right": "issue:45083", - "accept": false, - "reason": "Bamba torch.compile/SDPA failure and qwen3_omni_moe length helper behavior are different code-path problems." - }, - { - "left": "issue:45372", - "right": "issue:45588", - "accept": false, - "reason": "Gemma4 processor import failure and a flash_attention NoneAttribute crash are distinct loading vs runtime bugs." - }, - { - "left": "issue:43873", - "right": "issue:44493", - "accept": false, - "reason": "Quantization/offloading behavior is not the same as unexpected position-id key warnings." - }, - { - "left": "issue:37428", - "right": "issue:44987", - "accept": false, - "reason": "Missing flash_supports_window_size import and a model-loading failure for physical-intelligence/fast are unrelated." - }, - { - "left": "issue:32090", - "right": "issue:33357", - "accept": false, - "reason": "Trainer NoneType broadcast failure and MacOS bus error on a CLIP model are different failures." - }, - { - "left": "issue:43412", - "right": "issue:45588", - "accept": false, - "reason": "Gemma3n executorch export issues and a sink-less flash_attention AttributeError are separate problems." - }, - { - "left": "issue:43873", - "right": "issue:45229", - "accept": false, - "reason": "Quantization/offloading concerns are not the same as Gemma4 multi-GPU CUDA OOM." + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same underlying bug: both report transformers>=5.4.0 Qwen3.5 MoE from_pretrained failures for the same model family and code path." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate judgment for #45310/#45478 is plausible, but the analyst overstates key facts from the packet. #45310 is not open in the provided data, so the rationale that it is the \u201cearlier, open report\u201d and the \u201cbest representative\u201d is not grounded. Because the canonical-selection reasoning relies on that incorrect state claim, the response should be rejected." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -7992,18 +7792,18 @@ "issue:29942", "issue:30064", "issue:32090", + "issue:37428", "issue:41720", "issue:42915", "issue:42994", "issue:43065", "issue:43066", - "issue:43257", "issue:43278", "issue:43329", "issue:43381", - "issue:43421", + "issue:43412", "issue:43653", - "issue:43756", + "issue:43873", "issue:43976", "issue:43994", "issue:44186", @@ -8019,6 +7819,7 @@ "issue:44857", "issue:44871", "issue:44977", + "issue:44987", "issue:45198", "issue:45229", "issue:45245", @@ -8027,13 +7828,15 @@ "issue:45412", "issue:45440", "issue:45459", - "issue:45491", "issue:45520", "issue:45563", "issue:45588", "issue:45823" ], "soft_pairs": [ + "issue:37428|issue:44987", + "issue:43412|issue:45588", + "issue:43873|issue:45229", "issue:45245|issue:45563", "issue:43278|issue:43329", "issue:41720|issue:45229", @@ -8045,221 +7848,224 @@ "issue:43065|issue:43994", "issue:43066|issue:44462", "issue:44857|issue:45588", - "issue:43065|issue:45412", "issue:44448|issue:44466", "issue:43278|issue:43381", + "issue:43065|issue:45412", "issue:43653|issue:44743", "issue:45588|issue:45823", "issue:25251|issue:32090", "issue:44186|issue:45198", - "issue:43994|issue:45412", "issue:44206|issue:44479", + "issue:43994|issue:45412", "issue:44451|issue:45520", "issue:44361|issue:45198", "issue:44871|issue:44977", "issue:42915|issue:42994", "issue:44265|issue:45198", - "issue:29942|issue:30064", - "issue:43756|issue:45440", - "issue:43421|issue:45491", - "issue:43257|issue:43278" + "issue:29942|issue:30064" ], - "prepared_review_unit_hash": "df45945ffe9ef777b4dc850e38397735fe0151fc20c576e9cce2d03a44a7f7c3", + "prepared_review_unit_hash": "8d78d47bab9b9743fc316086ec9a5485a218973e891e16ebf51d2e137fd2bd7a", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36447, - "estimated_input_tokens": 9112, - "estimated_eval_tokens": 18480 + "serialized_chars": 36377, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items do not form a single duplicate cluster; they span many unrelated tokenizer, generation, model-loading, vision, and Gemma/Qwen-specific issues. None of the soft pairs look like the same underlying bug or change, so all are rejected.", - "confidence": 0.95, - "canonical_issue_reason": "No clear canonical issue: the set is heterogeneous and the soft pairs do not converge on one shared bug report.", - "canonical_pr_reason": "No pull requests are present.", - "best_issue_reason": "No single issue is a good global representative because the items cover many distinct code paths and model families.", - "best_pr_reason": "No pull requests are present.", + "summary": "This cluster is highly heterogeneous: the soft-similarity pairs span unrelated tokenizer, export, quantization, flash-attention, model-loading, and training issues. None look like the same underlying bug/change, so all soft edges should be rejected and there is no meaningful duplicate canonical for the cluster.", + "confidence": 0.96, + "canonical_issue_reason": "No strong canonical issue: the items do not form a single duplicate set, so there is no sensible anchor issue for this cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global representative because the cluster is a mix of unrelated regressions across different subsystems and models.", + "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "One is a flash-attention import error; the other is a model loading failure. Different failure modes and code paths." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n export with erfinv/training-guard issues is unrelated to flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading behavior and Gemma4 multi-GPU CUDA OOM are different problems, not the same bug." + }, { "left": "issue:45245", "right": "issue:45563", "accept": false, - "reason": "Different problems: category cardinality limit during modeling vs a stale generate() warning." + "reason": "Category-count limits in image handling are unrelated to a stale generate warning about num_return_sequences." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "Unrelated bugs: dtype drift in eval vs a missing video-path helper in multimodal token counting." + "reason": "A dtype regression between train/eval is unrelated to a multimodal token-counting bug in video branches." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Both involve large-model inference, but one is a cuda assert in Qwen3 auto-mapping and the other is Gemma4 multi-GPU OOM." + "reason": "Both involve GPU memory/runtime failures, but one is Qwen3 device-mapping assert and the other is Gemma4 OOM; not the same code path." }, { "left": "issue:10000", "right": "issue:15354", "accept": false, - "reason": "Completely different model/framework failures: DistilBertModel issue vs torch.jit GeneratorExp export support." + "reason": "German DistilBertModel issue and torch.jit generator-exp export failure are unrelated." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Different code paths and symptoms: Python version compatibility vs DeepSeekV3MoE parity divergence." + "reason": "Python-version compatibility and DeepSeekV3MoE divergence are distinct bugs." }, { "left": "issue:28218", "right": "issue:29127", "accept": false, - "reason": "Tokenizer spacing behavior vs LayoutLMv3 box validation error; not the same bug." + "reason": "Tokenizer spacing after added tokens is unrelated to a LayoutLMv3 error-message clarity issue." }, { "left": "issue:45372", "right": "issue:45406", "accept": false, - "reason": "Both touch Gemma4 processing, but one is an import dependency break and the other is a missing _tokenizer attribute in serve." + "reason": "Gemma4 processor import failure and Gemma4Processor missing _tokenizer during serve are different failures." }, { "left": "issue:45372", "right": "issue:45459", "accept": false, - "reason": "Unrelated: Gemma4 processor loading import failure vs tokenizer error masking when protobuf is absent." + "reason": "A mistral_common import breakage is unrelated to tokenizer error masking when protobuf is absent." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "Different models and failures: dummy Conv2d in Sam3PixelDecoder vs bad outputs from siglip2 AutoModel/pipeline." + "reason": "Dummy Conv2d in Sam3PixelDecoder and SigLIP2 nonsensical outputs are unrelated model-specific bugs." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Both concern tokenizer loading/serialization, but one is a decoder-type mismatch in v5 and the other ignores tokenizer.json from the repo." + "reason": "Wrong tokenizer decoder type and AutoTokenizer ignoring tokenizer.json are both tokenizer-related but clearly different defects." }, { "left": "issue:44857", "right": "issue:45588", "accept": false, - "reason": "Different runtime failures: LwDetrImageLoss AMP/CUDA crash vs flash_attention.py handling of s_aux=None." - }, - { - "left": "issue:43065", - "right": "issue:45412", - "accept": false, - "reason": "No common bug: Sam3PixelDecoder dummy layer vs RT-DETR memory not released on deletion." + "reason": "LwDetrImageLoss AMP crash and flash_attention s_aux=None crash are different code paths and symptoms." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "Both are serialization/output regressions, but one is Pegasus generation drift and the other is lm_head.weight tied-weight serialization." + "reason": "Pegasus output drift across versions is unrelated to lm_head.weight serialization inconsistency." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Different behavior classes: embedding dtype mismatch vs gradient checkpointing restriction in eval mode." + "reason": "Embedding dtype regression and gradient-checkpointing-in-eval failure are separate behaviors." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "A placeholder Conv2d implementation is unrelated to RT-DETR memory not being released after deletion." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "Tokenizer special-token registration vs Qwen3.5 recurrent-state cache reset; unrelated." + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated." }, { "left": "issue:45588", "right": "issue:45823", "accept": false, - "reason": "Both mention Gemma4/attention-related behavior, but one is a flash-attention attribute crash and the other is a PLE device mismatch." + "reason": "flash_attention s_aux=None crash and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:25251", "right": "issue:32090", "accept": false, - "reason": "Pipeline top_k nesting bug vs Trainer GPU broadcast NoneType error; unrelated." + "reason": "Pipeline top_k nesting behavior and Trainer _gpu_broadcast_one NoneType error are unrelated." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "Different tokenizer/model families and failure modes: LayoutLMv2 NER padding crash vs Wav2Vec2 save/tokenization failure." + "reason": "LayoutLMv2 tokenizer crashes and Wav2Vec2 save/tokenization failures are different tokenizer/model issues." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Feature extractor unsupported-argument crash vs Qwen video-input regression; not the same underlying bug." + "reason": "Unsupported center arg in LasrFeatureExtractor and Qwen2.5/3 VL video regression do not share the same root cause." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 incorrect outputs and RT-DETR memory leak are unrelated." }, { "left": "issue:44451", "right": "issue:45520", "accept": false, - "reason": "Different subsystem and symptom: ScandiBERT loading failure vs flash_attn import KeyError on Python 3.13." + "reason": "Loading ScandiBERT and a flash_attn import KeyError on Python 3.13 are different issues." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "Both are tokenizer-related, but MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure are distinct." + "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure are unrelated." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Different models and issues: Gemma-3 eos_token_id inconsistency vs Qwen3.5 flash-attention generation regression." + "reason": "Gemma eos_token_id config mismatch is not the same as Qwen3.5 flash-attention generation regression." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "Separate problems: Qwen3 MoE weight conversion/loading vs quantized model save failure." + "reason": "Qwen3Moe FineGrainedFP8 failure and quantized model saving failure are different quantization-related bugs." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "Unrelated: torch.export with torch_compilable_check vs Wav2Vec2 save/tokenization failure." + "reason": "torch.export failing on torch_compilable_check and Wav2Vec2 save/tokenization failure are unrelated." }, { "left": "issue:29942", "right": "issue:30064", "accept": false, - "reason": "Different test and model areas: Flash Attention 2 failures vs image processor void segmentation-map handling." - }, - { - "left": "issue:43756", - "right": "issue:45440", - "accept": false, - "reason": "Both are model-architecture issues, but Smollm3 RoPE-layer dropping and DeepSeekV3MoE divergence are distinct." - }, - { - "left": "issue:43421", - "right": "issue:45491", - "accept": false, - "reason": "Tokenizer post-processor update bug vs Gemma3 NaN embeddings from sliding-window attention; different code paths." - }, - { - "left": "issue:43257", - "right": "issue:43278", - "accept": false, - "reason": "Different issues: Qwen3 MoE loading/conversion vs embedding dtype mismatch during evaluation." + "reason": "Failing Flash Attention 2 tests and void segmentation-map processing are unrelated test/product bugs." } ] }, @@ -8294,16 +8100,19 @@ "issue:39692", "issue:42915", "issue:43232", + "issue:43257", "issue:43262", + "issue:43278", "issue:43334", "issue:43381", "issue:43388", + "issue:43421", "issue:43441", "issue:43526", "issue:43531", - "issue:43575", "issue:43653", "issue:43701", + "issue:43756", "issue:43761", "issue:43824", "issue:43931", @@ -8314,8 +8123,6 @@ "issue:44442", "issue:44464", "issue:44610", - "issue:44743", - "issue:44811", "issue:44857", "issue:44977", "issue:45072", @@ -8332,16 +8139,19 @@ "issue:45907" ], "soft_pairs": [ + "issue:43756|issue:45440", + "issue:43257|issue:43278", + "issue:43421|issue:45491", "issue:33357|issue:39692", "issue:29127|issue:34567", - "issue:44610|issue:45412", "issue:32090|issue:33290", + "issue:44610|issue:45412", "issue:45103|issue:45372", "issue:45440|issue:45593", "issue:43824|issue:44279", "issue:43653|issue:44442", - "issue:43761|issue:45859", "issue:43381|issue:43701", + "issue:43761|issue:45859", "issue:43526|issue:45834", "issue:29942|issue:36010", "issue:43526|issue:44186", @@ -8356,200 +8166,204 @@ "issue:43526|issue:44857", "issue:43232|issue:43388", "issue:43441|issue:45588", - "issue:44062|issue:45907", - "issue:44743|issue:44811", - "issue:43575|issue:43701" + "issue:44062|issue:45907" ], - "prepared_review_unit_hash": "c2f6f99d6f20d6fbbb4e70b360e134997d7585df1c0eb363dfb852f9aecdf1ee", + "prepared_review_unit_hash": "018ca9623e7ca17f1c06409133fb7a8607d7430f76edf3a7beda930d4cc88baa", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 27, - "serialized_chars": 35306, - "estimated_input_tokens": 8827, - "estimated_eval_tokens": 17910 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36212, + "estimated_input_tokens": 9053, + "estimated_eval_tokens": 18362 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly soft-similarity noise: the issues share generic Transformers themes, but they describe different models, code paths, and failure modes. None of the proposed soft pairs look like the same underlying bug or change.", - "confidence": 0.89, - "canonical_issue_reason": "issue:45103 is the strongest anchor: it is open, has inbound references, and describes a concrete, reproducible crash in a shared utility. The rest of the set is heterogeneous and not duplicate-compatible.", + "summary": "This cluster is mostly heterogeneous: the paired items share broad transformer-adjacent vocabulary but do not describe the same underlying bug or change. I would not collapse these into one duplicate set; each soft pair is a false positive.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 43441 is the clearest self-contained bug report in the set, with a specific model, version context, and runtime failure path. It is a reasonable anchor if one issue must be chosen, but it does not represent the rest of the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45103 is the most suitable representative issue overall because it is active, specific, and likely to attract/fix follow-up work. It is still not a duplicate of the other reports.", + "best_issue_reason": "Issue 43441 is the best single issue artifact here because it is concrete and actionable, but only for its own flash-attention failure; the cluster as a whole is too mixed for a broader canonical issue.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Both mention MoE/DeepSeek, but one is about dropped RoPE layers while the other is about native-vs-remote implementation divergence; not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43278", + "accept": false, + "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed is unrelated to an embedding dtype change between training and evaluation." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime post-processor updates for special tokens are unrelated to Gemma3 NaNs from sliding-window attention and all-padding windows." + }, { "left": "issue:33357", "right": "issue:39692", "accept": false, - "reason": "Both involve vision/processor/model loading problems, but one is a MacOS bus error and the other is a doc example mismatch/quantization failure; different bugs and code paths." + "reason": "A MacOS bus error in CLIP loading is not the same as documentation example errors for SigLIP2." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "One is a LayoutLMv3 box-content error message issue; the other is TrainerState token counting not updating. No shared failure mode." + "reason": "LayoutLMv3 error-message clarity and TrainerState token-count updates are different problems in different code paths." }, { - "left": "issue:44610", - "right": "issue:45412", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "OmDet processor/model input-size mismatch vs RT-DETR memory not released; unrelated symptoms and fixes." + "reason": "A NoneType broadcast crash in Trainer is not the same as an Adafactor/deepspeed OOM issue." }, { - "left": "issue:32090", - "right": "issue:33290", + "left": "issue:44610", + "right": "issue:45412", "accept": false, - "reason": "Trainer GPU broadcast NoneType error and Deepspeed Adafactor OOM are different training stack failures." + "reason": "Processor/model input-size mismatch is unrelated to RT-DETR memory not being released on deletion." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "Both are import/processing regressions, but one is an auto-docstring annotations crash and the other is a mistral_common import breakage in Gemma processor loading." + "reason": "auto_docstring annotation handling and Gemma 4 processor loading/import errors are different failures." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "DeepseekV3MoE divergence from remote implementation is a model parity bug; D-FINE missing auxiliary losses is a training-loss configuration issue." + "reason": "DeepSeekV3 implementation divergence does not match D-FINE's missing auxiliary losses when denoising is off." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "Importing Qwen2.5-VL and a generic dependency issue are not the same bug; the latter is too broad and unspecified." + "reason": "A missing Qwen2.5-VL import and a generic dependency issue are too different to be duplicates." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "BigBirdTokenizer special-token registration and AutoTokenizer loading FastSpeech2ConformerTokenizer are separate tokenizer bugs." + "reason": "BigBird mask-token special registration and FastSpeech2Conformer tokenizer loading are different tokenizer bugs." }, { - "left": "issue:43761", - "right": "issue:45859", + "left": "issue:43381", + "right": "issue:43701", "accept": false, - "reason": "CLIPVisionModel hidden_states regression and Qwen3_5Moe CPU offload incompatibility affect different model internals." + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate trainer behaviors." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:43761", + "right": "issue:45859", "accept": false, - "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are unrelated training workflow issues." + "reason": "CLIPVision hidden_states regression and Qwen3_5Moe CPU-offload incompatibility affect different models and code paths." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug and Kosmos2.5 OCR index error are different preprocessing/model bugs." + "reason": "BeitImageProcessorFast reduce_labels behavior is unrelated to Kosmos2.5 OCR long-input index errors." }, { "left": "issue:29942", "right": "issue:36010", "accept": false, - "reason": "Flash Attention 2 test failures are a CI/test regression; GenerationMixin import failure is an API/package import problem." + "reason": "FlashAttention 2 test failures and a GenerationMixin import error are distinct regressions." }, { "left": "issue:43526", "right": "issue:44186", "accept": false, - "reason": "Both are tokenizer/processor-area bugs, but one is BeitImageProcessorFast label handling and the other is LayoutLMv2Tokenizer crashing on NER padding/truncation." + "reason": "Both are tokenizer/processor-adjacent, but one is a label-reduction bug and the other is a LayoutLMv2Tokenizer crash on NER/padding." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 box-content error messaging and CLIP MacOS bus error are completely different model/runtime problems." + "reason": "An error-message issue in LayoutLMv3 is not the same as a CLIP MacOS bus error." }, { "left": "issue:43334", "right": "issue:43531", "accept": false, - "reason": "Qwen3-VL pad_token_id loading failure and Qwen3-MoE sliding_window issue are both Qwen-family issues but on different configs and code paths." + "reason": "Missing pad_token_id on Qwen3-VL text config is unrelated to sliding_window behavior in Qwen3-MoE." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "LayoutLMv3 validation messaging and embedding reinitialization after resize are unrelated model setup bugs." + "reason": "LayoutLMv3 error messaging and token-embedding reinitialization after resize are different bugs." }, { "left": "issue:44464", "right": "issue:45491", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward and Gemma3 NaN embeddings from sliding-window all-padding windows are different generation vs attention-path issues." + "reason": "Chunked generation inconsistencies with compiled forward are not the same as Gemma3 NaNs from sliding-window attention." }, { "left": "issue:44977", "right": "issue:45245", "accept": false, - "reason": "Qwen3.5 flash-attention generation failure is model/runtime-specific; the 2^24 category limit is a separate embedding/category-size constraint." + "reason": "Qwen3.5 flash-attention generation issues are unrelated to the category-count runtime limit." }, { "left": "issue:42915", "right": "issue:43931", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen3-VL weight-shape mismatch are different Qwen model loading regressions." + "reason": "FineGrainedFP8Config failures for Qwen3Moe and weight-shape mismatches for Qwen3-VL are different model-loading bugs." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "AutoTokenizer loading FastSpeech2ConformerTokenizer and dtype mismatches in SwitchTransformers/TimmWrapperModel are not the same failure." + "reason": "Tokenizer loading failure and bfloat16 dtype mismatches affect different subsystems and failure modes." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Audio chat-template sample-rate defaulting and LasrFeatureExtractor center-arg crash are unrelated processor API bugs." + "reason": "Audio chat-template sampling-rate defaults and a feature-extractor unsupported-argument crash are not the same bug." }, { "left": "issue:43526", "right": "issue:44857", "accept": false, - "reason": "BeitImageProcessorFast label reduction and LwDetrImageLoss AMP/CUDA crash affect different components and are not mergeable as one fix." + "reason": "BeitImageProcessorFast label reduction and LwDetrImageLoss AMP/CUDA crashes are unrelated." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs update after sync_gpus and gather_for_metrics dropping tuple label elements are separate distributed/runtime bugs." + "reason": "Generation kwargs update after sync_gpus and metric-gathering label truncation are separate distributed-training issues." }, { "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "Ministral flash-attention failure and flash_attention.py crashing on s_aux=None are both flash-attention-related, but they target different models and failure conditions." + "reason": "Both involve flash attention, but one is a Ministral-3 failure in Transformers v5 RC while the other is an s_aux=None crash for sink-less models; not the same concrete bug." }, { "left": "issue:44062", "right": "issue:45907", "accept": false, - "reason": "AddedToken special kwarg duplication and list_repo_files missing revision forwarding are unrelated API bugs." - }, - { - "left": "issue:44743", - "right": "issue:44811", - "accept": false, - "reason": "Qwen3.5 modular recurrent-state reset and Whisper batch_decode skip_special_tokens handling are different model behavior issues." - }, - { - "left": "issue:43575", - "right": "issue:43701", - "accept": false, - "reason": "Qwen2-57B-A14B-Instruct tensor-parallel OOM and checkpoint key mismatch are unrelated training/loading failures." + "reason": "An AddedToken special-argument conflict is unrelated to list_repo_files failing to forward revision." } ] }, @@ -8577,7 +8391,6 @@ "issue:34567", "issue:34689", "issue:35141", - "issue:36010", "issue:36331", "issue:39692", "issue:41628", @@ -8591,12 +8404,12 @@ "issue:43611", "issue:43643", "issue:43646", + "issue:43701", "issue:43720", "issue:43723", "issue:43749", "issue:43761", "issue:43827", - "issue:43856", "issue:43873", "issue:43931", "issue:44077", @@ -8608,6 +8421,7 @@ "issue:44479", "issue:44496", "issue:44743", + "issue:44811", "issue:44871", "issue:44898", "issue:44913", @@ -8623,6 +8437,8 @@ "issue:45803" ], "soft_pairs": [ + "issue:44743|issue:44811", + "issue:43575|issue:43701", "issue:43646|issue:44077", "issue:43575|issue:44387", "issue:45003|issue:45440", @@ -8637,10 +8453,10 @@ "issue:33290|issue:34567", "issue:34689|issue:43611", "issue:39692|issue:45491", - "issue:43381|issue:43873", - "issue:43931|issue:45216", "issue:33290|issue:35141", + "issue:43381|issue:43873", "issue:33290|issue:34689", + "issue:43931|issue:45216", "issue:43493|issue:43827", "issue:43550|issue:44743", "issue:44077|issue:44479", @@ -8651,224 +8467,216 @@ "issue:43749|issue:43931", "issue:33290|issue:36331", "issue:44190|issue:44496", - "issue:43873|issue:44871", - "issue:34689|issue:36010", - "issue:43381|issue:43856" + "issue:43873|issue:44871" ], - "prepared_review_unit_hash": "514d7e0c6cbebfd0a2b48df9294fd300aa23fcb17b6e28b00e08fc473c6862ed", + "prepared_review_unit_hash": "0b13ed7c4c6876dd90cbbcff021a75a3c0d63f06522e49722a5c036e46f89f75", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37225, - "estimated_input_tokens": 9307, - "estimated_eval_tokens": 18870 + "serialized_chars": 37215, + "estimated_input_tokens": 9304, + "estimated_eval_tokens": 18864 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a noisy mix of unrelated Transformers issues spanning loading, config, tokenizer, training, quantization, and docs regressions. None of the soft pairs look like true duplicates, so there is no strong canonical item for the whole set.", - "confidence": 0.94, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a loose collection of Transformers v5 regressions around loading, config round-tripping, and serialization, but the proposed soft pairs are not true duplicates of the same underlying bug/change.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43646 is the broadest representative bug here: a v5 regression breaking custom model initialization, which aligns best with the cluster\u2019s general loading/config-compatibility theme.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:43646 is the best single issue to anchor the cluster because it captures a broad, high-impact initialization regression rather than a narrow model-specific failure.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different bugs: recurrent-state reset with cache vs Whisper decode parameter handling." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Model-load OOM with tensor parallelism is unrelated to resume_from_checkpoint key mismatches." + }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "Both concern model initialization behavior in v5, but one is a broad custom-model init regression and the other is a specific `post_init` policy change; not the same bug." + "reason": "Different components and failures: custom model init regression vs patchtsmixer post_init policy change." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "Both mention OOM, but one is TP load-time OOM for a specific model and the other is increased reserved memory under int4 quantization; different code paths." + "reason": "Both mention OOM, but one is TP model loading and the other is int4 quantization memory inflation; not the same bug." }, { "left": "issue:45003", "right": "issue:45440", "accept": false, - "reason": "`sys.modules` access safety vs DeepSeekV3 implementation divergence are unrelated bugs." + "reason": "sys.modules access safety bug is unrelated to DeepseekV3MoE implementation divergence." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "One is a CLIP hidden-states regression; the other is incorrect `ModelOutput` key assignment. Different layers of the stack." + "reason": "Specific CLIP hidden_states regression is not the same as generic ModelOutput key assignment behavior." }, { "left": "issue:43723", "right": "issue:45479", "accept": false, - "reason": "Tokenizer loading failure and degenerate sequence-classification loss are unrelated." + "reason": "Tokenizer loading in v5 and degenerate classification loss are separate issues." }, { "left": "issue:44442", "right": "issue:45803", "accept": false, - "reason": "AutoTokenizer loading and a bare-except in Fuyu tensor conversion are separate issues." + "reason": "Tokenizer loading failure is unrelated to Fuyu tensor conversion swallowing exceptions." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "CLIP hidden-states behavior and attention-kernel divergence under `torch.compile` do not describe the same bug." + "reason": "CLIP hidden_states bug and torch.compile attention-kernel divergence are different code paths." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Both are config/API complaints, but they affect different model families and different missing fields." + "reason": "Both are config-related, but they concern different models and different missing/discrepant fields." }, { "left": "issue:44479", "right": "issue:45290", "accept": false, - "reason": "Video-input regression for Qwen models is unrelated to chat-template crashes on tool-call assistant messages." + "reason": "Video-input regression and chat-template tool-call crash are not the same underlying bug." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Both involve save/load flows, but one is about saving dequantized models and the other about loading the wrong custom module; too different to merge." + "reason": "Both involve save/load flows, but one is dequantized Ministral/Devstral saving and the other is wrong custom-module loading after save_pretrained." }, { "left": "issue:44442", "right": "issue:44898", "accept": false, - "reason": "Tokenizer loading failure vs Perceiver image-classification failure are unrelated." + "reason": "Tokenizer loading failure is unrelated to Perceiver image classification with non-default resolution." }, { "left": "issue:33290", "right": "issue:34567", "accept": false, - "reason": "AdaFactor/deepspeed OOM and `num_input_tokens_seen` not updating are separate training bugs." + "reason": "AdaFactor/deepspeed OOM and num_input_tokens_seen not updating are unrelated trainer bugs." }, { "left": "issue:34689", "right": "issue:43611", "accept": false, - "reason": "A specific model-loading breakage and a generic `base_model_prefix` regression are not the same concrete problem." + "reason": "Both are model-loading regressions, but they affect different mechanisms and failure modes." }, { "left": "issue:39692", "right": "issue:45491", "accept": false, - "reason": "Docs-example errors for SigLIP2 and Gemma3 NaN embeddings are unrelated." + "reason": "Documentation/example errors for SigLIP2 are unrelated to Gemma3 NaN embeddings." }, { - "left": "issue:43381", - "right": "issue:43873", + "left": "issue:33290", + "right": "issue:35141", "accept": false, - "reason": "Gradient checkpointing in eval mode and quantization offloading are different behaviors." + "reason": "Optimizer OOM and output embedding reinitialization in post_init are distinct problems." }, { - "left": "issue:43931", - "right": "issue:45216", + "left": "issue:43381", + "right": "issue:43873", "accept": false, - "reason": "Weight-shape mismatch when loading a model and incorrect checkpoint saving are different save/load failures." + "reason": "Eval-mode gradient-checkpointing restriction is not the same as quantization offloading failure." }, { "left": "issue:33290", - "right": "issue:35141", + "right": "issue:34689", "accept": false, - "reason": "OOM during training and embedding reinitialization on resize are unrelated." + "reason": "Different bugs: optimizer memory blowup vs Llama model-loading regression." }, { - "left": "issue:33290", - "right": "issue:34689", + "left": "issue:43931", + "right": "issue:45216", "accept": false, - "reason": "Training OOM and Llama 3.2 loading regression are unrelated." + "reason": "Weight-shape mismatch at load time is separate from a save_pretrained regression." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation drift and outdated docs references to `pipeline()` are unrelated." + "reason": "SigLIP2 implementation drift is unrelated to docs still referencing removed pipelines." }, { "left": "issue:43550", "right": "issue:44743", "accept": false, - "reason": "A `torch.compile` SDPA failure in Bamba is unrelated to Qwen recurrent-state reset behavior." - }, - { - "left": "issue:44077", - "right": "issue:44479", - "accept": false, - "reason": "`post_init` API enforcement and Qwen video-input regression are not the same issue." + "reason": "torch.compile SDPA failure and recurrent-state reset with cache are different code-path bugs." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "BitNet packed-weight loading and FastSpeech2Conformer tokenizer loading are unrelated." - }, - { - "left": "issue:43316", - "right": "issue:45759", - "accept": false, - "reason": "Both are config/API mismatches, but they concern different model types and different unwrapping behavior." + "reason": "BitNet accelerated-loading unpacking bug is unrelated to tokenizer loading failure." }, { "left": "issue:41628", "right": "issue:45372", "accept": false, - "reason": "Missing `AutoImageProcessor` import and Gemma 4 processor loading failing on `ReasoningEffort` are separate import errors." + "reason": "Both are import errors, but they involve different symbols and different processor-loading paths." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "Missing fields with `trust_remote_code=True` and `rotary_pct` resetting on reload are different config serialization bugs." + "reason": "Config missing-fields on trust_remote_code load is not the same as rotary_pct not persisting on reload." }, { "left": "issue:43749", "right": "issue:43931", "accept": false, - "reason": "FSDP CPU RAM-efficient loading and Qwen3-VL weight mismatch are unrelated loading problems." + "reason": "FSDP CPU-RAM-efficient loading and Qwen3-VL weight mismatch are unrelated." }, { - "left": "issue:33290", - "right": "issue:36331", + "left": "issue:43493", + "right": "issue:43827", "accept": false, - "reason": "Training OOM and a custom trainer signature mismatch are unrelated." + "reason": "Implementation discrepancy and docs pipeline removal are unrelated." }, { - "left": "issue:44190", + "left": "issue:43749", "right": "issue:44496", "accept": false, - "reason": "Local dataset loading in an example script and an unrecognized model/config error are unrelated." - }, - { - "left": "issue:43873", - "right": "issue:44871", - "accept": false, - "reason": "Quantization offloading behavior and Gemma eos-token inconsistency are unrelated." + "reason": "Broken FSDP loading and unrecognized model config are different issues." }, { - "left": "issue:34689", - "right": "issue:36010", + "left": "issue:44190", + "right": "issue:44496", "accept": false, - "reason": "Model loading breakage and missing `GenerationMixin` import are different import/load failures." + "reason": "Example script dataset loading failure is unrelated to config/model-type recognition." }, { - "left": "issue:43381", - "right": "issue:43856", + "left": "issue:43873", + "right": "issue:44871", "accept": false, - "reason": "Gradient checkpointing eval-mode behavior and MoE training memory use are different training/runtime issues." + "reason": "Quantization offloading behavior and Gemma eos_token_id config inconsistency are separate bugs." } ] }, @@ -8896,6 +8704,8 @@ "issue:22355", "issue:25251", "issue:28282", + "issue:34689", + "issue:36010", "issue:41950", "issue:42175", "issue:42617", @@ -8908,7 +8718,6 @@ "issue:43441", "issue:43452", "issue:43475", - "issue:43504", "issue:43540", "issue:43550", "issue:43575", @@ -8920,6 +8729,7 @@ "issue:43742", "issue:43761", "issue:43825", + "issue:43856", "issue:44077", "issue:44162", "issue:44479", @@ -8927,7 +8737,6 @@ "issue:44514", "issue:44534", "issue:44617", - "issue:44734", "issue:44743", "issue:44792", "issue:44855", @@ -8937,15 +8746,15 @@ "issue:45127", "issue:45137", "issue:45200", - "issue:45406", - "issue:45584", - "issue:45698" + "issue:45584" ], "soft_pairs": [ + "issue:43381|issue:43856", + "issue:34689|issue:36010", "issue:44855|issue:45137", "issue:43381|issue:43742", - "issue:43316|issue:44877", "issue:43540|issue:44514", + "issue:43316|issue:44877", "issue:43673|issue:44743", "issue:42175|issue:43316", "issue:43404|issue:43441", @@ -8954,219 +8763,210 @@ "issue:44162|issue:44792", "issue:43388|issue:43688", "issue:43295|issue:43825", - "issue:41950|issue:43475", "issue:16998|issue:22355", "issue:44077|issue:44534", + "issue:41950|issue:43475", "issue:45127|issue:45200", "issue:44871|issue:45584", "issue:43550|issue:43701", "issue:43761|issue:44479", "issue:43606|issue:43701", "issue:43295|issue:44871", - "issue:43452|issue:44496", "issue:43575|issue:43742", + "issue:43452|issue:44496", "issue:42617|issue:43575", - "issue:43646|issue:44913", - "issue:44734|issue:45406", - "issue:43504|issue:45698", - "issue:44617|issue:45584" + "issue:43646|issue:44913" ], - "prepared_review_unit_hash": "27fc3a7707980c14717cabc8c5430ea821438962bb42fe51e5b1dce7e0dd3f75", + "prepared_review_unit_hash": "de38021d5b862389defef0917faba0f0ba4bad3cbe25b097c44ed0f146113c21", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36009, - "estimated_input_tokens": 9003, - "estimated_eval_tokens": 18262 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35060, + "estimated_input_tokens": 8765, + "estimated_eval_tokens": 17786 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of mostly unrelated Transformers issues: loading/config regressions, processor API changes, model-specific bugs, training/runtime errors, and a few packaging/import problems. None of the soft pairs look like the same concrete bug, so I reject all of them. The best representative issue is the broad custom-module load/save regression.", - "confidence": 0.36, - "canonical_issue_reason": "issue:45698 is the broadest and most representative active loading-path regression in the set, centered on custom module resolution across save_pretrained/from_pretrained.", + "summary": "This cluster is heterogeneous: mostly unrelated Transformers issues spanning model loading, config quirks, generation regressions, and pipeline/model-specific bugs. The soft matches are mostly broad topical overlap, not the same underlying defect or change.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43646 is the broadest representative of the cluster\u2019s v5 regression / model-initialization failures, though the overall set is too mixed to form a strong duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45698 has the highest discussion activity and is the clearest umbrella issue for nearby custom-loading/configuration problems.", + "best_issue_reason": "issue:43646 is the best single issue to anchor this cluster because it most generally reflects the recurring v5 breakage theme.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Different problems: gradient checkpointing in eval mode vs Qwen3 MoE training memory usage." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Unrelated failures: Llama 3.2 Vision model loading vs missing GenerationMixin import." + }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 JIT parsing/decorator syntax error and DeepSpeed ZeRO3 deque failure are unrelated failure modes." + "reason": "Different code paths: Python 3.13 Deberta import parsing vs DeepSpeed ZeRO3 deque error." }, { "left": "issue:43381", "right": "issue:43742", "accept": false, - "reason": "Gradient checkpointing in eval mode and MobileLLM config key errors affect different code paths and root causes." + "reason": "Checkpointing/eval-mode behavior does not match a MobileLLM loading key error." }, { - "left": "issue:43316", - "right": "issue:44877", + "left": "issue:43540", + "right": "issue:44514", "accept": false, - "reason": "Both are config-related, but one is a Gemma3 API mismatch and the other is strict config loading for granite_speech; different bugs." + "reason": "Both mention multimodal/video processing, but one is a video-input validation error and the other is a batched chat-template padding crash." }, { - "left": "issue:43540", - "right": "issue:44514", + "left": "issue:43316", + "right": "issue:44877", "accept": false, - "reason": "One is a Qwen3OmniMoe video-input ValueError, the other a Qwen2_5_VL batched chat-template crash; different models and entrypoints." + "reason": "Different config bugs in different models: Gemma3TextConfig API mismatch vs granite_speech config strictness." }, { "left": "issue:43673", "right": "issue:44743", "accept": false, - "reason": "Both mention cache/state handling, but one is missing GenerationMixin cache in chunked_prefill and the other resets recurrent states in modular_qwen3_5.py." + "reason": "Both involve caching/state, but one is missing generation cache in v5 and the other is recurrent state reset in a specific Qwen module." }, { "left": "issue:42175", "right": "issue:43316", "accept": false, - "reason": "Packaging/backend dependency issue vs a Gemma3TextConfig API discrepancy; not the same bug." + "reason": "Package/backend installation issue vs model config API discrepancy; no shared bug." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Weight tying in Mistral3ForConditionalGeneration and FlashAttention failure in Ministral-3 are distinct model bugs." + "reason": "Weight tying bug in Mistral3ForConditionalGeneration is not the same as a FlashAttention failure in Ministral-3." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline top_k output nesting and missing PyTorch import are unrelated behavior vs environment detection issues." + "reason": "Pipeline output nesting bug vs missing PyTorch import; entirely different failures." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "Sam3PixelDecoder dummy Conv2d and Sam3Video CUDA OOM are different problems; one is architecture/code, the other memory pressure." + "reason": "SAM3 pixel decoder dummy layer issue is not the same as a SAM3Video OOM." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "ESM2 breakage and a janus image-generation test failure do not share the same underlying bug." + "reason": "ESM2 breakage and janus image-generation test failure are unrelated." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Metric gathering label truncation and auxiliary-loss normalization are both training issues, but they are separate mechanisms." + "reason": "Training/evaluation utility bug vs auxiliary-loss normalization bug; different mechanisms." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Processor.tokenizer regression is not the same as a pipeline error message about unsupported translation tasks." - }, - { - "left": "issue:41950", - "right": "issue:43475", - "accept": false, - "reason": "Video-classification image-processor lookup and missing fpn_position_embeddings in SAM3 video are different subsystems and fixes." + "reason": "Both are v5 regressions, but one is processor/tokenizer behavior and the other is a pipeline error message." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "A model_max_length question and a missing transformers.onnx import are unrelated." + "reason": "DeBERTa model_max_length question and ONNX import error are unrelated." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "Optional post_init validation for patchtsmixer and junk-filled non-persistent buffers are separate v5 issues." + "reason": "PatchTSMixer post_init policy is not the same as non-persistent buffer initialization corruption." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor selection and Sam3Video encoder attribute access are different bugs." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "LoRA merge collapse with extended vocab and Gemma 4 mm_token_type_ids defaults are distinct model-specific bugs." + "reason": "LoRA merge / tied embeddings issue is unrelated to Gemma 4 token-type defaults." }, { "left": "issue:44871", "right": "issue:45584", "accept": false, - "reason": "Gemma-3 eos_token_id inconsistency and Whisper empty-transcription handling after align_special_tokens are different tokenization/decoding problems." + "reason": "Gemma eos_token_id config mismatch is not the same as Whisper empty-transcription generation failure." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "torch.compile SDPA failure in Bamba and resume_from_checkpoint key mismatch are unrelated." + "reason": "torch.compile + SDPA failure is unrelated to resume_from_checkpoint key mismatch." }, { "left": "issue:43761", "right": "issue:44479", "accept": false, - "reason": "CLIPVisionModel hidden_states regression and Qwen video-input regression are different model/API failures." + "reason": "CLIPVision hidden_states regression and Qwen VL video-input regression affect different models and paths." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch and checkpoint key mismatch do not share the same root cause." + "reason": "CPU offload device mismatch is unrelated to checkpoint resume key mismatch." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Processor.tokenizer regression and Gemma-3 eos_token_id mismatch are separate API/config issues." + "reason": "Different v5-era config regressions with different symptoms and models." }, { - "left": "issue:43452", - "right": "issue:44496", + "left": "issue:43575", + "right": "issue:43742", "accept": false, - "reason": "gguf_file loading regression and missing model_type in config.json are both load failures, but with different causes and code paths." + "reason": "Both are loading problems, but one is tensor-parallel OOM and the other is a MobileLLM key error." }, { - "left": "issue:43575", - "right": "issue:43742", + "left": "issue:43452", + "right": "issue:44496", "accept": false, - "reason": "Tensor-parallel OOM for Qwen2-57B and MobileLLM key errors are unrelated." + "reason": "GGUF loading breakage is not the same as an unrecognized config/model_type error." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "A script/runtime failure in 3d_parallel.py and an OOM during tensor-parallel model loading are different issues." + "reason": "3d_parallel.py runtime issue and Qwen2 load-time OOM are unrelated." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Broad custom model initialization regression and a specific GPTNeoXConfig reload/default issue are related only at a high level, not the same bug." - }, - { - "left": "issue:44734", - "right": "issue:45406", - "accept": false, - "reason": "KV-cache continuation tensor indexing in serve and Gemma4Processor missing _tokenizer are different runtime failures." - }, - { - "left": "issue:43504", - "right": "issue:45698", - "accept": false, - "reason": "Legacy-field BEiT preset loading and wrong custom-module resolution after save_pretrained are both loading issues, but not the same concrete bug." - }, - { - "left": "issue:44617", - "right": "issue:45584", - "accept": false, - "reason": "SAM3 video OOM and Whisper empty-transcription behavior are unrelated." + "reason": "Custom model initialization breakage is not the same as GPTNeoXConfig rotary_pct not persisting on reload." } ] }, @@ -9196,8 +8996,8 @@ "issue:43065", "issue:43335", "issue:43388", + "issue:43504", "issue:43575", - "issue:43582", "issue:43761", "issue:43827", "issue:43873", @@ -9205,7 +9005,6 @@ "issue:43931", "issue:43994", "issue:44060", - "issue:44077", "issue:44079", "issue:44190", "issue:44242", @@ -9215,6 +9014,7 @@ "issue:44521", "issue:44534", "issue:44610", + "issue:44617", "issue:44734", "issue:44737", "issue:44792", @@ -9222,7 +9022,6 @@ "issue:44871", "issue:44913", "issue:44933", - "issue:44938", "issue:44964", "issue:44977", "issue:45127", @@ -9231,226 +9030,234 @@ "issue:45357", "issue:45362", "issue:45381", - "issue:45542", + "issue:45406", "issue:45563", "issue:45584", "issue:45588", "issue:45600", + "issue:45698", "issue:45803", "issue:45907" ], "soft_pairs": [ + "issue:44734|issue:45406", + "issue:43504|issue:45698", "issue:43575|issue:43873", - "issue:43761|issue:44242", + "issue:44617|issue:45584", "issue:43931|issue:44060", - "issue:45563|issue:45584", + "issue:43761|issue:44242", "issue:32090|issue:35141", + "issue:45563|issue:45584", "issue:44464|issue:44871", "issue:44792|issue:44977", - "issue:43065|issue:44610", "issue:44479|issue:44734", + "issue:43065|issue:44610", "issue:44521|issue:45381", "issue:43827|issue:43994", - "issue:44464|issue:45563", "issue:43906|issue:45290", + "issue:44464|issue:45563", "issue:44534|issue:44913", "issue:44933|issue:45362", "issue:44190|issue:44964", "issue:41720|issue:45245", "issue:44792|issue:45600", - "issue:44737|issue:45803", "issue:44855|issue:45588", - "issue:43388|issue:43582", + "issue:44737|issue:45803", + "issue:43388|issue:44079", "issue:45127|issue:45357", "issue:44462|issue:45907", - "issue:43335|issue:44855", - "issue:44938|issue:45542", - "issue:44077|issue:44079" + "issue:43335|issue:44855" ], - "prepared_review_unit_hash": "84708b1df145877ad46392b9ff7ae41d55a01ca6b6ef3f60e3a5fcabac4a6f52", + "prepared_review_unit_hash": "3375ef60bbe5e5295191826186b4790a8338d015895d74246414147d1058f449", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 26, - "serialized_chars": 35673, - "estimated_input_tokens": 8919, - "estimated_eval_tokens": 18094 + "soft_pair_count": 27, + "serialized_chars": 35993, + "estimated_input_tokens": 8999, + "estimated_eval_tokens": 18254 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: it mixes unrelated bugs across loading, generation, multimodal chat templating, quantization/offloading, docs, and config regressions. None of the soft pairs look like the same concrete fix, so all soft edges are rejected.", - "confidence": 0.16, - "canonical_issue_reason": "Issue 43873 is the broadest and most active unresolved thread in the set (open, highest discussion/inbound activity), so it is the best cluster anchor even though the surrounding items are not true duplicates.", + "summary": "These are mostly unrelated issues with only superficial title overlap; I don\u2019t see any soft pair that is confidently the same underlying bug. The clearest standalone issue here is the ModelOutput None-key bug.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44079 is a precise, self-contained framework bug with clear failure semantics and broad utility as a representative issue.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the most representative open issue by activity and breadth; the rest of the items are too diverse to form a real duplicate cluster.", + "best_issue_reason": "issue:44079 is the most canonical-looking issue: specific, actionable, and not tied to a model-specific edge case or a flaky environment report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both touch `transformers serve`, but they describe different failures: KV-cache tensor indexing vs a missing `_tokenizer` attribute." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "One is a legacy preset-loading field problem; the other is loading the wrong custom module after save/load. Different code paths." + }, { "left": "issue:43575", "right": "issue:43873", "accept": false, - "reason": "Different symptoms and likely different layers: TP/OOM on a specific model load vs generic offloading/quantization behavior." + "reason": "OOM under tensor parallelism and offloading failures with quantization are different resource-management problems." }, { - "left": "issue:43761", - "right": "issue:44242", + "left": "issue:44617", + "right": "issue:45584", "accept": false, - "reason": "CLIPVision hidden_states regression is unrelated to MoE load-balancing loss computation." + "reason": "Sam3Video CUDA OOM and Whisper empty-transcription failure are unrelated model-specific bugs." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Weight-shape mismatch on Qwen3-VL loading is not the same bug as a tied-weights warning involving dt_bias." + "reason": "Weight-shape mismatch when loading Qwen3-VL is not the same as an incorrect tied-weights warning in Qwen3-Next." }, { - "left": "issue:45563", - "right": "issue:45584", + "left": "issue:43761", + "right": "issue:44242", "accept": false, - "reason": "A stale generation warning and a Whisper empty-transcription crash are different code paths and failures." + "reason": "CLIP hidden-states regression and missing load-balancing loss are different forward-pass behaviors in different model families." }, { "left": "issue:32090", "right": "issue:35141", "accept": false, - "reason": "Trainer GPU broadcast TypeError and embedding reinitialization after resize are unrelated issues." + "reason": "Trainer GPU broadcast `NoneType` error and embedding reinitialization after resize are separate mechanisms." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale generation warning is unrelated to Whisper failing on empty transcription after token alignment." }, { "left": "issue:44464", "right": "issue:44871", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is unrelated to Gemma eos_token_id configuration." + "reason": "Chunked generation inconsistency with compiled forward and Gemma EOS config mismatch are different bugs." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation bug are different model/runtime problems." + "reason": "A Janus image-generation test failure and Qwen3.5 flash-attention generation issues are not the same defect." }, { - "left": "issue:43065", - "right": "issue:44610", + "left": "issue:44479", + "right": "issue:44734", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo image-size mismatch are unrelated preprocessing/model-shape issues." + "reason": "Video-input regression in Qwen multimodal models and a serve KV-cache crash are unrelated." }, { - "left": "issue:44479", - "right": "issue:44734", + "left": "issue:43065", + "right": "issue:44610", "accept": false, - "reason": "Video-input regression in Qwen VL models is not the same as KV-cache continuation indexing in serve responses." + "reason": "A dummy Conv2d implementation and a processor/model input-size mismatch are different issues." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "All-zero assistant masks from apply_chat_template and wrong vision_position_ids are different multimodal templating failures." + "reason": "Zero assistant masks in multimodal chat templates and wrong video vision_position_ids are distinct multimodal bugs." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs still referencing pipeline() is unrelated to SigLIP2 producing nonsense outputs." + "reason": "Docs still mentioning `pipeline()` and SigLIP2 nonsense outputs are different problems." }, { - "left": "issue:44464", - "right": "issue:45563", + "left": "issue:43906", + "right": "issue:45290", "accept": false, - "reason": "These target different behaviors: compiled-forward chunked generation vs a stale warning in paged generate()." + "reason": "`43906` is only a reproduction marker for another issue; the titles alone do not establish the same concrete bug as the chat-template/tool-call crash." }, { - "left": "issue:43906", - "right": "issue:45290", + "left": "issue:44464", + "right": "issue:45563", "accept": false, - "reason": "An isolated reproduction of an earlier issue is not enough to treat this as the same bug as the tool-call/chat-template crash." + "reason": "Generation inconsistency under compilation and a stale warning about `num_return_sequences` are not the same code-path problem." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer junk on v5 upgrade and GPTNeoX rotary_pct reload loss are different configuration/state bugs." + "reason": "Non-persistent buffer junk values and GPTNeoX config reload behavior are unrelated regressions." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "A missing import from image_utils is unrelated to Qwen3.5 chat crashes." + "reason": "Missing `image_utils` import and a Qwen3.5 chat crash do not describe the same failure." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Local dataset loading in a no-trainer script and Phi-4 multimodal model loading are unrelated." + "reason": "Local dataset loading failure in a training script and model loading failure for Phi-4 multimodal are different subsystems." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "CUDA assert during Qwen3 auto device mapping is unrelated to the category-cardinality RuntimeError." + "reason": "CUDA assert on auto device mapping and a category-cardinality runtime limit are unrelated." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "A Janus generate-images test failure and stale auto_mappings references are different failures in different parts of the tree." + "reason": "Janus test failure and stale auto-mapping references to removed configs are different CI/model issues." }, { - "left": "issue:44737", - "right": "issue:45803", + "left": "issue:44855", + "right": "issue:45588", "accept": false, - "reason": "CPU allocation in XLNet forward is unrelated to Fuyu's bare-except error handling." + "reason": "Python parsing failure in DebertaV2 import and flash-attention `s_aux=None` crash are unrelated." }, { - "left": "issue:44855", - "right": "issue:45588", + "left": "issue:44737", + "right": "issue:45803", "accept": false, - "reason": "Python 3.13 decorator parsing failure in DebertaV2Model is unrelated to flash-attention's s_aux=None crash." + "reason": "CPU-bound rotary-position computation and a bare-except bug in tensor conversion are different defects." }, { "left": "issue:43388", - "right": "issue:43582", + "right": "issue:44079", "accept": false, - "reason": "Incorrect metric gathering for tuple labels is unrelated to Apple Silicon caching allocator warmup type errors." + "reason": "Metric gathering drops tuple labels in the last batch; ModelOutput key assignment with prior `None` is a separate data-structure bug." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA merge collapse with extended vocabulary is not the same as saving incorrect visual encoder keys." + "reason": "LoRA/vocabulary merge collapse and incorrect visual encoder keys during save_pretrained are different persistence bugs." }, { "left": "issue:44462", "right": "issue:45907", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json is unrelated to list_repo_files failing to forward revision." + "reason": "Ignoring `tokenizer.json` and failing to forward `revision` are separate Hub-loading issues." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and DebertaV2 Python 3.13 import parsing are unrelated." - }, - { - "left": "issue:44938", - "right": "issue:45542", - "accept": false, - "reason": "Python 3.14 import/load failure is unrelated to the TensorBoard-without-TensorFlow backend error." - }, - { - "left": "issue:44077", - "right": "issue:44079", - "accept": false, - "reason": "An optional post_init policy issue in patchtsmixer is unrelated to ModelOutput key assignment when a prior value was None." + "reason": "Sparse-layer construction in SwitchTransformers and a Python 3.13 import parse error are unrelated." } ] }, @@ -9481,21 +9288,20 @@ "issue:41628", "issue:41950", "issue:42175", - "issue:42907", "issue:43065", "issue:43295", "issue:43335", "issue:43404", - "issue:43441", "issue:43479", "issue:43540", "issue:43550", "issue:43643", "issue:43644", "issue:43650", - "issue:43854", "issue:43976", "issue:44016", + "issue:44077", + "issue:44079", "issue:44112", "issue:44186", "issue:44190", @@ -9509,21 +9315,24 @@ "issue:44734", "issue:44855", "issue:44913", + "issue:44938", "issue:44964", "issue:45003", "issue:45216", "issue:45245", "issue:45276", - "issue:45335", "issue:45375", "issue:45406", "issue:45446", + "issue:45542", "issue:45584", "issue:45600", "issue:45632", "issue:45803" ], "soft_pairs": [ + "issue:44938|issue:45542", + "issue:44077|issue:44079", "issue:43976|issue:44855", "issue:43644|issue:44913", "issue:43550|issue:44190", @@ -9537,15 +9346,15 @@ "issue:45245|issue:45584", "issue:43295|issue:45276", "issue:43479|issue:45375", - "issue:36246|issue:41628", "issue:41950|issue:44016", + "issue:36246|issue:41628", "issue:36246|issue:38617", "issue:44186|issue:44734", "issue:38175|issue:43065", "issue:44683|issue:45446", "issue:44466|issue:44913", - "issue:42175|issue:44279", "issue:44734|issue:45632", + "issue:42175|issue:44279", "issue:44016|issue:44220", "issue:43404|issue:45216", "issue:44561|issue:45632", @@ -9553,237 +9362,222 @@ "issue:43643|issue:44496", "issue:45406|issue:45632", "issue:44964|issue:45600", - "issue:44112|issue:44964", - "issue:42907|issue:43441", - "issue:43854|issue:44964", - "issue:43295|issue:45335" + "issue:44112|issue:44964" ], - "prepared_review_unit_hash": "9b6125ad1e1e02fa87060173ace4d693b31433b074ec81c85ca5f86bdbd57c00", + "prepared_review_unit_hash": "170330efc71c1dd7d5bdc9ba9e466b581c7b122e298b5aa7acd1e653e789c034", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37714, - "estimated_input_tokens": 9429, - "estimated_eval_tokens": 19114 + "soft_pair_count": 32, + "serialized_chars": 37420, + "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 a loose mix of unrelated Transformers regressions and model-specific bugs. The apparent similarities are mostly superficial API/import/config themes, so none of the soft pairs look like the same underlying issue. I would keep the issues separate.", - "confidence": 0.88, - "canonical_issue_reason": "issue:41628 is the broadest, clearest user-facing regression in the set: a direct import breakage that reflects the common API-stability theme without being overly model-specific.", + "summary": "These are mostly unrelated issues across different Transformers features and regressions. I don\u2019t see any soft pair that is strong enough to treat as the same underlying bug/change, so I\u2019d keep the items separate. The broad v5 serialization regression is the best cluster representative among the issues listed.", + "confidence": 0.41, + "canonical_issue_reason": "issue:43644 is the broadest, most central regression here: it targets a core v5 save/load behavior with wider model impact than the narrower model-, pipeline-, or environment-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41628 is the best representative issue here because it is concrete, high-signal, and broadly relevant to the shared import/regression theme across a few nearby reports.", + "best_issue_reason": "issue:43644 is the strongest representative issue in the set because it describes a concrete, high-impact core regression rather than a niche model-specific failure.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Different failure classes: Python 3.14 import/load issue versus a missing TensorFlow backend dependency when only tensorboard is installed." + }, + { + "left": "issue:44077", + "right": "issue:44079", + "accept": false, + "reason": "Unrelated bugs: one is about an optional post_init constraint in patchtsmixer, the other about ModelOutput key assignment semantics." + }, { "left": "issue:43976", "right": "issue:44855", "accept": false, - "reason": "Different failure modes: Python package compatibility vs JIT parsing/IndentationError on import." + "reason": "Both are version-related import problems, but they affect different code paths and root causes (package compatibility vs. JIT parsing of a model file)." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "Unrelated bugs: non-persistent buffer initialization vs config serialization/reload of rotary_pct." + "reason": "One is a v5 buffer serialization regression; the other is GPTNeoXConfig losing rotary_pct on reload. Different model/config paths." }, { "left": "issue:43550", "right": "issue:44190", "accept": false, - "reason": "Different subsystems and symptoms: torch.compile/SDPA model failure vs local dataset loading in an example script." + "reason": "Different subsystems and symptoms: Bamba torch.compile/SDPA failure versus dataset loading in an image-classification example script." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Both are multimodal/video-serving adjacent, but one is video input validation and the other is KV-cache continuation indexing in serve." + "reason": "Different products and code paths: Qwen3OmniMoe video input validation versus transformers serve KV-cache indexing in /v1/responses." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "CI test flake in GraniteSpeech vs stale auto_mappings references to removed configs; same general CI area, not same bug." + "reason": "Both touch CI, but one is a stale GraniteSpeech test and the other is auto_mappings referencing removed Sam3LiteText configs." }, { "left": "issue:44496", "right": "issue:45803", "accept": false, - "reason": "Unrecognized model/config parsing vs a bare-except tensor conversion bug; unrelated root causes." + "reason": "An unrecognized-model/config issue is not the same bug as a bare-except swallowing interrupts/errors in FuyuBatchFeature." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Different code paths: Gemma4 token embedding resize propagation vs Whisper empty-transcription generation after token alignment." + "reason": "Different model behaviors: Gemma4 embedding resize bug versus Whisper empty-transcription generation failure." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "SwitchTransformers config creation bug vs an image-classification example script issue." + "reason": "SwitchTransformers config-layer construction and a local dataset loading failure are unrelated." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "Placeholder issue with no technical content vs audio feature extraction bug." + "reason": "The first is a non-informative placeholder issue; the second is an fbank feature extraction bug. No shared underlying change." }, { "left": "issue:39692", "right": "issue:44190", "accept": false, - "reason": "SigLIP2 doc/example mismatches and quantization errors are unrelated to the local dataset loading failure." + "reason": "Documentation/example errors in SigLIP2 are not the same as a dataset-loading script failure." }, { "left": "issue:45245", "right": "issue:45584", "accept": false, - "reason": "Category cardinality runtime error is unrelated to Whisper empty transcription handling." + "reason": "Category-count overflow and Whisper special-token generation are unrelated." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Processor/tokenizer regression and Gemma4 embedding-resize bug are different API breakages." + "reason": "Processor/tokenizer regression and Gemma4 embedding-resize behavior are different problems." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Both are config-default issues, but they affect different model families and different missing/default fields." + "reason": "Different config classes and fields; these are separate multimodal config initialization bugs." }, { - "left": "issue:36246", - "right": "issue:41628", + "left": "issue:41950", + "right": "issue:44016", "accept": false, - "reason": "Both are import errors, but they involve different symbols and modules; no evidence of the same broken code path." + "reason": "A video-classification pipeline processor lookup bug is unrelated to a notebook syntax error." }, { - "left": "issue:41950", - "right": "issue:44016", + "left": "issue:36246", + "right": "issue:41628", "accept": false, - "reason": "Video-classification pipeline processor lookup vs a notebook syntax error are unrelated." + "reason": "Both are import errors, but they concern different symbols and likely different fixes: Qwen2_5_VLImageProcessor versus AutoImageProcessor." }, { "left": "issue:36246", "right": "issue:38617", "accept": false, - "reason": "Separate import failures for different names from different modules." + "reason": "Different missing imports in different modules; no evidence of a shared root cause." }, { "left": "issue:44186", "right": "issue:44734", "accept": false, - "reason": "Tokenizer padding/truncation crash vs serve KV-cache continuation crash; different components and failure points." + "reason": "LayoutLMv2 tokenizer padding/NER crashes are unrelated to serve KV-cache continuation indexing." }, { "left": "issue:38175", "right": "issue:43065", "accept": false, - "reason": "SigLIP2 probability output issue is unrelated to a dummy Conv2d in Sam3PixelDecoder." + "reason": "A SigLIP2 probability anomaly and a dummy Conv2d in Sam3PixelDecoder are different model-level bugs." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Torch >=2.9 flex_attention compilation issue is unrelated to a PyTorch version guard around AuxRequest import." + "reason": "Both mention flex_attention, but one is a compiled flex_attention failure on torch>=2.9 while the other is a version-check/import bug for AuxRequest; not enough to merge." }, { "left": "issue:44466", "right": "issue:44913", "accept": false, - "reason": "Both concern tied/serialized weights and config reload, but one is serialization inconsistency and the other is a GPTNeoX config field reset." + "reason": "Serialization of tied weights depending on device is a different issue from GPTNeoXConfig rotary_pct not persisting on reload." }, { - "left": "issue:42175", - "right": "issue:44279", + "left": "issue:44734", + "right": "issue:45632", "accept": false, - "reason": "Packaging/backend dependency complaint vs a generic dependency issue; too vague and not the same problem." + "reason": "Serve KV-cache tensor indexing and trust_remote_code cache-path collisions are unrelated code paths." }, { - "left": "issue:44734", - "right": "issue:45632", + "left": "issue:42175", + "right": "issue:44279", "accept": false, - "reason": "Serve KV-cache indexing crash vs trust_remote_code cache-path collision; different features and root causes." + "reason": "A TensorFlow backend packaging problem is not the same as a generic dependency issue." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error vs audio feature extraction bug." + "reason": "Notebook syntax error and audio feature extraction bug are unrelated." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Weight-tying bug in Mistral3 vs Qwen3.5 save_pretrained serialization regression; different model behaviors." + "reason": "lm_head weight tying in Mistral3ForConditionalGeneration is a different bug from Qwen3.5 checkpoint save_pretrained serialization." }, { "left": "issue:44561", "right": "issue:45632", "accept": false, - "reason": "Removal of is_torch_fx_available breaks remote-code models at import/runtime, while the other is a local cache-path collision." + "reason": "Removal of is_torch_fx_available breaking trust_remote_code models is separate from local-model cache-path collisions." }, { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Import slowness is a performance problem; sys.modules access is a correctness/safety bug." - }, - { - "left": "issue:43643", - "right": "issue:44496", - "accept": false, - "reason": "Missing fields with trust_remote_code config loading vs unrecognized model/config parsing are different config-loading failures." + "reason": "Slow import and unsafe sys.modules access are different issues, even if both concern import-time behavior." }, { "left": "issue:45406", "right": "issue:45632", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to local-model cache collisions." + "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to trust_remote_code cache path collisions." }, { "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "Phi-4 multimodal model loading failure vs stale Sam3LiteText auto-mapping references; different breakages." + "reason": "Loading Phi-4-multimodal-instruct and auto_mappings referencing removed Sam3LiteText configs are different breakages." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "CI stale device override test vs model loading failure; not the same bug." - }, - { - "left": "issue:42907", - "right": "issue:43441", - "accept": false, - "reason": "Dequantized save failure for Ministral/Devstral is unrelated to FlashAttention failure in Ministral-3." - }, - { - "left": "issue:43854", - "right": "issue:44964", - "accept": false, - "reason": "Unit-test loading issue for GLM-4.7-Flash is unrelated to Phi-4 multimodal loading." - }, - { - "left": "issue:43295", - "right": "issue:45335", - "accept": false, - "reason": "Both involve processor/tokenizer or embedding-related regressions, but they affect different models and different fields/components." + "reason": "A stale CI test failure and a model-loading regression are not the same underlying bug." } ] }, @@ -9810,9 +9604,11 @@ "issue:38175", "issue:39692", "issue:41950", + "issue:42907", "issue:43295", "issue:43299", "issue:43381", + "issue:43441", "issue:43479", "issue:43525", "issue:43526", @@ -9831,17 +9627,15 @@ "issue:44336", "issue:44351", "issue:44368", - "issue:44462", "issue:44496", "issue:44521", "issue:44617", "issue:44683", - "issue:44779", "issue:44933", + "issue:44964", "issue:44987", "issue:45020", "issue:45042", - "issue:45081", "issue:45200", "issue:45216", "issue:45278", @@ -9857,6 +9651,9 @@ "issue:45823" ], "soft_pairs": [ + "issue:42907|issue:43441", + "issue:43854|issue:44964", + "issue:43295|issue:45335", "issue:44016|issue:44062", "issue:43976|issue:44987", "issue:43673|issue:44336", @@ -9875,227 +9672,231 @@ "issue:44496|issue:45759", "issue:44521|issue:45325", "issue:44242|issue:45200", - "issue:41950|issue:45020", "issue:43976|issue:45405", + "issue:41950|issue:45020", "issue:43479|issue:43525", - "issue:41950|issue:45823", "issue:43577|issue:45216", + "issue:41950|issue:45823", "issue:38175|issue:44617", "issue:44279|issue:45405", - "issue:44683|issue:45542", - "issue:43299|issue:43854", "issue:39692|issue:43901", - "issue:43295|issue:44779", - "issue:44462|issue:45081" + "issue:43299|issue:43854", + "issue:44683|issue:45542" ], - "prepared_review_unit_hash": "6ba0a52895986f48e19eac324c09345968542b20b1904de001b737e59f18d444", + "prepared_review_unit_hash": "38275d3f886c76e2757f0075006a8870a86accd1dfbe02d9b188ecf4c8bb379c", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36711, - "estimated_input_tokens": 9178, - "estimated_eval_tokens": 18612 + "soft_pair_count": 31, + "serialized_chars": 36903, + "estimated_input_tokens": 9226, + "estimated_eval_tokens": 18708 }, "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 issues, with several broad regression/import reports but no soft pair that clearly represents the same underlying bug. No PRs are present.", - "confidence": 0.74, - "canonical_issue_reason": "Issue 45020 is the broadest umbrella-style report here: it describes recent Transformers versions breaking remote-code models, which is the closest thing to a cluster-level anchor among these heterogeneous regressions.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions and docs/import issues. The only strong thematic center is recent-version breakage around model loading and custom/remote-code models, so I\u2019d use that as the representative issue, but I would not deduplicate any of the soft pairs on titles alone.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45020 is the broadest and most representative report in the set: it frames the common theme of recent Transformers versions breaking model loading for custom/remote_code models. It is the best umbrella issue for this mixed regression cluster, even though it is not a strict duplicate of most others.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the best representative issue because it is open, broad, and captures the general \u2018recent version breaks model loading\u2019 theme better than the more narrow or model-specific reports.", + "best_issue_reason": "issue:45020 best captures the shared cluster theme of recent-version compatibility regressions and model-loading breakages, and it is broad enough to serve as the main triage anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Same model family, but one is about saving dequantized Ministral/Devstral models and the other is a FlashAttention loading failure. Different bugs and code paths." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different root causes. Not the same underlying bug." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor/tokenizer regression versus resize_token_embeddings affecting decoder.embed_tokens. Different subsystems and failure modes." + }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error and an AddedToken keyword conflict are unrelated failures." + "reason": "Notebook syntax error and AddedToken keyword conflict are unrelated." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "One is a Python-version compatibility complaint; the other is a specific model-loading failure." + "reason": "Python-version support issue versus a specific model loading failure. Too different to merge." }, { "left": "issue:43673", "right": "issue:44336", "accept": false, - "reason": "Cache handling in generation and ANSI color output in loading_report are different subsystems and bugs." + "reason": "Generation cache regression and ANSI escape handling in loading_report are unrelated." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and Sam3LiteText auto_mappings CI breakage are unrelated." + "reason": "Qwen3-MoE sliding-window behavior versus auto_mappings cleanup for Sam3LiteText configs. Different problems." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized model config and PIL/torchvision dependency issues are different problem types." + "reason": "Unrecognized-model config error versus PIL backend torchvision dependency regression. Different code paths." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "A processor label-reduction bug and stale pipeline docs are not the same bug." + "reason": "BeitImageProcessorFast label reduction bug versus stale pipeline documentation. Not the same bug." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Both are import errors, but they concern different symbols and modules with no evidence of the same root cause." + "reason": "Both are import errors, but they concern different missing symbols and different modules; no clear shared underlying defect." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Qwen3VL model loading regression and Qwen3.5 save_pretrained checkpoint corruption are different code paths." + "reason": "Qwen3VL MoE loading regression versus Qwen3.5 save_pretrained checkpoint corruption. Related model family, but different concrete bugs." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "Embedding resize propagation in t5gemma and Whisper empty-transcription generation are unrelated." + "reason": "Embedding resize bug and Whisper empty-transcription generation failure are unrelated." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Unused SGD args and a tie_word_embeddings warning are not the same issue." + "reason": "Unused SGD args and a warning about tie_word_embeddings are not the same issue." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "Python compatibility breakage and a generic dependency issue do not point to the same bug." + "reason": "Python compatibility regression and a vague dependency issue do not point to the same concrete defect." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Docs mentioning return_all_scores and model config loading failure are unrelated." + "reason": "Pipeline docs mismatch versus unrecognized model/config error. Different artifacts and different fixes." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing eval-mode behavior and D-FINE auxiliary-loss handling are different model/training bugs." + "reason": "Gradient checkpointing in eval mode and missing auxiliary losses are unrelated training behaviors." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "These are both Qwen3.5-related, but one is a loading regression and the other is incorrect visual encoder keys on save; not the same concrete bug." + "reason": "Qwen3VL loading regression versus save_pretrained visual-encoder-key regression. Same family, but not the same underlying bug." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "A missing HybridCache import may be one symptom of broad import breakage, but the reports are too different and one is much more general." + "reason": "A specific missing import and a broad collection of import errors are too weakly linked without evidence of the same root cause." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Unrecognized model config and AutoModelForCausalLM not unwrapping text_config are different loading failures." + "reason": "Unrecognized model config versus AutoModelForCausalLM.from_config not unwrapping text_config. Different API paths." }, { "left": "issue:44521", "right": "issue:45325", "accept": false, - "reason": "All-zero assistant masks and still-image rope-index scaling are different multimodal bugs." + "reason": "All-zero assistant masks in multimodal apply_chat_template is unrelated to Qwen2.5-VL rope position scaling." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "Missing load-balancing loss and mm_token_type_ids defaults are unrelated training/data-path issues." + "reason": "Missing load-balancing loss when router logits are off versus mm_token_type_ids defaulting behavior. Different model-training bugs." }, { - "left": "issue:41950", - "right": "issue:45020", + "left": "issue:43976", + "right": "issue:45405", "accept": false, - "reason": "A video-classification pipeline processor lookup bug is not the same as the broad remote_code regression report." + "reason": "Python-version support regression and MIN_PEFT_VERSION bump are unrelated." }, { - "left": "issue:43976", - "right": "issue:45405", + "left": "issue:41950", + "right": "issue:45020", "accept": false, - "reason": "Version/Python compatibility and an unreleased PEFT minimum version bump are unrelated." + "reason": "Video-classification pipeline processor selection and remote_code model loading are different problems." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Default multimodal config initialization and missing pad_token_id on Llama4Config are different config bugs." + "reason": "Phi4MultimodalConfig default-initialization bug and missing pad_token_id on Llama4Config are distinct configuration issues." }, { - "left": "issue:41950", - "right": "issue:45823", + "left": "issue:43577", + "right": "issue:45216", "accept": false, - "reason": "Pipeline processor selection and Gemma4 PLE device placement are unrelated." + "reason": "BLIP2 dtype propagation and Qwen3.5 save_pretrained corruption are unrelated." }, { - "left": "issue:43577", - "right": "issue:45216", + "left": "issue:41950", + "right": "issue:45823", "accept": false, - "reason": "Blip2 dtype propagation and Qwen3.5 save_pretrained checkpoint keys are different code paths." + "reason": "Video pipeline image-processor lookup and Gemma4 PLE device mismatch are different failure modes." }, { "left": "issue:38175", "right": "issue:44617", "accept": false, - "reason": "Unexpected zero probabilities and CUDA OOM are clearly unrelated." + "reason": "Unexpected zero probabilities in SigLIP2 and Sam3Video CUDA OOM are unrelated." }, { "left": "issue:44279", "right": "issue:45405", "accept": false, - "reason": "A generic dependency issue and a PEFT version bump are not the same underlying bug." - }, - { - "left": "issue:44683", - "right": "issue:45542", - "accept": false, - "reason": "Torch flex_attention compilation failure and a TensorBoard/TensorFlow backend issue are unrelated." - }, - { - "left": "issue:43299", - "right": "issue:43854", - "accept": false, - "reason": "Qwen3VL loading regression and GLM-4.7-Flash test loading failure concern different models and failures." + "reason": "Generic dependency issue and unreleased PEFT minimum version are not the same bug." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "Both mention docs, but one is a broken example with quantization/model mismatch and the other is stale API wording." + "reason": "SigLIP2 docs example errors and stale pipeline docs mention are both documentation-related, but they are different documents and different errors." }, { - "left": "issue:43295", - "right": "issue:44779", + "left": "issue:43299", + "right": "issue:43854", "accept": false, - "reason": "Custom processor/tokenizer regression and DeepSeek tokenizer correctness are different bugs." + "reason": "Qwen3VL loading regression and GLM-4.7 flash-loading test failure are unrelated." }, { - "left": "issue:44462", - "right": "issue:45081", + "left": "issue:44683", + "right": "issue:45542", "accept": false, - "reason": "Ignoring tokenizer.json and crashing in _patch_mistral_regex are distinct tokenizer-loading problems." + "reason": "Compiled flex_attention on torch>=2.9 and tensorboard/TensorFlow backend detection are unrelated." } ] }, @@ -10124,8 +9925,8 @@ "issue:42175", "issue:42994", "issue:43065", - "issue:43122", "issue:43262", + "issue:43295", "issue:43317", "issue:43441", "issue:43450", @@ -10169,6 +9970,8 @@ "issue:45907" ], "soft_pairs": [ + "issue:43295|issue:44779", + "issue:44462|issue:45081", "issue:41950|issue:43650", "issue:33290|issue:45656", "issue:44297|issue:44462", @@ -10193,200 +9996,205 @@ "issue:44361|issue:45593", "issue:44671|issue:44779", "issue:44016|issue:45823", - "issue:42175|issue:43976", "issue:43901|issue:43994", - "issue:43122|issue:43650" + "issue:42175|issue:43976" ], - "prepared_review_unit_hash": "1a90bea2e1e2a518f21be1bcef75ae696b09893588be0140f1a41b9eabffc018", + "prepared_review_unit_hash": "19a395c2956d0ff4f86ae84dae2bc8e87d6b42e143617334c6bffd19e38a7d65", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 35748, - "estimated_input_tokens": 8937, - "estimated_eval_tokens": 18130 + "soft_pair_count": 28, + "serialized_chars": 36089, + "estimated_input_tokens": 9023, + "estimated_eval_tokens": 18302 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: the items span tokenizer loading, processor/chat-template behavior, quantization, model-specific regressions, CI cleanup, and unrelated optimizer/deepspeed bugs. I do not see any soft pair that looks like the same underlying bug/change, so all soft edges are rejected. If one issue must anchor the cluster, the best representative is the tokenizer-loading regression in issue 44462.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 44462 is the closest representative because it describes a concrete, user-visible tokenizer-loading regression in v5 with clear reproduction and broad applicability. That said, the surrounding items are too diverse to treat this as a true duplicate cluster.", + "summary": "The cluster is heterogeneous: the soft pairs mostly share a subsystem label (tokenizers, processors, deepspeed, CI, model loading) but not the same concrete bug. None of the candidate pairs look mergeable as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 44462 is the strongest standalone issue in this set: it is specific, reproducible, and broadly relevant, unlike the others which are mostly unrelated regressions or narrow model-specific bugs.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 tokenizer regressions, but they affect different models and failure modes; not the same bug." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "One is AutoTokenizer ignoring tokenizer.json; the other is a Mistral regex patch crash. Different load paths." + }, { "left": "issue:41950", "right": "issue:43650", "accept": false, - "reason": "Video-classification processor bug vs. an unrelated placeholder/data request; no shared underlying code path." + "reason": "Unrelated topics: pipeline processor lookup vs a vague data request." }, { "left": "issue:33290", "right": "issue:45656", "accept": false, - "reason": "Adafactor OOM in DeepSpeed is a different optimizer/memory issue from optimizer.step being called twice." + "reason": "Both mention deepspeed/optimizer, but one is OOM and the other is double-stepping. Different defects." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Tokenizer class mismatch on save is a serialization metadata bug, while 44462 is about AutoTokenizer ignoring tokenizer.json during load." + "reason": "Both concern tokenizer persistence/loading, but one is save_pretrained metadata mismatch and the other is repository tokenizer.json being ignored." }, { "left": "issue:45081", "right": "issue:45907", "accept": false, - "reason": "Mistral regex patch crash and list_repo_files revision forwarding are unrelated bugs in different parts of the stack." + "reason": "Different bugs: Mistral tokenizer regex crash vs list_repo_files not forwarding revision." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error is unrelated to the init_empty_weights/_is_hf_initialized loading regression." + "reason": "Notebook syntax error and init_empty_weights argument error are unrelated." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Both are CI/dummy-config related, but one is stale auto_mappings references and the other is leftover dummy classes; different fixes." + "reason": "Both are repo-maintenance/CI issues, but they concern different files and failures." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "FlashAttention failure for Ministral-3 and OCR index errors in Kosmos2.5 are distinct model-specific bugs." + "reason": "Different model-specific failures: FlashAttention on Ministral vs OCR index error in Kosmos2.5." }, { "left": "issue:43450", "right": "issue:44279", "accept": false, - "reason": "Batched video processor shape bug is unrelated to a dependency/install issue." + "reason": "Incorrect batched video shape vs generic dependency issue; no shared code-path bug." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm and torch.split return handling in GlmMoeDsaIndexer are different low-level bugs." + "reason": "Different model internals: layernorm epsilon mismatch vs torch.split return handling." }, { "left": "issue:43065", "right": "issue:44898", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and Perceiver interpolation failure are different image-model issues." + "reason": "Different model components and symptoms; one is a dummy Conv2d, the other is Perceiver resolution handling." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Quantized model saving failure and dequantized model loading with device_map=auto are different quantization/offload paths." + "reason": "Both involve quantization/loading, but save failure and offload loading failure are distinct." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Inconsistent lm_head.weight serialization and resize_token_embeddings not affecting Gemma4 embeddings are not the same bug." + "reason": "Serialization of tied weights and resize_token_embeddings propagation are different bugs." }, { "left": "issue:43441", "right": "issue:44683", "accept": false, - "reason": "Both involve attention kernels, but Ministral FlashAttention and compiled flex_attention on torch>=2.9 are different failure modes." + "reason": "Different attention backends and environments; not the same concrete failure." }, { "left": "issue:43650", "right": "issue:45290", "accept": false, - "reason": "A vague placeholder issue is not the same as apply_chat_template crashing on tool-call assistant messages." + "reason": "Vague data request vs apply_chat_template crash on tool-call assistant messages." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Missing files in processor.save_pretrained and Gemma-4 zero3 from_pretrained are unrelated serialization vs loading bugs." + "reason": "Processor save_pretrained missing files and zero3 from_pretrained are unrelated." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained file loss is unrelated to a tokenizer-filtering performance regression." + "reason": "Missing processor save artifacts vs tokenizer throughput regression; different bug classes." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "BigBirdTokenizer special-token registration and D-FINE auxiliary losses are unrelated model/tokenizer behaviors." + "reason": "Tokenizer special-token registration and auxiliary-loss behavior are unrelated." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "AutoTokenizer loading regression and ESM2 broken are both broad release regressions, but they are not the same underlying bug." + "reason": "Tokenizer loading problem and ESM2 model breakage are not the same defect." }, { "left": "issue:43262", "right": "issue:44521", "accept": false, - "reason": "Audio chat-template sampling-rate default and multimodal assistant_masks being all-zero are different apply_chat_template issues." + "reason": "Both touch apply_chat_template, but one is audio sampling-rate defaulting and the other is multimodal assistant masks." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "Generation cache missing during chunked_prefill and CLIPVisionModel hidden_states being None are unrelated v5 regressions." + "reason": "Generation cache handling and CLIP hidden_states output are unrelated code paths." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Import NameError for sentence-transformers and undefined tf backend from tensorboard-only install are different dependency/import problems." + "reason": "Both are optional-backend import issues, but one is nn missing and the other is tf backend detection." }, { "left": "issue:44361", "right": "issue:45593", "accept": false, - "reason": "MLukeTokenizer AttributeError and D-FINE auxiliary-loss behavior do not share a code path." + "reason": "Tokenizer AttributeError and D-FINE auxiliary-loss behavior are unrelated." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "CamemBERT masked-LM prediction regression and DeepSeek tokenizer incorrect results are both tokenizer/model-output issues, but not the same bug." + "reason": "Both are v5 model regressions, but CamemBERT masked LM outputs and Deepseek tokenizer behavior are different bugs." }, { "left": "issue:44016", "right": "issue:45823", "accept": false, - "reason": "Notebook syntax error is unrelated to Gemma4 device-mismatch during forward." - }, - { - "left": "issue:42175", - "right": "issue:43976", - "accept": false, - "reason": "Missing TensorFlow in the torch extra and Python-version compatibility for Transformers 5.1.0 are different dependency/platform issues." + "reason": "Notebook syntax error vs Gemma4 device mismatch; no shared underlying issue." }, { "left": "issue:43901", "right": "issue:43994", "accept": false, - "reason": "Docs mentioning return_all_scores and SigLIP2 nonsensical outputs are not the same defect." + "reason": "Documentation mismatch in pipeline docs and wrong SigLIP2 outputs are not the same bug." }, { - "left": "issue:43122", - "right": "issue:43650", + "left": "issue:42175", + "right": "issue:43976", "accept": false, - "reason": "Tokenizer behavior regression and a placeholder 'ADD THE DATA' issue are unrelated." + "reason": "TensorFlow dependency packaging issue and Python version compatibility issue are distinct." } ] }, @@ -10415,6 +10223,7 @@ "issue:41950", "issue:42491", "issue:42757", + "issue:43122", "issue:43299", "issue:43352", "issue:43450", @@ -10423,6 +10232,7 @@ "issue:43531", "issue:43596", "issue:43643", + "issue:43650", "issue:43673", "issue:43824", "issue:43881", @@ -10444,14 +10254,12 @@ "issue:44589", "issue:44617", "issue:44749", - "issue:44779", "issue:44857", "issue:44871", "issue:44987", "issue:45003", "issue:45200", "issue:45250", - "issue:45292", "issue:45362", "issue:45397", "issue:45542", @@ -10460,6 +10268,7 @@ "issue:45593" ], "soft_pairs": [ + "issue:43122|issue:43650", "issue:44987|issue:45542", "issue:43824|issue:43976", "issue:44297|issue:45397", @@ -10468,8 +10277,8 @@ "issue:43525|issue:43643", "issue:43450|issue:44617", "issue:42491|issue:43299", - "issue:44589|issue:45362", "issue:43901|issue:44246", + "issue:44589|issue:45362", "issue:45561|issue:45563", "issue:41628|issue:43976", "issue:44336|issue:44373", @@ -10488,224 +10297,223 @@ "issue:44246|issue:44871", "issue:44242|issue:44857", "issue:44162|issue:45200", - "issue:36246|issue:43596", "issue:44295|issue:44749", - "issue:44779|issue:45292" + "issue:36246|issue:43596" ], - "prepared_review_unit_hash": "6d19bce909f9dfa32e7b4e9358a6f72cd90fe545b7aadd58f83d7443195a5909", + "prepared_review_unit_hash": "8c282ab5eaea7574b317f7ee6cdc1fde24cb3e030d104d68a5c3965f6811aa9d", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36672, - "estimated_input_tokens": 9168, - "estimated_eval_tokens": 18592 + "serialized_chars": 36619, + "estimated_input_tokens": 9155, + "estimated_eval_tokens": 18566 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a grab bag of Transformers v5-era compatibility regressions, import failures, and model-specific loading bugs. They are thematically related, but most soft pairs are not true duplicates of the same concrete failure path.", - "confidence": 0.63, - "canonical_issue_reason": "issue:44561 is the best umbrella representative: it describes a broad v5 compatibility regression that breaks trust_remote_code models, which matches the cluster\u2019s dominant import/load-breakage theme better than the narrower model-specific reports.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers bug reports. A small subset concerns import regressions, but the rest span tokenizer behavior, model-loading failures, docs, performance, and model-specific training/runtime bugs. None of the soft pairs look like true duplicates.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 41628 is the best single representative because it is a clear, user-facing top-level import regression and is closest to the recurring import/error-failure theme in the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44561 is the strongest global anchor for triage because it captures a version-wide breakage pattern rather than a single model or doc/test symptom, making it the most useful canonical issue for this mixed cluster.", + "best_issue_reason": "41628 is the most suitable cluster anchor: concrete, reproducible, and broadly understandable, while still being representative of the import-regression subset better than the model-specific or docs-only reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43122", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer behavior regression versus a placeholder/low-signal issue; no shared bug or code path." + }, { "left": "issue:44987", "right": "issue:45542", "accept": false, - "reason": "Different failures: one is a transformers>=5.1.0 model-loading break, the other is a TensorFlow backend/tensorboard installation issue." + "reason": "Different failure modes: model loading on transformers>=5.1.0 versus TensorFlow backend discovery when only tensorboard is installed." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "One is a Qwen2.5-VL import/export problem; the other is a Python-version compatibility complaint. Same library, different bug." + "reason": "Specific Qwen2_5_VL import failure versus a Python version compatibility problem; unrelated root causes." }, { "left": "issue:44297", "right": "issue:45397", "accept": false, - "reason": "Tokenizer config mismatch vs Gemma-4 zero3 from_pretrained failure; unrelated root causes." + "reason": "Tokenizer serialization mismatch and Gemma-4 ZeRO3 loading are different bugs in different paths." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Missing pad_token_id on Llama4Config is not the same as glm-4v loading failure." + "reason": "Llama4Config missing pad_token_id is unrelated to glm-4v-9b loading failure." }, { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "torch.compile attention-kernel divergence and ANSI codes in loading_report are unrelated." + "reason": "Torch.compile attention-kernel divergence and ANSI output in loading_report are completely unrelated." }, { "left": "issue:43525", "right": "issue:43643", "accept": false, - "reason": "Different code paths: Llama4Config attribute absence vs trust_remote_code AutoConfig field loss." + "reason": "Both mention config fields, but one is a missing model attribute and the other is a trust_remote_code config hydration issue." }, { "left": "issue:43450", "right": "issue:44617", "accept": false, - "reason": "Batched video processor shape bug is not the same as a CUDA OOM in Sam3Video." + "reason": "Batched video processor shape bug versus CUDA OOM in Sam3Video; not the same underlying issue." }, { "left": "issue:42491", "right": "issue:43299", "accept": false, - "reason": "Both involve Qwen MoE models, but they describe different version/model loading regressions and are not the same concrete bug." + "reason": "Both involve Qwen/moe models, but one is a HF4.x to HF5.x LoRA incompatibility and the other is a sliding-window loading regression." }, { - "left": "issue:44589", - "right": "issue:45362", + "left": "issue:43901", + "right": "issue:44246", "accept": false, - "reason": "Float8 storage lookup failure and Qwen3.5 chat crash are distinct issues." + "reason": "Docs mismatch for return_all_scores versus import-time slowness; different categories and symptoms." }, { - "left": "issue:43901", - "right": "issue:44246", + "left": "issue:44589", + "right": "issue:45362", "accept": false, - "reason": "Documentation mismatch about return_all_scores is not the same as intermittent slow import behavior." + "reason": "Float8 storage type error and Qwen3.5 chat crash are unrelated runtime failures." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "pytest-xdist file race and stale generate() warning are unrelated." + "reason": "Pytest-xdist file race and a stale generation warning are different parts of the codebase and different bugs." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "AutoImageProcessor import error and Python3.9+/3.10+ support issue are different failure modes." + "reason": "Top-level import error versus Python 3.9+/3.10+ support matrix issue; not the same defect." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "ANSI codes in loading_report and a wrong docstring for position_ids are separate documentation/UX problems." + "reason": "Terminal ANSI formatting bug and a wrong docstring for position_ids do not share a concrete code path." }, { "left": "issue:44242", "right": "issue:45593", "accept": false, - "reason": "Load-balancing loss gating and D-FINE auxiliary-loss behavior are different model-loss bugs." + "reason": "Both concern missing losses, but they affect different models and gating conditions; too broad to treat as one bug." }, { "left": "issue:43352", "right": "issue:45250", "accept": false, - "reason": "Generic Flash Attention 2.0 topic and NemotronHForCausalLM not supporting FA2 are not the same concrete issue." + "reason": "Nemotron Flash Attention 2 support gap is not the same as a generic Flash Attention 2 topic issue." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "Position_ids buffer read error and gemma-4 zero3 from_pretrained failure are unrelated." + "reason": "Position_ids buffer access bug versus Gemma-4 ZeRO3 loading failure; unrelated." }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Video-classification pipeline using image processors is not the same as _torch_extract_fbank_features()." + "reason": "Video-classification processor selection bug versus audio fbank extraction issue." }, { "left": "issue:44246", "right": "issue:44496", "accept": false, - "reason": "Slow import sometimes versus unrecognized model/config missing model_type are different bugs." + "reason": "Import-time performance issue and unrecognized model/config loading are unrelated." }, { "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Qwen3-MoE sliding_window issue and Qwen3.5 chat crash are different model-specific regressions." + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5 chat crash are different problems." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but for different symbols and packages; no shared underlying bug is evident." + "reason": "Both are import errors, but they concern different symbols from different packages and different fixes." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug and unsafe sys.modules access are unrelated." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to unsafe sys.modules access in modeling_utils." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Missing Qwen2_5_VLImageProcessor export and missing HybridCache export are both import failures, but they are different missing symbols with different causes." + "reason": "Different missing imports for different APIs; same symptom class, not the same bug." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI-code output issue is not the GLM-5 RoPE implementation problem." + "reason": "ANSI code emission and GLM-5 RoPE implementation are unrelated." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "Unsafe sys.modules access and xdist captured_info.txt race are separate infrastructure bugs." + "reason": "Unsafe sys.modules access and a pytest worker race do not overlap in code path or fix." }, { "left": "issue:43673", "right": "issue:44561", "accept": false, - "reason": "Chunked-prefill cache missing and removal of is_torch_fx_available breakage are different v5 regressions." + "reason": "Chunked-prefill cache handling and removal of is_torch_fx_available are different v5 regressions." }, { "left": "issue:44246", "right": "issue:44871", "accept": false, - "reason": "Slow import and Gemma-3 eos_token_id inconsistency are unrelated." + "reason": "Import slowness and eos_token_id configuration inconsistency are unrelated." }, { "left": "issue:44242", "right": "issue:44857", "accept": false, - "reason": "Load-balancing loss omission and LwDetrImageLoss AMP crash are different model issues." + "reason": "Load-balancing loss gating and a float16 AMP crash in LwDetrImageLoss are different bugs." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids defaulting are unrelated model bugs." - }, - { - "left": "issue:36246", - "right": "issue:43596", - "accept": false, - "reason": "Qwen2.5-VL import error and Deepspeed/BertModel index error are different failure paths." + "reason": "ESM2 breakage and Gemma-4 mm_token_type_ids defaults are unrelated model-specific issues." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "position_ids buffer access bug and tokenizer/filtering slowdown are not duplicates." + "reason": "Position_ids buffer access and slower filtering after upgrade are unrelated performance/runtime issues." }, { - "left": "issue:44779", - "right": "issue:45292", + "left": "issue:36246", + "right": "issue:43596", "accept": false, - "reason": "Deepseek tokenizer regression and resize_token_embeddings not updating output_embeddings are different tokenizer vs embedding-sync bugs." + "reason": "Qwen2_5_VL image-processor import error and BertModel zero3 index error are unrelated." } ] }, @@ -10737,7 +10545,6 @@ "issue:43408", "issue:43475", "issue:43540", - "issue:43646", "issue:43653", "issue:43673", "issue:43901", @@ -10753,6 +10560,7 @@ "issue:44496", "issue:44623", "issue:44734", + "issue:44779", "issue:44869", "issue:44877", "issue:44933", @@ -10779,6 +10587,7 @@ "issue:45759" ], "soft_pairs": [ + "issue:44779|issue:45292", "issue:44877|issue:45375", "issue:45292|issue:45584", "issue:45292|issue:45701", @@ -10786,8 +10595,8 @@ "issue:42886|issue:45229", "issue:43232|issue:44336", "issue:44496|issue:45003", - "issue:43065|issue:43408", "issue:44987|issue:45459", + "issue:43065|issue:43408", "issue:45375|issue:45759", "issue:43257|issue:45137", "issue:43262|issue:44246", @@ -10807,217 +10616,216 @@ "issue:44869|issue:45356", "issue:43232|issue:43673", "issue:44297|issue:44623", - "issue:43673|issue:45278", - "issue:43646|issue:44162" + "issue:43673|issue:45278" ], - "prepared_review_unit_hash": "de5b1d7355b370cd1e2f21581b6c185a22acc7510114b90cbc8e958c098df262", + "prepared_review_unit_hash": "f2016cf240b0e95c9506908e1f173f567e9e077d2c0874a217e207c7c55321ef", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36658, - "estimated_input_tokens": 9165, - "estimated_eval_tokens": 18586 + "serialized_chars": 36678, + "estimated_input_tokens": 9170, + "estimated_eval_tokens": 18596 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated issue reports across tokenizers, generation, model loading, and multimodal models. I found one plausible duplicate pair around `resize_token_embeddings` propagation; the rest look like separate bugs. No PRs were provided.", - "confidence": 0.66, - "canonical_issue_reason": "issue:45278 is the broadest umbrella item here and the best single representative of the cross-version regression/import-breakage theme; the others are narrower subsystem- or model-specific reports rather than near-duplicates.", + "summary": "This cluster is not a true duplicate set: it mixes unrelated tokenizer, config, generation, Deepspeed, and SAM/Qwen/Gemma issues. None of the soft pairs appear to be the same underlying bug or change.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45701 is the broadest, most cluster-like issue: a version-dependent tokenization regression. It is the closest representative of the tokenizer/version-regression subset, even though it is not a duplicate of the other items.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the least overfit choice for a cluster anchor because it is broad, current, and captures the general upgrade regression pattern better than the more specialized titles.", + "best_issue_reason": "issue:45701 is the best single issue to represent the loose theme of version-induced tokenizer behavior changes across the set. It is still only a thematic anchor, not a duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Deepseek tokenizer correctness regression vs resize_token_embeddings behavior; different code paths and symptoms." + }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Both involve config/strictness, but one is a load-time failure and the other is a missing-field serialization issue; different concrete bugs." + "reason": "Both involve config handling/strictness, but they target different models and different failure modes." }, { "left": "issue:45292", "right": "issue:45584", "accept": false, - "reason": "`resize_token_embeddings` vs Whisper empty-transcription failure after `align_special_tokens`; different code paths and symptoms." + "reason": "Embedding resize bug and Whisper empty-transcription generation crash are unrelated." }, { "left": "issue:45292", "right": "issue:45701", "accept": false, - "reason": "One is an embedding-resize bug, the other is a broad tokenization-change report; too general to be the same underlying issue." + "reason": "Resize-token-embeddings behavior is not the same as a tokenization change across versions." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "Docs mismatch for `return_all_scores` versus unsafe `sys.modules[]` access; unrelated bugs." + "reason": "Docs mismatch for a pipeline option is unrelated to unsafe sys.modules access in modeling_utils." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Cache/offline tokenizer loading versus Gemma4 multi-GPU OOM; no shared code-path." + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are different bugs." }, { "left": "issue:43232", "right": "issue:44336", "accept": false, - "reason": "Generation kwargs/sync-gpus bug versus ANSI output in loading report; unrelated." + "reason": "Generation kwargs sync issue is unrelated to ANSI output in loading_report." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Model-type/config loading failure versus `sys.modules` access bug; different root causes." + "reason": "Unrecognized model/config loading error and sys.modules access bug do not share the same code defect." }, { - "left": "issue:43065", - "right": "issue:43408", + "left": "issue:44987", + "right": "issue:45459", "accept": false, - "reason": "Different SAM 3 issues: dummy conv layer versus model-type warning." + "reason": "Model loading failure for a specific repo and protobuf-related tokenizer error masking are distinct." }, { - "left": "issue:44987", - "right": "issue:45459", + "left": "issue:43065", + "right": "issue:43408", "accept": false, - "reason": "Loading a specific model on transformers>=5.1.0 versus tokenizer error masking when protobuf is absent; not the same bug." + "reason": "Both are SAM3-related, but one is a dummy Conv2d concern and the other is a model-type mismatch warning." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Missing config field under strict loading versus `from_config` not unwrapping `text_config`; different model/config problems." + "reason": "Missing config field under strict handling and text_config unwrapping for composite configs are different loading bugs." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Qwen3 MOE weight conversion with accelerate+deepspeed versus ZeRO3 deque underflow; separate training/runtime failures." + "reason": "Deepspeed weight conversion issue and ZeRO3 deque underflow are unrelated runtime failures." }, { "left": "issue:43262", "right": "issue:44246", "accept": false, - "reason": "Audio sampling-rate default bug versus slow import complaint; unrelated." + "reason": "Audio sampling-rate default mismatch is unrelated to slow import performance." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "Stale indexer cache on second forward pass versus Gemma4 mm_token_type_ids defaulting; different mechanisms." + "reason": "Stale indexer cache on second forward pass is different from missing mm_token_type_ids defaulting." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Tied-weights warning bug versus patchtsmixer `post_init` allowance; only loosely thematically related." + "reason": "Incorrect tied-weights warning and patchtsmixer post_init validation are unrelated." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "Both are video/Qwen/SAM adjacent, but one is a missing attribute in SAM 3 video and the other is a video-input validation error in Qwen3OmniMoe." + "reason": "SAM3 video encoder output attribute error and Qwen3OmniMoe video-input processing failure are different model bugs." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Both touch embedding/tied-weight behavior, but one is about inconsistent serialization by device and the other is about `resize_token_embeddings` not updating decoder embeddings; not clearly the same bug." + "reason": "Serialization inconsistency for lm_head.weight is not the same as resize_token_embeddings not updating decoder.embed_tokens." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBird tokenizer CI decode bug versus ESM2 breakage; separate models and failure modes." + "reason": "BigBirdTokenizer special-token registration bug and ESM2 breakage are not the same defect." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "Stale cache bug in GlmMoeDsa versus FlashAttention `s_aux=None` crash; unrelated." + "reason": "Second-forward stale cache crash is unrelated to flash_attention handling of s_aux=None." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "Qwen3.5 tie-word-embeddings warning versus stale `num_return_sequences` warning in paged generate; different warnings and causes." + "reason": "Both mention warnings, but they concern different generation behaviors and different warning conditions." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "Warning about tied embeddings versus flash-attention generation malfunction; not the same underlying issue." + "reason": "Tie-word-embeddings warning is not the same as flash-attention generation failure." }, { "left": "issue:45397", "right": "issue:45656", "accept": false, - "reason": "Both mention Deepspeed, but one is a zero3 `from_pretrained` issue and the other is double optimizer stepping; different layers of the stack." + "reason": "Gemma-4 ZeRO3 loading and double optimizer step under deepspeed are different issues." }, { "left": "issue:43673", "right": "issue:44734", "accept": false, - "reason": "Both concern generation/cache, but one is missing cache in v5 chunked prefill and the other is wrong tensor indexing in serve continuation; not a duplicate." + "reason": "Chunked-prefill cache missing and KV-cache continuation tensor indexing are both generation/cache related but not the same bug." }, { "left": "issue:45137", "right": "issue:45600", "accept": false, - "reason": "Both touch training/runtime behavior, but one is ZeRO3 deque errors and the other is stale `auto_mappings` references breaking CI." + "reason": "ZeRO3 deque pop error and stale Sam3LiteText references in auto_mappings are unrelated." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Shape mismatch loading Qwen3-VL versus Qwen3.5 tie-embeddings warning; different model families and failures." + "reason": "Qwen3-VL weight-shape mismatch and tie_word_embeddings warning are different loading/model-config problems." }, { "left": "issue:44933", "right": "issue:45459", "accept": false, - "reason": "Missing import from `image_utils` versus tokenizer error hiding when protobuf is missing; unrelated." + "reason": "Missing import from image_utils is unrelated to tokenizer error masking when protobuf is absent." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper word-timestamp decode crash versus Kimi-K2.5 tokenizer regression; different model and code path." + "reason": "Whisper trailing-replacement decode crash and Kimi tokenizer codec regression are separate tokenizer bugs." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "Both are generation-related, but one is `update_model_kwargs_for_generation` after `sync_gpus` and the other is chunked-prefill cache handling; not the same bug." + "reason": "Both are generation-related, but one is kwargs syncing and the other is cache availability in chunked prefill." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Tokenizer save-pretrained mismatch versus processor save-pretrained missing files; both serialization, but different artifact types and fixes." + "reason": "tokenizer.save_pretrained metadata mismatch and processor.save_pretrained missing files are different save-path bugs." }, { "left": "issue:43673", "right": "issue:45278", "accept": false, - "reason": "Generation cache issue versus broad import-error regression after upgrade; too different." - }, - { - "left": "issue:43646", - "right": "issue:44162", - "accept": false, - "reason": "Transformers 5.0.0 custom model init regression versus ESM2 breakage; different scope and likely fixes." + "reason": "Generation cache missing in chunked prefill and broad import failures after upgrade are not the same underlying issue." } ] }, @@ -11056,19 +10864,19 @@ "issue:43475", "issue:43482", "issue:43525", - "issue:43526", "issue:43643", + "issue:43646", "issue:43673", "issue:43704", "issue:43873", "issue:44038", + "issue:44162", "issue:44164", "issue:44188", "issue:44222", "issue:44242", "issue:44351", "issue:44361", - "issue:44496", "issue:44623", "issue:44661", "issue:44749", @@ -11085,286 +10893,281 @@ "issue:45381", "issue:45397", "issue:45478", - "issue:45561", "issue:45656", "issue:45712", "issue:45753" ], "soft_pairs": [ + "issue:43646|issue:44162", "issue:44361|issue:44749", "issue:44623|issue:45656", "issue:44749|issue:45656", "issue:28282|issue:30990", "issue:44749|issue:45310", "issue:42994|issue:43482", - "issue:42371|issue:45092", "issue:43475|issue:43643", + "issue:42371|issue:45092", "issue:43475|issue:45381", "issue:45230|issue:45397", "issue:44038|issue:44623", "issue:41628|issue:44933", "issue:45081|issue:45103", - "issue:43475|issue:44661", "issue:42994|issue:43525", + "issue:43475|issue:44661", "issue:45375|issue:45478", "issue:42673|issue:43704", "issue:43232|issue:44188", "issue:45310|issue:45375", "issue:43873|issue:44945", - "issue:41628|issue:44351", "issue:42831|issue:44222", + "issue:41628|issue:44351", "issue:44945|issue:45753", "issue:44222|issue:45310", "issue:42222|issue:45200", "issue:43673|issue:44188", - "issue:36010|issue:41628", "issue:42994|issue:44164", + "issue:36010|issue:41628", "issue:42175|issue:45712", "issue:43232|issue:44792", "issue:44038|issue:45310", - "issue:43377|issue:44242", - "issue:43526|issue:45561", - "issue:44496|issue:45561" + "issue:43377|issue:44242" ], - "prepared_review_unit_hash": "9acbcdb57f81667edb262718daa46753974ecdea5669f40f5ab94da8d08a48ef", + "prepared_review_unit_hash": "75fcb2a0ff93d32c98527c4950fddebcc1afea3f916f3479f84cada4193c160a", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37594, - "estimated_input_tokens": 9399, - "estimated_eval_tokens": 19054 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36602, + "estimated_input_tokens": 9151, + "estimated_eval_tokens": 18558 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed set of unrelated Transformers bug reports. A few pairs are superficially similar around imports, save/load, or Qwen3.5/Qwen3.5-MoE regressions, but none are strong enough to merge as duplicates under a strict triage standard. No PRs are present.", - "confidence": 0.63, - "canonical_issue_reason": "Issue 45310 is the clearest representative of the recurring Qwen3.5 MoE from_pretrained regression theme; it is specific, reproducible, and matches the most obvious repeated symptom among the set.", + "summary": "Mostly a heterogeneous set of Transformers-v5 regression reports across imports, model init/load, quantization, and generation. Only one soft edge looks like a true duplicate: the repeated Qwen3ForCausalLM VRAM-leak report. The Qwen3.5 MoE-related items are similar in theme but still appear to be distinct bugs rather than the same concrete failure.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the broadest umbrella report in the set: a Transformers 5.0 custom-model initialization regression. It is the best single anchor for the overall theme, even though the cluster contains several distinct sub-bugs.", "canonical_pr_reason": null, - "best_issue_reason": "45310 is the best single issue to anchor the cluster because it captures a concrete model-loading failure for a widely recurring model family, while the other issues are either broader, more specific variants, or unrelated regressions.", + "best_issue_reason": "issue:43646 is the most representative issue because it describes the highest-level regression pattern rather than one model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Both are v5-init regressions, but one is generic custom-model init breakage and the other is ESM2-specific; not clearly the same bug." + }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Different failures: tokenizer/task AttributeError vs data-filtering slowdown after upgrade." + "reason": "Tokenizer AttributeError vs upgrade-induced slowdown; different failure modes." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Save-pretrained file handling vs Deepspeed double optimizer-step bug; unrelated code paths." + "reason": "save_pretrained missing files vs deepspeed optimizer stepping twice; different code paths." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "One is a performance regression, the other is training-step duplication; not the same bug." + "reason": "Performance regression in data filtering is unrelated to a duplicated optimizer step." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "PyTorch ImportError vs sentence-transformer loading hang; different symptoms and causes." + "reason": "PyTorch ImportError vs Sentence Transformers loading hang; separate runtime issues." }, { "left": "issue:44749", "right": "issue:45310", "accept": false, - "reason": "Filtering slowdown is unrelated to Qwen3.5 MoE from_pretrained failure." + "reason": "Slow filtering after upgrade is not the same as Qwen3.5 MoE from_pretrained failure." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized save failure vs GGUF loading failure; opposite directions and different mechanisms." + "reason": "Quantized saving failure vs GGUF loading failure; save/load bugs, not one issue." }, { - "left": "issue:42371", - "right": "issue:45092", + "left": "issue:43475", + "right": "issue:43643", "accept": false, - "reason": "TF32 settings warning is unrelated to remote-code/meta-init incompatibility." + "reason": "SAM3 attribute error vs missing fields from trust_remote_code config; different objects and causes." }, { - "left": "issue:43475", - "right": "issue:43643", + "left": "issue:42371", + "right": "issue:45092", "accept": false, - "reason": "Missing attribute on SAM 3 output vs missing fields from AutoConfig with trust_remote_code." + "reason": "TF32 API warning is unrelated to InternVL2 meta-initialization incompatibility." }, { "left": "issue:43475", "right": "issue:45381", "accept": false, - "reason": "Different model families and different vision-token issues." + "reason": "Different models and different attribute/position-id failures." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "Generic report vs Gemma-4 zero3 from_pretrained bug; no shared underlying failure." + "reason": "Generic bug report vs gemma-4 zero3 from_pretrained issue; no clear overlap." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe issue vs processor save_pretrained missing files; not the same bug." + "reason": "Qwen3-VL-Moe from_pretrained bug and processor.save_pretrained missing files are different components." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Both are import-related, but they concern different symbols and likely different API breakages." + "reason": "Both involve imports, but AutoImageProcessor and image_utils failures are different symbols/code paths." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Tokenizer regex patch crash vs auto_docstring annotation crash; separate helper paths." + "reason": "Tokenizer regex crash vs auto_docstring annotations crash; unrelated." }, { - "left": "issue:43475", - "right": "issue:44661", + "left": "issue:42994", + "right": "issue:43525", "accept": false, - "reason": "SAM output attribute bug and TOKENIZER_MAPPING_NAMES model registration bug are unrelated." + "reason": "Quantized save failure vs missing pad_token_id on Llama4Config; distinct regressions." }, { - "left": "issue:42994", - "right": "issue:43525", + "left": "issue:43475", + "right": "issue:44661", "accept": false, - "reason": "Quantized saving bug vs missing pad_token_id on Llama4Config; different failures." + "reason": "SAM3 output attribute error vs tokenizer-mapping add-new-model-like failure; unrelated." }, { "left": "issue:45375", "right": "issue:45478", "accept": false, - "reason": "Both are Qwen3.5-MoE-adjacent, but one is a specific config-field drop and the other is a broader from_pretrained error; not clearly the same concrete bug." + "reason": "Missing config field in Qwen3_5MoeVisionConfig vs Qwen3.5 MoE from_pretrained error; same family, different bug." }, { "left": "issue:42673", "right": "issue:43704", - "accept": false, - "reason": "Same title about VRAM leaks, but the later report is a separate report without evidence it is the same underlying defect." + "accept": true, + "reason": "Exact same title and same Qwen3ForCausalLM VRAM-leak symptom in multiple dataloader threads; this is the same report." }, { "left": "issue:43232", "right": "issue:44188", "accept": false, - "reason": "Generation kwarg sync issue vs torch.compile attention-kernel divergence; different code paths." + "reason": "Generation kwargs update issue vs torch.compile attention-kernel branching issue; different code paths." }, { "left": "issue:45310", "right": "issue:45375", "accept": false, - "reason": "Qwen3.5 MoE from_pretrained failure is not the same as missing deepstack_visual_indexes in a vision config." + "reason": "Qwen3.5 MoE from_pretrained error and missing vision config field are related but not the same concrete failure." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Quantization/offloading bug vs incorrect pipeline-parallel output; too different to merge." + "reason": "Quantization/offloading behavior vs incorrect pipeline-parallel output; different problems." }, { - "left": "issue:41628", - "right": "issue:44351", + "left": "issue:42831", + "right": "issue:44222", "accept": false, - "reason": "Both import errors, but different missing symbols with no evidence of one root cause." + "reason": "FP8 accuracy regression vs FP8 save_pretrained failure; different symptoms." }, { - "left": "issue:42831", - "right": "issue:44222", + "left": "issue:41628", + "right": "issue:44351", "accept": false, - "reason": "FineGrainedFP8 accuracy regression vs FP8 save_pretrained/moe serialization bug." + "reason": "AutoImageProcessor import error vs HybridCache import error; separate missing exports." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Pipeline-parallel output issue vs Qwen3_5 infinite loop on a specific image; unrelated symptoms." + "reason": "Pipeline-parallel output bug vs infinite loop on a specific image; unrelated." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "FP8 save-pretrained issue is not the same as Qwen3.5 MoE from_pretrained loading failure." + "reason": "FP8 save_pretrained on MoE vs Qwen3.5 MoE from_pretrained error; different operation and failure mode." }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "VitPose breakage vs Gemma 4 token-type defaulting; different model families and bugs." + "reason": "VitPose breakage vs Gemma 4 mm_token_type_ids defaulting; unrelated model families." }, { "left": "issue:43673", "right": "issue:44188", "accept": false, - "reason": "Chunked-prefill cache regression vs attention kernel branching divergence; different failure modes." + "reason": "Chunked_prefill cache regression vs torch.compile bidirectional-skip branching issue; distinct generation bugs." }, { - "left": "issue:36010", - "right": "issue:41628", + "left": "issue:42994", + "right": "issue:44164", "accept": false, - "reason": "Both are import errors, but for different APIs and likely separate compatibility regressions." + "reason": "Quantized model saving failure vs extra_state save/from_pretrained handling; serialization-related but different defects." }, { - "left": "issue:42994", - "right": "issue:44164", + "left": "issue:36010", + "right": "issue:41628", "accept": false, - "reason": "Quantized model save failure vs extra_state handling in save/from_pretrained; not the same concrete issue." + "reason": "GenerationMixin import error vs AutoImageProcessor import error; different missing symbols." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Packaging/backend dependency issue vs leftover dummy class leakage; unrelated maintenance problems." + "reason": "TensorFlow backend packaging issue vs leftover dummy PyTorch classes; different packaging problems." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Generation kwargs update bug vs failing janus image-generation test; no shared root cause." + "reason": "Generation kwargs bug vs janus generate test failure; not the same underlying issue." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Different Qwen model variants and different reported failures; too broad to merge." + "reason": "Different Qwen MoE variants and different failure modes; not a clear duplicate." }, { "left": "issue:43377", "right": "issue:44242", "accept": false, - "reason": "Padding-mask batch inconsistency vs router-logits load-balancing loss bug; unrelated model behavior." - }, - { - "left": "issue:43526", - "right": "issue:45561", - "accept": false, - "reason": "Image processor label reduction bug vs xdist file race; separate areas." - }, - { - "left": "issue:44496", - "right": "issue:45561", - "accept": false, - "reason": "Model-type/config loading failure vs test-file race condition; unrelated." + "reason": "MIMI batching/padding-mask bug vs router-logits load-balancing-loss bug; unrelated architecture issues." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, - "accepted_nontrivial_soft_edge": false, + "evaluator_result": { + "accept": false, + "feedback": "The cluster-level summary is slightly inconsistent with the structured verdicts: it says only two soft edges look like true duplicates, but only one pair is actually accepted. More importantly, the Qwen3.5 MoE pair is described as a duplicate-like case in the summary even though the verdict marks it false, so the final writeup is not fully grounded in the packet. The rest of the conservative rejections look reasonable." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -11385,15 +11188,14 @@ "issue:30990", "issue:33290", "issue:35141", - "issue:41628", "issue:42175", - "issue:42757", "issue:42831", "issue:42898", "issue:43116", "issue:43278", "issue:43479", "issue:43504", + "issue:43526", "issue:43644", "issue:43723", "issue:43784", @@ -11410,6 +11212,7 @@ "issue:44297", "issue:44373", "issue:44448", + "issue:44496", "issue:44568", "issue:44623", "issue:44734", @@ -11430,14 +11233,16 @@ "issue:45812" ], "soft_pairs": [ + "issue:43526|issue:45561", + "issue:44496|issue:45561", "issue:43504|issue:45561", "issue:44964|issue:45137", "issue:42898|issue:44568", "issue:42175|issue:44933", "issue:30990|issue:35141", "issue:43479|issue:44869", - "issue:43976|issue:44933", "issue:43479|issue:44079", + "issue:43976|issue:44933", "issue:42898|issue:45812", "issue:43784|issue:43824", "issue:44188|issue:44373", @@ -11460,229 +11265,234 @@ "issue:43479|issue:43644", "issue:44206|issue:45103", "issue:44222|issue:44623", - "issue:42831|issue:44623", - "issue:41628|issue:42757" + "issue:42831|issue:44623" ], - "prepared_review_unit_hash": "e1178a2569d79f736885b2edab480e0d3d9dc781bc13ba2bde83ffc5f5a3015c", + "prepared_review_unit_hash": "9ef9e86958d1a1a7427bda638dd6080cf321840f7ca15836accd358d611f2a68", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 32, - "serialized_chars": 36785, - "estimated_input_tokens": 9197, - "estimated_eval_tokens": 18650 + "soft_pair_count": 33, + "serialized_chars": 37179, + "estimated_input_tokens": 9295, + "estimated_eval_tokens": 18846 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes tokenizer regressions, import/load failures, multimodal model/config issues, and a few unrelated runtime bugs. No soft pair looks like a true duplicate under a strict same-code-path standard.", - "confidence": 0.72, - "canonical_issue_reason": "issue:42898 is the broadest, most representative tokenizer-regression report in the set (`clean_up_tokenization_spaces` behavior change in v5), and it has the strongest thematic overlap with several other tokenizer-related issues.", + "summary": "The cluster is highly heterogeneous: model loading, tokenizer regressions, training/runtime errors, and tooling issues. None of the soft pairs look like the same underlying bug, so all are rejected.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43824 is the strongest canonical candidate: it has high discussion/inbound-reference activity and a clear, user-facing, concrete failure mode with enough detail to anchor duplicate triage.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the best single issue to anchor this cluster because it is concrete, reproducible, and broadly scoped enough to represent the tokenizer-behavior regression subgroup better than the more narrowly named reports.", + "best_issue_reason": "issue:43824 is the best standalone issue in this set because it is well-scoped, highly referenced, and describes a specific reproducible import failure.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "Different subsystems and failures: BEiT label reduction vs a pytest-xdist file race." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Unrecognized-model/config loading error vs test harness race on captured_info.txt." + }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "Pretrained-model legacy-field loading bug vs pytest-xdist file-race in testing utils; unrelated subsystems and failure modes." + "reason": "Pretrained preset loading/legacy field issue is unrelated to xdist test-worker file contention." }, { "left": "issue:44964", "right": "issue:45137", "accept": false, - "reason": "Phi-4 multimodal model loading error vs DeepSpeed ZeRO3 deque underflow; no shared code path." + "reason": "Model loading failure for Phi-4 multimodal is unrelated to DeepSpeed ZeRO3 deque underflow." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer regressions, but one is a `clean_up_tokenization_spaces` behavior change and the other is BOS/EOS injection failure for mdeberta; different bugs." + "reason": "Both involve tokenization, but one is a v5 behavior change while the other is tokenizer-specific BOS/EOS insertion regression." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "Missing TensorFlow backend in install extras vs a nonexistent `image_utils` import; different import/package issues." + "reason": "Backend install packaging issue vs missing/nonexistent image_utils import; different failure classes." }, { "left": "issue:30990", "right": "issue:35141", "accept": false, - "reason": "Sentence-transformers loading hang vs embedding resize/post-init reinitialization; unrelated." + "reason": "Sentence-Transformers loading hang is unrelated to token embedding reinitialization in post_init." }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Multimodal config default handling vs Whisper timestamp decoding crash; different model paths." + "reason": "Multimodal config default initialization bug vs Whisper word-timestamp decode crash; different code paths." }, { - "left": "issue:43976", - "right": "issue:44933", + "left": "issue:43479", + "right": "issue:44079", "accept": false, - "reason": "Python version compatibility regression vs missing import from `image_utils`; unrelated." + "reason": "Config initialization with None is unrelated to ModelOutput key assignment semantics." }, { - "left": "issue:43479", - "right": "issue:44079", + "left": "issue:43976", + "right": "issue:44933", "accept": false, - "reason": "Config initialization bug vs `ModelOutput` key assignment bug; not the same underlying defect." + "reason": "Python version compatibility regression vs missing import from image_utils." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Both involve tokenizers, but one is a v5 cleanup-space behavior change and the other is wrong Granite token IDs; different concrete regressions." + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; not the same concrete bug." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Import-time `nn` NameError vs missing `Qwen2_5_VLForConditionalGeneration` export; distinct import failures." + "reason": "Sentence-transformers import NameError is unrelated to Qwen2.5-VL missing export/import error." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "Attention-kernel divergence under `torch.compile` vs docstring text issue; no code-path overlap." + "reason": "torch.compile attention-kernel divergence and a docstring typo are unrelated." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Processor save/export missing files vs Qwen3.5 MoE `from_pretrained` failure; different loading/saving paths." + "reason": "Missing processor-save files vs Qwen3.5-Moe from_pretrained failure; different loading/saving paths." }, { "left": "issue:43278", "right": "issue:45137", "accept": false, - "reason": "Dtype mismatch between train/eval vs DeepSpeed ZeRO3 deque error; unrelated behavior." + "reason": "Embedding dtype drift during eval is unrelated to DeepSpeed ZeRO3 deque errors." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "Serving KV-cache indexing crash vs auto-docstring crash under future annotations; unrelated." + "reason": "KV-cache continuation crash in serve is unrelated to auto_docstring crashing on future annotations." }, { "left": "issue:30333", "right": "issue:33290", "accept": false, - "reason": "MLflow reporting failure on job failure vs OOM with Adafactor+DeepSpeed; different training issues." + "reason": "MLFlow reporting on failed training jobs is unrelated to adafactor/deepspeed OOM." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "Generic tokenizer loading issue vs a specific BOS/EOS special-token regression; not the same bug." + "reason": "General AutoTokenizer loading failure is not the same bug as mdeberta BOS/EOS handling regression." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer regressions: cleanup-space behavior vs Kimi codec/regex handling." + "reason": "Both are tokenizer regressions, but the affected models and codec/warning behavior differ; no same underlying bug evidence." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Wrong tied-weights warning in Qwen3-Next vs missing `_tp_plan` in Qwen3.5 MoE; separate model-definition issues." + "reason": "Tied-weights warning bug in Qwen3-Next is unrelated to missing _tp_plan in Qwen3.5-Moe." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Missing image-processor method vs remote-code meta-init incompatibility; distinct multimodal loading problems." + "reason": "Missing image-patch method in GLM46V is unrelated to old InternVL2 remote-code meta-init incompatibility." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Different tokenizer regressions: cleanup-space behavior vs Deepseek tokenization correctness." + "reason": "Different tokenizer regressions affecting different models; no indication they are the same fix." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "Dependency version bump to unreleased PEFT vs leftover dummy classes; unrelated maintenance issues." + "reason": "PEFT version pin mismatch is unrelated to dummy class cleanup/check_repo failures." }, { "left": "issue:42898", "right": "issue:44448", "accept": false, - "reason": "Both mention v5 output differences, but one is a tokenizer behavior change and the other is model output drift for Pegasus; not the same underlying bug." + "reason": "Pegasus output drift and clean_up_tokenization_spaces behavior change are different regression reports." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Buffer-read error for `position_ids` vs optimizer step being called twice under DeepSpeed; unrelated." + "reason": "Buffer read error for position_ids is unrelated to DeepSpeed double optimizer-step behavior." }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "Missing import from `image_utils` vs stale dummy classes leaking into `dir(transformers)`; different code paths." + "reason": "Missing image_utils import and leftover dummy classes are unrelated packaging/codegen issues." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "FP8 save/load issue for MoE vs tokenizer metadata mismatch on save; unrelated serialization bugs." + "reason": "FP8 MoE save_pretrained bug is unrelated to tokenizer_class metadata mismatch on save_pretrained." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Example script returning empty multi-label results vs local dataset loading failure in a different example script; not duplicates." + "reason": "Multi-label example returning empty results is unrelated to local dataset loading in another example script." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer save metadata bug vs DeepSpeed optimizer-step bug; unrelated." + "reason": "Tokenizer save metadata bug is unrelated to DeepSpeed optimizer-step duplication." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Phi4 multimodal config default handling vs non-persistent buffer initialization regression; different causes." + "reason": "Multimodal config defaults and junk-filled non-persistent buffers are different initialization problems." }, { "left": "issue:44206", "right": "issue:45103", "accept": false, - "reason": "Feature extractor argument incompatibility vs auto-docstring annotation crash; unrelated." + "reason": "Feature extractor center-arg crash is unrelated to auto_docstring annotation handling." }, { "left": "issue:44222", "right": "issue:44623", "accept": false, - "reason": "MoE FP8 save_pretrained bug vs processor save_pretrained missing files; different save/export paths." + "reason": "FP8 MoE save_pretrained issue is not the same as processor.save_pretrained missing files." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FP8 accuracy regression vs processor export missing files; unrelated." - }, - { - "left": "issue:41628", - "right": "issue:42757", - "accept": false, - "reason": "Missing `AutoImageProcessor` export vs missing `is_offline_mode` import; both are import errors but from different packages and fixes." + "reason": "FineGrainedFP8 accuracy regression is unrelated to processor save_pretrained file omissions." } ] }, @@ -11755,6 +11565,7 @@ "issue:45834" ], "soft_pairs": [ + "issue:41628|issue:42757", "issue:22355|issue:28282", "issue:30333|issue:36331", "issue:43232|issue:45834", @@ -11765,8 +11576,8 @@ "issue:28282|issue:30333", "issue:30333|issue:35141", "issue:44263|issue:45092", - "issue:44483|issue:45406", "issue:44263|issue:44360", + "issue:44483|issue:45406", "issue:44373|issue:44485", "issue:44514|issue:45103", "issue:44295|issue:44623", @@ -11787,220 +11598,226 @@ "issue:43756|issue:44060", "issue:43232|issue:45593" ], - "prepared_review_unit_hash": "8e214b43c02ce334ca50d51cb2b4671bd4835ac2b44fea3c0cd588376d326368", + "prepared_review_unit_hash": "887ba8a34e842c77239a2b062eee7b302cf3b78b7d5aaab97487d7847dcc112e", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36138, - "estimated_input_tokens": 9035, - "estimated_eval_tokens": 18326 + "soft_pair_count": 32, + "serialized_chars": 36440, + "estimated_input_tokens": 9110, + "estimated_eval_tokens": 18476 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose collection of Transformers issues centered on import/loading/tokenizer/model-regression failures, but most items are distinct root causes rather than true duplicates. The broadest representative is the v5 upgrade/import-regression report.", - "confidence": 0.43, - "canonical_issue_reason": "issue:45278 is the broadest umbrella report here: it captures the general v5 upgrade/import-failure theme better than the narrower class/model-specific bugs.", + "summary": "The set is not a true duplicate cluster: it mixes import/load regressions, tokenizer/model-specific bugs, save_pretrained issues, and unrelated training/runtime problems. Most soft pairs are only superficially similar and should be kept separate.", + "confidence": 0.37, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella here, covering multiple import regressions after a Transformers upgrade, which matches the largest visible theme in the cluster better than the narrower model-specific bugs.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best single representative for the cluster because it is the most generic user-facing regression and most likely to subsume the shared import/load failure pattern.", + "best_issue_reason": "Issue 45278 is the best single representative because it is the most general and open-ended report in the set; the rest are too specific or unrelated to serve as a canonical duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Different failing imports from different packages/components: AutoImageProcessor in transformers vs is_offline_mode from huggingface_hub." + }, { "left": "issue:22355", "right": "issue:28282", "accept": false, - "reason": "Both are import errors, but one is about transformers.onnx and the other about missing PyTorch; different missing-dependency failures." + "reason": "One is missing transformers.onnx; the other is missing PyTorch for AutoModel. Different dependency failures." }, { "left": "issue:30333", "right": "issue:36331", "accept": false, - "reason": "MLflow reporting on failed training jobs is unrelated to CustomTrainer.compute_loss signature changes." + "reason": "Unrelated problems: MLflow reporting on failed training vs CustomTrainer.compute_loss signature breakage." }, { "left": "issue:43232", "right": "issue:45834", "accept": false, - "reason": "Generation state update after sync_gpus is unrelated to Kosmos2.5 long OCR input indexing." + "reason": "Different generation/pathology issues: sync_gpus model kwargs handling vs Kosmos2.5 long OCR index error." }, { "left": "issue:30333", "right": "issue:34689", "accept": false, - "reason": "Training-job reporting and Llama 3.2 Vision model loading are different code paths and failures." + "reason": "Training-job reporting bug vs model loading regression for Llama 3.2 vision are not the same bug." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 save_pretrained for MoE models is not the same as a question about max_shard_size defaults." + "reason": "Both touch save_pretrained, but one is FP8/MoE saving correctness and the other is shard-size behavior." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer class metadata mismatch on save is different from Qwen3.5 Moe from_pretrained loading failure." + "reason": "Tokenizer config mismatch on save_pretrained is unrelated to Qwen3.5-Moe from_pretrained loading errors." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Both mention Qwen3.5/tokenizer-related loading issues, but one is generic tokenizer loading in v5 and the other is a flash-attention generation bug." + "reason": "Generic tokenizer loading in v5 vs flash-attention generation failure are different code paths and symptoms." }, { "left": "issue:28282", "right": "issue:30333", "accept": false, - "reason": "Missing PyTorch import and MLflow status reporting are unrelated failures." + "reason": "Missing PyTorch import dependency and MLflow reporting on failed training are unrelated." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "Training-job reporting is unrelated to embedding resize/post_init reinitialization." + "reason": "Training failure reporting vs token-embedding reinitialization in post_init are different bugs." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "DSA torch.split indexing behavior is unrelated to remote-code checkpoint meta initialization." + "reason": "Different model internals: GLM DSA indexer split behavior vs InternVL2 meta-initialization incompatibility." }, { - "left": "issue:44483", - "right": "issue:45406", + "left": "issue:44263", + "right": "issue:44360", "accept": false, - "reason": "Chat/completions request acceptance and Gemma4Processor missing _tokenizer are different bugs." + "reason": "Same broad DSA area, but one is torch.split return values and the other is a missing ReLU; not the same concrete bug." }, { - "left": "issue:44263", - "right": "issue:44360", + "left": "issue:44483", + "right": "issue:45406", "accept": false, - "reason": "Both concern the DSA indexer, but one is about torch.split return values and the other about a missing ReLU; not the same defect." + "reason": "API request handling in chat/completions is unrelated to Gemma4Processor crashing in serve due to missing _tokenizer." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Docstring issues for position_ids and GLM-5 RoPE implementation are not the same underlying bug." + "reason": "Wrong docstring for position_ids is documentation-only; GLM-5 RoPE implementation is a code behavior issue." }, { "left": "issue:44514", "right": "issue:45103", "accept": false, - "reason": "Qwen2_5_VLProcessor batched chat-template crash is unrelated to the auto_docstring AttributeError." + "reason": "Batched chat-template crash with padding=False is unrelated to AttributeError in auto_docstring with future annotations." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Position_ids buffer read error and processor.save_pretrained missing files are different save/load behaviors." + "reason": "Reading position_ids after buffer registration and processor.save_pretrained missing files are different problems." }, { "left": "issue:43723", "right": "issue:44871", "accept": false, - "reason": "Generic tokenizer loading in v5 and Gemma-3 eos_token_id config mismatch are distinct issues." + "reason": "Tokenizer loading regression and eos_token_id inconsistency are different classes of tokenizer/model config issues." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA q_a_layernorm epsilon precision issue is unrelated to InternVL2 meta initialization incompatibility." + "reason": "MLA layernorm epsilon precision issue is unrelated to remote-code checkpoint/meta-init compatibility." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Meta-initialization incompatibility and incorrect visual encoder keys on save_pretrained are different defects." + "reason": "Meta-initialization incompatibility and incorrect visual encoder keys on save_pretrained are different bugs." }, { "left": "issue:43723", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer load failures, but they target different checkpoints and likely different root causes." + "reason": "Both are tokenizer-loading related, but they involve different repositories/models and likely different root causes." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Sentence Transformers load hang and SigLIP2 zero probabilities are different model/runtime problems." + "reason": "Sentence Transformers hanging on load and SigLIP2 zero probabilities are not the same underlying issue." }, { "left": "issue:43723", "right": "issue:44442", "accept": false, - "reason": "Generic tokenizer loading in v5 and FastSpeech2ConformerTokenizer import/load failure are not the same bug." + "reason": "AutoTokenizer loading failure for one tokenizer is not the same as FastSpeech2ConformerTokenizer import/load failure." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "AutoImageProcessor import failure and Qwen2_5_VLForConditionalGeneration import failure are separate missing-export issues." + "reason": "Different import errors for different symbols/models: AutoImageProcessor vs Qwen2_5_VLForConditionalGeneration." }, { "left": "issue:43479", "right": "issue:44913", "accept": false, - "reason": "Phi4MultimodalConfig None-handling and GPTNeoXConfig rotary_pct reload behavior are unrelated config bugs." + "reason": "Config default-initialization bug and GPTNeoX rotary_pct not persisting on reload are different config issues." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy issues and Qwen3.5 Moe from_pretrained errors are different problems." + "reason": "FineGrainedFP8 accuracy regression is unrelated to Qwen3.5-Moe from_pretrained loading failure." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Qwen3 tokenizer offset_mapping is unrelated to mdeberta add_special_tokens behavior." + "reason": "Different tokenizer bugs: wrong offset_mapping vs missing BOS/EOS insertion." }, { "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 loading and quantized model saving failure are unrelated." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "DSA indexer torch.split behavior and position_ids docstring are not duplicates." + "reason": "DSA indexer split behavior and a docstring issue for position_ids are not the same bug." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "Both involve import errors after upgrade, but one is a specific huggingface_hub symbol removal and the other is a broad transformers v5 import-regression report." + "reason": "45278 is a broad import-regression report, but 42757 is a specific missing symbol in huggingface_hub; not the same concrete failure." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "FineGrainedFP8 accuracy and Qwen3-VL-Moe bug reports are different issues." + "reason": "FP8 accuracy issue and Qwen3-VL-Moe transformers v5 bug are different model/runtime failures." }, { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Smollm3 RoPE-layer dropping and Qwen3-Next tied-weights warning are unrelated model-architecture bugs." + "reason": "RoPE layer-dropping in Smollm3 and tied-weights warning in Qwen3-Next are different model implementation bugs." }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Generation after sync_gpus and D-FINE auxiliary-loss behavior are unrelated." + "reason": "Generation kwargs update after sync_gpus is unrelated to D-FINE auxiliary-loss behavior when denoising is off." } ] }, @@ -12049,6 +11866,7 @@ "issue:43976", "issue:44038", "issue:44117", + "issue:44222", "issue:44261", "issue:44263", "issue:44295", @@ -12083,8 +11901,8 @@ "issue:30990|issue:33357", "issue:43232|issue:44743", "issue:42907|issue:44945", - "issue:44704|issue:45290", "issue:44295|issue:44297", + "issue:44704|issue:45290", "issue:41628|issue:45278", "issue:43976|issue:44945", "issue:43122|issue:44514", @@ -12103,222 +11921,229 @@ "issue:43122|issue:44779", "issue:44038|issue:44295", "issue:42907|issue:43452", - "issue:36296|issue:43116" + "issue:36296|issue:43116", + "issue:44038|issue:44222" ], - "prepared_review_unit_hash": "26a6bbc76d82cecf8948fb5848cab7836f99d6594af7dd968744cb7d310c023a", + "prepared_review_unit_hash": "d20d62990132357af9e9e324e1ef5ca53ebdf7f55dedc987b445b84f1ee5a801", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 35955, - "estimated_input_tokens": 8989, - "estimated_eval_tokens": 18234 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36782, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated transformers issues; the soft pairs share broad themes like tokenization, loading, or generation, but not the same concrete bug, so none should be merged.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45278 is the broadest current umbrella in the set (multiple import failures after upgrading), so it is the most plausible anchor for a noisy cluster, even though many other items are unrelated.", + "summary": "This is a heterogeneous issue pile rather than a true duplicate cluster. There is a loose subgroup around Transformers v5 import/loading regressions, but none of the soft pairs look like the same underlying bug or change.", + "confidence": 0.94, + "canonical_issue_reason": "No true canonical issue: the set spans unrelated import, tokenizer, generation, multimodal, FP8, and training bugs; at most, issue:45278 is a loose umbrella for post-upgrade import regressions.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best issue candidate because it is open, recent, and phrased broadly enough to represent the cluster\u2019s general upgrade/import-regression theme.", + "best_issue_reason": "issue:45278 is the broadest representative candidate because it captures the only clear mini-theme (many import errors after upgrading), but it still does not unify the overall set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "Both involve v5 breakage, but one is the env command and the other is a sentence-transformers import NameError; different failures and code paths." + "reason": "`transformers env` CLI breakage vs a sentence-transformers import-time `nn` NameError; different code paths." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "FineGrainedFP8 accuracy regression and a Qwen3_5 infinite-loop-on-image bug are unrelated behaviors." + "reason": "FP8 accuracy regression vs Qwen3_5 image infinite loop; unrelated bugs." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline cache loading failure and wrong tokenizer decoder type are different tokenizer bugs with different failure modes." + "reason": "Offline cache loading failure vs wrong tokenizer decoder type in v5; both tokenizer-related but not the same defect." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "Missing transformers.onnx module import and tokenizer spacing after added token are unrelated issues." + "reason": "Missing `transformers.onnx` module import vs extra space after added token; unrelated." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 API guidance and a GLM-MoE torch.split return-value issue do not share the same underlying bug." + "reason": "TF32 warning/settings issue vs `torch.split` return handling in DSA indexer; different subsystems." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Both mention tokenization/chat templates, but one is a version-to-version tokenization change and the other is multimodal assistant-mask generation." + "reason": "Cross-version tokenization drift vs multimodal assistant mask bug; not the same bug." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None is a tokenizer registry bug, while AutoProcessor kwargs not reaching cached_file is an unrelated processor-loading bug." + "reason": "Tokenizer mapping `None` handling vs AutoProcessor kwargs forwarding; different failure points." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "A stuck sentence-transformers load and a MacOS bus error on CLIP are different runtime failures." + "reason": "Sentence-Transformers loading hang vs MacOS bus error loading CLIP; unrelated symptoms." }, { "left": "issue:43232", "right": "issue:44743", "accept": false, - "reason": "Both relate to generation state, but one is sync_gpus/update_model_kwargs and the other is recurrent-state reset in modular_qwen3_5; not the same defect." + "reason": "Generation kwargs update after `sync_gpus` vs Qwen3.5 recurrent-state reset with cache; different code paths." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Saving dequantized models and incorrect outputs under pipeline parallelism are distinct problems." + "reason": "Dequantized model save failure vs incorrect pipeline-parallel outputs; unrelated." }, { - "left": "issue:44704", - "right": "issue:45290", + "left": "issue:44295", + "right": "issue:44297", "accept": false, - "reason": "Processor kwargs propagation and chat-template crashes with tool-call assistant messages are separate code paths." + "reason": "`position_ids` buffer access error vs tokenizer class mismatch on save; different save/load bugs." }, { - "left": "issue:44295", - "right": "issue:44297", + "left": "issue:44704", + "right": "issue:45290", "accept": false, - "reason": "Reading a buffer-registry position_ids error and tokenizer_class mismatch on save_pretrained are unrelated." + "reason": "AutoProcessor cached-file kwargs bug vs chat-template crash on tool-call messages; different APIs." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "Both are import-related, but one is a specific AutoImageProcessor import error and the other is a broad post-upgrade import-error report; not the same bug." + "reason": "Specific `AutoImageProcessor` import failure vs broad post-upgrade import errors; overlap is too general to be a duplicate." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python version compatibility and pipeline-parallel inference correctness are unrelated." + "reason": "Python-version compatibility issue vs pipeline-parallel output bug; unrelated." }, { "left": "issue:43122", "right": "issue:44514", "accept": false, - "reason": "Both are tokenizer/chat issues, but one is a version regression in tokenization and the other is batched apply_chat_template crashing with padding=False." + "reason": "Tokenizer behavior drift vs `apply_chat_template` crash on batched multimodal input; different bugs." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Qwen3 offset_mapping is a tokenizer alignment bug, while loading cjvt/sleng-bert is a model-loading failure." + "reason": "Qwen3 offset-mapping bug vs failure to load a specific model; not the same issue." }, { "left": "issue:43784", "right": "issue:45600", "accept": false, - "reason": "Sentence-transformers import failure and broken auto_mappings references to removed SAM3 configs are unrelated." + "reason": "Import-time `nn` NameError in sentence-transformers vs stale `auto_mappings` CI break; both import-ish but not the same bug." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Dequantized save failure and invalid GenerationConfig for GLM-5 are different issues." + "reason": "Model save failure vs invalid `GenerationConfig`; unrelated." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "A SAM3 model-type warning and a missing fpn_position_embeddings attribute are related to the same family but are not the same concrete bug." + "reason": "SAM3 type warning vs missing `fpn_position_embeddings` attribute; related family, but different concrete failures." }, { "left": "issue:44261", "right": "issue:44373", "accept": false, - "reason": "A missing rms_norm_eps precision issue and a wrong docstring for position_ids are unrelated." + "reason": "Layernorm epsilon precision bug vs wrong docstring; functional bug and documentation issue are not duplicates." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "DSA indexer activation choice and GLM-5 RoPE implementation are separate model-internal bugs." + "reason": "DSA indexer missing ReLU vs GLM-5 RoPE implementation; unrelated model logic." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "num_labels propagation and flash_attention_3 degenerate training are not the same code-path problem." + "reason": "Qwen3.5 `num_labels` propagation bug vs degenerate training with flash_attention_3; different problems." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Tokenizer offset mapping and multimodal assistant mask generation are different bugs despite both touching tokenization." + "reason": "Tokenizer offset mapping vs all-zero assistant masks for multimodal inputs; different processing stages." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Example-script multi-label output emptiness and mm_token_type_ids defaulting are unrelated training/inference issues." + "reason": "Example script returning empty labels vs Gemma 4 token-type defaulting issue; unrelated." }, { "left": "issue:29127", "right": "issue:30990", "accept": false, - "reason": "LayoutLMv3 error-message clarity and Sentence Transformers loading hang are unrelated." + "reason": "LayoutLMv3 error-message clarity vs Sentence-Transformers loading stall; no overlap." }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals and loading Phi-4 multimodal instruct are different model-loading/save paths." + "reason": "Dequantized save bug vs failure to load Phi-4 multimodal; different model-loading/saving paths." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "A Qwen3-VL-Moe bug and tokenizer_class mismatch on save_pretrained are not the same underlying issue." + "reason": "Qwen3-VL-Moe bug vs tokenizer config class mismatch; unrelated." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and different behaviors." + "reason": "General tokenization drift vs Deepseek-specific incorrect tokenization; not the same concrete bug." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Qwen3-VL-Moe behavior and position_ids buffer access error are unrelated." + "reason": "Qwen3-VL-Moe issue vs `position_ids` buffer access error; unrelated." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Dequantized save failure and gguf_file loader breakage affect different loading/saving flows." + "reason": "Dequantized save failure vs `gguf_file` breakage in from_pretrained; different code paths." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor-parallel training bug and multi-label classification example output are not the same issue." + "reason": "Tensor-parallel training bug vs example-script multi-label output bug; unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Qwen3-VL-Moe bug vs FP8 `save_pretrained` bug; not the same defect." } ] }, @@ -12392,7 +12217,6 @@ ], "soft_pairs": [ "issue:43824|issue:44351", - "issue:44038|issue:44222", "issue:45216|issue:45478", "issue:42757|issue:44351", "issue:43278|issue:44945", @@ -12424,58 +12248,52 @@ "issue:43937|issue:44977", "issue:42673|issue:43906" ], - "prepared_review_unit_hash": "8679de61982b64296a6072915075e897be8b05ad571401af7f9c17cf710877c1", + "prepared_review_unit_hash": "51a44ef657dd386e77a684959b9e8ec9e4663f3c1d4188aacba3d23548b52122", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 32, - "serialized_chars": 36280, - "estimated_input_tokens": 9070, - "estimated_eval_tokens": 18396 + "soft_pair_count": 31, + "serialized_chars": 35975, + "estimated_input_tokens": 8994, + "estimated_eval_tokens": 18244 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: a few v5-era import/tokenizer regressions, but the paired items are generally different concrete bugs and should stay separate.", - "confidence": 0.92, - "canonical_issue_reason": "issue:45278 is the broadest umbrella-like issue in the set, centered on post-upgrade import breakage; it best represents the loose theme, though most items are not true duplicates.", + "summary": "This cluster is mostly a mix of Transformers v5 regressions and model-specific bugs, not a single duplicate thread. Issue 45278 is the broadest umbrella for the import-error subset, but the rest are too heterogeneous for strong deduping. No PRs were provided.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45278 is the broadest, most cluster-like report: it frames a general upgrade regression with multiple import errors after moving to 5.5.0, which best matches the shared v5 breakage theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most general and cluster-representative issue; it covers the broadest user-facing regression pattern among the listed items.", + "best_issue_reason": "issue:45278 is the best representative because it is the most general symptom statement and can plausibly cover several import-related reports in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Different missing imports in different modules (`Qwen2_5_VLForConditionalGeneration` vs `HybridCache`); same symptom class, not the same bug." - }, - { - "left": "issue:44038", - "right": "issue:44222", - "accept": false, - "reason": "One is a Qwen3-VL-Moe loading regression, the other is FP8 save_pretrained behavior; different code paths and failures." + "reason": "Both are import errors, but they name different missing symbols and likely different code paths." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Save-pretrained checkpoint corruption and from_pretrained loading error are opposite directions with different reported failures." + "reason": "One is about save_pretrained checkpoint correctness, the other about from_pretrained loading; related family, but not the same bug." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Different import errors from different libraries (`huggingface_hub` vs `transformers`); no shared underlying bug." + "reason": "Different missing imports from different modules; no evidence of one underlying defect." }, { "left": "issue:43278", @@ -12487,157 +12305,151 @@ "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both involve tokenizers, but one is decoder type selection and the other is Deepseek tokenization regression; not the same concrete defect." + "reason": "Both tokenizer regressions, but they affect different models and different failure modes." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "SigLIP2 implementation discrepancy and a Qwen3.5 infinite-loop-on-image bug are unrelated." + "reason": "Different model families and symptoms; no sign of the same underlying bug." }, { "left": "issue:42907", "right": "issue:44222", "accept": false, - "reason": "Both touch saving/quantization, but one is dequantized model saving and the other is FP8 save_pretrained; different failures." + "reason": "Both involve saving models, but one is dequantized Ministral/Devstral and the other is FP8 MoE; too different to merge." }, { "left": "issue:43122", "right": "issue:44977", "accept": false, - "reason": "A general tokenization difference across versions is not the same as a Qwen3.5 flash-attention generation bug." + "reason": "Tokenizer-version difference vs flash-attention generation failure are separate problems." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "Missing embed_tokens weight in from_pretrained and passing kwargs to cached_file are different loader issues." + "reason": "A missing weight in from_pretrained is not the same as AutoProcessor failing to forward kwargs." }, { "left": "issue:43784", "right": "issue:45446", "accept": false, - "reason": "Sentence-transformers import NameError and a PyTorch version-check bug in flex_attention are unrelated." + "reason": "Import-time NameError in sentence-transformers is unrelated to a PyTorch version check in flex_attention." }, { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "Both mention import breakage, but one is a specific missing symbol and the other is a broad upgrade-wide report; too generic to treat as duplicate." + "reason": "Generic import failure vs broad upgrade import regressions; similar theme but not demonstrably the same bug." }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "VRAM leak in multi-threaded generation is unrelated to CI dtype mismatch during bfloat16 inference." + "reason": "VRAM leak in dataloader threads is unrelated to dtype mismatch in bfloat16 inference." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "GenerationConfig validation failure and GLM-5 RoPE implementation are different bugs in different parts of the stack." + "reason": "GenerationConfig validation and RoPE implementation are different GLM-5 issues." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "Missing `transformers.onnx` module and a LayoutLMv3 error-message clarity issue are unrelated." + "reason": "A missing module import and a LayoutLMv3 error message issue are unrelated." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 API warning guidance and missing `get_number_of_image_patches` are different problems." + "reason": "TF32 API guidance and a missing image-patch method are unrelated." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "Wrong optimizer states on non-zero ranks and missing GenerationMixin cache in chunked prefill are different training/generation bugs." + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to GenerationMixin cache handling." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "Broken VitPose models and Ministral FlashAttention failures are separate model-specific regressions." + "reason": "Broken VitPose imports and Ministral flash-attention failures are different model-specific regressions." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Python version support mismatch is unrelated to the flex_attention import version check." + "reason": "Python-version support issue and flex_attention version check bug are not the same underlying defect." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Ignored SGD args and trainer.evaluate() failing after train are different training-loop issues." + "reason": "Unused SGD args and evaluate-after-train failure are separate training bugs." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Multi-label classification example output and SigLIP2 implementation discrepancy are unrelated." + "reason": "Example-script multi-label output bug is unrelated to SigLIP2 implementation discrepancy." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Example-script multi-label behavior and pipeline deprecation/error messaging are different issues." + "reason": "Example-script classification output bug and pipeline() error messaging are different issues." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Same model family, but different versions/models and different load errors; not enough to call the same bug." - }, - { - "left": "issue:44222", - "right": "issue:45478", - "accept": false, - "reason": "FP8 save_pretrained behavior and Qwen3.5 from_pretrained loading failure are different directions and symptoms." + "reason": "Related Qwen3 MoE area, but one is a broad v5/moe bug and the other is a specific from_pretrained error; not enough to merge." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "Feature-extractor length computation bug and a PyTorch import version-check bug are unrelated." + "reason": "Feature-extract length helper behavior is unrelated to an import version check." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "GenerationConfig invalid and a stale num_return_sequences warning are not the same issue." + "reason": "GLM-5 generation config validation and a stale warning about num_return_sequences are different code paths." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Tokenizer loading regression and Qwen3 tokenizer offset_mapping regression are different concrete tokenizer problems." + "reason": "Tokenizer loading regression and offset-mapping regression are distinct tokenizer issues." }, { "left": "issue:44117", "right": "issue:45588", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None and a flash-attention AttributeError are unrelated." + "reason": "Tokenizer mapping fallback issue and flash_attention crash on s_aux=None are unrelated." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and failure modes; not clearly the same underlying bug." + "reason": "Both tokenizer regressions, but for different models and different symptoms." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still referencing pipeline() and a flex_attention import version check are unrelated." + "reason": "Docs still referencing pipeline() is unrelated to a runtime import/version-check bug." }, { - "left": "issue:42673", - "right": "issue:43906", + "left": "issue:43906", + "right": "issue:42673", "accept": false, - "reason": "VRAM leak in dataloader-thread generation and an isolated reproduction of another issue are not the same bug." + "reason": "A repro of an older issue and a VRAM leak in dataloader threads are unrelated." } ] }, @@ -12724,15 +12536,15 @@ "issue:42673|issue:43582", "issue:42371|issue:44261", "issue:43408|issue:43742", - "issue:44933|issue:45278", - "issue:42222|issue:45542", "issue:42617|issue:45341", - "issue:42673|issue:43577", + "issue:42222|issue:45542", + "issue:44933|issue:45278", "issue:43937|issue:44496", + "issue:42673|issue:43577", "issue:42175|issue:42222", + "issue:44075|issue:45656", "issue:38617|issue:44220", "issue:44483|issue:45464", - "issue:44075|issue:45656", "issue:42371|issue:42831", "issue:43066|issue:43122", "issue:43012|issue:43577", @@ -12742,214 +12554,232 @@ "issue:43874|issue:44261", "issue:43097|issue:44493" ], - "prepared_review_unit_hash": "b1868a186d99663aaaf84a52dfab490f30464e4f9e85c44f6ee5380dd5bb2796", + "prepared_review_unit_hash": "cb3c2f4cb71d2d6916846567d0f0cbe9f8f52c5f9d5e5338dcae274059c491c3", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36475, - "estimated_input_tokens": 9119, - "estimated_eval_tokens": 18494 + "serialized_chars": 36472, + "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 a broad mix of unrelated Transformers issues; the soft-similarity pairs mostly share only a subsystem or failure shape, not the same underlying bug. I would not collapse the set into a single canonical issue.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a loose v5-regression theme (imports, tokenizers, model loading, backend/config behavior), not a tight duplicate set. The best representative issue is the broad import-regression tracker.", + "confidence": 0.72, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report ('Many import errors after update from 4.57.0 to 5.5.0') and best matches the cluster's overall migration/import-breakage theme.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:45278 is the most representative top-level tracker: broad scope, current, and explicitly about multiple import errors after upgrade.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "Both are version-5 breakages, but one is an init_empty_weights argument error and the other is PreTrainedConfig type-checking; different code paths and fixes." + "reason": "Both are v5 breakages, but they describe different failures in different code paths (_is_hf_initialized vs PreTrainedConfig type checking)." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "Unrelated dependency/version problems: PEFT minimum version pin vs a PyTorch version check for flex_attention imports." + "reason": "Unrelated problems: PEFT version pinning vs a PyTorch version check in flex_attention." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Both involve video/processor flows, but one is a batched video shape bug and the other is an AutoProcessor crash when torchvision is missing." + "reason": "Both touch video processing, but one is a batched-shape bug and the other is an AutoProcessor/torchvision import crash." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "Both touch testing utilities, but one is a bug in testing_utils.py and the other is a pytest-xdist race writing captured_info.txt." + "reason": "Same file area, but the reports are different: a generic testing_utils bug vs an xdist race on captured_info.txt." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "Tokenizer-related, but one is TOKENIZER_MAPPING_NAMES returning None and the other is a Mistral regex patch crash on backend_tokenizer access." + "reason": "Both are tokenizer-loading issues, but they target different mechanisms (mapping fallback vs Mistral regex patch crash)." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols from different modules and likely different regressions." + "reason": "Both are import errors, but for different symbols and likely different regression causes." }, { "left": "issue:39401", "right": "issue:43421", "accept": false, - "reason": "Both are tokenizer behavior issues, but offset_mapping for Qwen3 is not the same as runtime post-processor updates when special tokens change." + "reason": "Tokenizer-related, but one is wrong offset mapping and the other is runtime post-processor update behavior." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Different model-specific failures: missing attn projections in qwen3next vs a Qwen3.5 MoE from_pretrained error." + "reason": "Different Qwen-family model issues with different failing paths; not the same bug." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "One is a flash_attention_3 training degeneration, the other is a save_pretrained regression for Qwen3.5 checkpoints." + "reason": "Training instability with flash_attention_3 is unrelated to the Qwen3.5 save_pretrained checkpoint regression." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "These are unrelated: max_shard_size default behavior versus a Qwen3.5 MoE loading error." + "reason": "Chat/completions request rejection and from_pretrained model-loading failure are different code paths." }, { "left": "issue:42673", "right": "issue:43582", "accept": false, - "reason": "Completely different failures: multi-thread VRAM leak in Qwen3.5 versus an Apple Silicon TypeError in caching_allocator_warmup." + "reason": "VRAM leak in multi-threaded dataloader use is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "Both mention numeric/precision settings, but TF32 configuration warnings and MLA rms_norm_eps precision errors are distinct bugs." + "reason": "TF32 API guidance and MLA precision/config epsilon issues are unrelated." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "A model-type mismatch warning for sam3_* is not the same as a key error loading facebook/MobileLLM-125M." + "reason": "Different model-loading warnings/errors for different model families." }, { - "left": "issue:44933", - "right": "issue:45278", + "left": "issue:38617", + "right": "issue:44220", "accept": false, - "reason": "The newer issue is a broad import-regression report; the older one is a specific missing import from image_utils. Similar symptom, not the same confirmed bug." + "reason": "ImportError for a missing symbol and an audio feature-extraction issue are not the same underlying bug." }, { - "left": "issue:42222", - "right": "issue:45542", + "left": "issue:44483", + "right": "issue:45464", "accept": false, - "reason": "Vitpose model breakage and missing TensorFlow backend in a torch-only install are unrelated." + "reason": "Both mention chat/completions, but one is request validation/acceptance and the other is streaming inference failure." }, { - "left": "issue:42617", - "right": "issue:45341", + "left": "issue:42371", + "right": "issue:42831", "accept": false, - "reason": "A 3d_parallel.py runtime failure and a testing_utils bug are different areas and different code paths." + "reason": "TF32 backend behavior and FineGrainedFP8 accuracy are separate numeric/configuration issues." }, { - "left": "issue:42673", + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both concern tokenizer behavior in v5, but the symptoms are different and there is no clear same-root-cause evidence." + }, + { + "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "VRAM leakage in Qwen3.5 multi-threaded use is unrelated to BLIP2 dtype remaining float32 after loading." + "reason": "PyTorch precision warning on compile and BLIP2 dtype propagation are distinct bugs." }, { - "left": "issue:43937", - "right": "issue:44496", + "left": "issue:43673", + "right": "issue:45341", "accept": false, - "reason": "Both are model-loading/config issues, but one is invalid GenerationConfig and the other is an unrecognized model due to missing model_type." + "reason": "Generation cache behavior during chunked_prefill is unrelated to a small testing_utils bug." }, { - "left": "issue:42175", - "right": "issue:42222", + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen chat/completions or from_pretrained, but they are different failures and likely different fixes." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen model variants and different failure modes; not mergeable as one bug." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing image-patch helper and MLA norm epsilon precision error are unrelated." + }, + { + "left": "issue:43097", + "right": "issue:44493", "accept": false, - "reason": "TensorFlow backend installation behavior and vitpose model breakage are different issues." + "reason": "A removed config option and unexpected position-id keys are separate v5 regressions." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "An import error for layer_type_validation is unrelated to a _torch_extract_fbank_features() failure." + "reason": "A missing import symbol and a feature-extraction helper issue do not share the same code path." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "Both hit chat/completions-style serving, but one is an API request acceptance problem and the other is a streaming inference failure on Qwen3.5-0.8B." - }, - { - "left": "issue:44075", - "right": "issue:45656", - "accept": false, - "reason": "Both concern optimizers, but SGD args being ignored is not the same bug as an optimizer step being called twice with deepspeed." + "reason": "Endpoint request acceptance and streaming inference behavior are different problems." }, { "left": "issue:42371", "right": "issue:42831", "accept": false, - "reason": "TF32 warning behavior and FineGrainedFP8 accuracy degradation are distinct precision-related issues." + "reason": "No overlap beyond general numerical/backend concerns." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both are tokenizer regressions in v5, but one is a decoder-type mismatch and the other is broader tokenization differences; not clearly the same fix." + "reason": "The tokenizer decoder-type issue is not clearly the same as the broader tokenization-difference report." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "A compile-time bfloat16 warning is unrelated to BLIP2 staying float32 when loaded with half/bfloat16." + "reason": "Different precision/dtype bugs in different model flows." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "GenerationMixin cache handling in chunked_prefill is unrelated to a testing_utils bug." + "reason": "Generation cache bug and testing utility bug are unrelated." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "max_shard_size default behavior and a Qwen3.5 MoE from_pretrained error are unrelated." + "reason": "Similar surface area, but different model-loading/API failures." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Missing qkv/gate projections in qwen3next and a Qwen3.5 MoE loading error are different model-specific problems." + "reason": "Different Qwen regressions with different root causes." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "One is a missing image-patch-count method in a processor, the other is an rms_norm_eps precision mismatch in MLA; different components." + "reason": "Different multimodal/image-patch vs norm-epsilon precision issues." }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Removing tie_embeddings_and_encoder_decoder and unexpected position-id keys are different regression classes with different remedies." + "reason": "Different regression symptoms and code paths." } ] }, @@ -13033,8 +12863,8 @@ "issue:45003|issue:45341", "issue:43335|issue:45071", "issue:43097|issue:43452", - "issue:43986|issue:44484", "issue:43986|issue:45341", + "issue:43986|issue:44484", "issue:44117|issue:44987", "issue:43408|issue:44368", "issue:43874|issue:44263", @@ -13058,244 +12888,244 @@ "issue:43066|issue:43906", "issue:43097|issue:43901" ], - "prepared_review_unit_hash": "0164f49c8435d14d889f566d3b8b0fcab4731c93719b843c3f1fbf3e76a27141", + "prepared_review_unit_hash": "cfefcf720ef82fdd46256997845345596f37e44d6d0caeff1e1bd0a03e94fec4", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 35, - "serialized_chars": 37662, - "estimated_input_tokens": 9416, - "estimated_eval_tokens": 19088 + "serialized_chars": 37656, + "estimated_input_tokens": 9414, + "estimated_eval_tokens": 19084 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of unrelated Transformers regressions and import/config bugs. None of the soft pairs look like true duplicates or the same underlying fix, so all soft edges are rejected.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue: the items span distinct bugs (different missing imports, model-specific shape/config problems, tokenizer/loading regressions, and unrelated runtime warnings).", - "canonical_pr_reason": null, - "best_issue_reason": "issue:43986 is the best representative anchor because it is a concrete, well-scoped crash report with clear impact and reproducibility, unlike the more fragmentary one-off regressions in the set.", - "best_pr_reason": null, + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues that share broad keywords like v5, import errors, Qwen/GLM, or flash attention. None of the soft pairs looks like the same underlying bug/change strongly enough to merge.", + "confidence": 0.86, + "canonical_issue_reason": "No single canonical issue stands out: the items span unrelated import failures, model-specific regressions, training bugs, and documentation/warning reports, with only superficial title similarity.", + "canonical_pr_reason": "No PRs are present in the cluster.", + "best_issue_reason": "No strong global best issue for deduping; any representative would be misleading because the cluster is not centered on one concrete code path.", + "best_pr_reason": "No PRs are present in the cluster.", "soft_edge_verdicts": [ { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "Different GLM internals: a torch.split indexing issue vs a RoPE implementation question; not the same bug or fix." + "reason": "Different bugs: torch.split return handling in a GLM MoE indexer vs RoPE implementation details." }, { "left": "issue:43012", "right": "issue:43408", "accept": false, - "reason": "Unrelated problems: bfloat16 compilation warning vs sam3_video/sam3_tracker model-type mismatch." + "reason": "Both mention warnings, but one is a bfloat16 precision warning and the other is a model-type mismatch warning." }, { "left": "issue:42222", "right": "issue:45459", "accept": false, - "reason": "VitPose model breakage and a tokenizer protobuf exception are different code paths and symptoms." + "reason": "Unrelated subsystems: vitpose model breakage vs tokenizer protobuf error masking." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "AutoImageProcessor import failure and PretrainedConfig type checking breakage are separate regressions." + "reason": "Import error for AutoImageProcessor is unrelated to PretrainedConfig type-checking breakage." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "FlashAttention failure for Ministral-3 is unrelated to flash_attention_3 training degeneration in sequence classification." + "reason": "Both involve flash attention, but one is a load-time failure and the other is degenerate training behavior." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "Config type checking vs tensorboard/TensorFlow backend detection are unrelated bugs." + "reason": "PretrainedConfig typing regression is unrelated to TensorBoard/TensorFlow backend detection." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch is not the same as missing attn projections in qwen3next." + "reason": "Different Qwen3-family failures: weight-shape mismatch vs missing attention projections in a specific layer." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Both are Qwen/VL-related, but one is a loading shape mismatch and the other is a transformers 5.0 compatibility bug; not the same fix." + "reason": "Both are Qwen model issues, but one is a loading shape mismatch and the other is a Qwen3-VL-Moe regression." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "Unsafe sys.modules access and a testing_utils bug are distinct implementation issues." + "reason": "sys.modules access bug is unrelated to a testing_utils bug." }, { "left": "issue:43335", "right": "issue:45071", "accept": false, - "reason": "Sparse-layer construction in SwitchTransformersConfig is unrelated to PretrainedConfig type-checking regressions." + "reason": "SwitchTransformers sparse-layer construction is not the same issue as PretrainedConfig typing." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "Missing tie_embeddings guidance in v5 and gguf_file tokenizer/model loading breakage do not share a concrete code path." + "reason": "Removed config flag warning and gguf_file loading breakage are different code paths." }, { "left": "issue:43986", - "right": "issue:44484", + "right": "issue:45341", "accept": false, - "reason": "Missing torchvision crash in AutoProcessor is unrelated to save_pretrained shard-size behavior." + "reason": "Video AutoProcessor crash and a testing_utils bug do not match." }, { "left": "issue:43986", - "right": "issue:45341", + "right": "issue:44484", "accept": false, - "reason": "AutoProcessor/torchvision loading crash and a testing utility bug are unrelated." + "reason": "AutoProcessor/torchvision crash is unrelated to save_pretrained shard size behavior." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES None handling and a transformers>=5.1.0 model-loading failure are different loader defects." + "reason": "Tokenizer mapping None-handling is a different failure than loading physical-intelligence/fast." }, { "left": "issue:43408", "right": "issue:44368", "accept": false, - "reason": "sam3 model-type warning and tie_word_embeddings warning are different warnings from different model families." + "reason": "Different warnings in different contexts: model-type mismatch vs tie_word_embeddings advice." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Missing image-patch method in GLM46V and torch.split return handling in GlmMoeDsaIndexer are unrelated." + "reason": "Different GLM bugs: missing image-patch helper vs torch.split return values." }, { "left": "issue:43976", "right": "issue:45071", "accept": false, - "reason": "Python version support breakage and PretrainedConfig type checking are unrelated compatibility issues." + "reason": "Python version support issue is unrelated to a config typing regression." }, { "left": "issue:44829", "right": "issue:45245", "accept": false, - "reason": "Flash attention training degeneration and category-count runtime limit are different failure modes." + "reason": "Flash-attention training degeneration is unrelated to category cardinality limits." }, { "left": "issue:43986", "right": "issue:45372", "accept": false, - "reason": "Missing torchvision crash and ReasoningEffort import failure affect different processor stacks and dependencies." + "reason": "Video AutoProcessor/torchvision crash is unrelated to ReasoningEffort import failure in Gemma 4 loading." }, { "left": "issue:43531", "right": "issue:45341", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and a testing_utils bug do not overlap at the code-path level." + "reason": "Qwen3-MoE sliding_window bug is unrelated to a testing_utils bug." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols from different libraries; not the same underlying bug." + "reason": "Both are import errors, but for different missing symbols from different packages." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm and GLM-5 RoPE implementation are different model math issues." + "reason": "RMS norm epsilon precision bug does not match the GLM-5 RoPE implementation issue." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption and Liger Kernel not applied via model_init are separate training pipeline problems." + "reason": "Different training-time problems: optimizer state/rank mismatch vs Liger Kernel application during model init." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Missing projections in qwen3next and a silently dropped Qwen3_5MoeVisionConfig field are unrelated model-definition bugs." + "reason": "Both are Qwen config/model issues, but one is missing projections and the other is a missing vision config field." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "layer_type_validation missing from configuration_utils and HybridCache missing from transformers are different import regressions." + "reason": "Distinct import failures for different symbols: layer_type_validation vs HybridCache." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "Tie-embeddings deprecation and flash_attention.py crashing on s_aux=None are unrelated." + "reason": "Removed tie_embeddings warning is unrelated to flash_attention.py crashing on s_aux=None." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "AddedToken keyword duplication and TOKENIZER_MAPPING_NAMES None handling are separate tokenizer-loader bugs." + "reason": "Tokenizer AddedToken argument collision is unrelated to TOKENIZER_MAPPING_NAMES returning None." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Qwen3-VL-Moe compatibility and qwen3next missing projections are different model-specific regressions." + "reason": "Both mention Qwen3 variants, but the concrete failures and code paths differ." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "AutoProcessor cached_file kwargs forwarding and auto_docstring annotation handling are unrelated utility bugs." + "reason": "AutoProcessor cached_file kwargs forwarding is unrelated to auto_docstring annotation handling." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "local_files_only network leakage and UMT5Encoder missing embed_tokens.weight are different loading defects." + "reason": "Local-files-only network leak and missing embed_tokens.weight are unrelated loading issues." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Missing torchvision guard in AutoProcessor and tokenizer mapping None handling are not the same problem." + "reason": "Video AutoProcessor crash and tokenizer mapping None-handling do not share the same bug." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "A v5 config/API deprecation and an offloading-with-quantization issue are unrelated." + "reason": "Removal of a config flag is unrelated to offloading/quantization behavior." }, { "left": "issue:43824", "right": "issue:43986", "accept": false, - "reason": "Qwen2.5-VL import failure and AutoProcessor/torchvision crash are distinct processor/model import paths." + "reason": "Qwen2.5-VL import error and video-model AutoProcessor crash are different model/loading problems." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "Wrong tokenizer decoder type and a reproduction of issue 38071 are not the same concrete bug." + "reason": "Wrong tokenizer decoder type is not the same as the separate reproduced issue linked by 43906." }, { "left": "issue:43097", "right": "issue:43901", "accept": false, - "reason": "tie_embeddings_and_encoder_decoder removal and pipeline docs mismatch are different documentation/API issues." + "reason": "A removed config API warning is unrelated to a documentation mismatch in TextClassificationPipeline." } ] }, @@ -13407,20 +13237,20 @@ "issue:43611|issue:44655", "issue:43976|issue:44938" ], - "prepared_review_unit_hash": "e7a54364d1f7c4bd37ad1f7a98828b30fefab1e7c5f351ef0a8af852c85d1cf1", + "prepared_review_unit_hash": "0a2daaa890d35229510251a3943cd2fb0d66d01b574ce0f8d2830e59018cc1e3", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 37, - "serialized_chars": 38514, + "serialized_chars": 38515, "estimated_input_tokens": 9629, "estimated_eval_tokens": 19514 }, @@ -13429,204 +13259,228 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous mix of unrelated Transformers issues: import errors, model-specific loading/saving bugs, chat/template regressions, backend/compile problems, and documentation issues. None of the soft-edge pairs look like the same underlying bug or change.", - "confidence": 0.92, - "canonical_issue_reason": "No single clear canonical issue exists because the items span many distinct regressions. If one must be chosen, issue 44483 is the broadest, most user-visible v5 API breakage in the set.", + "summary": "This set is not a true duplicate cluster: it mixes unrelated import errors, model-loading regressions, tokenizer/pipeline behavior changes, save_pretrained issues, and model-specific bugs. The closest thematic hub is pipeline/model saving, but none of the soft pairs look like the same concrete fix.", + "confidence": 0.48, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 44483 is the best representative only in the sense that it is broad and high-impact; however, it is not truly a duplicate umbrella for the rest of the cluster.", + "best_issue_reason": "issue:44655 is the best representative hub because several soft similarities point toward save_pretrained/pipeline serialization, even though the overall set is too heterogeneous to treat as one duplicate cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "Different problems: pipeline task-support/error messaging vs saving Pipeline objects." + "reason": "Translation-task help text vs pipeline save_pretrained failure; unrelated." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Model weight loading mismatch vs a question about the default shard size in save_pretrained()." + "reason": "Missing embed_tokens.weight vs shard-size question; unrelated." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "Different code paths and symptoms: unsupported Flash Attention 2 for one model vs ignored attention scaling in GPT-2 under SDPA/FlashAttention." + "reason": "Both are attention/backend bugs, but they affect different models and different code paths." }, { "left": "issue:43208", "right": "issue:43550", "accept": false, - "reason": "Unrelated model/training bugs: xLSTM training blockers vs Bamba compile+SDPA failure." + "reason": "xLSTM training bugs vs Bamba torch.compile/SDPA issue; unrelated." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "A warning about bfloat16 compilation precision is not the same bug as a model failing under torch.compile with SDPA." + "reason": "PyTorch precision warning vs a compile-time SDPA failure; not the same bug." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "Precision warning during compile and a tied-embeddings warning in Qwen3.5 fine-tuning are distinct issues." + "reason": "Different warning symptoms in different feature paths; unrelated." }, { "left": "issue:42886", "right": "issue:43381", "accept": false, - "reason": "Offline cache loading failure is unrelated to gradient checkpointing in eval mode." + "reason": "Offline cache loading vs gradient checkpointing in eval mode; unrelated." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Saving dequantized models and an inference infinite-loop bug are different underlying problems." + "reason": "Saving dequantized models vs Qwen3_5 image looping; unrelated." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Model saving failure is not the same as saving Pipeline objects." + "reason": "Both are saving-related, but one is quantized model serialization and the other is pipeline save_pretrained; separate fixes." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Attention scaling backend regression is unrelated to degenerate zero-loss classification behavior." + "reason": "GPT2 attention-scaling backend bug vs sequence-classification zero-loss bug; unrelated." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Flash Attention support for a model vs a Python-version compatibility break are unrelated." + "reason": "Model-specific attention support issue vs Python-version compatibility problem; unrelated." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "Distributed optimizer-state corruption is not the same bug as attention scaling being ignored." + "reason": "FSDP2/PEFT optimizer-state bug vs attention-scaling backend bug; unrelated." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "Multiple xLSTM training bugs do not match a BeitImageProcessorFast label reduction bug." + "reason": "xLSTM training issues vs BeitImageProcessorFast label reduction bug; unrelated." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3Config missing vocab_size and an Apple Silicon allocator TypeError are unrelated." + "reason": "Missing Gemma3Config attribute vs Apple Silicon allocator TypeError; unrelated." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Both touch chat-related APIs, but one is a global /v1/chat/completions acceptance issue and the other is a processor batched-input crash." + "reason": "Chat completions endpoint rejection vs batched chat_template crash; different layers and failures." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIP output fields missing and GPT-2 attention scaling under backends are different regressions." + "reason": "CLIP output regression vs GPT2 attention-scaling bug; unrelated." }, { "left": "issue:43618", "right": "issue:43761", "accept": false, - "reason": "Both are CLIP-related, but one is missing attentions and the other missing hidden_states; not clearly the same code-path bug." + "reason": "Both are CLIP regressions, but one drops attentions and the other hidden_states; not clearly the same fix." }, { "left": "issue:43012", "right": "issue:44683", "accept": false, - "reason": "A float32 precision warning during bfloat16 compilation is not the same as compiled flex_attention failure on torch>=2.9." + "reason": "A precision warning is not the same as a flex_attention compilation failure." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Different missing imports from different packages and symbols." + "reason": "Two import errors, but for different symbols from different modules; unrelated." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "ImportError for a configuration utility and unwanted API requests despite local_files_only=True are unrelated." + "reason": "ImportError in configuration utils vs network calls despite local_files_only=True; unrelated." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Both mention Qwen3-VL/MoE, but one is a loading regression and the other is fp8 inference support; not the same bug." + "reason": "Qwen3-VL-Moe loading regression vs fp8 inference support request; not the same bug." }, { "left": "issue:44483", "right": "issue:45290", "accept": false, - "reason": "A chat-completions API rejection is not the same as apply_chat_template crashing on assistant messages with tool calls." + "reason": "Chat API acceptance regression vs assistant tool-call template crash; different code paths." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Qwen3-VL-Moe loading breakage and sorted-state_dict load errors are unrelated." + "reason": "Qwen3VL-Moe loading breakage vs sorted state_dict load error; unrelated." }, { "left": "issue:44483", "right": "issue:45381", "accept": false, - "reason": "Different Qwen2.5-VL regressions: API request rejection vs incorrect vision_position_ids." + "reason": "Chat completions API regression vs Qwen2.5-VL video position_ids bug; unrelated." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Both mention RoPE, but they concern different models and different implementations/results." - }, - { - "left": "issue:43299", - "right": "issue:45753", - "accept": false, - "reason": "Model loading breakage and an infinite-loop generation bug are not the same underlying issue." + "reason": "Both mention RoPE, but they concern different models and likely different implementations." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Pipeline docs mismatch is unrelated to saving Pipeline objects." + "reason": "Docs mismatch for return_all_scores vs pipeline save_pretrained failure; unrelated." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Different attribute-missing model issues in different architectures." + "reason": "Gemma3 config missing vocab_size vs SAM3 encoder output missing an attribute; unrelated." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model saving failure is broader/different from saving Pipeline objects." + "reason": "Both saving-related, but quantized-model save failure and pipeline save_pretrained are separate code paths." }, { "left": "issue:43352", "right": "issue:43937", "accept": false, - "reason": "A Flash Attention support error and a GenerationConfig validation error are different failures." + "reason": "FlashAttention support bug vs GLM-5 GenerationConfig validation error; unrelated." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "Generic state_dict ordering load error and a Qwen3.5 save_pretrained regression are unrelated." + "reason": "State_dict ordering load error vs checkpoint save_pretrained regression; different directions and symptoms." }, { "left": "issue:44077", "right": "issue:44380", "accept": false, - "reason": "A patchtsmixer post_init constraint issue does not match GPT-2 attention scaling under backend implementations." + "reason": "PatchTSMixer post_init policy issue vs GPT2 attention scaling; unrelated." + }, + { + "left": "issue:42898", + "right": "issue:43066", + "accept": false, + "reason": "Tokenizer cleanup behavior change vs wrong tokenizer decoder type; related area but distinct defects." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Import errors from different packages and symbols; unrelated." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Gemma3Config attribute bug vs PatchTSMixer post_init policy change; unrelated." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "base_model_prefix load regression vs pipeline save_pretrained issue; unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python compatibility reports, but they involve different versions and different failure modes." } ] }, @@ -13734,242 +13588,233 @@ "issue:43502|issue:44062", "issue:43010|issue:44075" ], - "prepared_review_unit_hash": "4e68e8cfa0424ba84525bb2b724eb1233b027f568a09073950f2f90bc774d6b0", + "prepared_review_unit_hash": "b8a9f17214c8e31e093982cc53633f778e4da6be9e73b06d5b2c337079c957b5", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37115, - "estimated_input_tokens": 9279, - "estimated_eval_tokens": 18814 + "serialized_chars": 37119, + "estimated_input_tokens": 9280, + "estimated_eval_tokens": 18816 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and looks like a noisy similarity group rather than a true duplicate set. The only plausible duplicate pair is the import-time slowdown reports, centered on `import transformers` being slow due to model import patterns.", - "confidence": 0.56, - "canonical_issue_reason": "`issue:41669` is the most concrete and actionable report in the only plausible subcluster: it identifies a specific cause (`import *` usage in models) and quantifies the import-time impact. `issue:44246` is the broader symptom report, but 41669 better captures the underlying bug.", + "summary": "Mostly unrelated issues spanning different models, config attributes, training paths, import regressions, and test flakes. I don\u2019t see any soft pair that looks like the same underlying bug/change, so this cluster does not collapse cleanly into duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:45278 is the broadest, most umbrella-like report here: multiple import regressions after upgrading versions. It\u2019s the closest thing to a representative issue for the set, even though the rest are not true duplicates of it.", "canonical_pr_reason": null, - "best_issue_reason": "`issue:41669` is the best representative issue for the cluster\u2019s only credible duplicate theme because it ties the symptom to a concrete root cause. The rest of the items are largely unrelated model/config/test regressions.", + "best_issue_reason": "issue:45278 best represents the cluster because it captures the broad upgrade/regression theme, rather than a narrow model-specific or test-only failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44263", "right": "issue:45440", "accept": false, - "reason": "Different models and different failures: DSA indexer split semantics vs DeepSeekV3 MoE divergence." + "reason": "Both concern model internals, but one is a GlmMoe DSA indexer return issue and the other is DeepSeekV3 MoE divergence; different code paths and symptoms." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Both are missing-config-attribute bugs, but on different configs and different attributes; not the same underlying defect." + "reason": "Both are missing config attributes, but they affect different configs (`Gemma3Config.vocab_size` vs `Llama4Config.pad_token_id`) and different failure points." }, { "left": "issue:43352", "right": "issue:45593", "accept": false, - "reason": "Unrelated model features: Flash Attention support gating vs D-FINE auxiliary-loss behavior." + "reason": "Nemotron FlashAttention support gating is unrelated to D-FINE auxiliary-loss behavior when denoising is off." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "Different code paths: CLIP output attentions regression vs Qwen3.5 save_pretrained checkpoint corruption." + "reason": "CLIPOutput attentions assignment and Qwen3.5 checkpoint key saving are separate bugs in different parts of the stack." }, { "left": "issue:44188", "right": "issue:44380", "accept": false, - "reason": "Both mention attention/backend behavior, but the actual bugs are different: compile-time branching divergence vs ignored GPT2 scaling config." + "reason": "Both touch attention, but one is a torch.compile branching divergence bug and the other is backend-specific GPT2 attention scaling being ignored." }, { "left": "issue:36683", "right": "issue:44242", "accept": false, - "reason": "Different models and different functionality: config attribute missing vs MoE load-balancing loss omission." + "reason": "Missing `vocab_size` on Gemma3Config is not the same as load-balancing loss being omitted when `output_router_logits=False`." }, { "left": "issue:43064", "right": "issue:44492", "accept": false, - "reason": "Training/optimizer-state bug and a cache-strategy typo are unrelated." + "reason": "A distributed training optimizer-state bug is unrelated to a documentation/typing typo in cache strategies." }, { "left": "issue:43352", "right": "issue:43931", "accept": false, - "reason": "Different load-time issues for different models: Flash Attention 2 unsupported vs Qwen3-VL shape mismatch." + "reason": "One is unsupported FlashAttention 2.0 for Nemotron; the other is a Qwen3-VL weight-shape mismatch during model loading." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "Cache no_grad decoration bug is unrelated to the broad import-error report." + "reason": "`no_grad` decoration on cache updates is a code-path fix; widespread import errors after upgrading versions are a different regression class." }, { "left": "issue:43992", "right": "issue:45310", "accept": false, - "reason": "Both are model loading errors, but for different models and different failure modes." + "reason": "Both are from loading models, but they are model-specific missing-key/from_pretrained failures with different root causes." }, { "left": "issue:43867", "right": "issue:45375", "accept": false, - "reason": "Different checkpoint/config problems: state_dict ordering load failure vs missing vision config field." + "reason": "A sorted-state_dict loading error and a missing `deepstack_visual_indexes` config field are distinct failure modes." }, { "left": "issue:41669", "right": "issue:44246", - "accept": true, - "reason": "These look like the same import-time slowdown problem: one report names the broad symptom, the other identifies the `import *` cause in model imports." + "accept": false, + "reason": "Import-time performance from `import *` usage is not the same bug as intermittent slow `import transformers`." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Both are import/parse failures, but one is generic compiled-Python errors while the other is a specific Python 3.13 `torch.jit.script` parsing issue." + "reason": "Compiled Python file errors and a Python 3.13 `@torch.jit.script` parsing/IndentationError are different import-time issues." }, { "left": "issue:36683", "right": "issue:43064", "accept": false, - "reason": "Different subsystems and failures: config attribute missing vs optimizer state sharding/training bug." + "reason": "A missing config attribute on Gemma3 has nothing to do with FSDP2/PEFT optimizer-state corruption in training." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Both are test flakiness, but one is missing `set_seed()` in SamHQ tests and the other is an xdist file-race issue." + "reason": "Missing `set_seed()` in a test file is unrelated to an xdist race writing `captured_info.txt`." }, { "left": "issue:44938", "right": "issue:45071", "accept": false, - "reason": "Python 3.14 load failure and `PretrainedConfig` type-checking regression are different issues." + "reason": "Python 3.14 import/load compatibility and PretrainedConfig type-checking breakage are separate regressions." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Both involve loading/saving checkpoints, but one is state_dict ordering and the other is incorrect visual encoder key serialization." + "reason": "State-dict sorting/load failure is not the same as saving incorrect visual encoder keys in Qwen3.5." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "Cache update no_grad decoration bug is unrelated to the generic \u201cmany models break\u201d report." + "reason": "A specific `@torch.no_grad` cache update bug is not the same as a broad complaint that v5 breaks too many models." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Same general family of loading issues, but different models and distinct causes." + "reason": "Different Qwen3.5-related from_pretrained issues, but the reported missing key/error context does not match closely enough to be the same bug." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "Cache-layer update decoration issue is unrelated to Python-version compatibility." + "reason": "Cache update decoration and Python version support for v5.1.0 are unrelated." }, { "left": "issue:44938", "right": "issue:45381", "accept": false, - "reason": "Different regressions: Python 3.14 import/load failure vs Qwen2.5-VL video position id bug." + "reason": "A Python 3.14 load failure is not the same as a Qwen2.5-VL video position-id bug." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "CLIP attentions regression and patchtsmixer `post_init` restriction are unrelated." + "reason": "CLIPOutput attentions missing and patchtsmixer `post_init` policy are unrelated model API issues." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "Different model internals and different mathematical behavior; no shared bug." + "reason": "A missing ReLU in the DSA indexer is not the same as DeepSeekV3 MoE divergence from the remote implementation." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Tensor-parallel training bug and TP OOM on a specific model are not the same issue." + "reason": "Tensor-parallel training bug and Qwen2-57B-A14B-Instruct OOM are both TP-related but not the same underlying defect." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "ImportError about missing PyTorch and tokenizer-loading regression are unrelated." + "reason": "PyTorch import-missing errors and tokenizer-loading regression are different startup problems with different triggers." }, { "left": "issue:43867", "right": "issue:44484", "accept": false, - "reason": "Checkpoint load ordering bug and max shard size question are unrelated." + "reason": "A load-model state_dict ordering error is unrelated to the `max_shard_size` default in save_pretrained()." }, { "left": "issue:43618", "right": "issue:44977", "accept": false, - "reason": "CLIP attentions regression and Qwen3.5 flash-attention generation bug are unrelated." + "reason": "CLIP output attentions assignment and Qwen3.5 flash-attention generation issues are separate code paths." }, { "left": "issue:43618", "right": "issue:44857", "accept": false, - "reason": "Different models and different failures: missing attentions vs AMP/CUDA crash." + "reason": "Different failures: missing CLIP attentions vs LwDetr float16 AMP CUDA crash." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Test seed flakiness is unrelated to a multimodal model loading regression." + "reason": "Test seeding flakiness is unrelated to a Phi-4 multimodal model-loading regression." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Different CI flake causes: missing seed setup vs stale device override test." + "reason": "A flaky integration test due to missing seeds is not the same as a GraniteSpeech CI device-override issue." }, { "left": "issue:42890", "right": "issue:45600", "accept": false, - "reason": "Unrelated test flake and auto-mapping breakage." - }, - { - "left": "issue:43502", - "right": "issue:44062", - "accept": false, - "reason": "Local-files-only network leak and AddedToken argument collision are different bugs." + "reason": "Test flakiness and stale auto_mappings references are different classes of failures." }, { "left": "issue:43010", "right": "issue:44075", "accept": false, - "reason": "Cache no_grad decoration issue is unrelated to optimizer argument handling." + "reason": "Cache update decoration and unused SGD args are unrelated." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. The only clearly supported duplicate theme is the import-time slowdown pair (`issue:41669` / `issue:44246`), and the rest of the pairwise rejections are cautious rather than overstated. No PR-mergeability issues apply here since this is an issue-only cluster." - }, - "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 }, @@ -14037,8 +13882,8 @@ ], "soft_pairs": [ "issue:43723|issue:44556", - "issue:42890|issue:45341", "issue:44222|issue:45125", + "issue:42890|issue:45341", "issue:44387|issue:44556", "issue:42548|issue:44351", "issue:43010|issue:43065", @@ -14057,8 +13902,8 @@ "issue:30333|issue:36296", "issue:44355|issue:45071", "issue:44222|issue:44230", - "issue:44355|issue:44655", "issue:43827|issue:44961", + "issue:44355|issue:44655", "issue:33453|issue:34689", "issue:42757|issue:43576", "issue:33453|issue:36296", @@ -14077,20 +13922,20 @@ "issue:36296|issue:43526", "issue:45362|issue:45850" ], - "prepared_review_unit_hash": "c6f40b8f5ea4904a6f1defb612603b17f314f567db8d2dd890e317c03dae7ead", + "prepared_review_unit_hash": "10a8091e81d7185058e30d9d3ae5dfb90e1c988302afb71dc981bcc60b235dab", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 40, - "serialized_chars": 38986, + "serialized_chars": 38987, "estimated_input_tokens": 9747, "estimated_eval_tokens": 19750 }, @@ -14099,210 +13944,210 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of Transformers v5 regression reports (imports, loading, model behavior, docs/tests), not one shared concrete bug. The soft pairs are mostly high-level similarities only, so I reject them all.", - "confidence": 0.46, - "canonical_issue_reason": "issue:45850 is the broadest umbrella for the cluster\u2019s main theme: many v5-related breakage reports across models and APIs.", + "summary": "This cluster is dominated by heterogeneous Transformers v5 regressions, import/load failures, and a few unrelated docs/test bugs. The soft pairs are mostly not duplicates of the same underlying defect; they just share broad subsystem or version-related similarity.", + "confidence": 0.62, + "canonical_issue_reason": "issue:44556 is the best concrete umbrella among the set: it clearly captures a version-upgrade loading regression and is specific enough to be actionable, while many other issues are narrower variants around v5 breakage.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 is the most representative single issue because it captures the widest version-regression concern, even though it is vague.", + "best_issue_reason": "issue:44556 is the strongest representative issue because it describes a concrete, user-facing reload failure after upgrading versions, which aligns with the broader v5 regression theme in this cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Both involve loading regressions after v5, but one is tokenizer loading and the other is checkpoint reload; different failure modes." + "reason": "Both are loading-related, but one is tokenizer-from_pretrained failure and the other is checkpoint reload after upgrade; different bugs." }, { - "left": "issue:42890", - "right": "issue:45341", + "left": "issue:44222", + "right": "issue:45125", "accept": false, - "reason": "One is a flaky integration test, the other is a generic testing_utils bug; not enough evidence of the same underlying defect." + "reason": "FP8 save_pretrained/MoE bug vs missing _tp_plan for tensor parallelism; related area, but not the same defect." }, { - "left": "issue:44222", - "right": "issue:45125", + "left": "issue:42890", + "right": "issue:45341", "accept": false, - "reason": "FP8 save_pretrained vs missing tensor-parallel plan are different code paths." + "reason": "Both mention tests, but one is flakiness from missing set_seed and the other is a separate testing_utils bug." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Quantization OOM and checkpoint reload breakage are unrelated symptoms." + "reason": "Quantization OOM from reserved memory is a different failure mode than checkpoint reload regression." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different missing-export fixes." + "reason": "Both are import errors, but they target different missing symbols and likely different code paths." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "Cache update no_grad behavior and a dummy Conv2d in Sam3PixelDecoder are unrelated." + "reason": "Cache no_grad decoration and dummy Conv2d in Sam3PixelDecoder are unrelated bugs." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "Local-files-only network leakage and GLM-5 RoPE are unrelated." + "reason": "Offline-mode network leakage and GLM-5 RoPE implementation are unrelated." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled-file import errors and a performance regression during filtering are not the same bug." + "reason": "Compiled-file import/runtime errors vs a performance regression after upgrading; not the same issue." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "Loading hang and documentation confusion are different issues." + "reason": "Sentence Transformers loading hang vs documentation confusion in model.forward; different problem class." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Both mention generation/cache internals, but the concrete problems are different." + "reason": "Both are generation/cache adjacent, but they describe different cache-update bugs in different paths." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Same model family, but tensor-parallel metadata and from_pretrained loading error are distinct regressions." + "reason": "Both involve Qwen3.5 MoE loading/tensor-parallel concerns, but one is missing TP plan and the other is a from_pretrained error; too distinct to merge." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Missing tp-plan and missing config field are related only at a high level, not the same bug." + "reason": "Tensor parallel plan issue vs missing config field silently dropped; not the same bug." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLflow reporting failure and tokenizer loading regression are unrelated." + "reason": "MLflow status reporting after failed training vs tokenizer loading regression; unrelated." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "API requests despite local_files_only and FP8 save_pretrained touch different areas." + "reason": "Offline network requests vs FP8 save_pretrained bug; same broad area but not the same concrete change." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Qwen3VL MoE loading and generic reload-after-upgrade breakage are not the same concrete failure." + "reason": "Both are v5 loading regressions, but one is Qwen3VL MoE-specific and the other is generic checkpoint reload failure." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel application and DSA indexer ReLU are unrelated." + "reason": "model_init Liger-kernel application and DSA indexer ReLU are unrelated." }, { "left": "issue:42831", "right": "issue:43874", "accept": false, - "reason": "FP8 accuracy issues and a missing image-processor method are distinct." + "reason": "FP8 accuracy issue and missing image-patch method are different bugs." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "Local file request leakage and a standalone 'racoon' issue are clearly unrelated." + "reason": "The second item is effectively unrelated/too vague to match the local_files_only issue." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLflow job-status reporting and tensor-parallel training are different problems." + "reason": "MLflow reporting after failed training is unrelated to tensor-parallel training behavior." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled-file errors and PretrainedConfig type checking are unrelated." + "reason": "Compiled Python file failures vs PretrainedConfig type-checking breakage are different issues." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "FP8 save_pretrained and FP8 inference support are different change requests." + "reason": "Both are FP8/MoE related, but one is save_pretrained and the other is inference support; not the same code-path bug." }, { - "left": "issue:44355", - "right": "issue:44655", + "left": "issue:43827", + "right": "issue:44961", "accept": false, - "reason": "Compiled-file errors and pipeline save_pretrained are different bugs." + "reason": "Docs still referencing pipeline() is unrelated to the unrelated/placeholder item." }, { - "left": "issue:43827", - "right": "issue:44961", + "left": "issue:44355", + "right": "issue:44655", "accept": false, - "reason": "Docs referencing pipeline() and an unrelated issue title do not match." + "reason": "Compiled-file errors and inability to save Pipeline objects are different defects." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Tokenizer loading regression and Llama 3.2 Vision model loading breakage are related only broadly." + "reason": "Tokenizer-loading regression vs Llama 3.2 model-loading regression; similar surface area, different underlying bug." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "ImportError from huggingface_hub and a broken env command are different failures." + "reason": "ImportError for is_offline_mode and the transformers env command are unrelated." }, { "left": "issue:33453", "right": "issue:36296", "accept": false, - "reason": "Tokenizer loading and tensor-parallel training are unrelated." + "reason": "Tokenizer loading regression and tensor-parallel training bug are unrelated." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "Broken env command and config type-checking are unrelated." + "reason": "Broken env command and PretrainedConfig type-checking breakage are separate v5 issues." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "Env command breakage and a Qwen2.5-VL import error are different issues." + "reason": "Env command failure and Qwen2_5_VL import error are unrelated." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Docs confusion and an ImportError for layer_type_validation do not share a concrete bug." + "reason": "Documentation ambiguity in forward args vs missing layer_type_validation import; different problems." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "Uninformative racoon issue and Gemma-4 audio positional encoding are unrelated." + "reason": "Placeholder title versus Gemma-4 audio positional encoding bug; no duplicate evidence." }, { "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Both are root-package import errors, but for different missing symbols and not proven to share one fix." + "reason": "AutoImageProcessor import error and PreTrainedModel import error are different missing exports." }, { "left": "issue:38617", "right": "issue:43576", "accept": false, - "reason": "A missing import symbol and the env command issue are unrelated." + "reason": "Missing layer_type_validation import and env command breakage do not share the same bug." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Generation cache behavior and a cache-strategy typo are not the same bug." + "reason": "Chunked-prefill cache bug vs a typo in cache strategies; not the same issue." }, { "left": "issue:43576", @@ -14314,37 +14159,37 @@ "left": "issue:44908", "right": "issue:44961", "accept": false, - "reason": "Scheduler kwargs handling and the racoon issue are unrelated." + "reason": "Scheduler kwargs bug and unrelated placeholder item are not duplicates." }, { "left": "issue:44485", "right": "issue:44961", "accept": false, - "reason": "GLM-5 RoPE implementation and the racoon issue are unrelated." + "reason": "GLM-5 RoPE implementation bug and unrelated placeholder item are not duplicates." }, { "left": "issue:36296", "right": "issue:43901", "accept": false, - "reason": "Tensor-parallel training bug and pipeline docs mismatch are different issues." + "reason": "Tensor-parallel training bug vs pipeline docs mention return_all_scores; unrelated." }, { "left": "issue:41628", "right": "issue:44355", "accept": false, - "reason": "Importing AutoImageProcessor and compiled-file errors are unrelated." + "reason": "Import error for AutoImageProcessor and compiled Python file errors are unrelated." }, { "left": "issue:36296", "right": "issue:43526", "accept": false, - "reason": "Tensor-parallel training and BeitImageProcessorFast label handling are different code paths." + "reason": "Tensor-parallel training bug and BeitImageProcessorFast reduce_labels bug are different code paths." }, { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "These both sound broadly v5-related, but they are not the same concrete bug." + "reason": "Specific Qwen3.5 chat crash vs broad complaint that v5 breaks too many models; too general to treat as the same bug." } ] }, @@ -14429,8 +14274,8 @@ "issue:41669|issue:43673", "issue:44556|issue:45092", "issue:44492|issue:45850", - "issue:43572|issue:45850", "issue:42371|issue:44492", + "issue:43572|issue:45850", "issue:44704|issue:44908", "issue:42890|issue:44908", "issue:45278|issue:45850", @@ -14458,14 +14303,14 @@ "issue:45230|issue:45310", "issue:44273|issue:44908" ], - "prepared_review_unit_hash": "66724294b292c455c06811c7f1b4c44dc7b7e7df5a40d9cd4f692bee81805354", + "prepared_review_unit_hash": "bf502ef058a435bada82af1187f28ba1057ccec0f5d1e79f7d628cd3b7dee6e6", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 47, @@ -14480,252 +14325,252 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of Transformers v5 regressions and model-specific breakages. The strongest unifying theme is upgrade-related loading/import incompatibilities, but the individual reports point to different code paths and symptoms, so they are not duplicates.", - "confidence": 0.67, - "canonical_issue_reason": "issue:45278 is the best umbrella candidate: it broadly describes many import errors after upgrading from 4.57.0 to 5.5.0, which matches the cluster\u2019s general v5 regression theme better than the narrower model- or API-specific reports.", + "summary": "This cluster is mostly a mix of unrelated Transformers v5-era regression reports (imports, config fields, model loading, schedulers, docs, tests). The soft edges do not look like true duplicates, so I rejected them all.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45278 is the best representative issue: it is broad, concrete, and matches the recurring theme of upgrade/import breakage better than the more vague umbrella complaints.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most representative issue in the set because it captures the broad upgrade/regression pattern across multiple failures, whereas most others are narrower single-model or single-feature bugs.", + "best_issue_reason": "issue:45278 has the clearest actionable scope among the cluster\u2019s broad regression reports and best captures the common v5 compatibility/import problem pattern.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44261", "right": "issue:44360", "accept": false, - "reason": "Both are config-related bugs, but they affect different components and failure modes: MLA precision/eps handling vs DSA indexer behavior." + "reason": "Different bugs: RMS norm epsilon precision/config concern vs DSA indexer ReLU implementation issue." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "One is a specific processor.save_pretrained file-saving bug; the other is an unspecific bug report with no concrete matching code path." + "reason": "One is a specific processor save_pretrained file-saving bug; the other is just a placeholder title with no matching concrete failure." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Lazy loading failures and TensorBoard-without-TF backend errors are unrelated subsystems." + "reason": "Lazy-loading/import behavior and a TensorFlow backend dependency error are different failure modes." }, { "left": "issue:44273", "right": "issue:45003", "accept": false, - "reason": "These point to different causes: lazy loading mechanics vs unsafe sys.modules access in modeling_utils." + "reason": "Possible thematic overlap around imports, but not enough to treat lazy loading symptoms as the same bug as a specific sys.modules access issue." }, { "left": "issue:44393", "right": "issue:45753", "accept": false, - "reason": "Both involve Qwen3-VL/Qwen3_5, but one is a 2D bbox hallucination and the other is an infinite loop on a specific image; not the same bug." + "reason": "Same model family, but the reported failures differ: 2D bbox hallucination/error vs infinite loop on a specific image." }, { "left": "issue:43576", "right": "issue:44556", "accept": false, - "reason": "Both are v5-related, but one is the env command and the other is checkpoint reload compatibility; different code paths." + "reason": "Both involve v5 regressions, but one breaks the env command and the other breaks checkpoint reload compatibility; not the same bug." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Both mention Qwen3.5/Qwen3-VL-Moe, but one is a model bug and the other is missing tensor-parallel plan metadata." + "reason": "Both mention Qwen3/Qwen3-VL-Moe, but one is a general bug report and the other is a missing tensor-parallel plan; different code-path issues." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Qwen3-VL output corruption is unrelated to PretrainedConfig type-checking breakage." + "reason": "Model output/vision bug versus PretrainedConfig type-checking regression; unrelated code paths." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "A missing pad_token_id attribute in one config is not the same as a broad complaint that many models break in v5." + "reason": "A specific missing pad_token_id attribute is not the same as a broad complaint that v5 breaks many models." }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-time slowdown from import * usage is different from a missing GenerationMixin cache during chunked prefill." + "reason": "Import-time slowdown from import * removal is distinct from a missing GenerationMixin cache during chunked_prefill." }, { "left": "issue:44556", "right": "issue:45092", "accept": false, - "reason": "Both are v5 checkpoint-loading incompatibilities, but they involve different model families and failure mechanisms, so they are not the same concrete bug." + "reason": "Both are checkpoint-loading regressions after v5, but they affect different model families and likely different root causes." }, { "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "A typo in cache strategies is too specific to match a generic 'many models broken' report." + "reason": "A typo in cache strategy naming is not the same as a vague broad regression report." }, { - "left": "issue:43572", - "right": "issue:45850", + "left": "issue:42371", + "right": "issue:44492", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is a specific config regression, not the broad multi-model breakage described here." + "reason": "TF32 API migration guidance and a cache-strategy typo are unrelated." }, { - "left": "issue:42371", - "right": "issue:44492", + "left": "issue:43572", + "right": "issue:45850", "accept": false, - "reason": "TF32 settings guidance and a cache-strategy typo are unrelated." + "reason": "Missing pad_token_idx in StableLmConfig is a concrete config regression, not the same as a generic v5 breakage complaint." }, { "left": "issue:44704", "right": "issue:44908", "accept": false, - "reason": "AutoProcessor kwargs forwarding and inverse_sqrt scheduler kwargs are different APIs with different bugs." + "reason": "Processor.from_pretrained kwargs plumbing and inverse_sqrt scheduler kwargs handling are different APIs and bugs." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "A flaky integration test due to missing seeds is unrelated to scheduler keyword handling." + "reason": "Test seeding flakiness and scheduler kwargs propagation are unrelated." }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "Both are broad complaints about v5 regressions, but they are not the same specific underlying bug or change." + "reason": "Both are broad v5 complaints, but one is a concrete import-error regression and the other is too vague to be the same issue." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "A model-type mismatch warning and a Liger-kernel/model_init issue do not share a concrete code path." + "reason": "Model type mismatch warning and Liger Kernel application with model_init are different problems." }, { "left": "issue:44246", "right": "issue:44273", "accept": false, - "reason": "Slow import and broken lazy loading are related at a high level, but they are not clearly the same bug." + "reason": "General import slowness and lazy loading not working are adjacent topics, but not clearly the same underlying bug." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python version compatibility and Qwen3-VL output errors are unrelated." + "reason": "Python version incompatibility is unrelated to a Qwen3-VL output/hallucination bug." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention 2 support gating is unrelated to 2D bbox output hallucinations." + "reason": "Flash Attention 2 support gating and a Qwen3-VL bbox output issue are different failures." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "The env command regression and scheduler kwargs bug are separate features." + "reason": "The env command regression and scheduler kwargs bug touch different subsystems." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer class metadata mismatch is unrelated to a generic bug report." + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to an empty placeholder bug report." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs handling and PretrainedConfig type checking are different issues." + "reason": "Scheduler kwarg handling and PretrainedConfig type-checking are different issues." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "Buffer registration/position_ids reading and a generic bug report do not match." + "reason": "position_ids buffer access error is specific and unrelated to the placeholder report." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import * performance overhead and an import failure for HybridCache are distinct problems." + "reason": "Import * performance regression and cannot-import-HybridCache are different import-time bugs." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Import slowdown is not the same as a small testing_utils bug." + "reason": "Model import performance regression is unrelated to a testing_utils bug." }, { "left": "issue:44492", "right": "issue:44945", "accept": false, - "reason": "A cache-strategy typo and incorrect pipeline-parallel output are unrelated." + "reason": "Cache strategy typo and incorrect pipeline-parallel output are not the same code-path." }, { "left": "issue:43874", "right": "issue:45850", "accept": false, - "reason": "A missing method in one image processor is too specific to be the same as a broad v5 breakage complaint." + "reason": "A missing image-patch helper in GLM46V is far more specific than the generic v5 breakage complaint." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Qwen3-VL bbox hallucination and Qwen3.5 Moe from_pretrained errors are different model-level failures." + "reason": "Both are Qwen3.5/VL-related, but one is a bbox output issue and the other is a from_pretrained error; not clearly the same bug." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "Optional post_init behavior in patchtsmixer and a tied-weight-key AttributeError are unrelated." + "reason": "PatchTSMixer post_init policy and tied-weight-key AttributeError are unrelated internals." }, { "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Docs still mentioning pipeline() and scheduler kwargs handling do not reflect the same bug." + "reason": "Docs still referencing pipeline() and scheduler kwargs are unrelated." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger-kernel application with model_init and flash_attention_3 training degeneration are different code paths." + "reason": "Liger Kernel not applied with model_init and degenerate training under flash_attention_3 are different concrete problems." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Documentation confusion in model.forward and tokenizer loading regression are not the same issue." + "reason": "Documentation confusion in model.forward and a tokenizer loading regression are not the same bug." }, { "left": "issue:41669", "right": "issue:44908", "accept": false, - "reason": "Import performance and scheduler kwargs are unrelated." + "reason": "Import-time performance regression and scheduler kwarg handling are unrelated." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "The env command bug and lazy loading malfunction are different features." + "reason": "Broken env command and lazy loading problems are distinct upgrade regressions." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Saving Pipeline objects and tied-weight-key crashes affect different functionality." + "reason": "Saving Pipeline objects and tied-weight key handling are separate features/issues." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "DSA indexer ReLU concerns and Gemma-4 audio positional encoding are unrelated." + "reason": "DSA indexer ReLU concern and Gemma4AudioRelPositionalEncoding are unrelated model internals." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "A generic bug report does not duplicate the specific Qwen3.5 Moe from_pretrained error." + "reason": "Placeholder bug report versus Qwen3.5 Moe from_pretrained regression; no shared concrete bug." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading problems and inverse_sqrt scheduler kwargs are unrelated bugs." + "reason": "Lazy loading failure and inverse_sqrt scheduler kwargs are unrelated." } ] }, @@ -14820,20 +14665,20 @@ "issue:31515|issue:33453", "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "9612df844668f8b1711f5bb5053baff70f44748a9be43aac7732fda88d7cb574", + "prepared_review_unit_hash": "b437dc525c33d54a474984b6ae324656e000bba886af10470b1afd7598cb5fb0", "component_budget": { "node_count": 397, "item_count": 397, "soft_pair_count": 1488, - "serialized_chars": 767102, - "estimated_input_tokens": 191776, - "estimated_eval_tokens": 383808 + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 }, "budget": { "node_count": 36, "item_count": 36, "soft_pair_count": 32, - "serialized_chars": 29799, + "serialized_chars": 29800, "estimated_input_tokens": 7450, "estimated_eval_tokens": 15156 }, @@ -14842,204 +14687,204 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a noisy cluster around Transformers v5/model-loading and tokenizer regressions; the soft pairs are related by broad subsystem, but none look like the same concrete bug or fix.", - "confidence": 0.72, - "canonical_issue_reason": "issue:45092 is the clearest concrete representative of the core theme: a Transformers v5 meta-initialization/loading incompatibility affecting checkpoints.", + "summary": "This cluster is mostly heterogeneous: a loose subgroup concerns Transformers v5 loading/tokenizer/config regressions, but the soft pairs are not the same underlying bug/change. I would not merge any of the candidate pairs.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 33453 is the most central concrete loading-regression report in this set and the best umbrella anchor for the tokenizer/model-loading subgroup.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45092 best captures the cluster\u2019s dominant failure mode with a specific, reproducible loading regression.", + "best_issue_reason": "33453 is the strongest cluster anchor: it is specific, actionable, and closest to the recurring v5 loading/tokenizer complaints without being too vague.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Different model components: timestamp calculation in Qwen3-VL processor vs GLM-5 RoPE implementation." + "reason": "Different domains: Qwen3VL timestamp math vs GLM-5 RoPE implementation." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Processor timestamp bug and inverse_sqrt scheduler kwargs bug are unrelated." + "reason": "Processor timestamp bug vs scheduler kwargs handling; unrelated code paths." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion vs a concrete model-loading regression; not the same bug." + "reason": "Documentation confusion vs a model-loading regression; not the same issue." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import-star cleanup/perf issue is unrelated to the Qwen3CausalLM VRAM leak." + "reason": "Import-star cleanup/perf issue is unrelated to a Qwen3ForCausalLM VRAM leak." }, { "left": "issue:31795", "right": "issue:37428", "accept": false, - "reason": "Docs ambiguity about forward args does not match an ImportError in flash-attention utils." + "reason": "Docs ambiguity vs flash-attention import error; different symptoms and fixes." }, { "left": "issue:44315", "right": "issue:45092", "accept": false, - "reason": "Liger Kernel application in model_init and remote-code/meta-init incompatibility are different code paths." + "reason": "Liger Kernel application in model_init is not the same as meta-init incompatibility for old remote-code checkpoints." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Both touch positional encodings, but they are different models and distinct defects." + "reason": "Both are model-specific, but one is RoPE logic and the other is audio positional encoding; distinct bugs." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak in threaded inference is unrelated to RoPE implementation." + "reason": "VRAM leak in dataloader threads vs RoPE implementation error; no shared code-path." }, { "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "TF32 configuration behavior and a multi-thread VRAM leak are unrelated." + "reason": "TF32 settings warning is unrelated to a memory leak." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "local_files_only network leakage is unrelated to timestamp computation." + "reason": "Unexpected network access vs incorrect timestamp calculation; unrelated behavior." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Processor timestamp math and missing save_pretrained files are different bugs." + "reason": "Processor timestamp bug vs missing save_pretrained files; different API surfaces." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Timestamp calculation in a processor is unrelated to a VRAM leak." + "reason": "Processor timestamp bug vs VRAM leak; not the same failure mode." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "torch.split return handling in GLM-MoE indexing is unrelated to Gemma-4 audio positional encoding." + "reason": "torch.split handling in GLM MoE indexing is unrelated to Gemma audio positional encoding." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch and audio positional encoding are different failure modes." + "reason": "CPU offload device mismatch is a runtime placement bug, not an audio encoding issue." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "RMS norm epsilon precision issue is unrelated to Gemma-4 audio positional encoding." + "reason": "RMS norm epsilon precision issue does not match Gemma audio positional encoding." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both concern tokenizer v5 behavior, but one is broad behavior drift and the other is a specific decoder-type bug; not clearly the same issue." + "reason": "Both involve tokenizer v5 changes, but one is broad behavior drift and the other is a specific decoder-type bug." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs bug is unrelated to an off-topic/irrelevant issue." + "reason": "Scheduler kwargs bug and an unrelated unemployment issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weight key handling and chat-template tool-call crashes are different code paths." + "reason": "Tied-weight key handling is unrelated to chat template crashes on tool-call assistant messages." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "A cache-strategy typo is unrelated to the off-topic issue." + "reason": "Cache-strategy typo is unrelated to an off-topic unemployment issue." }, { "left": "issue:44998", "right": "issue:45850", "accept": false, - "reason": "Off-topic unemployment issue is unrelated to a broad Transformers v5 complaint." + "reason": "One is nonsensical/off-topic; the other is a broad v5 complaint, but they are not the same bug." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression is unrelated to the off-topic issue." + "reason": "CLIPOutput attentions regression is unrelated to the off-topic unemployment issue." }, { "left": "issue:44998", "right": "issue:45341", "accept": false, - "reason": "Testing-utils bug and the off-topic issue are unrelated." + "reason": "Unrelated off-topic issue versus a testing_utils bug." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight key crashes." + "reason": "Missing pad_token_id on Llama4Config is unrelated to tied-weight key list handling." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "StableLmConfig pad_token_idx regression is unrelated to tied-weight key handling." + "reason": "StableLmConfig pad_token_idx regression is unrelated to tied-weight key extraction." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Checkpoint loading slowness and an ImportError in configuration_utils are not the same bug." + "reason": "Slow checkpoint loading is not the same as an ImportError for layer_type_validation." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "fp8 inference support request is unrelated to missing processor save files." + "reason": "FP8 inference support for Qwen3-VL/Qwen3.5 MoE is unrelated to processor save_pretrained file emission." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow reporting for failed training jobs is unrelated to slow checkpoint loading." + "reason": "MLflow failure reporting and slow from_pretrained checkpoint loading are different problems." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "from_pretrained slowness and tensor-parallel training bugs are different problems." + "reason": "Checkpoint loading slowdown vs tensor-parallel training bug; different code paths." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Checkpoint load performance and forward-doc confusion do not share a concrete defect." + "reason": "Loading performance issue vs documentation confusion." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "fp8 support for Qwen3-VL and a Qwen3.5 MoE from_pretrained error are distinct issues." + "reason": "Both mention Qwen3.5 MoE, but one is fp8 inference support and the other is a from_pretrained regression." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Slow checkpoint loading is not the same as a tokenizer-loading regression." + "reason": "Both concern loading, but one is checkpoint load speed and the other is a tokenizer-loading regression." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching support for audio models and fp8 inference support are different feature gaps." + "reason": "Continuous batching audio-model support and fp8 Qwen3-VL/Qwen3.5 MoE inference are unrelated." } ] }, @@ -15071,12 +14916,12 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "e0e40c5e11f93eda21729991f4c8fd20df3c2c5ee3f2f287b8b82b604d1effda", + "prepared_review_unit_hash": "79543dcba06d67c60dc8ac378e42e419e6e9c38926a34e1fdd57c5d6d292fa04", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2479, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -15084,7 +14929,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2479, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -15093,24 +14938,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues do not look like true duplicates: one is about TAPAS tokenization mismatching TensorFlow, one is about the error message for unauthenticated private-model downloads, and one is a generic \"can't load the model\" report. The soft links are too broad to merge.", - "confidence": 0.38, - "canonical_issue_reason": "issue:16618 is the broadest user-facing model-loading report, but it still appears distinct from the TAPAS tokenization bug and the private-model auth/error-message bug.", + "summary": "This cluster appears to contain three unrelated issue themes: tokenization mismatch, private model download/auth error messaging, and a generic model loading failure. None of the proposed soft pairs look like the same underlying bug.", + "confidence": 0.93, + "canonical_issue_reason": "Issue #16618 is the broadest and most representative of a core model-loading failure; the others are separate concerns about tokenization behavior and auth/error messaging.", "canonical_pr_reason": null, - "best_issue_reason": "issue:16618 is the best representative of the cluster only because it is the most general model-loading failure; the other issues are clearly more specific and not the same underlying problem.", + "best_issue_reason": "#16618 is the best single issue to anchor the cluster because it matches the most general user-facing failure mode and is the most likely duplicate target among these candidates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "TAPAS tokenization discrepancy is a different code path and symptom than a generic model-loading failure; no evidence they are the same bug." + "reason": "Tokenization mismatch vs. model loading failure are different bugs affecting different code paths; not the same underlying issue." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Private-model download/auth error-message issue is about login handling, not the same underlying failure as loading bert-base-uncased." + "reason": "One is an auth/download error message problem for private models, the other is a generic inability to load a public model; related area but not the same bug." } ] }, @@ -15140,12 +14985,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "6ed4e1bc6c62768dffb3ea93fa15f945fe03015c328ca3e9434445de7a4bad96", + "prepared_review_unit_hash": "8caa4d8a8429a71fdcb0167054a661205e24526ea34b0b016480cdd5ebbb57f7", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1590, + "serialized_chars": 1589, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -15153,7 +14998,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1590, + "serialized_chars": 1589, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -15162,18 +15007,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two separate feature-request issues: one for Flash Attention 2 support across more architectures, the other for GGUF support across more architectures. They share a broad \u201ccommunity contribution\u201d theme, but not the same underlying bug or change.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "These are not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They target different features and code paths.", + "confidence": 0.98, + "canonical_issue_reason": "Issue #26350 is the stronger representative only because it has much higher discussion activity, but it is not a duplicate of #33260.", "canonical_pr_reason": null, - "best_issue_reason": "No single canonical issue: the requests target different features and should remain separate. If forced to pick a representative, #26350 is slightly older and has higher discussion activity, but it is not a duplicate of #33260.", + "best_issue_reason": "#26350 is the best issue to anchor the cluster if one must be chosen, due to higher activity and clearer community interest; however, the pair should not be merged.", "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 support. Similar wording around architecture support is too broad to treat as the same issue." + "reason": "Different underlying requests: Flash Attention 2 vs GGUF support. They concern separate model/runtime integration paths and are not the same bug or change." } ] }, @@ -15210,7 +15055,7 @@ "issue:43834|issue:44365", "issue:44365|issue:44370" ], - "prepared_review_unit_hash": "79c7623d7a98026c6e8096ed9f238c23186bbb705c7662247346c434dfef599b", + "prepared_review_unit_hash": "1aaeeb0c9aa25514d5a4730a6293b3715b238718db942ab0432f38966e129d8c", "component_budget": { "node_count": 4, "item_count": 4, @@ -15232,48 +15077,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are all i18n documentation-translation issue templates, but only one is concrete (Turkish). The others look like placeholder instances for different languages, so they should not be merged as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "issue 27088 is the only language-specific, concrete request ([i18n-TR] Translating docs to Turkish); the others are generic placeholder-style issues and likely refer to other languages.", + "summary": "All four issues are highly similar translation-request templates, but the concrete language target differs or is unspecified, so they do not look like the same underlying issue. The Turkish issue is the most specific and best representative of the cluster.", + "confidence": 0.73, + "canonical_issue_reason": "issue:27088 is the only concrete, fully specified request ('Turkish'); the others are generic placeholder/template variants for some language and are less suitable as the canonical representative.", "canonical_pr_reason": null, - "best_issue_reason": "issue 27088 is the best representative because it is specific, complete, and not a template placeholder.", + "best_issue_reason": "issue:27088 is the best issue candidate because it is concrete and actionable, while the others are generic template-style duplicates with unspecified language targets.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Same issue template, but not the same underlying request; 27088 is Turkish while 44370 is a generic placeholder that could be a different language." + "reason": "Same translation-request pattern, but one is explicitly Turkish and the other is an unspecified language placeholder; different underlying request." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Both are translation-template issues, but only 27088 is concretely Turkish; the placeholder issue is not clearly the same request." + "reason": "The titles indicate different language targets (Turkish vs placeholder generic language); not the same issue." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "The titles match the template pattern, but 27088 is a specific Turkish translation request and 43834 is a generic placeholder for some other language." + "reason": "Concrete Turkish request does not match a generic '/' placeholder issue; likely separate translation requests." }, { "left": "issue:43834", "right": "issue:44370", "accept": false, - "reason": "These are similar template issues, but the placeholder text does not establish the same language or same concrete task." + "reason": "Both are template-style translation issues, but the available titles do not show the same language target or same concrete request." }, { "left": "issue:43834", "right": "issue:44365", "accept": false, - "reason": "Highly similar wording, but no evidence they are the same underlying translation request rather than separate language placeholders." + "reason": "The two issues appear to be separate instances of the same template, not evidence of one identical underlying request." }, { "left": "issue:44365", "right": "issue:44370", "accept": false, - "reason": "The issues appear to be template-generated translation requests, not the same specific bug or change." + "reason": "Very similar template text, but no concrete shared language target or identical request is visible; keep them separate." } ] }, @@ -15350,244 +15195,244 @@ "issue:37015|issue:45027", "issue:43837|issue:45177" ], - "prepared_review_unit_hash": "dfb9403b449b761df206315a105863c5aebb9086a213ad84661e729a627d7a87", + "prepared_review_unit_hash": "077d6d0e6e6b78f775764d673a72560220f2dc066bb262bbdfcf60c2435a49e1", "component_budget": { "node_count": 15, "item_count": 15, "soft_pair_count": 35, - "serialized_chars": 20711, - "estimated_input_tokens": 5178, - "estimated_eval_tokens": 10612 + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 }, "budget": { "node_count": 15, "item_count": 15, "soft_pair_count": 35, - "serialized_chars": 20711, - "estimated_input_tokens": 5178, - "estimated_eval_tokens": 10612 + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a loose topic group of model-addition requests, not true duplicates. The items span unrelated model families and even different modalities (BERT-like encoders, audio/TTS/ASR, OCR, sparse transformers, vision). No soft-edge pair should be merged.", - "confidence": 0.81, - "canonical_issue_reason": "Issue 41115 is the broadest active request in the set ('Add Model Architecture for MiniCPM3') and is the closest generic representative, but the cluster is heterogeneous rather than duplicate-heavy.", + "summary": "These are all separate model-request issues; the title similarity is mostly generic \"add/support model\" wording, not the same underlying bug or change. None of the soft pairs look mergeable as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:43837 is the strongest hub in this cluster by soft-similarity and is a clear, open request; if one issue must represent the group, this is the best fit, though it is still a distinct Qwen3-ASR request rather than a duplicate of the others.", "canonical_pr_reason": null, - "best_issue_reason": "41115 is the best cluster anchor only in a loose thematic sense because it is a general model-architecture request; however, it does not actually subsume the other model-specific issues.", + "best_issue_reason": "issue:43837 is the most representative single issue: it is specific, open, and centrally connected to several of the audio/model-request items, but it does not actually subsume the other model additions.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different requested models; EuroBert and DEIMv2 are unrelated additions." + "reason": "EuroBERT config vs DEIMv2: different model families and different requests; only the generic \"add model\" framing matches." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "BERT model config request vs Qwen3-TTS audio support; different modality and code path." + "reason": "EuroBERT config vs Qwen3-TTS: unrelated model families and tasks." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "EuroBert config vs RF-DETR model; distinct model families." + "reason": "EuroBERT config vs RF-DETR: one is a config entry request, the other is a specific model addition." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are different models, not the same fix/request." + "reason": "RF-DETR vs DEIMv2: different architectures and unrelated model additions." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "RF-DETR vision model vs LFM2.5 Audio request; unrelated." + "reason": "RF-DETR vs LFM2.5 Audio 1.5B: object-detection model request vs audio model request." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR vs weight-sparse transformer request; different underlying work." + "reason": "RF-DETR vs OpenAI Weight-Sparse Transformer: unrelated model families and change scopes." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR vision model vs Qwen3-ASR audio support; not duplicates." + "reason": "RF-DETR vs Qwen3-ASR: different model families and tasks." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are different model families/modalities." + "reason": "RF-DETR vs Qwen3-TTS: different model families and tasks." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT request and Qwen3-ASR support are unrelated." + "reason": "NeoBERT vs Qwen3-ASR: BERT-style text model request vs ASR support request." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT vs sparse transformer support; distinct code paths." + "reason": "NeoBERT vs OpenAI Weight-Sparse Transformer: different model families and different requests." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT encoder model vs audio model request; not the same issue." + "reason": "NeoBERT vs LFM2.5 Audio 1.5B: unrelated model families and tasks." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR vision model and NeoBERT are separate model additions." + "reason": "RF-DETR vs NeoBERT: unrelated model additions with no shared concrete change." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are both BERT-like, but they are different model requests." + "reason": "NeoBERT vs ModernVBERT models: similar naming, but different BERT variants and separate requests." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture request vs Qwen3-ASR support; different modality and target model." + "reason": "MiniCPM3 vs Qwen3-ASR: different model families and different tasks." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "Generic model-architecture request vs sparse transformer request; not the same underlying change." + "reason": "MiniCPM3 vs OpenAI Weight-Sparse Transformer: unrelated model requests." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are separate model support requests." + "reason": "MiniCPM3 vs ModernVBERT models: different model families and no shared concrete code change." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 architecture vs Voxtral TTS support; different model and task." + "reason": "MiniCPM3 vs Voxtral-4B-TTS-2603: unrelated model families and modalities." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated additions." + "reason": "DEIMv2 vs Qwen3-TTS: different model families and tasks." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are different model requests." + "reason": "ModernVBERT models vs Qwen3-ASR: unrelated model families and tasks." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT encoder-style model vs Qwen3-TTS audio model; not duplicates." + "reason": "ModernVBERT models vs Qwen3-TTS: different model families and different requests." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT request is model-specific; BERT-like models with RoPE is a broader architecture feature request." + "reason": "ModernVBERT models vs BERT-like models with RoPE: related wording, but one is a specific model request and the other is a general architectural feature request." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE support and Voxtral TTS support are unrelated code paths." + "reason": "BERT-like models with RoPE vs Voxtral TTS: unrelated model families and scopes." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are different tasks; same vendor name is not enough." + "reason": "Qwen3-TTS vs Qwen3-ASR: same product line, but different modalities/tasks and not the same change." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Two different TTS model requests; different model families." + "reason": "Qwen3-TTS vs Voxtral-4B-TTS-2603: both TTS-related, but different model families and separate support requests." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated modalities." + "reason": "Qwen3-TTS vs DeepSeek-OCR2: different modalities and unrelated requests." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "TTS support request vs sparse transformer request; not the same bug/change." + "reason": "Qwen3-TTS vs OpenAI Weight-Sparse Transformer: unrelated model families." }, { "left": "issue:43671", "right": "issue:43909", "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio are separate audio model requests, not duplicates." + "reason": "Qwen3-TTS vs LFM2.5 Audio 1.5B: both audio-related, but different model families and separate additions." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are different audio-model requests." + "reason": "Qwen3-ASR vs LFM2.5 Audio 1.5B: same broad audio domain, but different models and tasks." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "ASR support vs EUPE vision/encoder request; different model families." + "reason": "Qwen3-ASR vs EUPE: ASR support request vs a perception-encoder model request; not the same change." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "ASR support and sparse transformer support are unrelated." + "reason": "Qwen3-ASR vs OpenAI Weight-Sparse Transformer: unrelated model families and scopes." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different model additions." + "reason": "LFM2.5 Audio 1.5B vs SAM3-LiteText: audio model request vs multimodal perception/text model request." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Weight-sparse transformer request vs EUPE encoder request; not the same underlying change." + "reason": "OpenAI Weight-Sparse Transformer vs EUPE: different model families and separate requests." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are different audio tasks/models." + "reason": "Qwen3-ASR vs Voxtral-4B-TTS-2603: both audio-related, but ASR and TTS are different concrete changes." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are unrelated requests." + "reason": "NeoBERT vs Voxtral-4B-TTS-2603: unrelated model families and modalities." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "ASR support and OCR model support are different code paths and modalities." + "reason": "Qwen3-ASR vs DeepSeek-OCR2: ASR support vs OCR model support; different underlying changes." } ] }, @@ -15628,82 +15473,82 @@ "issue:43668|issue:44031", "issue:43668|issue:43813" ], - "prepared_review_unit_hash": "3f1f143a32e4df4430ae55ec14835a5d42f2382cfa86ba4f4adbd8f124fb89d2", + "prepared_review_unit_hash": "20d02a86283c3c6a7671907080253a8ffea65c6938bf8f1df54a64934e8a2309", "component_budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6292, - "estimated_input_tokens": 1573, - "estimated_eval_tokens": 3402 + "serialized_chars": 6288, + "estimated_input_tokens": 1572, + "estimated_eval_tokens": 3400 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6292, - "estimated_input_tokens": 1573, - "estimated_eval_tokens": 3402 + "serialized_chars": 6288, + "estimated_input_tokens": 1572, + "estimated_eval_tokens": 3400 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is not a duplicate set: the issues describe distinct bugs across unrelated areas (model norms, loss plumbing, tokenizer warnings, a typo, and a type-hint problem). The soft similarities are mostly lexical.", - "confidence": 0.95, - "canonical_issue_reason": "No clear canonical issue: the reports span different subsystems and failure modes, so they do not appear to share one underlying bug.", + "summary": "These issues are only loosely similar at a surface level (model/config/loss/tokenizer bugs), but they describe different code paths and distinct defects. No soft-edge pair looks like the same underlying bug.", + "confidence": 0.82, + "canonical_issue_reason": "No clear canonical duplicate emerges; the issues target different components and symptoms rather than one shared bug.", "canonical_pr_reason": null, - "best_issue_reason": "No strong global representative exists; 43668 is the closest to a generic code/metadata bug, but the cluster is too heterogeneous for a true best issue.", + "best_issue_reason": "Issue 43668 is the cleanest self-contained report, but it is not a duplicate representative of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:37934", - "right": "issue:43668", + "left": "Is Llama4TextL2Norm meant to be RMS norm?", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", "accept": false, - "reason": "Different bugs: one questions Llama4 norm semantics, the other is a ModernBERT type-hint error." + "reason": "Different models and problems: one questions Llama4 norm semantics, the other is a ModernBERT typing issue." }, { - "left": "issue:42936", - "right": "issue:43240", + "left": "Mask2former model's ignore_value not used after definition", + "right": "kwargs are not passed to loss calculation function.", "accept": false, - "reason": "Unrelated areas: Mask2Former ignore_value handling vs. loss kwargs plumbing." + "reason": "Different bug locations and failure modes: Mask2former ignore_value handling vs loss kwargs plumbing." }, { - "left": "issue:42936", - "right": "issue:44031", + "left": "Mask2former model's ignore_value not used after definition", + "right": "All tokenizers raise incorrect regex pattern warning after version 4.57.3?", "accept": false, - "reason": "Different components and failures: Mask2Former ignore_value vs. tokenizer regex warning." + "reason": "Different subsystems entirely: model loss configuration vs tokenizer warning regression." }, { - "left": "issue:42936", - "right": "issue:43813", + "left": "Mask2former model's ignore_value not used after definition", + "right": "Typo - should be \"orig_conversion.quantization_operation\"", "accept": false, - "reason": "One is a model-loss/config bug, the other is a naming typo in conversion code." + "reason": "Unrelated issues: Mask2former ignore_value usage vs a typo in a quantization-related attribute name." }, { - "left": "issue:42936", - "right": "issue:43668", + "left": "Mask2former model's ignore_value not used after definition", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", "accept": false, - "reason": "No shared underlying bug; these are separate configuration/code-quality issues in different models." + "reason": "Different models and defect types; one is unused ignore_value logic, the other is a config type hint problem." }, { - "left": "issue:43240", - "right": "issue:43668", + "left": "kwargs are not passed to loss calculation function.", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", "accept": false, - "reason": "Loss kwargs not forwarded is unrelated to a ModernBERT type annotation mistake." + "reason": "Different concerns: missing kwargs propagation into loss computation vs incorrect type annotation." }, { - "left": "issue:43668", - "right": "issue:44031", + "left": "ModernBERTConfig `norm_eps` type hint is incorrect", + "right": "All tokenizers raise incorrect regex pattern warning after version 4.57.3?", "accept": false, - "reason": "Distinct problems: incorrect type hint vs. tokenizer regex pattern warning after a version change." + "reason": "Different areas and symptoms: ModernBERT config typing vs tokenizer warning regression." }, { - "left": "issue:43668", - "right": "issue:43813", + "left": "ModernBERTConfig `norm_eps` type hint is incorrect", + "right": "Typo - should be \"orig_conversion.quantization_operation\"", "accept": false, - "reason": "Type-hint correction and a typo in a different symbol are not the same bug." + "reason": "A type-hint fix and a string typo in another code path are not the same underlying bug." } ] }, @@ -15733,7 +15578,7 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "61215756f5be66e2d993edf1cf97e45c05ff659823d658dcc9f9460c14bb3f42", + "prepared_review_unit_hash": "f998e09d4a92338d7511caa5151185e8978a0cb5c117200e0998db1e961d1db0", "component_budget": { "node_count": 2, "item_count": 2, @@ -15755,18 +15600,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are not duplicates: one is about block decoding behavior, the other is about adding DeepSeek v3.2 support. They describe different feature/problem areas with no concrete overlap.", - "confidence": 0.97, - "canonical_issue_reason": "Issue #41084 is the better canonical representative only because it is more generic and technical, but it is not the same underlying request as #43037.", + "summary": "These are unrelated open issues: one asks for block decoding, the other asks for DeepSeek v3.2 support. They do not look like the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical duplicate issue here; the requests are about different concerns (decoding behavior vs model/version support).", "canonical_pr_reason": null, - "best_issue_reason": "#41084 is the stronger anchor for the cluster if one must be chosen, since it describes a broader code-path topic; however, the items are not duplicates.", + "best_issue_reason": "43037 is the more concrete feature request and is better suited as a standalone representative issue, but it is not a duplicate of 41084.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different topics: block decoding vs. DeepSeek v3.2 model support. No evidence they refer to the same bug or change." + "reason": "Different topics and code paths: block decoding vs DeepSeek v3.2 support. Shared similarity is only broad ML/runtime context, not the same underlying change." } ] }, @@ -15796,7 +15641,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "d372a1e7d9488c4240f750945b37edc3af8b0ac172c51aa0259544f62b024006", + "prepared_review_unit_hash": "f2e08bb507e354cd34a781faf7f8ffa5c7881bf3be5b65d6cbb4e33dcb447b1c", "component_budget": { "node_count": 2, "item_count": 2, @@ -15818,18 +15663,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related only at a high level (Trainer customizability), but they describe different problems: one is about `remove_unused_column`, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", - "confidence": 0.87, - "canonical_issue_reason": "#43598 is the better representative issue because it focuses on a general Trainer customization limitation, which is the broader of the two themes.", + "summary": "The two issues are related at a high level to custom trainers and Trainer behavior, but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", + "confidence": 0.76, + "canonical_issue_reason": "Issue 43598 is the broader, more direct Trainer customization concern and better represents the cluster theme.", "canonical_pr_reason": null, - "best_issue_reason": "#43598 is the most central issue for this cluster since it directly targets a Trainer behavior flag (`remove_unused_column`) and frames the customization concern more concretely.", + "best_issue_reason": "Issue 43598 is the best issue candidate because it addresses the underlying Trainer customization behavior rather than a narrower derived symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Different underlying behavior: `remove_unused_column` handling versus FLOPS estimation for custom trainers. They share a broad customization theme but not the same bug or fix path." + "reason": "They share a general custom-trainer context, but the concrete problems differ: one concerns `remove_unused_column` customizability, the other concerns FLOPS accounting. Not the same underlying bug." } ] }, @@ -15859,7 +15704,7 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "c17dc91bc338ef9c0ec695c4e22fe37493eb9b0f06288012833eb6e4921027c3", + "prepared_review_unit_hash": "595cd49557777a2a8b20649268c7cdbb735ed142430ac16dbd4f5f337d7afe00", "component_budget": { "node_count": 2, "item_count": 2, @@ -15881,18 +15726,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues are about different model families and different API surface areas: one requests registration for Qwen3-Omni in AutoModel/AutoModelForConditionalGeneration, the other requests AutoModelForSequenceClassification support for Qwen3.5. They do not look like the same underlying bug or change.", + "summary": "These issues are related only at a broad Transformers/model-registration level, but they target different model families and different APIs: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs. Qwen3.5 sequence-classification support. They should not be merged as duplicates.", "confidence": 0.97, - "canonical_issue_reason": null, + "canonical_issue_reason": "Issue 43684 is the cleaner representative because it is a concrete registration request with explicit target classes and a narrowly defined model family.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 43684 is the best issue to keep as the cluster representative since it is specific and actionable, while the other issue is a separate feature request for a different config/class combination.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different targets and use cases: Qwen3-Omni model registration vs Qwen3.5 sequence-classification support. These are separate feature requests, not a duplicate fix." + "reason": "Different underlying requests: one asks for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, the other asks for AutoModelForSequenceClassification support for Qwen3.5 (Qwen3_5Config). They do not describe the same bug or change." } ] }, @@ -15929,7 +15774,7 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "85848dab40d4822b7913cb47b659f38abbc4fe7ce8c2469c09e79fc4e9ad2ee3", + "prepared_review_unit_hash": "733719b45928aa04790a1997019c14cad10173f1e38498632c66b76d6cd6afc6", "component_budget": { "node_count": 4, "item_count": 4, @@ -15951,54 +15796,54 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three issues share the same minimal title \"huggingface \" and are very likely duplicates; the \"Fett\" issue looks unrelated. With no bodies or code references, this is a title-based dedupe only.", - "confidence": 0.83, - "canonical_issue_reason": "issue:43845 is one of the duplicated \"huggingface \" reports and is the earliest of that subgroup, so it is the safest canonical issue among the visible items.", + "summary": "Three issues share the exact same vague title \"huggingface \" and were opened close together, so they likely represent one duplicate cluster; \"Fett\" looks unrelated.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface \" reports, so it is the best representative for the duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43845 best represents the cluster because it matches the repeated title exactly and appears earliest; 43859 has a tiny bit more discussion but no extra specificity.", + "best_issue_reason": "Issue 43845 is the cleanest canonical choice because it matches the other duplicate titles and was created first.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no shared technical detail; \"Fett\" does not look like the same underlying issue as \"huggingface \"." + "reason": "Different title/content signals; \"Fett\" does not appear to describe the same issue as the \"huggingface \" reports." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no evidence they describe the same bug or request." + "reason": "Different title/content signals; no evidence these refer to the same underlying bug." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no supporting details tying them to the same underlying issue." + "reason": "Different title/content signals; no evidence these refer to the same underlying bug." }, { "left": "issue:43845", "right": "issue:43846", "accept": true, - "reason": "Same exact title \"huggingface \" and near-identical sparse metadata; these look like duplicate reports of the same issue." + "reason": "Same exact title and very close creation time, with no distinguishing details, strongly suggesting duplicates of the same issue." }, { "left": "issue:43845", "right": "issue:43859", "accept": true, - "reason": "Same exact title \"huggingface \" with no distinguishing details; likely duplicate reports." + "reason": "Same exact title and same vague report pattern; likely a duplicate of the same underlying issue." }, { "left": "issue:43846", "right": "issue:43859", "accept": true, - "reason": "Same exact title \"huggingface \" and no evidence of a distinct underlying issue." + "reason": "Same exact title and same vague report pattern; likely a duplicate of the same underlying issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet: three issues have the same exact title and similarly sparse metadata, while 43835 has a different title and no evidence tying it in. The soft-edge verdicts stay conservative and do not overclaim beyond title-based similarity. " + "feedback": "The summary is grounded in the packet: 43845/43846/43859 all share the exact title \"huggingface \" and appear close enough to plausibly form one duplicate cluster, while 43835 titled \"Fett\" is distinct. The canonical choice of 43845 is conservative and supported by it being the earliest of the three similar reports. The soft-edge verdicts are also appropriately cautious." }, "evaluator_used": true, "retried": false, @@ -16036,7 +15881,7 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "3f296fe36317a6cecd8c33280a9fd63f9b07004e8e461ab4954cdd4a3d103385", + "prepared_review_unit_hash": "2665af50fba4dcd2142dc08cace44ce1e4c4a3c568306408839281d473edf44c", "component_budget": { "node_count": 5, "item_count": 5, @@ -16058,66 +15903,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster appears to be spam/noise rather than real duplicate bug reports. The repeated \"\" issues are superficially similar, but there is no concrete shared defect or change request to merge.", - "confidence": 0.97, - "canonical_issue_reason": "Issue 44371 is the most representative placeholder of the cluster because it is one of the repeated \"\" reports, but none of the issues contain enough substance to establish a real canonical bug.", + "summary": "This cluster is mostly low-information spam/generic issues with no bodies or actionable details. The titles overlap superficially, but there is not enough evidence that they describe the same underlying problem, so I would not deduplicate them as true duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "If one issue must serve as canonical, issue #44371 is the earliest and has the same placeholder spam-style title as the closest matches, making it the most representative of the group\u2019s noise pattern.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44371 is the best cluster representative only by being a generic spam-like placeholder; it is not a meaningful canonical bug report.", + "best_issue_reason": "Issue #44371 is the best representative issue in this set because it is earliest and shares the same generic spam title form as the other placeholder issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are spam-like placeholders, but there is no evidence they describe the same underlying issue." + "reason": "Both are just '' placeholders with no substantive body; identical titles alone are not enough to prove the same underlying issue." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Same generic spam pattern, but no concrete shared bug, target, or reproduction to justify duplication." + "reason": "Same as above: generic spam titles and no issue details or shared targets/file context." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is a spam placeholder and the other is a documentation suggestion; these are unrelated." + "reason": "One is spam-placeholder text and the other is a vague documentation suggestion; these do not appear to be the same concrete issue." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "\"Check out \\\"Google LLC\\\"\" is unrelated to the spam placeholder reports and does not match the same issue." + "reason": "'' and 'Check out \"Google LLC\"' are different low-information titles with no shared technical context." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "Documentation enhancement and spam placeholder are not the same underlying issue." + "reason": "A generic documentation suggestion is not the same underlying problem as a spam placeholder issue." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Documentation enhancement and spam placeholder are unrelated." + "reason": "No concrete overlap beyond both being vague/low-detail issues." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders, but there is no evidence they refer to the same underlying bug or request." + "reason": "Although both are '' placeholders, there is no body, target, or context showing they report the same underlying bug rather than separate spam reports." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Different spam/ad-like titles with no shared concrete defect or change request." + "reason": "Different titles and no supporting details tie them to the same issue." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Different spam/ad-like titles with no evidence of a single underlying issue." + "reason": "Different low-information titles with no evidence of a shared root cause." } ] }, @@ -16147,12 +15992,12 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "e257e3f5e35d0243be0e502f4486b40dca5d878e376f516976ecc586b9ce429b", + "prepared_review_unit_hash": "76b0893a14957d42661b96f2c88b34d9c229dc0e7949298e49ee196b611d8dbc", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, + "serialized_chars": 1645, "estimated_input_tokens": 412, "estimated_eval_tokens": 1080 }, @@ -16160,7 +16005,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1646, + "serialized_chars": 1645, "estimated_input_tokens": 412, "estimated_eval_tokens": 1080 }, @@ -16169,28 +16014,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues have identical titles and very similar bodies, indicating a duplicate report for the same energy-optimization article/topic.", - "confidence": 0.99, - "canonical_issue_reason": "issue #45145 is the better canonical issue because it was created first and already has inbound reference activity, making it the more established thread for this duplicate topic.", + "summary": "These two closed issues have identical titles and similar size, so they are very likely related and may be duplicates, but the packet does not expose body content, so the evidence is not strong enough to confirm they are the same request. #45145 remains the best canonical issue because it has the stronger reference signal.", + "confidence": 0.82, + "canonical_issue_reason": "#45145 is the best canonical issue because it shares the exact title with #45151 and has the only inbound reference, making it the likelier primary thread.", "canonical_pr_reason": null, - "best_issue_reason": "issue #45145 is the best issue candidate overall: same title/body as #45151, earlier timestamp, and slightly more evidence of being the primary discussion target.", + "best_issue_reason": "#45145 is the strongest single issue representative due to the identical title and slightly higher linkage signal.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", - "accept": true, - "reason": "Same exact title and highly similar content; they appear to be duplicate reports of the same underlying issue/topic." + "accept": false, + "reason": "The titles are identical and the issues are probably related, but the packet does not show the issue bodies or concrete content overlap. That is enough for a likely same-topic pair, not enough to confidently accept a duplicate edge." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The conclusion is grounded in the packet: the two issues have identical titles, similar body lengths, and the pair is explicitly marked with soft similarity. The canonical choice is also conservative and supported by the earlier timestamp plus inbound references. No PR mergeability concerns apply here." + "accept": false, + "feedback": "The canonical choice is grounded, but the duplicate verdict is too strong for the evidence provided. The packet only shows identical titles, similar body lengths, and a soft-similarity signal; it does not expose the issue bodies, so phrases like \"clearly the same proposed energy-optimization work\" and \"straightforward duplicate pair\" overstate certainty. This should be framed more conservatively as a likely duplicate/same-topic pair, not a definite one." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -16267,40 +16112,40 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "6d15667a38457cd70c9d58994908241a20bb885242935be70debcb87d3403557", + "prepared_review_unit_hash": "b998c2ed9383e81edb09276e441c622e520fa75a1328595f94461ba314a79ddd", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1667, - "estimated_input_tokens": 417, - "estimated_eval_tokens": 1090 + "serialized_chars": 1669, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1667, - "estimated_input_tokens": 417, - "estimated_eval_tokens": 1090 + "serialized_chars": 1669, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues both occur in 4-bit inference, but they describe different failure modes and code paths: one is a torch.finfo TypeError on uint8 weights in Gemma4 audio modules, the other is an AssertionError caused by llm_int8_skip_modules clearing the default lm_head exclusion. They are related in area, not duplicates.", - "confidence": 0.9, - "canonical_issue_reason": "Issue #45674 is the better cluster representative because it states a clearer, more general BitsAndBytesConfig bug with a concrete reproduction and explicit configuration trigger.", + "summary": "Both items are 4-bit inference bugs, but they describe different root causes and different failing paths: Gemma4 audio-module uint8 weights hitting `torch.finfo()` versus BitsAndBytesConfig skip-module handling removing the default `lm_head` exclusion and triggering an AssertionError. They should not be deduplicated.", + "confidence": 0.96, + "canonical_issue_reason": "Issue #45674 is the broader, more central 4-bit inference regression and has stronger discussion/inbound activity, so it is the better representative if one issue must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "#45674 is the most actionable and broadly applicable issue in the cluster; it identifies a specific configuration regression rather than a model-specific symptom.", + "best_issue_reason": "Issue #45674 is the better cluster representative because it is more general and appears to be the more reusable description of a 4-bit BitsAndBytesConfig failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Different underlying bugs: torch.finfo TypeError on uint8 audio-module weights vs lm_head exclusion being removed and triggering an AssertionError. Same broad 4-bit area, but not the same fixable defect." + "reason": "Same high-level area (4-bit inference) but different bug mechanisms, code paths, and symptoms; not the same underlying issue." } ] }, @@ -16330,7 +16175,7 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "51767c263cb3bec0b1425a3af2403ddce620fc5a3eb897f82d57151eabd03f4f", + "prepared_review_unit_hash": "ec75034adf5686b15c4624bc5ad53fad7ad051a1ffb665e8999bbc9c7536e78c", "component_budget": { "node_count": 2, "item_count": 2, @@ -16352,18 +16197,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicates; they are two unrelated closed issues that only resemble each other by having a personal name plus a date in the title.", - "confidence": 0.99, - "canonical_issue_reason": null, + "summary": "These two issues are not duplicates; they are unrelated name/date entries with no shared technical problem or code path.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 45762 is the slightly better canonical placeholder only because it was created first, but there is no real duplicate relationship between the two issues.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Neither issue is a strong canonical duplicate target; if one must be chosen, 45762 is marginally better due to earlier creation, not because of content overlap.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Different people and different dates; no shared bug report, code path, or issue target, so they do not represent the same underlying issue." + "reason": "The titles are different personal-name/date strings and there is no evidence of a shared bug, request, or code-path change." } ] }, @@ -16393,40 +16238,40 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "2c48eb63bee78029e799f977007edfb57244b3f075c83a5778e3d92b49a70390", + "prepared_review_unit_hash": "8c24f8a175f8ca44266f737603101c473f42ed32fea0ee9d29cec92c3f965a37", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "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; they are empty, unrelated entries with different names and dates and no supporting context. This cluster looks more like spam or accidental submissions than a real duplicate set.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the titles refer to different people/dates, and there is no body, linkage, or shared technical context indicating the same underlying report.", + "summary": "The two closed issues appear unrelated and lack any substantive body or references; their titles are just different personal-name/date strings, so the soft similarity is not enough to treat them as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a good representative for a duplicate cluster because they do not describe the same problem and provide no actionable content.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different names and dates, with no description or shared context; they do not appear to be the same underlying issue." + "reason": "Reject: the titles only share a superficial pattern (a name plus date-like text). There is no body, discussion, target, or other evidence that they describe the same underlying bug or request." } ] }, @@ -16466,88 +16311,88 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "533b03576de0827ca146983b499cd0ee9448f28a6dd73b6c8ae13d657a5c7a9e", + "prepared_review_unit_hash": "73474d02abf376ca2abc414ae7451c1732bf1dfbe32cdbca6347acc65b3e7b01", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5721, - "estimated_input_tokens": 1431, - "estimated_eval_tokens": 3118 + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5721, - "estimated_input_tokens": 1431, - "estimated_eval_tokens": 3118 + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two exact duplicate issue pairs are present, but the cluster as a whole splits into different topics (dataset promo vs. two separate DataForge example/request variants), so there is no single true cluster-wide duplicate set.", - "confidence": 0.79, - "canonical_issue_reason": "Issue 45872 is the strongest representative of its duplicate pair: it has the more detailed title/body and is duplicated exactly by 45873; the other items are different topics rather than the same underlying issue.", + "summary": "There are two clear duplicate issue pairs: 45870\u219445871 and 45872\u219445873. The dataset-sales post (45869) is unrelated, and the two financial-time-series items (45870/71 vs 45872/73) are similar in theme but not the same underlying issue.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 45872 is the strongest representative: it is a substantive, product-specific request and has an exact duplicate in 45873, while the other similar items are either a different topic or less clearly central.", "canonical_pr_reason": null, - "best_issue_reason": "45872 is the best overall issue because it is the most specific and substantive of the open issues, and it anchors one of the exact duplicate pairs.", + "best_issue_reason": "45872 is the best issue candidate because it is the clearest, most concrete example of the underlying request and has the strongest duplicate match.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Different topics: a financial dataset bundle promotion versus a DataForge time-series example. Similar keywords are not enough to treat them as the same issue." + "reason": "Different subjects: a financial dataset listing versus an example about time series prediction. Similar finance wording is not enough to treat them as the same issue." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Different topics: dataset listing/promo versus a DataForge example. No evidence they describe the same underlying report." + "reason": "Different subjects: a financial dataset listing versus an example about time series prediction. Not the same underlying bug or request." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Exact same title and matching body size strongly indicate the same issue/example duplicated." + "reason": "Exact duplicate title and essentially identical content; these are the same issue." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Both mention financial/DataForge, but one is an example and the other adds a scraping service; these read like different requests, not one concrete issue." + "reason": "Related finance/DataForge theme, but the request types differ: one is a prediction example, the other adds a scraping service. Too different to merge as duplicates." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Same as 45870 vs 45872: related wording, but not the same underlying change or bug." + "reason": "Same as 45870\u219445872: related domain, but not the same concrete issue/request." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Different request scope: example/prediction versus time series + scraping service." + "reason": "Related theme only; the concrete asks differ, so they should not be merged as duplicates." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Different request scope: example/prediction versus time series + scraping service." + "reason": "Related theme only; the concrete asks differ, so they should not be merged as duplicates." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Exact same title and matching body size strongly indicate duplicate issues for the same request." + "reason": "Exact duplicate title and essentially identical content; these are the same issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The verdicts are conservative and supported by the packet: the two exact duplicate pairs are justified by identical titles and matching body lengths, while the non-duplicate calls rely on clearly different titles/topics and do not overreach. The cluster-level summary is also grounded in the evidence that these split into two separate topic groups rather than one unified duplicate set." + "feedback": "The conclusions are grounded in the packet: 45870/45871 and 45872/45873 have identical titles and matching body lengths, while 45869 is clearly a different topic. The non-duplicate soft-edge verdicts are conservative given only title-level evidence." }, "evaluator_used": true, "retried": false, @@ -16611,12 +16456,12 @@ "pull_request:43823|pull_request:45899", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "ca849e6a6ad209d54d73f4e11fb3873df7d8ac6b3f940ff0d53d943a8d833990", + "prepared_review_unit_hash": "7127ece962ac4784c8f66b9c46c4b4b547789748fd32e492b10660522cd939bf", "component_budget": { "node_count": 14, "item_count": 14, "soft_pair_count": 26, - "serialized_chars": 39790, + "serialized_chars": 39789, "estimated_input_tokens": 9948, "estimated_eval_tokens": 20152 }, @@ -16624,7 +16469,7 @@ "node_count": 14, "item_count": 14, "soft_pair_count": 26, - "serialized_chars": 39790, + "serialized_chars": 39789, "estimated_input_tokens": 9948, "estimated_eval_tokens": 20152 }, @@ -16633,174 +16478,174 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: it contains unrelated PRs for new models, docs, refactors, and small bug fixes. The only clear duplicate pair is the two `fixed_cross_entropy` PRs; everything else should stay separate.", - "confidence": 0.97, + "summary": "Most items here are unrelated PRs that only share broad Transformers themes (new models, docs, refactors). The only true duplicate-style match is the pair of `fixed_cross_entropy` fixes, which target the same helper and same bug in the loss path.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43251 is the best canonical representative because it is the only true near-duplicate pair with 43254, it explicitly targets issue 43240, and its diff directly fixes `fixed_cross_entropy` by forwarding supported kwargs to `nn.functional.cross_entropy`.", + "canonical_pr_reason": "PR 43251 is the best canonical representative: it explicitly links issue 43240 and implements the fuller `cross_entropy` kwargs pass-through in the shared loss helper; 43254 is just a narrower variant of the same change.", "best_issue_reason": null, - "best_pr_reason": "PR 43251 is the strongest overall candidate in this cluster: it has a concrete bug target, the smallest and most focused code change, and a direct duplicate counterpart (43254).", + "best_pr_reason": "PR 43251 is the clearest triage anchor for the only real duplicate pair, with the explicit issue reference and the more complete fix in the exact same code path.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "Unrelated PRs: VideoPrism model addition vs an ExecuTorch dynamic-shapes test. Same broad area of transformers, but not the same change or bug." + "reason": "Videoprism model addition vs executorch dynamic-shape test; unrelated code paths." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Mistral tokenizer method additions vs adding MobileLLM. Different model families and different change scopes." + "reason": "Tokenizer API additions vs adding a new MobileLLM model; not the same change." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Tokenizer API additions and an ExecuTorch export test are unrelated." + "reason": "Tokenizer methods vs executorch export test; different subsystems and goals." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Processor auto-registration/robustness changes for audioflamingo3, lasr, phi4 multimodal, etc. are not the same change as adding MobileLLM." + "reason": "Processor auto-loading robustness vs new model integration; only broad transformer overlap." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "`fixed_cross_entropy` kwargs handling is unrelated to introducing a new MobileLLM model." + "reason": "Loss-function kwargs fix vs model addition; unrelated." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Loss-function kwargs support is unrelated to Mistral4 test fixes and model refactoring." + "reason": "Cross-entropy kwargs fix vs Mistral4 test fix; not the same underlying bug." }, { "left": "pull_request:43424", "right": "pull_request:45899", "accept": false, - "reason": "ExecuTorch dynamic-shape export tests and continuous batching docs are different concerns." + "reason": "Executorch export test vs docs for continuous batching decode path; unrelated." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "ExecuTorch export tests do not match Mistral4 test/model changes." + "reason": "Executorch dynamic-shape test vs Mistral4 test fix; different changes." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Dynamic-shape export tests are unrelated to adding a new model." + "reason": "Executorch test vs model addition; no shared concrete bug." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "MobileLLM model addition and a `cls.config` resolution fix are unrelated." + "reason": "Adding MobileLLM vs fixing `cls.config` resolution order; unrelated." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "Adding MobileLLM is unrelated to Falcon output-interface refactoring." + "reason": "MobileLLM model addition vs Falcon output-collection refactor; different code paths." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon output-capture refactor and Mistral4 test fixes are different code paths." + "reason": "Falcon refactor vs Mistral4 test fixes; not mergeable as one fix." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon model refactor is unrelated to the model-creation skill PR." + "reason": "Falcon model refactor vs model-creation skill PR; unrelated." }, { "left": "pull_request:43995", "right": "pull_request:45899", "accept": false, - "reason": "Falcon output-interface refactor is unrelated to continuous batching documentation." + "reason": "Falcon refactor vs docs-only decode fast path update; unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45899", "accept": false, - "reason": "Mistral4 tests/model fixes and continuous batching docs do not address the same bug or change." + "reason": "Mistral4 test fixes vs continuous batching docs; unrelated despite both touching Transformers." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fixes are unrelated to the `layrnorm` -> `layernorm` renaming cleanup." + "reason": "Mistral4 tests vs `layrnorm` renaming across CLIP-like models; different bugs." }, { "left": "pull_request:44827", "right": "pull_request:45795", "accept": false, - "reason": "Mistral4 fixes are unrelated to docs for audio/video processors." + "reason": "Mistral4 test fix vs audio/video processor docs; unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4 test/model changes are unrelated to config-resolution ordering in `modeling_utils`." + "reason": "Mistral4 test fix vs config-resolution order fix; unrelated." }, { "left": "pull_request:45170", "right": "pull_request:45899", "accept": false, - "reason": "Weight-renaming cleanup for CLIP-like models is unrelated to continuous batching docs." + "reason": "Layernorm typo cleanup vs docs for decode fast path; unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "Mistral4 fixes and the model-creation skill PR are not the same underlying change." + "reason": "Mistral4 test fix vs model-creation skill PR; unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "Adding MobileLLM is unrelated to the model-creation skill scaffolding PR." + "reason": "MobileLLM addition vs model-creation skill PR; not the same change." }, { "left": "pull_request:45213", "right": "pull_request:45899", "accept": false, - "reason": "Model-creation skill tooling is unrelated to continuous batching documentation." + "reason": "Model-creation skill tooling/docs vs continuous batching docs; unrelated." }, { "left": "pull_request:45795", "right": "pull_request:45899", "accept": false, - "reason": "Docs for audio/video processors and docs for decode fast path are distinct documentation additions, not the same change." + "reason": "Docs for audio/video processors vs docs for decode fast path; different documentation topics." }, { "left": "pull_request:43823", "right": "pull_request:45795", "accept": false, - "reason": "MobileLLM model addition and processor docs are unrelated." + "reason": "MobileLLM model addition vs audio/video processor docs; unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45899", "accept": false, - "reason": "MobileLLM model addition is unrelated to continuous batching docs." + "reason": "MobileLLM addition vs continuous batching docs; unrelated." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "These are the same underlying bug fix: both modify `fixed_cross_entropy` in `loss_utils.py` to pass supported kwargs through to `nn.functional.cross_entropy`, and both target issue 43240." + "reason": "Both patches change `fixed_cross_entropy` to pass supported kwargs to `nn.functional.cross_entropy`; same underlying bug and same code path." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the only accepted pair (43251/43254) is a genuine same-bugfix duplicate that looks mergeable into one PR. The other pair judgments are conservative and do not overstate the evidence." + "feedback": "The summary is grounded in the packet, and the only accepted soft pair is the one with shared issue target and the same `fixed_cross_entropy` helper. The other verdicts are conservative and do not overclaim mergeability for clearly unrelated PRs." }, "evaluator_used": true, "retried": false, @@ -16827,7 +16672,7 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "0eff68878a082cbda4113a5c98f02e590eddf1e4676dfea06482468f2806b263", + "prepared_review_unit_hash": "6f8915a46745bd9085d5fd2ceaea267c1b3974c9f5ecfa0e495674df4b3db01d", "component_budget": { "node_count": 2, "item_count": 2, @@ -16849,18 +16694,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related VibeVoice feature PRs, but they are not duplicates: one implements the base VibeVoice model, while the other adds VibeVoice Realtime plus an acoustic tokenizer and generation changes. The overlap is in shared model scaffolding, not the same concrete change.", - "confidence": 0.88, + "summary": "Both PRs relate to VibeVoice, but they target different scopes: #40546 is the core VibeVoice model integration, while #42781 is a draft for VibeVoice Realtime with additional acoustic-tokenizer and generation work. Shared files reflect the same model family, not the same concrete change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #40546 is the better canonical PR because it appears to be the original, broader VibeVoice implementation with the core model, docs, and auto-class wiring, whereas #42781 is a later draft focused on realtime/tokenizer extensions.", + "canonical_pr_reason": "#40546 is the broader, original VibeVoice integration and is the better anchor for the cluster; it has the fuller model hookup and more review activity.", "best_issue_reason": null, - "best_pr_reason": "PR #40546 is the strongest single representative of the cluster: it covers the core VibeVoice integration and is more complete than the draft realtime extension in #42781.", + "best_pr_reason": "#40546 is the best representative because it implements the main VibeVoice support, whereas #42781 adds a separate realtime extension and related tokenizer components.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "They share VibeVoice infrastructure files, but the change sets target different features: #40546 implements VibeVoice itself, while #42781 adds VibeVoice Realtime and an acoustic tokenizer. That is related work, not the same underlying PR change." + "reason": "Reject: these are related VibeVoice PRs but not the same change. #40546 adds the base model integration; #42781 extends VibeVoice for realtime usage with acoustic tokenizer and extra generation/model files." } ] }, @@ -16890,7 +16735,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "d25530b4c76b83c5d0295cb12ee77cd7a30ea1d8414395bb4e53eb4143cfda46", + "prepared_review_unit_hash": "a514a5a070afe4faef212a4a71147d0e4dc902b0a38b0ffb0859931f558284cf", "component_budget": { "node_count": 2, "item_count": 2, @@ -16912,18 +16757,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs touch different models and different features: TimesFM covariate forecasting vs. DINOv3 image classification. They are not duplicates and should not be merged.", - "confidence": 0.98, + "summary": "The two pull requests 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.", + "confidence": 0.99, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "41224 is the more focused PR and a cleaner representative of the cluster, but it is still unrelated to 40755 and should not be treated as a duplicate/merge candidate.", + "best_pr_reason": "No single best PR for this cluster: the PRs address different features in different model families, so neither can serve as a canonical duplicate representative.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different models, different code paths, and different feature goals (TimesFM covariates vs. DINOv3 image classification). No concrete overlap that would justify merging them into one PR." + "reason": "Different underlying changes: TimesFM forecasting with covariates vs DINOv3 image-classification support. No shared bug or mergeable common fix." } ] }, @@ -17018,46 +16863,46 @@ "pull_request:41797|pull_request:45186", "pull_request:45186|pull_request:45630" ], - "prepared_review_unit_hash": "dfcf5285f30eaaa3ff9ad23f3c5aeb4aef08aa4343d19f344bdf9b740279590d", + "prepared_review_unit_hash": "5dbac1d316db05553a244f2278e144606cc21daf6adf0efac42ae62fbb5c851a", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10452, - "estimated_input_tokens": 2613, - "estimated_eval_tokens": 5482 + "serialized_chars": 10409, + "estimated_input_tokens": 2603, + "estimated_eval_tokens": 5462 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10452, - "estimated_input_tokens": 2613, - "estimated_eval_tokens": 5482 + "serialized_chars": 10409, + "estimated_input_tokens": 2603, + "estimated_eval_tokens": 5462 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate new-model addition PRs (DeepSeek OCR, Isaac, and Kimi2-6). They overlap heavily in boilerplate registration/docs, but they do not fix the same underlying bug or implement the same model change.", + "summary": "These are three separate model-addition pull requests (DeepSeek OCR, Isaac, and Kimi2-6). They share Transformers model-registration plumbing, but they target different models and different implementations, so they are not duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 45630 is the strongest representative of the cluster because it has the highest discussion/review activity and is a full model-addition change touching the expected registration, docs, tests, and model code.", "best_issue_reason": null, - "best_pr_reason": "PR 45630 is the most substantial and heavily reviewed of the three, so it is the best representative if one must be chosen, but it is not a duplicate of the others.", + "best_pr_reason": "PR 45630 is the best single representative for triage purposes, but only as an example of the broader model-addition pattern; it is not the same underlying change as the others.", "soft_edge_verdicts": [ { "left": "pull_request:41797", "right": "pull_request:45186", "accept": false, - "reason": "Different model integrations: DeepSeek OCR vs Isaac. Shared auto-registration and docs boilerplate is expected for new-model PRs and does not indicate the same change." + "reason": "Different models and code paths: DeepSeek OCR vs Isaac. The overlap is only shared Transformers integration files and docs, which is normal for separate model additions." }, { "left": "pull_request:45186", "right": "pull_request:45630", "accept": false, - "reason": "Different model integrations: Isaac vs Kimi2-6. They touch similar framework files, but the concrete model codepaths and target behavior are unrelated." + "reason": "Different model families: Isaac vs Kimi2-6. They share registry/auto-mapping updates, but they are distinct additions with different implementations and should not be merged as one PR." } ] }, @@ -17087,7 +16932,7 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "4b75d601b8bac44d5761c766d120fe76fe7a882cd1a41adf9ef66b444f149dd5", + "prepared_review_unit_hash": "a2e880e2466dbce3e7b644939dc578c92dff6eca9823626cdf7c58a71ffb6c0a", "component_budget": { "node_count": 2, "item_count": 2, @@ -17109,18 +16954,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related at a very broad GitHub Actions maintenance level, but they fix different actions in different workflows and do not look mergeable as one PR.", - "confidence": 0.96, + "summary": "Two open PRs both upgrade GitHub Actions, but they target different action families and maintenance goals: one is a Node 24 compatibility bump for workflow helper actions, the other updates Docker build actions. They are related in theme only, not the same underlying change.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the broader and more representative maintenance change: it upgrades multiple core workflow actions for Node 24 compatibility across many workflows, which better fits the cluster theme than the narrower docker-action bump in 42976.", + "canonical_pr_reason": "PR 42975 is the broader, more central workflow-actions upgrade focused on Node 24 compatibility across many files, so it is the better canonical representative if this cluster were treated as one change.", "best_issue_reason": null, - "best_pr_reason": "42975 is the best standalone PR for the cluster because it addresses the more central, cross-cutting workflow compatibility update rather than a single-action version bump limited to docker builds.", + "best_pr_reason": "PR 42975 is the better representative because it covers the wider GitHub Actions runtime-compatibility update, while PR 42976 is a separate Docker-action version bump.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Reject: both are workflow dependency upgrades, but 42975 updates checkout/cache/upload-artifact for Node 24 compatibility while 42976 only bumps docker/build-push-action in build workflows. Different code paths and different change intent." + "reason": "Different underlying changes: 42975 updates checkout/cache/upload-artifact and related workflow actions for Node 24 compatibility, while 42976 upgrades docker/build-push-action. They touch different action code paths and would not naturally be merged as one PR." } ] }, @@ -17163,12 +17008,12 @@ "pull_request:44601|pull_request:45838", "pull_request:45490|pull_request:45838" ], - "prepared_review_unit_hash": "3d16c60273db22d73391350255fa970610479cfa97b4c0b8811eb5f8b7618d5d", + "prepared_review_unit_hash": "affbe9c991d165666bd7eec81676c5605f9fba6c402b2d0c649a2b0647891f32", "component_budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 9, - "serialized_chars": 23542, + "serialized_chars": 23541, "estimated_input_tokens": 5886, "estimated_eval_tokens": 12028 }, @@ -17176,7 +17021,7 @@ "node_count": 7, "item_count": 7, "soft_pair_count": 9, - "serialized_chars": 23542, + "serialized_chars": 23541, "estimated_input_tokens": 5886, "estimated_eval_tokens": 12028 }, @@ -17185,66 +17030,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are separate PRs for different model additions and one distributed-loading feature; none appear to be duplicates or the same underlying change. The shared files are mostly registry/docs boilerplate, not evidence of a common code path.", - "confidence": 0.99, + "summary": "This cluster is heterogeneous: several unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM, LFM2 Audio, PP-OCRv6/PP-LCNetV4) plus one distributed pipeline-parallel infrastructure PR. The overlap is mostly boilerplate docs/auto-registration, not the same underlying change or bug, so none of the soft edges should be merged.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical PR: the items are not duplicates of one another; they span different model families and even a separate distributed-loading feature. Shared files are generic documentation/auto-mapping boilerplate.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "No single PR is a good global representative for dedupe triage, because there is no common concrete code-path change across the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different models: ViT NEPA vs Molmo. Overlap is only in docs/auto-registration scaffolding, not the same bug or feature." + "reason": "Different model additions (ViT NEPA vs Molmo); shared docs/auto-registration files are just boilerplate." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different models and implementations: ViT NEPA vs Molmo2. Shared registry files do not indicate one underlying change." + "reason": "Different model families (ViT NEPA vs Molmo2); not the same underlying feature or bug." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "ViT NEPA model addition is unrelated to CTSM model support; they touch different model families and code paths." + "reason": "Different model domains (ViT NEPA vs CTSM); only generic model-registration/docs overlap." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are separate model families/versions with distinct configs, processors, and modeling code; not mergeable as one fix." + "reason": "Molmo and Molmo2 are distinct model additions, not one concrete change; they could not be merged as a single fix." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo model support and CTSM model support are unrelated additions despite both updating the model registry." + "reason": "Molmo vs CTSM are unrelated model integrations; shared auto-mapping/docs are not enough." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 and CTSM are different models with different implementations and tests; no shared concrete bug or change." + "reason": "Molmo2 and CTSM address different model families and code paths; not duplicates." }, { "left": "pull_request:43973", "right": "pull_request:45838", "accept": false, - "reason": "Lfm2 audio support and PP-OCRv6 text-recognition support are different model additions with unrelated code paths." + "reason": "LFM2 audio support is an audio model add, while PP-OCRv6/PP-LCNetV4 is OCR/text recognition; different changes." }, { "left": "pull_request:44601", "right": "pull_request:45838", "accept": false, - "reason": "Pipeline-parallel loading infrastructure is a core distributed systems change, while PP-OCRv6 is a model-specific addition; they cannot plausibly be one PR." + "reason": "Pipeline-parallel loading infrastructure is unrelated to OCR model support; not the same code-path problem." }, { "left": "pull_request:45490", "right": "pull_request:45838", "accept": false, - "reason": "CTSM model support and PP-OCRv6 text-recognition support are distinct model integrations, not the same underlying change." + "reason": "CTSM and PP-OCRv6/PP-LCNetV4 are different model additions with only generic docs/registration overlap." } ] }, @@ -17277,52 +17122,52 @@ "pull_request:43291|pull_request:44815", "pull_request:43665|pull_request:44815" ], - "prepared_review_unit_hash": "c652ce3e32fdff6a7fb0fecfc8a501b1fb0a12c5fa975f78ddceca4deade25bd", + "prepared_review_unit_hash": "54f3a77fde9e5098062f333a58ac8ff13904abb19e12a183c70aa1e65a6c6632", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7091, - "estimated_input_tokens": 1773, - "estimated_eval_tokens": 3802 + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7091, - "estimated_input_tokens": 1773, - "estimated_eval_tokens": 3802 + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These pull requests are not duplicates: one is a Whisper tokenizer/test fix, one is a test-only change with no real bugfix signal, and one is an FP8 dequantization/loading fix across multiple files. No soft edges should be accepted.", - "confidence": 0.98, + "summary": "These PRs do not look like the same underlying change. #43291 is a Whisper tokenizer/test fix, #43665 is an unrelated test-only edit set across CLIP/ViT, and #44815 is a substantive dequant/FP8/model-loading fix. No duplicate pairs should be merged.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:44815 is the most substantive and coherent fix in the set, with a concrete code-path change in model loading and FP8 integration plus associated tests.", + "canonical_pr_reason": "#44815 is the best representative PR in this cluster because it contains the clearest concrete code-path fix (core model loading plus FP8 quantization logic) and is not just a test-only or obviously unrelated change.", "best_issue_reason": null, - "best_pr_reason": "pull_request:44815 best represents the cluster because it contains the clearest real bugfix work and broadest code impact; the other PRs are either unrelated or test noise.", + "best_pr_reason": "#44815 is the strongest standalone PR: it addresses a specific implementation bug with meaningful code changes and supporting tests, whereas #43291 is narrower and #43665 does not resemble a real fix for the same problem.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Whisper tokenizer decoding fixes are unrelated to the CLIP/ViT test edits; this looks like a test-only or noise PR, not the same underlying change." + "reason": "Whisper tokenizer/test changes are unrelated to CLIP/ViT test edits; same broad \"fix tests\" framing is not enough to imply the same bug or mergeable PR." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper decode logic and FP8/model-loading dequantization are different subsystems and different bugs; they would not plausibly merge into one PR." + "reason": "These touch different subsystems and problems: Whisper decoding vs FP8/core loading/dequant behavior. No concrete shared code-path issue." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "The CLIP/ViT test edits are unrelated to the FP8 dequantization and model-loading fix; shared 'tests' presence is too broad to indicate the same change." + "reason": "A test-only CLIP/ViT patch is not the same underlying bug as FP8 dequant/model-loading fixes, and the overlap is only superficial title similarity." } ] }, @@ -17352,7 +17197,7 @@ "soft_pairs": [ "pull_request:43488|pull_request:43532" ], - "prepared_review_unit_hash": "402598fcdc1d1789551a4ca6144c799b1fd8650dca42c440348208562e6bf2bf", + "prepared_review_unit_hash": "daefc180cb6afa6032a04c8957d63d045dd936dfb0765775ebea0c1c459d8978", "component_budget": { "node_count": 2, "item_count": 2, @@ -17374,18 +17219,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are both non-mergeable test/instrumentation PRs, but they address different things: #43488 is a small repo-bot/setup/version-format check, while #43532 changes CI/workflow and testing utilities. The shared Vit test file looks incidental, not a shared underlying bug or change.", - "confidence": 0.96, + "summary": "The two PRs are unrelated: one is a setup/version/dependency tweak with a vit test file, while the other is a CI/workflow and diff-display change with broader tooling edits. The shared test filename looks incidental, not a duplicate signal.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "#43532 is the better representative PR because it contains the broader substantive workflow/testing changes; #43488 is a small formatting/check PR and not the same change.", + "canonical_pr_reason": "No clear canonical PR: PR #43488 is about repo setup/version/dependency changes, while PR #43532 is about workflow/diff-display behavior and related tooling. They do not represent the same change.", "best_issue_reason": null, - "best_pr_reason": "#43532 is the strongest standalone PR in the cluster: it has the most concrete code-path impact and the richest change set, whereas #43488 is a narrow bot/check exercise.", + "best_pr_reason": "Neither PR is a good global best candidate for deduplication because they target different code paths and only overlap on an unrelated test filename.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Different underlying changes: #43488 is about repo-bot/version formatting and a small setup tweak, while #43532 rewires CI/workflow behavior and related testing utilities. Shared test filename is not enough to treat them as the same fix." + "reason": "Reject: the overlap is superficial. #43488 changes setup/version/deps and a vit test, while #43532 changes workflow/comment handling and diff display. They are not the same concrete fix and would not plausibly merge into one PR." } ] }, @@ -17415,7 +17260,7 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "04edf05ae4e7cb570e06e0cf36fd27b573bdcfc1b9224f1f9bc55a91dae5ef80", + "prepared_review_unit_hash": "09117cbc12278132adbf6cdd50cb5a572422cd11d088f9499725c9ede487bcba", "component_budget": { "node_count": 2, "item_count": 2, @@ -17437,18 +17282,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related add-new-feature PRs with overlapping scaffolding, but they implement different pipelines and different underlying segmentation tasks, so they should not be treated as duplicates.", - "confidence": 0.97, + "summary": "These are two related but distinct PRs adding separate pipelines for different segmentation tasks. They share registration/docs/test plumbing, but the underlying features and code paths are not the same, so they should not be deduplicated.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation; they target different features and code paths.", + "canonical_pr_reason": "No single canonical PR: #43612 adds Promptable Concept Segmentation, while #43613 adds Promptable Visual Segmentation; they target different tasks and model integrations.", "best_issue_reason": null, - "best_pr_reason": "43613 is the more comprehensive standalone PR in terms of model/test coverage, but it is still a distinct feature rather than a duplicate of 43612.", + "best_pr_reason": "#43613 is the broader of the two and touches more models/tests, but it is still a separate feature rather than a duplicate of #43612.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "They share documentation/pipeline scaffolding, but the actual feature is different: concept segmentation vs visual segmentation, with different task docs, pipeline modules, and model integrations." + "reason": "Different pipeline features with different task semantics and model support; overlap is limited to shared docs/registry scaffolding, not the same concrete bug or change." } ] }, @@ -17478,7 +17323,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "e666371523a501b662038b606174de4b9407c3d90e00a895a3716aecb73002ee", + "prepared_review_unit_hash": "609df94f1371bd8448d72f7f653be613c22ca5d9b4a5c17a3509787a40e8d0d8", "component_budget": { "node_count": 2, "item_count": 2, @@ -17500,18 +17345,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs are both in `trainer.py` but address different Trainer extension points: one changes gradient-accumulation loss scaling, the other refactors signature-based column filtering for custom trainers. They do not look like the same underlying bug or change.", + "summary": "The two PRs are both trainer.py changes, but they target different code paths: one adjusts gradient-accumulation loss scaling, the other adds a hook for signature-column filtering. They are not duplicates.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs are unrelated functional changes in Trainer and should remain separate.", + "canonical_pr_reason": "43651 is the more canonical representative of the cluster because it changes core training_step loss-scaling behavior and adds a dedicated control hook for GA scaling.", "best_issue_reason": null, - "best_pr_reason": "No single best PR for this cluster; neither PR subsumes the other because they solve different problems.", + "best_pr_reason": "43651 is the stronger representative: it is a concrete runtime behavior fix, while 43663 is a separate extensibility change for column filtering.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Different code paths and behaviors: 43651 changes loss normalization during training_step, while 43663 adds a hook for signature column filtering. Shared filename alone is not enough to treat them as duplicates." + "reason": "Different underlying changes in trainer.py: 43651 fixes/controls gradient-accumulation loss scaling, while 43663 adds a separate signature-column override mechanism. They do not appear mergeable as one PR." } ] }, @@ -17543,7 +17388,7 @@ "pull_request:43656|pull_request:43842", "pull_request:43836|pull_request:43842" ], - "prepared_review_unit_hash": "f8b0bc83a19762b0b6a91d6950f303dfaaf448d2eadc8251d063d75a3c947c0d", + "prepared_review_unit_hash": "4ecc12039bdc0838076b2005210b677faaf64372bc7369be80dd252b167d3b16", "component_budget": { "node_count": 3, "item_count": 3, @@ -17565,34 +17410,31 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "PR 43842 is still the best representative for the specific TypeAdapter NameError fix in serve.py. PR 43836 remains close enough to treat as a duplicate-change candidate, but PR 43656 looks more like a related/overlapping annotation cleanup in the same file than a clearly same-bug fix.", - "confidence": 0.76, + "summary": "All three PRs target TypeAdapter-related typing/runtime issues in `src/transformers/cli/serve.py`, but only one is the clean, narrowly scoped fix for the specific missing-pydantic NameError. The other two are broader or address a different annotation-evaluation path, so they should not be merged as duplicates.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43842 is the cleanest, most narrowly scoped fix for the TypeAdapter NameError in serve.py, with the most direct failure description and minimal surrounding changes.", + "canonical_pr_reason": "PR #43842 is the best canonical PR: it directly targets the explicit issue about `TypeAdapter` NameError when pydantic is not installed, with the smallest and most focused code change.", "best_issue_reason": null, - "best_pr_reason": "PR 43842 is the best representative because it directly targets the runtime NameError when pydantic is absent, without the broader refactor/annotation cleanup present in the others.", + "best_pr_reason": "PR #43842 is the strongest candidate overall because it is narrowly scoped, directly matches its linked issue, and fixes the concrete runtime failure without bundling unrelated typing/refactor changes.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "Same file and broadly similar typing context, but the packet evidence only supports soft similarity; the diff preview does not clearly show the same concrete TypeAdapter/pydantic fix, so this is better treated as related rather than duplicate-compatible." + "reason": "Both touch `serve.py` and mention `TypeAdapter`, but #43656 is a broader annotation/`from __future__` typing cleanup while #43842 specifically guards `TypeAdapter` when pydantic is absent. Different issue targets and different concrete failure modes." }, { "left": "pull_request:43836", "right": "pull_request:43842", - "accept": true, - "reason": "Both appear to address the same TypeAdapter NameError in transformers/cli/serve.py and are plausibly alternative fixes for the same concrete bug." + "accept": false, + "reason": "They share the same file and issue target, but #43836 is a larger refactor with a generic string-literals workaround, whereas #43842 is a small targeted fix for the missing-pydantic NameError. They are not the same concrete change and would not be a clean single PR." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The 43836\u219443842 pairing is reasonably supported, but the 43656\u219443842 verdict is a bit too strong for the evidence in the packet. For 43656, the support is mainly shared file/title and soft similarity; the diff preview does not clearly show the same TypeAdapter/pydantic fix, so calling it the same concrete bugfix is more confident than the packet justifies. A more conservative wording would be that it is related or possibly overlapping, not definitively duplicate-change-compatible." - }, - "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -17722,13 +17564,13 @@ "pull_request:44018|pull_request:44019", "pull_request:44019|pull_request:44071", "pull_request:44068|pull_request:44116", - "pull_request:44018|pull_request:44116", "pull_request:44068|pull_request:44722", + "pull_request:44018|pull_request:44116", "pull_request:44071|pull_request:44722", "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:44068|pull_request:44086", "pull_request:44059|pull_request:44086", "pull_request:44085|pull_request:44086", "pull_request:44098|pull_request:44101", @@ -17777,20 +17619,20 @@ "pull_request:44044|pull_request:44129", "pull_request:44044|pull_request:44154" ], - "prepared_review_unit_hash": "2d5c6f031df507889f32e98820f369812d7d877ef3f4b4a1d9c40ab77ad5395c", + "prepared_review_unit_hash": "9554032547e5cc50580aaf9f14a041a20da4d10f1355123b86d41af7387fbd8a", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84898, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84895, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 }, "budget": { "node_count": 27, "item_count": 27, "soft_pair_count": 72, - "serialized_chars": 75095, + "serialized_chars": 75093, "estimated_input_tokens": 18774, "estimated_eval_tokens": 37804 }, @@ -17799,366 +17641,438 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of model-specific output-tracing refactors tied to the same broader initiative, not true duplicates. Most soft links connect unrelated models that happen to share the same standardized-decorator theme or issue target. Only the GPT-Neo pair and the GPT-J pair(s) look like the same underlying change family.", - "confidence": 0.96, + "summary": "Mostly a set of separate per-model output-tracing refactors around the same umbrella issue (#43979). The only clear duplicate clusters are the GPT-J/CodeGen PRs and the GPT-Neo PR pair; the rest are same-theme but different code paths and should stay separate.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44066 is the clearest representative: it has an explicit issue target, a focused standardized output-tracing refactor, and concrete code changes in the GPT-J/CodeGen path.", + "canonical_pr_reason": "PR 44722 is the best canonical representative: it explicitly targets #43979 and covers the same concrete GPT-J/CodeGen output-tracing refactor as the earlier GPT-J PRs, making it the most consolidated version of that change.", "best_issue_reason": null, - "best_pr_reason": "PR 44066 is the strongest single representative of the cluster because it directly implements the shared output-tracing migration with explicit issue linkage and a coherent code-path change.", + "best_pr_reason": "PR 44722 is the strongest single representative of the cluster because it addresses the same concrete GPT-J/CodeGen tracing path, has the explicit issue target, and subsumes the earlier overlapping GPT-J-only PRs.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different models and code paths (FNet/CVT vs GPT-J); same umbrella refactor theme is not enough." + "reason": "Different model/code paths; sharing the same umbrella issue target is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different models (FNet/CVT vs DeBERTa); only the tracing refactor pattern is shared." + "reason": "Different model/code paths; both are output-tracing refactors but not the same bug/change." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both refactor GPT-J output tracing in the same file/path; 44066 is broader, but this looks like the same underlying GPT-J change family." + "reason": "Same GPT-J tracing refactor on the same concrete code path; these could plausibly be merged into one PR." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "ResNet vs EfficientNet are separate model implementations; not the same bug/change." + "reason": "Different vision backbones and code paths; same refactor theme only." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet and GPT-J/CodeGen are unrelated model paths." + "reason": "Different model families and code paths; only the tracing pattern overlaps." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen vs EfficientNet are unrelated changes despite similar refactor style." + "reason": "Different models/code paths; not the same underlying fix." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 vs DeBERTa are unrelated models." + "reason": "Different models and concrete code paths." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen vs MGP-STR are different code paths." + "reason": "Different models; same issue umbrella is too broad to treat as a duplicate." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen vs MPT are separate model refactors, not one shared fix." + "reason": "Different model families and code paths." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both target GPT-Neo output tracing in the same model/file; they look like overlapping implementations of the same refactor." + "reason": "Same GPT-Neo output-tracing refactor on the same code path; these look mergeable into one PR." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen and GPT-Neo are different models; shared initiative only." + "reason": "GPT-J/CodeGen versus GPT-Neo; different concrete changes." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different code paths." + "reason": "Different models and code paths." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet and GPT-J/CodeGen are unrelated." + "reason": "ResNet versus GPT-J/CodeGen; same refactor family only." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR vs GPT-J/CodeGen are different model implementations." + "reason": "Different models/code paths; not the same bug/change." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet vs GPT-J are unrelated models." + "reason": "Different models and different concrete fixes." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo vs ResNet are unrelated model refactors." + "reason": "Different model families and code paths." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet vs MPT are separate models and code paths." + "reason": "Different model families and code paths." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo vs Flaubert/XLM are unrelated models." + "reason": "GPT-Neo versus Flaubert; same generic refactor pattern only." }, { - "left": "pull_request:44018", - "right": "pull_request:44116", + "left": "pull_request:44068", + "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo vs Flaubert/XLM are unrelated models." + "reason": "GPT-Neo versus GPT-J/CodeGen; different concrete changes." }, { - "left": "pull_request:44068", - "right": "pull_request:44722", + "left": "pull_request:44018", + "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different models." + "reason": "Different models/code paths." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT vs GPT-J/CodeGen are unrelated model paths." + "reason": "Different model families and code paths." }, { "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo vs MGP-STR are unrelated." - }, - { - "left": "pull_request:44018", - "right": "pull_request:44085", - "accept": false, - "reason": "GPT-Neo vs GPT-J are different model refactors." + "reason": "Different models/code paths." }, { - "left": "pull_request:44068", + "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo vs MGP-STR are unrelated models." + "reason": "Different models/code paths." }, { - "left": "pull_request:44071", - "right": "pull_request:44085", + "left": "pull_request:44098", + "right": "pull_request:44101", "accept": false, - "reason": "MPT vs GPT-J are unrelated." + "reason": "Different model families; only shared output-handling style." }, { "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 vs MGP-STR are different model code paths." + "reason": "GPT-2 versus MGP-STR; not the same underlying change." }, { "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J vs MGP-STR are unrelated." + "reason": "Different models/code paths." }, { "left": "pull_request:44098", - "right": "pull_request:44101", + "right": "pull_request:44161", "accept": false, - "reason": "ViLT vs Flaubert/XLM are unrelated models." + "reason": "Different model families and code paths." }, { "left": "pull_request:44059", - "right": "pull_request:44722", - "accept": false, - "reason": "GPT-2 vs GPT-J/CodeGen are unrelated." - }, - { - "left": "pull_request:44018", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo vs MPT are different model implementations." + "reason": "GPT-2 versus MPT; same pattern only." }, { "left": "pull_request:44017", - "right": "pull_request:44071", + "right": "pull_request:44161", "accept": false, - "reason": "Segformer vs MPT are unrelated models." + "reason": "Segformer versus LongT5; different concrete changes." }, { "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "Segformer vs GPT-J/CodeGen are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "Segformer vs VisualBert are separate model refactors." + "reason": "Different models/code paths." }, { "left": "pull_request:44068", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo vs MPT are unrelated." + "reason": "GPT-Neo versus MPT; not the same fix." }, { "left": "pull_request:44071", "right": "pull_request:44086", "accept": false, - "reason": "MPT vs MGP-STR are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:43997", "right": "pull_request:44073", "accept": false, - "reason": "RegNet vs VisualBert are different models." + "reason": "Different model families and code paths." }, { "left": "pull_request:44017", - "right": "pull_request:44018", - "accept": false, - "reason": "Segformer vs GPT-Neo are unrelated model paths." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44056", + "right": "pull_request:44071", "accept": false, - "reason": "RegNet vs MPNet are separate models." + "reason": "Different model families and code paths." }, { - "left": "pull_request:44071", - "right": "pull_request:44076", + "left": "pull_request:44059", + "right": "pull_request:44071", "accept": false, - "reason": "MPT vs ImageGPT are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet vs Flaubert/XLM are different models." + "reason": "TextNet versus Flaubert/XLM; same refactor theme only." }, { "left": "pull_request:44056", "right": "pull_request:44161", "accept": false, - "reason": "MPNet vs LongT5 are unrelated." + "reason": "MPNet versus LongT5; different concrete changes." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet vs GPT-Neo are unrelated." - }, - { - "left": "pull_request:43997", - "right": "pull_request:44161", - "accept": false, - "reason": "RegNet vs LongT5 are unrelated models." + "reason": "Different model families and code paths." }, { "left": "pull_request:44074", "right": "pull_request:44161", "accept": false, - "reason": "TextNet vs LongT5 are unrelated." + "reason": "Different models/code paths." }, { - "left": "pull_request:44056", - "right": "pull_request:44101", + "left": "pull_request:43997", + "right": "pull_request:44071", "accept": false, - "reason": "MPNet vs Flaubert/XLM are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:44010", "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBert vs MPT are unrelated models." + "reason": "SqueezeBert versus MPT; not the same bug/change." }, { "left": "pull_request:44018", "right": "pull_request:44129", "accept": false, - "reason": "GPT-Neo vs SpeechT5 are separate model code paths." + "reason": "GPT-Neo versus SpeechT5; only the refactor style overlaps." }, { "left": "pull_request:44010", "right": "pull_request:44073", "accept": false, - "reason": "SqueezeBert vs VisualBert are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:43997", "right": "pull_request:44074", "accept": false, - "reason": "RegNet vs TextNet are unrelated model refactors." + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 versus MPT; same pattern only." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "Different model families and code paths." }, { "left": "pull_request:44018", + "right": "pull_request:44059", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44101", + "right": "pull_request:44161", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "GPT-Neo vs DeBERTa are different models." + "reason": "Different model families and code paths." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa vs TextNet are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Both touch GPT-J output tracing and overlap on the same underlying model refactor; 44722 appears to be an extension of the same GPT-J family." + "reason": "Same GPT-J/CodeGen tracing refactor; 44722 is the consolidated version of the same concrete change." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both are GPT-J output-tracing refactors in the same model/file, so they look like the same underlying change family." + "reason": "Same GPT-J output-tracing refactor on the same code path; mergeable as one PR." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J vs SpeechT5 are unrelated models." + "reason": "Different models/code paths." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "FNet/CVT vs ResNet are unrelated model paths." + "reason": "Different model families and code paths." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "FNet/CVT vs SpeechT5 are unrelated models." + "reason": "Different model families and code paths." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa vs SpeechT5 are unrelated." + "reason": "Different model families and code paths." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa vs VITS are different model implementations." + "reason": "Different models/code paths; same refactor pattern only." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The accepted pairs all appear to be within the same model/file family and plausibly mergeable into one bugfix PR, while the rejected pairs mostly involve clearly unrelated model implementations despite sharing the broader output-tracing initiative. The choice of PR 44066 as the canonical representative is also supported by the explicit issue link and focused GPT-J/CodeGen path." + "feedback": "The analyst summary is conservative and grounded in the packet. The accepted soft-edge pairs are limited to closely related PRs on the same model/code path, especially the GPT-J/CodeGen cluster and the GPT-Neo pair. The rejected pairs are consistently justified by different model families or different concrete fixes. I did not see an overstatement beyond the evidence provided." }, "evaluator_used": true, "retried": false, @@ -18196,20 +18110,20 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "7afb0ce801bb80308a2fd704af32687499ee39b339cb255fc6042cba50a4f1a2", + "prepared_review_unit_hash": "f72c58a76fe68b47871b44c5183c08f227bbd49c02bebf9c5806ff2729febb9f", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84898, - "estimated_input_tokens": 21225, - "estimated_eval_tokens": 42706 + "serialized_chars": 84895, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 6, - "serialized_chars": 17124, + "serialized_chars": 17123, "estimated_input_tokens": 4281, "estimated_eval_tokens": 8818 }, @@ -18218,48 +18132,48 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all model-specific refactors to standardized output capture/tracing, but they target different architectures and code paths. They look like the same refactor pattern, not the same underlying change, so they should not be merged as duplicates.", - "confidence": 0.92, + "summary": "These are all model-specific PRs in the same broad standardized-output-tracing theme, but they do not appear to be duplicates of one another. Each changes a different model implementation, so none of the soft edges should be merged.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44129 is the strongest representative of the group because it has the clearest scope and an explicit issue link, but it still addresses SpeechT5 only, not the broader set.", + "canonical_pr_reason": "PR 44129 is the strongest representative of the cluster because it has an explicit issue target and a clear, concrete refactor in a single model path, but it is still specific to SpeechT5 rather than a duplicate of the other PRs.", "best_issue_reason": null, - "best_pr_reason": "PR #44129 is the best single PR to represent the theme since it includes an explicit tracking issue and a substantial output-tracing cleanup in one model.", + "best_pr_reason": "PR 44129 is the best single representative: it is well-scoped, explicitly linked to an issue, and clearly applies the standardized output-capture refactor to one concrete code path.", "soft_edge_verdicts": [ { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "Both are standardized output-tracing refactors, but for different models (FocalNet vs ImageGPT). They do not fix the same concrete code-path problem." + "reason": "Both are output-tracing refactors, but they touch different model code paths (FocalNet vs ImageGPT) and are not the same underlying bug/change." }, { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Both migrate output handling to decorators/capture_outputs, but one is VisualBert and the other is TextNet; separate model implementations, not one shared bug." + "reason": "These are separate refactors for different architectures (VisualBert vs TextNet); same theme, but not a duplicate fix." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are different architectures with different tracing paths. Similar refactor theme is not enough to treat them as duplicates." + "reason": "MPNet and VisualBert are distinct model implementations with different tracing setup; not the same concrete change." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Both touch standardized output collection, but for different model backends (MPNet vs TextNet). Not the same underlying fix." + "reason": "Different model families and different forward/tracing code paths; too broad to treat as one change." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert changes are model-specific output-tracing updates. They are not the same code-path change and would not be merged as one PR." + "reason": "RegNet and SqueezeBert are unrelated model-specific output refactors, not a shared bug fix that could be merged into one PR." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert vs SpeechT5 affect different model internals; they share a refactor style but not a single bug or change request." + "reason": "Both concern standardized output capture, but they modify different models and different implementation details, so they are not duplicates." } ] }, @@ -18339,274 +18253,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "af1658b435caaf28d7ed442d1c7d4cd18b2bc5a3445e4a1e553d4d76df628c71", + "prepared_review_unit_hash": "77a4822b251bc6e9b710f0696c123e332e3f4fe9cefe11f527d2d1349f1119e4", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40156, - "estimated_input_tokens": 10039, - "estimated_eval_tokens": 20334 + "serialized_chars": 40168, + "estimated_input_tokens": 10042, + "estimated_eval_tokens": 20340 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40156, - "estimated_input_tokens": 10039, - "estimated_eval_tokens": 20334 + "serialized_chars": 40168, + "estimated_input_tokens": 10042, + "estimated_eval_tokens": 20340 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are all model-specific output-tracing refactors (`can_return_tuple`, `capture_outputs`, or related cleanup) across different architectures. They share a theme, but they do not describe the same concrete bug/change and should not be merged as duplicates.", - "confidence": 0.95, + "summary": "These are mostly parallel, model-specific output-tracing refactors across unrelated architectures. I don\u2019t see true duplicate PRs here; the similarities are mostly from a shared refactor pattern rather than the same concrete fix. The Mamba/FalconMamba PR is the best representative of the cluster\u2019s actual tracing plumbing change.", + "confidence": 0.81, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44003 is the strongest anchor for the cluster because it makes the broadest tracing-related change set and introduces the new output-capturing plumbing across closely related Mamba variants.", + "canonical_pr_reason": "PR 44003 is the strongest representative: it contains the most substantive tracing-plumbing changes, spans two closely related model implementations, and is broader/more concrete than the other single-model refactors.", "best_issue_reason": null, - "best_pr_reason": "PR #44003 is the best representative of the cluster\u2019s core refactor pattern: it is more substantial than the single-model follow-ups and includes the main tracing infrastructure changes.", + "best_pr_reason": "PR 44003 best captures the underlying theme of the cluster with real code-path changes (can_return_tuple/output recording) rather than just a model-specific mechanical refactor.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model implementations (`timm_backbone` vs `upernet`); same refactor style only, not the same code-path change." + "reason": "Different model implementations and code paths; both are output-tracing refactors, but not the same underlying bug/change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different models (`timm_backbone` vs `vision_text_dual_encoder`); similar output-wrapper cleanup, but not one underlying bug." + "reason": "Unrelated architectures (timm_backbone vs vision_text_dual_encoder); similar refactor theme only." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Different model families (`timm_backbone` vs `univnet`); only a shared refactor theme." + "reason": "Different models and runtime paths; same tracing motif, not a duplicate fix." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Separate architectures and files; both touch output tracing, but not the same concrete fix." + "reason": "These touch different model families; the shared output-tracing pattern is too broad to treat as one fix." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Different models (`timm_backbone` vs `rwkv`); output-tracing similarity is too broad to treat as a duplicate." + "reason": "Separate architectures and implementations; not the same concrete change." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Different models (`mobilenet_v1` vs `univnet`); both refactor outputs, but they do not fix the same path." + "reason": "MobileNetV1 and UnivNet are unrelated code paths; similarity is only the generic tracing refactor." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different architectures (`mobilenet_v1` vs `upernet`); same general refactor, not the same issue." + "reason": "Different model families and different forward paths; not mergeable as one duplicate change." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model targets (`mobilenet_v1` vs `depth_anything`); similar output tracing cleanup only." + "reason": "Separate architectures; both are model-specific output-tracing updates, not one underlying fix." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "Different code paths and model families; not plausibly one merged PR for the same bug." + "reason": "Different wrappers/models; same boilerplate refactor pattern only." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Different model implementations (`mobilenet_v1` vs `speech_encoder_decoder`); only a shared refactor pattern." + "reason": "Different code paths in different encoder-decoder wrappers; not the same concrete bug/change." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Different models (`vision_text_dual_encoder` vs `univnet`); output-tracing changes are unrelated at the bug level." + "reason": "Vision-text dual encoder and UnivNet are unrelated; shared refactor language is not enough." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model files and behavior surfaces; same general cleanup, not one concrete fix." + "reason": "Different architectures and model internals; not a duplicate PR." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "These are separate model-specific refactors; no shared failing code-path is evident." + "reason": "These affect distinct model families and cannot plausibly be one merged fix." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Different model families and implementations; too broad to merge as duplicates." + "reason": "Separate model implementations; same tracing refactor pattern only." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Different architectures (`univnet` vs `upernet`); similar API cleanup only." + "reason": "UnivNet and UperNet are unrelated changes; not the same code-path problem." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "Different model code paths (`univnet` vs `mamba`/`falcon_mamba`); not the same underlying bug." + "reason": "Different model families and forward logic; not a duplicate." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different model implementations; both are tracing refactors, but not one duplicate change." + "reason": "Different architectures and code paths; shared output-tracing theme is too broad." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "Different model families (`upernet` vs `rwkv`); same refactor theme, not the same fix." + "reason": "UperNet and RWKV are unrelated implementations; not one underlying fix." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Separate code paths and architectures; output-tracing similarity alone is insufficient." + "reason": "Different model families and different tracing code paths; not duplicates." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "`mamba`/`falcon_mamba` refactor vs `superpoint`; unrelated implementations." + "reason": "FalconMamba/Mamba vs SuperPoint are unrelated architectures; similar refactor pattern only." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "`mamba`/`falcon_mamba` output tracing vs `codegen` cache/output refactor; not the same concrete bug or change." + "reason": "Mamba/FalconMamba and CodeGen touch different model internals; not the same fix." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Different model families and output semantics; the changes are not interchangeable." + "reason": "Timm backbone and CodeGen are unrelated model code paths; not a duplicate." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "`univnet` and `codegen` touch unrelated forward paths; too broad to merge." + "reason": "UnivNet and CodeGen are different architectures; same refactor family only." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "Different architectures and output shapes; no shared underlying defect." + "reason": "UperNet and CodeGen are separate implementations; no shared concrete bug." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Different model code paths; same refactor language does not make them duplicates." + "reason": "Vision-text dual encoder and CodeGen are unrelated changes; not mergeable as one fix." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Different model families (`depth_anything` vs `vision_encoder_decoder`); related tooling, not the same fix." + "reason": "DepthAnything and VisionEncoderDecoder are different model families; only the tracing refactor pattern matches." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Different models and forward paths; same output-tracing cleanup only." + "reason": "Different code paths in different model families; not the same underlying issue." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "`depth_anything` vs `superpoint`; not the same concrete bug or code change." + "reason": "DepthAnything and SuperPoint are unrelated models; shared output-tracing refactor is too generic." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Different model implementations and tracing mechanisms; not mergeable as one PR." + "reason": "Different model families and implementation details; not a duplicate." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Different models (`depth_anything` vs `dpr`); similar refactor theme only." + "reason": "DepthAnything and DPR touch different model code paths; not one concrete fix." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "Different model families (`vision_encoder_decoder` vs `rwkv`); no shared failing path." + "reason": "VisionEncoderDecoder and RWKV are unrelated implementations; no duplicate code-path problem." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "Different encoder-decoder stacks and outputs; same pattern, not same change." + "reason": "These are two different encoder-decoder wrapper models; similar refactor style, but not the same underlying change." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Unrelated model implementations; output tracing refactors are independent." + "reason": "VisionEncoderDecoder and SuperPoint are unrelated; the similarity is only the tracing refactor pattern." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "Different model code paths (`vision_encoder_decoder` vs `dpr`); not duplicates." + "reason": "Different model families and different runtime paths; not duplicates." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Different model families (`speech_encoder_decoder` vs `superpoint`); no single concrete bug." + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated code paths; same broad theme only." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Different implementations and output surfaces; only the refactor motif matches." + "reason": "SpeechEncoderDecoder and RWKV are different architectures; not the same fix." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "`speech_encoder_decoder` and `dpr` are unrelated code paths; not one change." + "reason": "Different model implementations and code paths; not a duplicate." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "Different architectures (`superpoint` vs `rwkv`); not the same underlying issue." + "reason": "SuperPoint and RWKV are unrelated changes; shared refactor pattern is not enough." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations and outputs; too broad to merge." + "reason": "SuperPoint and DPR are different model families; not the same underlying bug/change." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "Different model families (`rwkv` vs `dpr`); same tracing theme only." + "reason": "RWKV and DPR are unrelated architectures; no shared concrete fix." } ] }, @@ -18639,12 +18553,12 @@ "pull_request:44662|pull_request:45075", "pull_request:45075|pull_request:45401" ], - "prepared_review_unit_hash": "8877a6f130026fc9ba82164629d1de54f22f4d6382c2bd20668eb6f1229bde70", + "prepared_review_unit_hash": "eecb7edf0ff49fdef2945977cfdca42d16bf7b37972e9c9ee9be591572f4dcf2", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 10914, + "serialized_chars": 10913, "estimated_input_tokens": 2729, "estimated_eval_tokens": 5714 }, @@ -18652,7 +18566,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 10914, + "serialized_chars": 10913, "estimated_input_tokens": 2729, "estimated_eval_tokens": 5714 }, @@ -18661,30 +18575,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three separate model-addition PRs with overlapping registration/docs scaffolding, but they target different architectures and use cases: PenguinVL, DeepSeek-OCR-2, and Voxtral TTS. No duplicate pair appears to fix the same concrete code-path problem.", - "confidence": 0.97, + "summary": "All three pull requests are separate model-addition efforts (PenguinVL, DeepSeek-OCR-2, and Voxtral TTS). They share Transformers registration/docs scaffolding, but they do not appear to fix the same bug or implement the same concrete change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45075 is the strongest representative if one must be chosen: it is a full model integration with explicit implementation, conversion tooling, tests, and an associated config update, plus substantial review activity. That said, it is not a duplicate of the other PRs.", + "canonical_pr_reason": "PR #45075 is the strongest representative: it is a complete, non-draft model addition with substantial implementation, conversion, processing, and test coverage, and it has the highest review activity.", "best_issue_reason": null, - "best_pr_reason": "PR #45075 is the best standalone PR in this cluster because it looks the most complete and self-contained, with broad integration work and explicit supporting tooling/tests. The other PRs are distinct model ports rather than alternative fixes for the same change.", + "best_pr_reason": "PR #45075 is the best standalone PR in the cluster because it looks the most mature and complete, with broader model infrastructure and test coverage than the others.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different models and different problem domains: PenguinVL is a vision-language model implementation, while Voxtral TTS is a text-to-speech model. Shared auto-registration and docs files are generic model-integration boilerplate, not evidence of the same underlying change." + "reason": "Different models and code paths: PenguinVL vs Voxtral TTS. Shared auto-registration and docs files are generic model-addition scaffolding, not evidence of the same underlying change." }, { "left": "pull_request:44662", "right": "pull_request:45075", "accept": false, - "reason": "Both are new model integrations, but they implement different models (PenguinVL vs DeepSeek-OCR-2) with different architectures, files, and expected behavior. Overlap is limited to standard HF registration/docs paths." + "reason": "Both are new model integrations, but they target different architectures and implementations (PenguinVL vs DeepSeek-OCR-2). Similar repo plumbing is not enough to treat them as duplicates." }, { "left": "pull_request:45075", "right": "pull_request:45401", "accept": false, - "reason": "DeepSeek-OCR-2 and Voxtral TTS are unrelated multimodal/speech model ports. They share only generic model-registration infrastructure, so they should not be merged as duplicates." + "reason": "Distinct model families and functionality: DeepSeek-OCR-2 versus Voxtral TTS. They share only generic Transformers boilerplate, so they should not be merged as the same PR." } ] }, @@ -18714,12 +18628,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "69714aa44840f8b7da91b5a0242a19e032683ff95b03a9b681855b10c065c141", + "prepared_review_unit_hash": "11d37c6c037fa8f13f0a51f4e23d07f66ffbd972c13c30af7c74ea7ab0cc2cd1", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4408, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -18727,7 +18641,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4408, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -18736,18 +18650,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related but not duplicates: one adds a generic Module Fusion API, the other implements n-to-1 kernel fusion via KernelConfig on top of the broader fusion machinery. They target different layers of the stack and would not naturally collapse into a single PR.", - "confidence": 0.88, + "summary": "Related fusion-feature PRs, but they target different layers: #44979 adds a generic Module Fusion API, while #45363 adds KernelConfig-driven n-to-1 kernel fusion integration and an example. They are not the same concrete change.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45363 is the more concrete, user-facing feature implementation: it wires fusion into `modeling_utils`/`integrations`, adds `KernelConfig`, and includes a runnable example. PR #44979 is the foundational API layer, but it is broader and more infrastructural.", + "canonical_pr_reason": "#44979 is the more canonical representative of the cluster because it introduces the core reusable fusion API, whereas #45363 is a narrower KernelConfig-based integration on top of that theme.", "best_issue_reason": null, - "best_pr_reason": "PR #45363 is the best representative of the cluster because it changes the actual kernel-fusion code path and surfaces the feature through `KernelConfig`; the other PR is a generic API scaffold rather than the same change.", + "best_pr_reason": "#44979 best represents the cluster\u2019s main subject since it defines the foundational module-fusion mechanism rather than a specific application of fusion.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "They are connected conceptually, but not the same underlying change. #44979 introduces a generic module-fusion API, while #45363 implements a specific n-to-1 kernel-fusion feature using `KernelConfig` and related integration points. They would not be merged as one duplicate PR." + "reason": "They are conceptually related, but not the same bug or change: one adds a general module-fusion API, the other implements n-to-1 kernel fusion via KernelConfig with different integration points and files." } ] }, @@ -18777,7 +18691,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "638e81c75ce2775c99bd159bd3aa937f2db9f9ca4b9f68eb83c2e6f00beda45f", + "prepared_review_unit_hash": "fd0bb7c38737398e30df6dc5011667ee7f088c4b19d195c39cd7bf609eed4ec6", "component_budget": { "node_count": 2, "item_count": 2, @@ -18799,18 +18713,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests target the same `_patch_mistral_regex` AttributeError in `src/transformers/tokenization_utils_tokenizers.py`; the second PR is essentially the same fix with added test coverage, so they are duplicates as PRs.", - "confidence": 0.97, + "summary": "Both pull requests implement the same Mistral regex fix in the same code path; the second PR appears to be a more complete version with tests, so they should be treated as duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45317 is the stronger canonical PR: it fixes the same code path as #45086 and also adds a test file, making it the more complete change set.", + "canonical_pr_reason": "PR 45317 is the stronger canonical PR because it contains the same fix as 45086 plus test coverage, making it the more complete candidate for the underlying bug.", "best_issue_reason": null, - "best_pr_reason": "PR #45317 is the best overall representative because it contains the same fix as #45086 plus tests and has more review activity, so it is the better standalone merge candidate.", + "best_pr_reason": "PR 45317 is the best overall PR candidate: it fixes the same AttributeError in `_patch_mistral_regex` and adds tests, so it better represents the intended change.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying bug and same code change in `_patch_mistral_regex`; #45317 appears to be an expanded version of #45086 with test coverage, so they can plausibly be merged into one PR." + "reason": "Same underlying AttributeError fix in the same file and code path, with identical line changes; 45317 is effectively a superset of 45086 with added tests." } ] }, @@ -18840,7 +18754,7 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "7789434d583ae1fd3e7520d5424e88b661ee26cc7802b3e0f82194a8029ccf75", + "prepared_review_unit_hash": "2db7f01b30d8df753ea503456d57b83df6d8e5fdfb705e761048f07cfb8e188a", "component_budget": { "node_count": 2, "item_count": 2, @@ -18862,18 +18776,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related NemotronH PRs, but they address different scopes: one is a narrower conversion/refactor PR, the other is a broader model split/implementation/docs PR. They should not be merged as duplicates.", - "confidence": 0.89, + "summary": "The two PRs overlap on NemotronH internals, but they are not the same change: #45115 is a narrow inheritance/refactor and conversion-mapping update, while #45569 is a broader model-split PR adding dense/sparse variants, docs, and tests. They should not be merged as duplicates.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45569 is the better canonical representative because it covers the substantive model work: new NemotronHDense/NemotronHSparse model files, auto-mapping updates, docs, and tests. PR #45115 is a narrower supporting refactor focused mainly on conversion-mapping and inheritance cleanup.", + "canonical_pr_reason": "PR #45569 is the better representative of the cluster because it covers the broader NemotronH family restructuring and adds the user-facing dense/sparse model support, while #45115 is a narrower internal refactor.", "best_issue_reason": null, - "best_pr_reason": "PR #45569 is the best overall representative of the cluster because it is the larger, more complete implementation of the NemotronH redesign. PR #45115 looks like a supporting infrastructure change rather than the same concrete change set.", + "best_pr_reason": "PR #45569 is the strongest standalone representative: it is the larger, more complete change set and subsumes the main NemotronH model-organization work more than #45115 does.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "They share the NemotronH area, but the actual changes differ: #45115 adds checkpoint conversion renames and compatibility plumbing, while #45569 introduces separate NemotronHDense/NemotronHSparse model artifacts, docs, and test updates. Different concrete change sets, not the same fix." + "reason": "Related NemotronH work, but different concrete changes and scopes: #45115 focuses on GraniteMoeHybrid-style inheritance and conversion key renames, while #45569 introduces separate nemotron_h_dense/sparse variants plus docs and tests. They are not the same fix and would not plausibly merge into one PR without losing scope clarity." } ] }, @@ -18903,12 +18817,12 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "cf5374a852b8716094d74603dd002fad622b3ba3bb42500665a062803eae508e", + "prepared_review_unit_hash": "bb445e52ebbba691fe11f6bd42ab53d3533101258835049478fecc0166e68f73", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5398, + "serialized_chars": 5397, "estimated_input_tokens": 1350, "estimated_eval_tokens": 2956 }, @@ -18916,7 +18830,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5398, + "serialized_chars": 5397, "estimated_input_tokens": 1350, "estimated_eval_tokens": 2956 }, @@ -18925,18 +18839,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related heterogeneous-config workstreams, but they are not duplicates: one PR adds per-layer config plumbing, the other adds heterogeneous model/runtime support on top of that foundation.", - "confidence": 0.88, + "summary": "Two related PRs in the same heterogeneity initiative: one adds per-layer heterogeneous config support, the other adds heterogeneous modeling/cache handling. They overlap in helper files, but they are not the same concrete change and should not be deduplicated.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45332 is the broader, more self-contained change set: it includes model/runtime integration, cache handling, and tests, so it is the best canonical PR in this cluster.", + "canonical_pr_reason": "PR 45332 is the broader umbrella change: it extends heterogeneous support into modeling and cache handling, while also relying on the config-layer helpers. It is the better representative PR for the cluster.", "best_issue_reason": null, - "best_pr_reason": "PR #45332 is the best overall representative because it covers the concrete modeling/path changes and depends on the config foundation introduced by #45333.", + "best_pr_reason": "PR 45332 is the best single representative of the cluster because it spans the largest portion of the new heterogeneous-model feature and subsumes the config work more than PR 45333 does.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "They share heterogeneous infrastructure files, but #45333 is configuration plumbing while #45332 is model/runtime support. They are complementary parts of the same feature area, not the same underlying bug or change, and would not normally be merged as one duplicate PR." + "reason": "Related feature work, but not the same bug/change: 45333 is per-layer config support, while 45332 is broader modeling/cache integration. Shared filenames alone are not enough to merge these PRs." } ] }, @@ -18966,7 +18880,7 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "a12fdb19b0920018b248e4a69018e9d7861f1003da1c5d8806c6edafad687132", + "prepared_review_unit_hash": "142753e821cc7c25478f0cd24e8b1bb5777f8d82101ef1830c6f16ebb2b3aae7", "component_budget": { "node_count": 2, "item_count": 2, @@ -18988,18 +18902,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs touch typing-related code, but they are not the same change: one is a broad repo-wide type-checking sweep, the other is a narrower typing/protocol follow-up around `_typing.py` and `modeling_utils` helpers.", - "confidence": 0.89, + "summary": "Both PRs are typing-related, but they are not duplicates: 45415 is a broad repo-wide type-checking sweep, while 45425 adds specific typing protocols/helpers for modeling and PEFT integration. They touch some of the same files, but not the same underlying change.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45425 is the more focused and self-contained typing change, centered on new typing protocols/helpers and specific modeling utilities. PR #45415 is a much broader pass enabling type checking across many `src/transformers/*.py` files, so it is less suitable as a duplicate representative.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR #45425 is the best single representative of the cluster because it has a narrower, clearer scope and more concrete code-path impact. PR #45415 is a broad style/typing sweep rather than the same underlying fix.", + "best_pr_reason": "45425 is the more focused, concrete typing-compatibility PR; 45415 is a much broader type-checking pass and should not be collapsed into the same change.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Both are typing-related, but they do not look like the same underlying bug/change. #45415 is a broad type-checking sweep across many files; #45425 adds typing protocols/helpers and targeted updates around `_typing.py`/`modeling_utils`. Shared files are not enough to treat them as duplicates." + "reason": "Shared typing files are incidental; 45415 is a wide type-checking sweep, while 45425 adds targeted typing infrastructure and compatibility helpers. Different underlying changes." } ] }, @@ -19031,12 +18945,12 @@ "pull_request:45487|pull_request:45843", "pull_request:45817|pull_request:45843" ], - "prepared_review_unit_hash": "8e827b1ea1beca37b469a10ae5dff3b5df3eaf92dddb53de7099b032764aac21", + "prepared_review_unit_hash": "e6595a1048cc9f2501e2b5b1f5564f3d5ed38e57d2dba26a91f7b1af6ea22952", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 7562, + "serialized_chars": 7561, "estimated_input_tokens": 1891, "estimated_eval_tokens": 4038 }, @@ -19044,7 +18958,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 7562, + "serialized_chars": 7561, "estimated_input_tokens": 1891, "estimated_eval_tokens": 4038 }, @@ -19053,24 +18967,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three open PRs all touch model-parallel handling, but they fix different model families and different failure modes. None should be merged as duplicates; the Gemma4 PR is the strongest representative of the cluster.", - "confidence": 0.95, + "summary": "These are three separate model-parallel fix PRs for different architectures (AltCLIP/ChineseCLIP, Gemma4, Deimv2). They share a broad theme but do not appear to fix the same concrete bug or be mergeable into one PR.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45817 is the best representative of the cluster: it fixes concrete model-parallel issues with a clear, self-contained code path change and test coverage, while the other PRs address separate models.", + "canonical_pr_reason": "PR #45817 is the strongest representative of the cluster theme: it has a clear, self-contained model-parallel fix, touches the relevant device-placement and no-split logic directly, and has the most discussion activity. The other PRs are separate model-specific fixes, not duplicates of it.", "best_issue_reason": null, - "best_pr_reason": "PR #45817 is the most substantial and broadly representative fix in this set, with direct device-placement/no-split-module changes and accompanying tests.", + "best_pr_reason": "PR #45817 is the best standalone PR in this set because it most directly addresses a concrete model-parallel failure mode with precise code-path changes. #45487 and #45843 are distinct fixes for different models and should not be conflated.", "soft_edge_verdicts": [ { "left": "pull_request:45487", "right": "pull_request:45843", "accept": false, - "reason": "Different models and different fixes: AltCLIP/ChineseCLIP device placement vs Deimv2 backbone _no_split_modules propagation. Same general theme, but not the same underlying bug." + "reason": "Both are model-parallel themed, but they target different model families and different code paths (AltCLIP/ChineseCLIP vs Deimv2). They are not the same underlying bug and would not plausibly merge into one PR." }, { "left": "pull_request:45817", "right": "pull_request:45843", "accept": false, - "reason": "Both concern model parallelism, but Gemma4 fixes specific Gemma4 code paths while Deimv2 fixes backbone module-splitting behavior. Distinct bugs and not mergeable as one PR." + "reason": "Gemma4-specific parallelization fixes are unrelated to Deimv2-specific backbone/module-splitting changes. Same broad subsystem, but different concrete bugs and implementations." } ] }, @@ -19100,7 +19014,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "fadbbd1772ea97d93bf0271705a85f1213d2bb6cbcd88a0eab486b4ae6180478", + "prepared_review_unit_hash": "9d0009d8c186bcb04337a0cd054fe0e0ab0d54e0c5ff5a248a4ac51a22449c84", "component_budget": { "node_count": 2, "item_count": 2, @@ -19122,18 +19036,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related by expert-parallelism/loading context, but they fix different bugs: one is config serialization for EP/DeepSpeed ZeRO-3, the other is a rank-0 overwrite issue during EP/FSDP2 loading. Not duplicates.", - "confidence": 0.81, + "summary": "These are related only at a broad EP/distributed-loading theme, but they fix different bugs and code paths: 45548 handles EP + DeepSpeed ZeRO-3 config/loading serialization, while 45662 fixes EP + FSDP2 expert tensors being overwritten during rank-0 broadcast.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45662 is the better representative because it targets a concrete loading corruption bug in the model-loading path and has the broader, more specific fix for EP behavior.", + "canonical_pr_reason": "No exact duplicate PR here; if one PR must represent the cluster, 45662 is the better anchor because it fixes the more concrete and severe expert-overwrite bug in the loading path.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the strongest standalone candidate: it fixes a distinct, concrete failure mode where experts can be silently overwritten during distributed loading.", + "best_pr_reason": "45662 is the stronger representative: it targets a specific corruption issue in distributed loading and spans the relevant code path more directly than 45548\u2019s narrower config/has_ep adjustment.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Different underlying bugs: #45548 removes distributed_config from serialization and adds has_ep for EP/ZeRO-3 loading, while #45662 fixes rank-0 broadcast overwriting experts during EP/FSDP2 loading. Same subsystem, not the same change." + "reason": "Same broad EP/distributed-loading area, but different underlying failures and fix paths; not plausible as one merged duplicate PR." } ] }, @@ -19163,7 +19077,7 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "3e34472b663b1c2751e3b536f231448d58c89a9dcb83aee617a9cb22f6d29882", + "prepared_review_unit_hash": "be0ad7f1b338599725a4d7c461ba48093cca7e72e0d95b7e31941d2ab7c36681", "component_budget": { "node_count": 2, "item_count": 2, @@ -19185,18 +19099,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs target the same xdist-safe captured_info logging problem; 45645 appears to subsume 45639 by adding the core fix plus CI log aggregation updates.", - "confidence": 0.87, + "summary": "Both PRs address the same xdist collision problem around captured_info output, with 45645 appearing to subsume 45639 and add the CI-side file handling needed to preserve logs.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45645 is the better canonical PR because it contains the same core xdist-safe captured_info fix as #45639 and also updates the CI/reporting path needed to consume the new per-worker log files.", + "canonical_pr_reason": "PR 45645 is the more complete change: it includes the xdist-safe testing_utils fix from 45639 and also updates CI/reporting code to consume multiple captured_info files, making it the better canonical representative.", "best_issue_reason": null, - "best_pr_reason": "PR #45645 is the strongest standalone change set: it addresses the bug in `testing_utils.py` and includes the supporting workflow/notification changes to preserve and surface the logs correctly.", + "best_pr_reason": "PR 45645 is the strongest standalone candidate because it covers the full end-to-end fix, not just the helper change in testing_utils.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying bug/change: both fix xdist collisions around `captured_info` output, share the same issue target and files, and the broader PR #45645 plausibly incorporates #45639\u2019s fix plus surrounding CI adjustments." + "reason": "Same underlying bug: xdist-safe handling of captured_info artifacts. 45645 extends the exact same code-path fix in testing_utils and adds the related CI output changes, so these could plausibly be merged into one PR." } ] }, @@ -19355,12 +19269,12 @@ "soft_pairs": [ "pull_request:45858|pull_request:45890" ], - "prepared_review_unit_hash": "33924db67c179c95e2e16cd11bc6df00bc84df3e186db8b1c46604a3ccb0e3e9", + "prepared_review_unit_hash": "43b423c313e03329c4798c8ce8a5e302c962ff893fc90d6de511f33af4986768", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4140, + "serialized_chars": 4139, "estimated_input_tokens": 1035, "estimated_eval_tokens": 2326 }, @@ -19368,7 +19282,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4140, + "serialized_chars": 4139, "estimated_input_tokens": 1035, "estimated_eval_tokens": 2326 }, @@ -19377,18 +19291,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests add XPU-specific test expectations, but they target different model test suites and different behavior, so they are not the same underlying fix/change.", - "confidence": 0.91, + "summary": "These are two unrelated model-specific test PRs that both add XPU expectations, but they touch different model test files and different code paths. The soft similarity is superficial rather than duplicate-level.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45890 is the better representative of the cluster because it includes an explicit XPU determinism gate plus new expectations, making it the more complete example of an XPU test-update PR.", + "canonical_pr_reason": "pull_request:45890 is the stronger representative of the cluster because it is a complete XPU test update with deterministic-for-XPU gating, while pull_request:45858 is a separate model's expectation-only patch.", "best_issue_reason": null, - "best_pr_reason": "PR 45890 is slightly stronger as a standalone representative since it modifies the test to account for XPU determinism in addition to adding expected values.", + "best_pr_reason": "pull_request:45890 is the best PR candidate overall for this cluster: it is self-contained, clearly scoped, and the added XPU-specific guard makes the test change more robust.", "soft_edge_verdicts": [ { "left": "pull_request:45858", "right": "pull_request:45890", "accept": false, - "reason": "Reject: these are model-specific test expectation updates for different files/models (hy_v3 vs exaone4_5), not the same concrete bug or code path. They could not plausibly be merged into one PR as a duplicate fix." + "reason": "Different models and different test files; both are just adding XPU expectations, not fixing the same concrete bug or change. This is too broad to treat as duplicates." } ] }, @@ -19418,40 +19332,40 @@ "soft_pairs": [ "pull_request:45861|pull_request:45903" ], - "prepared_review_unit_hash": "7390b63cd3dba59fd5e65437135c9973aeb0ee13f764c01fc995ff9448da528c", + "prepared_review_unit_hash": "1c8f486fe1121354bc015c0bf642b856e06835e917aeccec7f6cc5156af36b18", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5528, - "estimated_input_tokens": 1382, - "estimated_eval_tokens": 3020 + "serialized_chars": 5529, + "estimated_input_tokens": 1383, + "estimated_eval_tokens": 3022 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5528, - "estimated_input_tokens": 1382, - "estimated_eval_tokens": 3020 + "serialized_chars": 5529, + "estimated_input_tokens": 1383, + "estimated_eval_tokens": 3022 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs overlap on the same Qwen3-VL M-RoPE device placement fix, but PR 45903 also adds a separate meta/to_empty re-initialization change, so they are related rather than true duplicates.", - "confidence": 0.87, + "summary": "Both PRs target the same Qwen3-VL family M-RoPE/FSDP2 CPU-offload device mismatch, with 45903 expanding the earlier 45861 fix to also handle `meta` \u2192 `to_empty` re-init.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45903 is the better canonical PR because it includes the shared device-mismatch fix plus an additional, distinct meta/to_empty re-init handling change, making it the more complete patch.", + "canonical_pr_reason": "PR 45903 is the better canonical PR because it includes the same `inv_freq.to(x.device)` fix as 45861 and additionally covers the `meta`/`to_empty` re-initialization path, making it the more complete version of the same fix.", "best_issue_reason": null, - "best_pr_reason": "PR 45903 is the strongest standalone change set: it covers the shared FSDP2 CPU-offload device bug and also addresses the meta-device reinitialization path across the same model family.", + "best_pr_reason": "PR 45903 is the best representative because it addresses the full concrete failure mode in the same code path and subsumes 45861's narrower patch.", "soft_edge_verdicts": [ { "left": "pull_request:45861", "right": "pull_request:45903", - "accept": false, - "reason": "They share the same M-RoPE device-mismatch fix, but 45903 also introduces a separate `_apply`/`to_empty` re-init behavior. That makes it a superset, not the same concrete change that could cleanly merge as one duplicate PR." + "accept": true, + "reason": "Same underlying M-RoPE device/offload bug in the Qwen3-VL family; 45903 contains the 45861 change and extends it with the additional `meta`/`to_empty` fix, so they are overlapping versions of one fix." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index c72cc7033dd8b43caacf07b0a3892668acd02549..607e450d547dc6c58824518589764e4f19902a85 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260512t120027z", + "analysis_id": "hybrid-model-20260512t180027z", "archived_artifacts": { - "hybrid": "snapshots/20260512T120027Z/analysis-runs/hybrid-model-20260512t120027z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260512T120027Z/analysis-runs/hybrid-model-20260512t120027z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-05-12T12:09:56Z", + "published_at": "2026-05-12T18:10:17Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260512T120027Z", + "snapshot_id": "20260512T180027Z", "variant": "hybrid" } diff --git a/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..0e49465f3dc9c4617186ba73cbafa0699598e986 --- /dev/null +++ b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json @@ -0,0 +1,2279 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260512T180027Z", + "generated_at": "2026-05-12T18:10:15Z", + "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": 0.55, + "canonical_issue_number": 41211, + "canonical_pr_number": 44339, + "issue_numbers": [ + 41211 + ], + "pr_numbers": [ + 41356, + 44339 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41356, + "right_pr_number": 44339, + "code_similarity": 0.155, + "size_similarity": 0.094, + "file_overlap": 0.273, + "area_overlap": 0.078, + "patch_similarity": 0.183, + "shared_filenames": [ + "docs/source/en/model_doc/deimv2.md", + "src/transformers/models/deimv2/__init__.py", + "src/transformers/models/deimv2/configuration_deimv2.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "tests/models/deimv2/__init__.py", + "tests/models/deimv2/test_modeling_deimv2.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/deimv2.md", + "left_ranges": [ + [ + 1, + 132 + ] + ], + "right_ranges": [ + [ + 1, + 65 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/__init__.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/configuration_deimv2.py", + "left_ranges": [ + [ + 1, + 74 + ] + ], + "right_ranges": [ + [ + 1, + 266 + ] + ] + }, + { + "filename": "tests/models/deimv2/test_modeling_deimv2.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 1734 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-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-41115-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41115.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41115, + "canonical_pr_number": 45613, + "issue_numbers": [ + 41115 + ], + "pr_numbers": [ + 41116, + 45613 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41116, + "right_pr_number": 45613, + "code_similarity": 0.805, + "size_similarity": 0.885, + "file_overlap": 0.75, + "area_overlap": 0.854, + "patch_similarity": 0.693, + "shared_filenames": [ + "docs/source/en/_toctree.yml", + "docs/source/en/model_doc/minicpm3.md", + "src/transformers/models/auto/modeling_auto.py", + "src/transformers/models/minicpm3/__init__.py", + "src/transformers/models/minicpm3/configuration_minicpm3.py", + "src/transformers/models/minicpm3/modeling_minicpm3.py", + "src/transformers/models/minicpm3/modular_minicpm3.py", + "tests/models/minicpm3/__init__.py", + "tests/models/minicpm3/test_modeling_minicpm3.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/_toctree.yml", + "left_ranges": [ + [ + 713, + 720 + ] + ], + "right_ranges": [ + [ + 713, + 720 + ] + ] + }, + { + "filename": "docs/source/en/model_doc/minicpm3.md", + "left_ranges": [ + [ + 1, + 67 + ] + ], + "right_ranges": [ + [ + 1, + 45 + ] + ] + }, + { + "filename": "src/transformers/models/auto/modeling_auto.py", + "left_ranges": [ + [ + 285, + 291 + ], + [ + 700, + 706 + ], + [ + 1298, + 1304 + ] + ], + "right_ranges": [ + [ + 699, + 705 + ], + [ + 1301, + 1307 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/__init__.py", + "left_ranges": [ + [ + 1, + 29 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/configuration_minicpm3.py", + "left_ranges": [ + [ + 1, + 141 + ] + ], + "right_ranges": [ + [ + 1, + 126 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 544 + ] + ], + "right_ranges": [ + [ + 1, + 522 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modular_minicpm3.py", + "left_ranges": [ + [ + 1, + 444 + ] + ], + "right_ranges": [ + [ + 1, + 342 + ] + ] + }, + { + "filename": "tests/models/minicpm3/test_modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 109 + ] + ], + "right_ranges": [ + [ + 1, + 136 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45647-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45647.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 45647, + "canonical_pr_number": 45738, + "issue_numbers": [ + 45647 + ], + "pr_numbers": [ + 45737, + 45738 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 45737, + "right_pr_number": 45738, + "code_similarity": 0.526, + "size_similarity": 0.11, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.024, + "shared_filenames": [ + "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py", + "left_ranges": [ + [ + 491, + 497 + ] + ], + "right_ranges": [ + [ + 491, + 497 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43824-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43824.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 43824, + "canonical_pr_number": 43836, + "issue_numbers": [ + 43824 + ], + "pr_numbers": [ + 43836, + 43842 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "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-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-45561-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45561, + "canonical_pr_number": 45645, + "issue_numbers": [ + 45561 + ], + "pr_numbers": [ + 45639, + 45645 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45639, + "right_pr_number": 45645, + "code_similarity": 0.532, + "size_similarity": 0.619, + "file_overlap": 0.5, + "area_overlap": 0.543, + "patch_similarity": 0.451, + "shared_filenames": [ + "src/transformers/testing_utils.py", + "tests/utils/test_testing_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3529, + 3536 + ], + [ + 3753, + 3787 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 86 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + } + ] + } + ] + }, + { + "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-45861-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 45861, + "issue_numbers": [], + "pr_numbers": [ + 45861, + 45903 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45861, + "right_pr_number": 45903, + "code_similarity": 0.617, + "size_similarity": 0.516, + "file_overlap": 1.0, + "area_overlap": 0.377, + "patch_similarity": 0.546, + "shared_filenames": [ + "src/transformers/models/qwen3_5/modeling_qwen3_5.py", + "src/transformers/models/qwen3_5_moe/modeling_qwen3_5_moe.py", + "src/transformers/models/qwen3_omni_moe/modeling_qwen3_omni_moe.py", + "src/transformers/models/qwen3_vl/modeling_qwen3_vl.py", + "src/transformers/models/qwen3_vl/modular_qwen3_vl.py", + "src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", + "tests/models/qwen3_vl/test_modeling_qwen3_vl.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/qwen3_5/modeling_qwen3_5.py", + "left_ranges": [ + [ + 146, + 154 + ] + ], + "right_ranges": [ + [ + 146, + 154 + ], + [ + 161, + 183 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_5_moe/modeling_qwen3_5_moe.py", + "left_ranges": [ + [ + 142, + 150 + ] + ], + "right_ranges": [ + [ + 142, + 150 + ], + [ + 157, + 179 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_omni_moe/modeling_qwen3_omni_moe.py", + "left_ranges": [ + [ + 1315, + 1323 + ] + ], + "right_ranges": [ + [ + 1315, + 1323 + ], + [ + 1330, + 1352 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl/modeling_qwen3_vl.py", + "left_ranges": [ + [ + 354, + 362 + ] + ], + "right_ranges": [ + [ + 354, + 362 + ], + [ + 369, + 391 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl/modular_qwen3_vl.py", + "left_ranges": [ + [ + 294, + 302 + ] + ], + "right_ranges": [ + [ + 32, + 38 + ], + [ + 270, + 292 + ], + [ + 311, + 319 + ] + ] + }, + { + "filename": "src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py", + "left_ranges": [ + [ + 865, + 873 + ] + ], + "right_ranges": [ + [ + 865, + 873 + ], + [ + 880, + 902 + ] + ] + }, + { + "filename": "tests/models/qwen3_vl/test_modeling_qwen3_vl.py", + "left_ranges": [ + [ + 17, + 26 + ], + [ + 29, + 35 + ], + [ + 45, + 78 + ], + [ + 522, + 547 + ] + ], + "right_ranges": [ + [ + 17, + 26 + ], + [ + 29, + 35 + ], + [ + 45, + 79 + ], + [ + 523, + 582 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [], + "duplicate_prs": [ + { + "cluster_id": "cluster-41115-3", + "canonical_pr_number": 45613, + "duplicate_pr_numbers": [ + 41116 + ], + "target_issue_number": 41115, + "reason": "PRs in cluster-41115-3 are treated as duplicates because they converge on issue #41115 with closing_reference evidence." + }, + { + "cluster_id": "cluster-41211-3", + "canonical_pr_number": 44339, + "duplicate_pr_numbers": [ + 41356 + ], + "target_issue_number": 41211, + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43240-3", + "canonical_pr_number": 43251, + "duplicate_pr_numbers": [ + 43254 + ], + "target_issue_number": 43240, + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-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-43824-3", + "canonical_pr_number": 43836, + "duplicate_pr_numbers": [ + 43842 + ], + "target_issue_number": 43824, + "reason": "PRs in cluster-43824-3 are treated as duplicates because they converge on issue #43824 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-3 are treated as duplicates because they converge on issue #45561 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45647-3", + "canonical_pr_number": 45738, + "duplicate_pr_numbers": [ + 45737 + ], + "target_issue_number": 45647, + "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-45861-2", + "canonical_pr_number": 45861, + "duplicate_pr_numbers": [ + 45903 + ], + "target_issue_number": null, + "reason": "PRs in cluster-45861-2 are treated as duplicates because they share soft_similarity evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-45647-3", + "issue_number": 45647, + "reason": "Issue #45647 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 12 discussion comments plus 8 inbound references.", + "score": 162.0 + }, + "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/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..899172dcb2e6228f8444f5d7f3e4777c9908eedb --- /dev/null +++ b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,19380 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260512T180027Z", + "generated_at": "2026-05-12T18:10:15Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 1, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "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: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:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45070", + "issue:45081", + "issue:45161", + "issue:45237", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823", + "issue:45874", + "issue:45901" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:45237|issue:45685", + "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:44805|issue:45823", + "issue:43638|issue:44805", + "issue:43596|issue:44805", + "issue:44805|issue:45161", + "issue:44805|issue:44918", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43039|issue:45161", + "issue:43975|issue:44457", + "issue:44292|issue:44457", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:39290|issue:45081", + "issue:45823|issue:45874", + "issue:43828|issue:45161", + "issue:43425|issue:44292", + "issue:34928|issue:41720", + "issue:44918|issue:45161", + "issue:43716|issue:45685", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:45507|issue:45685", + "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:43828|issue:45685", + "issue:44918|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:43039|issue:44805", + "issue:43901|issue:44512", + "issue:41762|issue:43596", + "issue:39290|issue:41720", + "issue:45237|issue:45901", + "issue:44841|issue:45823", + "issue:43039|issue:44560", + "issue:43866|issue:44863", + "issue:43854|issue:43866", + "issue:41093|issue:43638", + "issue:39290|issue:43531", + "issue:43792|issue:45901", + "issue:43425|issue:45070", + "issue:43596|issue:45800" + ], + "prepared_review_unit_hash": "cd0b97f14548d2b625bf2900852fefd72a613f4f9a4704c8c2ae34b614a5ce30", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 63, + "serialized_chars": 45921, + "estimated_input_tokens": 11481, + "estimated_eval_tokens": 23218 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issues, with two small duplicate clusters: v5 docs cleanup for removed references, and DeepSpeed ZeRO-3 load-time index-0-out-of-bounds failures. The rest are only superficially similar title matches.", + "confidence": 0.72, + "canonical_issue_reason": "issue:43638 is the clearest representative of the shared DeepSpeed ZeRO-3 loading bug: the exact index-0-out-of-bounds failure appears here in a broad, model-agnostic form, while nearby reports look like variants of the same load path.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43638 is the strongest issue to keep as the cluster anchor because it states the concrete failure mode clearly and generalizes beyond a single model family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different models and failure modes: Qwen3.5 config attribute error vs streaming chat/completions issue." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated hardware/runtime failures: AMD GPU loading problem vs MPS histogram-on-int error." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different bugs: image-preprocessor dtype mismatch vs AMD GPU load failure." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Autocast dtype mismatch and AMD GPU device/runtime failure are not the same bug." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different code paths: GPU load failure vs invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "NVFP4 runtime error and AMD GPU load failure are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch vs ONNX export error; different subsystems and symptoms." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ONNX export failure and FineGrainedFP8 model failure are unrelated." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Different problems: export-time ONNX issue vs auto device-map cuda assert." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": true, + "reason": "Both are v5 docs cleanup for removed/deprecated references; they can be fixed in one documentation PR." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ONNX export issue vs import-time GenerationMixin regression; different failures." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask/tensor shape mismatch and Gemma4 PLE device mismatch are different runtime errors." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Different bugs: ZeRO-3 index-out-of-bounds during load vs generic mask shape mismatch." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Not the same underlying failure; one is ZeRO-3 loading, the other is a mask indexing mismatch." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Mask shape mismatch is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Different symptoms and code paths: mask indexing error vs Qwen3.5/TRl embedding unpacking." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "Processor voice_preset bug and ONNX export error are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Collective mismatch during ZeRO-3 and index-0-out-of-bounds loading are different failures." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds load failure; only the model family differs." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are not the same bug." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Detokenization regression and LoRA merge/save mismatch are different issues." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "NVFP4 runtime error vs LoRA save/reload inconsistency are unrelated." + }, + { + "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": "Different DeepSpeed problems: weight shape error vs collective mismatch." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "ZeRO-3 model-loading index error is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "Activation checkpointing/FSDP issue and CustomTrainer signature change are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "Different failures: ZeRO-3 loading bug vs GPT-OSS TP issue." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Dtype mismatch with autocast/image preprocessing is unrelated to an invalid method call." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Missing sliding_window_pattern and tokenizer backend_tokenizer crash are different regressions." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Device mismatch during forward and expensive reverse embedding lookup are different Gemma4 issues." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Autocast dtype mismatch and tensor-parallel failure are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen3 NVFP4 runtime error are different issues." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch is unrelated to cudaErrorAssert under auto device mapping." + }, + { + "left": "issue:44918", + "right": "issue:45161", + "accept": false, + "reason": "TRL SFT embedding unpacking and GPT-OSS TP failure are different code paths." + }, + { + "left": "issue:43716", + "right": "issue:45685", + "accept": false, + "reason": "Image-preprocessor dtype mismatch is unrelated to the MPS histogram int limitation." + }, + { + "left": "issue:45507", + "right": "issue:45685", + "accept": false, + "reason": "Invalid method call and MPS histogram int error are not the same bug." + }, + { + "left": "issue:43716", + "right": "issue:44292", + "accept": false, + "reason": "Different runtime failures: dtype mismatch vs NVFP4 model error." + }, + { + "left": "issue:44292", + "right": "issue:44918", + "accept": false, + "reason": "NVFP4 runtime error and Qwen3.5 embedding issue are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross_entropy issue is unrelated to TRL SFT input-embedding unpacking." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 index-0-out-of-bounds loading bug; the model differs, but the failure pattern is the same." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Both are stale v5 documentation references that need the same cleanup pass." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Different bugs: autocast dtype mismatch vs Qwen3-vl video StopIteration." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." + }, + { + "left": "issue:44918", + "right": "issue:45685", + "accept": false, + "reason": "TRL trainer embedding issue and MPS histogram int error are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Liger Kernel cross_entropy issue is not the same as Phi-tiny-MoE autocast dtype mismatch." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and trainer compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch and GenerationMixin import regression are different problems." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger Kernel cross_entropy issue is unrelated to the mask shape mismatch bug." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "return_all_scores docs behavior mismatch is not the same issue as the removed run-command docs reference." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": false, + "reason": "Same general ZeRO-3 theme, but not enough direct evidence to treat these as the same concrete duplicate pair." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "sliding_window_pattern missing and cudaErrorAssert under device mapping are different failures." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper model load failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch-version incompatibility and pydantic PretrainedConfig field breakage are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "ZeRO-3 load-time index error and torch/transformers incompatibility are different issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The accepted soft-edge pairs are mergeable as single fixes: the v5 docs references are both documentation cleanup, and the ZeRO-3 cases all share the same load-time index-out-of-bounds failure pattern despite different models. The rejected pairs also read as appropriately cautious rather than overstated." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 2, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36010", + "issue:39186", + "issue:39290", + "issue:40990", + "issue:41093", + "issue:41762", + "issue:43054", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43645", + "issue:43716", + "issue:43792", + "issue:43828", + "issue:43866", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44512", + "issue:44530", + "issue:44589", + "issue:44841", + "issue:44918", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45084", + "issue:45161", + "issue:45362", + "issue:45399", + "issue:45464", + "issue:45542", + "issue:45632", + "issue:45685", + "issue:45698", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:45901", + "issue:45902" + ], + "soft_pairs": [ + "issue:45859|issue:45902", + "issue:43975|issue:45901", + "issue:45362|issue:45464", + "issue:43792|issue:45070", + "issue:44292|issue:45901", + "issue:43404|issue:43866", + "issue:39290|issue:45823", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:36010|issue:39290", + "issue:44841|issue:45084", + "issue:41093|issue:41762", + "issue:45800|issue:45823", + "issue:43296|issue:43366", + "issue:43572|issue:45070", + "issue:44387|issue:45005", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:44291|issue:45070", + "issue:43366|issue:45464", + "issue:45020|issue:45800", + "issue:43531|issue:45070", + "issue:43541|issue:44589", + "issue:41093|issue:43596", + "issue:43054|issue:43994", + "issue:45542|issue:45800", + "issue:40990|issue:44960", + "issue:43716|issue:44918", + "issue:39186|issue:44918", + "issue:45632|issue:45698", + "issue:43404|issue:45005", + "issue:43645|issue:45399", + "issue:44530|issue:45005", + "issue:44360|issue:44512" + ], + "prepared_review_unit_hash": "dccfe0eb4d3044f2e9480206697ec53df0990a70b83154fc75818fe504204e19", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37849, + "estimated_input_tokens": 9463, + "estimated_eval_tokens": 19182 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: many unrelated model/runtime regressions, with only a couple of plausible duplicate sub-pairs. The strongest duplicate signals are around Siglip2 output regressions and remote_code/custom-module loading collisions.", + "confidence": 0.84, + "canonical_issue_reason": "There is no single true canonical issue for the whole cluster; it mixes unrelated bugs. If forced to pick an anchor, issue:45698 is the best representative because it is concrete, open, and sits in the most plausible duplicate sub-thread about custom-module/remote_code loading.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue:45698 is the clearest actionable report in the set and the best anchor for deduping the remote_code/custom-module loading cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Same model family, but different concrete failures: CPU offload compatibility vs uninitialized-memory access after meta->to_empty(cuda)." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Different subsystems and symptoms: detokenization bug vs table-question-answering crash." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Both mention Qwen3.5/chat, but one is a general crash and the other is a streaming API failure on a different model size; not the same bug." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper runtime failure and a PretrainedConfig/pydantic regression are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:45901", + "accept": false, + "reason": "Different model and failure mode; NVFP4 execution error is unrelated to table QA crash." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Tied lm_head weights in Mistral3 and a corrupted Ovis2 checkpoint are not the same underlying issue." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3 missing config attribute with vLLM vs Gemma4 device-mismatch during forward are different bugs." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Streaming chat API failure and an MPS histogram op limitation are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Different MoE issues: torch dynamo grouped_mm tracing failure vs MPS int histogram support." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Both are MoE/dtype-related, but one is grouped_mm tracing and the other is autocast dtype mismatch; too different to merge." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "ImportError for GenerationMixin and a Gemma3 config regression are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure for Voxtral and a template-node compile TypeError do not share the same code path." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are IndexError-shaped, but one is a mask-size mismatch and the other is ZeRO-3 model-loading out-of-bounds; not the same bug." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/Transformers version incompatibility and Gemma4 device placement mismatch are different regressions." + }, + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "PaddleOCR-VL load failure and GGUF gpt-oss support are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig and a pydantic PretrainedConfig regression are distinct config issues." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Int4 CUDA memory/OOM regression and tied-weights translation-model bug are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Both involve MoE/GPT-OSS, but one is a grouped_mm tracing bug and the other is tensor-parallel routing; not the same concrete issue." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Autocast dtype mismatch and streaming chat API failure are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized argument mismatch is a different loading-path bug from the pydantic PretrainedConfig regression." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference failure are unrelated." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Remote_code breakage across recent transformers versions and a specific torch/transformers incompatibility are not the same bug." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE sliding_window issue and the PretrainedConfig pydantic regression are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Grouped_mm tracing failure and Float8 storage lookup failure are different runtime errors." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both mention IndexError, but the shape mismatch and ZeRO-3 empty-tensor initialization are different failure modes." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": true, + "reason": "Both report Siglip2 producing bad/nonsensical text-embedding outputs; they look like the same underlying model behavior bug." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "Undefined tf backend error and torch/transformers incompatibility are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "High perplexity on gpt-oss and GLM5 are different model-specific issues." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Mistral-3 image-preprocessor dtype mismatch and Qwen3.5 SFT embedding unpacking failure are unrelated." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP weight-shape runtime error and TRL embedding unpacking failure are different code paths." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": true, + "reason": "Both point to remote_code/custom-module loading picking the wrong cached module; they are plausibly the same cache-key collision bug." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Mistral3 tied lm_head weights and translation-model tied-weights regressions are in the same broad area but not the same concrete bug." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Custom-model initialization in notebooks and flash-attn fallback blocking are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttention linear_attention crash and tied weights in translation models are unrelated." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "DSA indexer ReLU discussion and removed docs run command are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The cluster is clearly heterogeneous, and the two accepted soft pairs are phrased conservatively enough to be plausible duplicates rather than asserted matches. The only slight caution is the high confidence, but it is still defensible given the evidence provided." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 3, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36331", + "issue:38175", + "issue:39692", + "issue:41553", + "issue:41628", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43493", + "issue:43643", + "issue:43644", + "issue:43782", + "issue:43828", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:44292", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44492", + "issue:44512", + "issue:44534", + "issue:44589", + "issue:44863", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45237", + "issue:45399", + "issue:45507", + "issue:45520", + "issue:45701", + "issue:45800", + "issue:45859" + ], + "soft_pairs": [ + "issue:42915|issue:45399", + "issue:43873|issue:45005", + "issue:44863|issue:45005", + "issue:44492|issue:44512", + "issue:44918|issue:45507", + "issue:43975|issue:44292", + "issue:43950|issue:44534", + "issue:44292|issue:45507", + "issue:43644|issue:43950", + "issue:39692|issue:43054", + "issue:43828|issue:44928", + "issue:36331|issue:41762", + "issue:43866|issue:44403", + "issue:41628|issue:45800", + "issue:43054|issue:45701", + "issue:43872|issue:44589", + "issue:38175|issue:43054", + "issue:41553|issue:43883", + "issue:44910|issue:45859", + "issue:41553|issue:43881", + "issue:42947|issue:44387", + "issue:43054|issue:43493", + "issue:44403|issue:44451", + "issue:43643|issue:45020", + "issue:43975|issue:45520", + "issue:43975|issue:45237", + "issue:42947|issue:43856", + "issue:44403|issue:44488", + "issue:43782|issue:44960", + "issue:42491|issue:44368", + "issue:43881|issue:43883", + "issue:41762|issue:42915" + ], + "prepared_review_unit_hash": "203f12cbadc52545c81a3f563fa1af137c5e0812e20038874bf321f118de5a3b", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36368, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of unrelated Transformers issues, with only one clear duplicate-style pair around non-persistent buffer corruption. I\u2019d anchor on the 5.x buffer regression issue as the best representative.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43950 is the clearest, most directly stated report of the non-persistent-buffer regression, and it has the strongest same-bug overlap with issue:44534.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43950 is the best overall issue anchor: specific, technically clear, and it captures the concrete regression that appears duplicated elsewhere in the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Different problems: Qwen3 MoE fine-tuning failure vs flash-attn2 fallback blocked by unrelated checks." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior is unrelated to tied-weights translation-model warnings." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is a different code path than tied-weight handling." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A typo in cache strategies is not the same as stale docs mentioning a removed command." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Different model and failure mode: SFT embedding unpacking vs an invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Incorrect detokenization and Qwen-3 NVFP4 loading are distinct issues." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Both describe the same 5.x regression where non-persistent buffers get corrupted/junk values on load." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "No shared underlying bug; one is a model/runtime load failure, the other a method-call issue." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": false, + "reason": "Missing fields from trust_remote_code config is unrelated to non-persistent buffer corruption." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Both involve SigLIP2, but one is a docs/example error and the other is a model-quality complaint; not the same concrete bug." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different models and symptoms: dtype mismatch under autocast vs Qwen3.5 position_ids/SDPA behavior." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "CustomTrainer signature mismatch and Gemma3 ZeRO-3 load failure are unrelated." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Corrupted checkpoint vs noisy loading warnings are not the same bug." + }, + { + "left": "issue:41628", + "right": "issue:45800", + "accept": false, + "reason": "Import error for AutoImageProcessor is unrelated to torch/transformers version incompatibility." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "SigLIP2 embedding quality and tokenization changes are too broad and not the same concrete issue." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Different incompatibilities: bitsandbytes Int8Params kwarg mismatch vs missing Float8 storage object." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both concern SigLIP2 behavior, but the titles point to different symptoms without a clear shared code-path bug." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral and Molmo tied-weights attribute error are unrelated." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Flash-Attention packed-sequence misinterpretation and CPU-offload rotary-embedding incompatibility are different failures." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Voxtral loading UX problem and glm-4v loading failure do not share the same underlying bug." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Gradient checkpointing ineffectiveness is not the same issue as int4 quantization memory growth/OOM." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "Potentially related SigLIP2 implementation concerns, but not clearly the same concrete bug from the titles alone." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "General loading noise vs inability to load a specific model are different issues." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "A missing-field config return bug is not the same as broader remote_code breakages." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Detokenization bug and Python 3.13 flash_attn import error are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Detokenization and AMD GPU execution failure are distinct problems." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Different memory-related training issues; not the same code-path bug." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Unnecessary loading noise and failure to load a specific model are not the same underlying issue." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Qwen3VL weight_only load error and GLM5 are unrelated model-loading issues." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "LoRA compatibility on hf5.x and tied-word-embeddings warnings are different concerns." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Two separate model-loading failures with no clear shared bug." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Gemma3 ZeRO-3 load crash and Qwen3 MoE fine-tuning failure are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft edge, issue:43950 \u2194 issue:44534, is well-supported by the titles and the regression description. The rest of the pair verdicts appropriately avoid overclaiming shared bugs when the evidence is only superficial similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 4, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36331", + "issue:41093", + "issue:41553", + "issue:42915", + "issue:43257", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43638", + "issue:43742", + "issue:43792", + "issue:43824", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43872", + "issue:43940", + "issue:43950", + "issue:44155", + "issue:44220", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44560", + "issue:44661", + "issue:44805", + "issue:44863", + "issue:44928", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45305", + "issue:45313", + "issue:45356", + "issue:45399", + "issue:45538", + "issue:45584", + "issue:45588", + "issue:45663", + "issue:45685", + "issue:45759", + "issue:45823", + "issue:45902" + ], + "soft_pairs": [ + "issue:45685|issue:45823", + "issue:43824|issue:43872", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:44991|issue:45399", + "issue:43828|issue:45823", + "issue:43827|issue:44509", + "issue:45081|issue:45399", + "issue:43950|issue:45902", + "issue:45020|issue:45356", + "issue:43404|issue:43950", + "issue:43742|issue:45399", + "issue:43257|issue:43866", + "issue:42915|issue:45020", + "issue:43638|issue:45823", + "issue:43792|issue:44220", + "issue:43940|issue:45313", + "issue:43872|issue:44291", + "issue:42915|issue:43872", + "issue:43854|issue:44863", + "issue:41553|issue:44991", + "issue:43257|issue:45305", + "issue:45759|issue:45902", + "issue:44928|issue:45588", + "issue:36331|issue:41093", + "issue:41093|issue:44805", + "issue:43872|issue:45538", + "issue:44530|issue:44863", + "issue:41553|issue:44488", + "issue:44479|issue:44560", + "issue:24643|issue:30064", + "issue:43366|issue:43828", + "issue:45663|issue:45823", + "issue:44155|issue:45584", + "issue:43366|issue:43531" + ], + "prepared_review_unit_hash": "6d3efd902d4c17f0f10765e2801cf8b23d04549d0b76ab2f66c758f9a670b971", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38556, + "estimated_input_tokens": 9639, + "estimated_eval_tokens": 19534 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a loose semantic cluster around recent Transformers loading/runtime regressions, but the individual issues are usually about different models, error modes, and code paths. Only a few pairs are true duplicates on the same concrete bug.", + "confidence": 0.86, + "canonical_issue_reason": "issue:45020 is the broadest umbrella candidate: it targets recent-version breakage across multiple remote-code models, so it is the most plausible single issue to anchor a heterogeneous loading-regression cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the best overall representative because it is open, broadly scoped, and framed as a cross-model regression rather than a model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "Different failures: MPS histogram/int support in MoE code vs a Gemma4 device-mapping mismatch." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Import failure for Qwen2.5-VL is unrelated to bitsandbytes' unexpected `_is_hf_initialized` argument error." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Broad remote-code regression report vs a specific Gemma4 device mismatch; not the same bug." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer mapping edge case in `add-new-model-like` is a different code path from remote-code model breakage." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression for a specific model is unrelated to the flash-attn fallback gating issue." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Different model and different failure mode: dtype mismatch under autocast vs device placement mismatch." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Both report the same docs regression: old pipeline task names still referenced after v5 removals." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Mistral tokenizer regex patch crash is unrelated to flash-attn2 fallback blocking." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": false, + "reason": "Both involve buffer/materialization concerns, but one is a general `from_pretrained` regression and the other is a specific Qwen3-MoE uninitialized-memory bug." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Remote-code model breakage and Kimi tokenizer codec handling are different problems." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Tied-weight bug in Mistral3 is not the same as non-persistent buffer corruption." + }, + { + "left": "issue:43742", + "right": "issue:45399", + "accept": false, + "reason": "MobileLLM key error and flash-attn fallback gating are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43866", + "accept": false, + "reason": "Qwen3 MoE weight conversion under DeepSpeed is unrelated to a corrupted Ovis2 checkpoint." + }, + { + "left": "issue:42915", + "right": "issue:45020", + "accept": false, + "reason": "FineGrainedFP8Config failure is a specific quantization/config issue, not the broader remote-code regression." + }, + { + "left": "issue:43638", + "right": "issue:45823", + "accept": false, + "reason": "Zero-3 + non-pretrained BERT index error is unrelated to Gemma4 device mismatch." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Whisper model runtime failure is too generic to merge confidently with the `_torch_extract_fbank_features()` report." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Qwen3-Next missing weights under ZeRO-3 and Qwen3.5 language_model loading are distinct DeepSpeed loading bugs." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same underlying bug: loading under `init_empty_weights` hits the unexpected `_is_hf_initialized` keyword in bitsandbytes." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "Different root causes: FP8Config behavior vs bitsandbytes constructor incompatibility." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "GLM-4.7-Flash test loading issue and NemotronH checkpoint loading failure are unrelated model-specific problems." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-related, but they concern different models and different failure mechanisms." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "DeepSpeed weight conversion bug and gradient averaging bug are separate training issues." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "Composite config unwrapping and uninitialized rotary embedding memory are different code paths." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "RLHF gradient explosion from position IDs is unrelated to a flash-attention `s_aux=None` crash." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "CustomTrainer signature mismatch and mask/index shape mismatch are unrelated." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same shape-mismatch IndexError, with the later issue appearing to be the same underlying bug report." + }, + { + "left": "issue:43872", + "right": "issue:45538", + "accept": false, + "reason": "bitsandbytes init incompatibility is unrelated to CLIPTokenizer's `model_max_length` constant." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and NemotronH loading failure are different subsystems and symptoms." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Bad Voxtral tokenizer error message and failing to load `cjvt/sleng-bert` are separate tokenizer-loading issues." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Video-input regression across Qwen VL variants and a Whisper embedding `StopIteration` are different bugs." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "GGUF gpt-oss architecture support and Phi-tiny-MoE autocast dtype mismatch are unrelated." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states bug is distinct from Gemma4 device-map mismatch." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 track leakage and Whisper empty-transcription failure are different audio inference bugs." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Trainer API change vs tensor mask shape mismatch; not the same underlying issue." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression and flash-attn fallback gating do not share the same code path." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The accepted soft-edge pairs are well supported by near-identical titles and failure modes, and the rejected pairs are reasonably distinguished by model, subsystem, or error type. No overclaim stands out." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 5, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43366", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43824", + "issue:43854", + "issue:43931", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44756", + "issue:44805", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44960", + "issue:44977", + "issue:45030", + "issue:45276", + "issue:45313", + "issue:45357", + "issue:45488", + "issue:45685", + "issue:45812", + "issue:45823", + "issue:45834" + ], + "soft_pairs": [ + "issue:43950|issue:45834", + "issue:43550|issue:44910", + "issue:44560|issue:45685", + "issue:43645|issue:43646", + "issue:43824|issue:43957", + "issue:43366|issue:44589", + "issue:42915|issue:43957", + "issue:42617|issue:43366", + "issue:44560|issue:44918", + "issue:36010|issue:42915", + "issue:44661|issue:45823", + "issue:38175|issue:43994", + "issue:43632|issue:45834", + "issue:44560|issue:44805", + "issue:44877|issue:45030", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:44661|issue:44805", + "issue:43317|issue:44756", + "issue:43644|issue:44534", + "issue:38175|issue:39692", + "issue:44960|issue:44977", + "issue:42491|issue:43720", + "issue:45313|issue:45357", + "issue:45488|issue:45812", + "issue:44451|issue:44488", + "issue:41720|issue:42915", + "issue:43708|issue:44075", + "issue:43854|issue:43931", + "issue:42491|issue:45276", + "issue:43257|issue:43854", + "issue:42915|issue:44756", + "issue:42491|issue:43931", + "issue:39692|issue:43994" + ], + "prepared_review_unit_hash": "59866b35d4a5d134e9c2bfc54978ddf8574eece4a4c5b8c89959c15ccbfba431", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37806, + "estimated_input_tokens": 9452, + "estimated_eval_tokens": 19160 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports rather than one unified duplicate family. The clearest duplicate subcluster is the non-persistent-buffer corruption reports; there is also a likely related BERT-loading pair, but the rest are broader subsystem/theme overlaps rather than the same bug.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43644 is the cleanest canonical for the non-persistent-buffer regression subcluster: its title states the bug directly, and issue:44534 appears to be a near-rephrasing of the same failure mode.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the best overall issue representative because it is a concrete, well-scoped v5 regression with a solid duplicate counterpart; the cluster as a whole is too heterogeneous for a single broader canonical beyond that subcluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Different bugs: non-persistent buffer corruption versus Kosmos2.5 OCR index failure." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Both are Qwen/attention-related, but the failure modes differ: SDPA/torch.compile versus Flash Attention with 3D position_ids." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated model and backend issues: video StopIteration versus an MPS histogram implementation error." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": false, + "reason": "Both involve custom models, but one is notebook-specific and the other is a broader custom-model initialization regression." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Both are loading-related, but the specific failures differ: missing import versus meta-device loading behavior." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "GGUF/gpt-oss support is unrelated to the Float8 storage lookup error." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3Moe FP8 failure and meta-device loading failures are different code-path problems." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "Running 3d_parallel.py and GGUF gpt-oss support are unrelated issues." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Both touch Qwen3.5 video/input flows, but the observed failures are different: StopIteration versus TRL embedding unpacking." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to Qwen3Moe FP8 loading." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping edge case and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": false, + "reason": "Both are SigLIP2-related, but one is a zero-probabilities bug and the other is an AutoModel/pipeline output issue; not the same underlying defect." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "_is_hf_initialized flag regression is unrelated to Kosmos2.5 long-ocr index errors." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different failures: video StopIteration versus a mask/tensor shape mismatch." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Strict config blocking granite_speech loading and tiny-random glm4v config validation are separate config issues." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Zero3 training with non-pretrained BertModel and tokenizer-mapping-name handling are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Zero3 index error in BertModel is not the same bug as add-new-model-like failing inside TOKENIZER_MAPPING_NAMES." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Tokenizer mapping behavior and a mask shape mismatch are different problems." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "device_map=auto offload/dequantized-model loading and disabling mmap on Strix Halo are not the same underlying issue." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same regression: Transformers v5 corrupts non-persistent buffers during loading; 44534 is a reworded duplicate report." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Both mention SigLIP2, but one is a runtime output bug and the other is documentation/example errors plus quantization failure." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Different models and different failure modes: GLM5 versus Qwen3.5 flash-attention generation." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Qwen3 MoE LoRA compatibility and BitNet packed-weight unpacking are unrelated loading issues." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "DeepSpeed ZeRO-3 load failure for language_model and incorrect visual encoder keys during save_pretrained are different regression paths." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Tokenizer pre-tokenizer override in LlamaTokenizer is not the same as Granite token-ID misgeneration." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": true, + "reason": "Both report current/latest Transformers failing to load specific BERT checkpoints, which is plausibly the same underlying loading regression." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert and Qwen3Moe FP8 failures are separate issues." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint-resume step calculation and SGD args not being used are unrelated trainer/optimizer bugs." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Unit-test loading failure for GLM-4.7-Flash and Qwen3-VL weight-shape mismatch are different model-loading problems." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Qwen3 MoE LoRA compatibility and Gemma4 resize_token_embeddings affecting embeddings are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "MoE weights not converted with accelerate+deepspeed and GLM-4.7-Flash unit-test loading are different bugs." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3Moe FP8 failure and mmap OOM on Strix Halo are unrelated." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3 MoE LoRA incompatibility and Qwen3-VL weight-shape mismatch are different model-specific load failures." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Both are SigLIP2-related, but one is an example/documentation issue and the other is a runtime model-output issue." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The summary is mostly conservative, but it overstates the evidence in two places: (1) calling issue:43644 the cluster\u2019s \"strongest duplicate match\" is not grounded by the packet, since issue:44451/issue:44488 has higher similarity metrics; and (2) the BERT pair is better described as a likely related loading class rather than a confirmed duplicate bug. The non-persistent-buffer pair looks solid, but the representative/canonical choice should be phrased more cautiously." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 6, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42947", + "issue:43335", + "issue:43381", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43957", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44488", + "issue:44743", + "issue:44756", + "issue:44910", + "issue:44912", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45362", + "issue:45406", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45859", + "issue:45902" + ], + "soft_pairs": [ + "issue:43856|issue:43873", + "issue:44488|issue:44991", + "issue:43644|issue:43645", + "issue:45362|issue:45406", + "issue:43856|issue:44387", + "issue:44910|issue:45588", + "issue:42491|issue:45292", + "issue:36010|issue:39404", + "issue:45276|issue:45335", + "issue:43404|issue:43454", + "issue:44164|issue:44291", + "issue:41720|issue:44155", + "issue:44402|issue:44488", + "issue:39692|issue:44960", + "issue:42947|issue:45704", + "issue:43425|issue:43957", + "issue:43873|issue:44387", + "issue:43335|issue:45663", + "issue:45663|issue:45759", + "issue:45663|issue:45859", + "issue:44291|issue:44756", + "issue:44155|issue:45823", + "issue:44387|issue:44756", + "issue:43854|issue:45216", + "issue:44292|issue:44912", + "issue:42947|issue:43381", + "issue:44743|issue:44960", + "issue:43957|issue:44756", + "issue:45276|issue:45292", + "issue:45684|issue:45698", + "issue:43381|issue:44928", + "issue:45663|issue:45902", + "issue:44164|issue:44756", + "issue:43646|issue:43708" + ], + "prepared_review_unit_hash": "49b71b5c3e81dcc0f4aead9a7ca0d291e920e2ec72f2c3c09824b7091bad5779", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38091, + "estimated_input_tokens": 9523, + "estimated_eval_tokens": 19302 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of Transformers regressions around custom model loading/init, quantization, attention, and model-specific bugs. I only see one near-duplicate pair: a shared `resize_token_embeddings`/`output_embeddings` regression. The rest are too model- or symptom-specific to merge safely.", + "confidence": 0.53, + "canonical_issue_reason": "Issue 43646 is the broadest umbrella report in the set: a generic Transformers 5.0 custom-model initialization breakage. It best represents the cluster\u2019s main recurring theme compared with the narrower model-specific save/load or embedding reports.", + "canonical_pr_reason": null, + "best_issue_reason": "43646 is the most representative and actionable issue because it is the most general regression description and can subsume several narrower custom-model initialization failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Both mention memory/quantization concerns, but one is Qwen3 MoE training memory usage and the other is quantized offloading behavior; different bugs and fixes." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer/model-load failures, but they affect different models with different failure modes; no clear shared code-path bug." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Related to Transformers 5.0 breakages, but one is junk non-persistent buffers and the other is Jupyter/custom model initialization; not the same underlying issue." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Different model families and crash sites: Qwen3.5 chat vs Gemma4Processor in serve; too different to be duplicates." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both are memory-related, but one is training memory efficiency and the other is int4 reserved-memory OOM; separate regressions." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both touch flash attention, but one is a packed-sequence/position_ids crash and the other is an `s_aux=None` AttributeError; different code paths." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "LoRA checkpoint compatibility vs resize_token_embeddings propagation; unrelated bugs." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for `GenerationMixin` and Whisper pipeline `return_language` regression are unrelated." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": false, + "reason": "Both concern resize_token_embeddings, but they describe different model-specific embedding surfaces; not enough evidence they are the same bug." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Similar symptom of untied `lm_head`, but they affect different model classes and are likely separate model-specific omissions." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different lifecycle bugs: extra_state save/load handling versus an unexpected `_is_hf_initialized` constructor argument." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different problems: Qwen3 auto device mapping CUDA assert vs AudioFlamingo3 batched inference token leak." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "The latter is a model load failure that may involve vocab/tokenizer issues, but the former is a specific pad/unk vocab_size mismatch; not the same bug." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Unrelated: SigLIP2 doc-example errors vs GLM5 model issue." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing/PEFT LoRA vs T5 apex RMSNorm memory leak are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version incompatibility and meta-device model loading are different regressions." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both involve memory pressure, but one is quantization offloading and the other is int4 reserved-memory growth; different fixes." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and Gemma4 FSDP2 shared_kv_states are unrelated model bugs." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states vs Qwen multimodal config unwrapping; different code paths." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Gemma4 FSDP2 bug and Qwen3_5Moe CPU offload incompatibility are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "`init_empty_weights` constructor arg issue vs Strix Halo mmap OOM; not the same bug." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "AudioFlamingo3 track leakage and Gemma4 device mismatch are separate failures." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both are memory-related, but one is int4 reserved memory and the other is mmap on Strix Halo; different root causes." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "GLM-4.7-Flash load failure in tests and Qwen3.5 save_pretrained regression are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Different quantization/load problems on different models (NVFP4 vs MXFP4); not the same concrete bug." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is ineffective LoRA checkpointing and the other is eval-mode prohibition; separate issues." + }, + { + "left": "issue:44743", + "right": "issue:44960", + "accept": false, + "reason": "Qwen recurrent-state reset bug and GLM5 issue are unrelated." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading and Strix Halo mmap OOM are different initialization/memory issues." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying `resize_token_embeddings` regression: resized embeddings are not propagated to output embeddings; 45276 is just a Gemma4-specific manifestation with an extra symptom." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/from_pretrained for custom models, but one is file permissions and the other is wrong module resolution; different bugs." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Gradient checkpointing in eval mode and Qwen3.5 3D position_ids / SDPA fallback NaNs are unrelated." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states and Qwen3_5Moe rotary embedding uninitialized memory are different model internals." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Save/load extra_state handling and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Generic custom model initialization breakage and Trainer resume max_steps calculation are separate issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted near-duplicate pair, `45276`/`45292`, is reasonably supported by the titles as the same `resize_token_embeddings`/`output_embeddings` regression with a Gemma4-specific manifestation. The other pair verdicts appropriately avoid overstating similarity across model-specific or symptom-specific issues." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 7, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:40990", + "issue:42947", + "issue:43257", + "issue:43299", + "issue:43425", + "issue:43452", + "issue:43596", + "issue:43611", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44991", + "issue:45005", + "issue:45084", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45701", + "issue:45759", + "issue:45823", + "issue:45902" + ], + "soft_pairs": [ + "issue:44479|issue:44877", + "issue:45588|issue:45663", + "issue:45292|issue:45335", + "issue:43257|issue:44863", + "issue:43645|issue:44164", + "issue:44402|issue:44451", + "issue:43828|issue:43957", + "issue:43425|issue:43828", + "issue:39404|issue:45701", + "issue:44589|issue:44756", + "issue:44683|issue:45823", + "issue:43931|issue:44863", + "issue:43906|issue:44530", + "issue:43611|issue:43645", + "issue:43425|issue:45823", + "issue:44530|issue:45084", + "issue:43425|issue:43716", + "issue:43901|issue:44509", + "issue:39404|issue:44912", + "issue:44265|issue:44898", + "issue:43299|issue:43931", + "issue:43994|issue:44079", + "issue:44479|issue:44805", + "issue:39404|issue:44671", + "issue:45005|issue:45902", + "issue:40990|issue:42947", + "issue:44589|issue:45084", + "issue:44671|issue:44991", + "issue:43425|issue:44918", + "issue:44164|issue:45759", + "issue:36032|issue:43452", + "issue:43856|issue:44756", + "issue:43596|issue:44683" + ], + "prepared_review_unit_hash": "884edecc5efa642e2df19c63c60e8b47af34aa55df8e309b431ca5e314b5dc82", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37611, + "estimated_input_tokens": 9403, + "estimated_eval_tokens": 19062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is largely heterogeneous and mostly not duplicate-related: it spans unrelated loading, dtype, quantization, docs, and model-specific regressions. The only clear near-duplicate pair is the resize_token_embeddings bug affecting output/decoder embeddings.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 45292 is the broadest, most general statement of the shared bug in the one true duplicate-like pair: resize_token_embeddings not propagating to output embeddings.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45292 is the best representative of the cluster\u2019s only concrete duplicate theme and is phrased more generally than the t5gemma-specific report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Different regressions in different model families: Qwen video-input behavior vs granite_speech config loading." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Different code paths and failures: flash_attention s_aux=None crash vs Gemma4 FSDP2 shared_kv_states KeyError." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same underlying bug: resize_token_embeddings is not updating downstream embeddings; 45335 is the t5gemma-specific manifestation." + }, + { + "left": "issue:43257", + "right": "issue:44863", + "accept": false, + "reason": "Both are checkpoint/loading complaints, but they involve different model implementations and failure modes." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Custom-model notebook initialization is unrelated to save/from_pretrained extra_state handling." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "A tokenizer vocab mismatch is not the same as a model-specific load failure for ScandiBERT." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Different bugs: autocast dtype mismatch vs meta-device loading breakage." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Generic torch version incompatibility is unrelated to the Phi MoE autocast dtype issue." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to a general tokenization change report." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap OOM issue." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention failure on torch>=2.9 is unrelated to Gemma4 device placement mismatch." + }, + { + "left": "issue:43931", + "right": "issue:44863", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch is a different loading problem than NemotronH checkpoint incompatibility." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "An isolated reproduction of an older issue is not the same bug as the Qwen3.5 PagedAttentionCache crash." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both are loading-related, but base_model_prefix breakage and Jupyter custom-model initialization are different defects." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as the Gemma4 PLE device-map forward mismatch." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3.5 linear_attention cache crash is unrelated to the non-template-node compilation TypeError." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Version incompatibility and image-preprocessor dtype mismatch are different issues." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs issues, but they cover different removed/deprecated pipeline docs and are not the same change." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline behavior is unrelated to MXFP4 quantization fallback on gpt-oss-20b." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure on torch_compilable_check is unrelated to Perceiver non-default-resolution classification failure." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Different Qwen3-VL loading problems with different symptoms: model-specific v5 breakage vs weight-shape mismatch." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Incorrect predictions for SigLIP2 are unrelated to ModelOutput key assignment logic." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression for Qwen models is unrelated to the mask/tensor shape IndexError." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline regression and CamemBERT masked-LM prediction issues are distinct." + }, + { + "left": "issue:45005", + "right": "issue:45902", + "accept": false, + "reason": "Tied-weights translation-model loading and Qwen3.5 rotary embedding materialization are different bugs." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Perplexity regression on gpt-oss-20b is unrelated to gradient checkpointing with PEFT LoRA." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to the compile-time template-node error." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT prediction regression and tokenizer-loading failure for est-roberta are different defects." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility is unrelated to TRL SFT unpacking of Qwen3.5 embeddings." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "extra_state save/load handling is unrelated to composite Qwen config unwrapping." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer special-token conflict is unrelated to gguf_file breaking AutoTokenizer/AutoModel loading." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "MoE training memory inefficiency is unrelated to the Strix Halo mmap OOM workaround." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO-3 init IndexError is unrelated to compiled flex_attention on newer torch." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The one accepted soft pair (45292/45335) is well-supported by the titles, and the rejected pairs do not overstate duplicate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 8, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42915", + "issue:43262", + "issue:43317", + "issue:43377", + "issue:43404", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43541", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43883", + "issue:43994", + "issue:44164", + "issue:44206", + "issue:44220", + "issue:44451", + "issue:44466", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44610", + "issue:44661", + "issue:44683", + "issue:44737", + "issue:44811", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45507", + "issue:45584", + "issue:45859" + ], + "soft_pairs": [ + "issue:32090|issue:39404", + "issue:44991|issue:45081", + "issue:44737|issue:45859", + "issue:44164|issue:44479", + "issue:43582|issue:43632", + "issue:44488|issue:45081", + "issue:40990|issue:43421", + "issue:44661|issue:44683", + "issue:43611|issue:43644", + "issue:44488|issue:44987", + "issue:44206|issue:44220", + "issue:43716|issue:45084", + "issue:43643|issue:44530", + "issue:43883|issue:44488", + "issue:34567|issue:40444", + "issue:44811|issue:45584", + "issue:43644|issue:43646", + "issue:36032|issue:42915", + "issue:43452|issue:43482", + "issue:44451|issue:44987", + "issue:43541|issue:45084", + "issue:42915|issue:44530", + "issue:44849|issue:45325", + "issue:43317|issue:43482", + "issue:36032|issue:43645", + "issue:43404|issue:44466", + "issue:44987|issue:44991", + "issue:43883|issue:44991", + "issue:43262|issue:43377", + "issue:45084|issue:45507", + "issue:43646|issue:44534", + "issue:43994|issue:44610", + "issue:43645|issue:44534" + ], + "prepared_review_unit_hash": "3be1cac12bda73ff62d503adb74c18009a057c5a32e66b0c1bb83e67e3b38df9", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37700, + "estimated_input_tokens": 9425, + "estimated_eval_tokens": 19106 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous, with issues spanning unrelated Transformers subsystems. I found one likely duplicate pair around GGUF loading in v5; the rest are distinct bugs and should stay separate.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue for the whole set: the items cover many unrelated regressions across training, tokenizers, audio, vision, quantization, and model loading.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43452 is the strongest canonical issue candidate among the near-duplicates because it describes the broader GGUF loading regression directly, while issue:43482 looks like a specific model instance of the same failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Different failures: GPU broadcast of NoneType in Trainer vs Whisper pipeline return_language behavior." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Both involve loading/tokenizer code, but one is a tokenizer load failure for a specific model and the other is a Mistral regex patch/backend_tokenizer error." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "Different code paths: XLNet positional encoding device placement vs Qwen3.5-MoE CPU-offload incompatibility." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "Save/load extra_state handling is unrelated to the Qwen video-input regression." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and the _is_hf_initialized regression are separate bugs." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Model-loading failure for cjvt/sleng-bert is unrelated to the Mistral tokenizer backend_tokenizer crash." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression on gpt-oss is not the same as runtime post-processor updates in TokenizersBackend." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Tokenizer mapping-name bug and compiled flex_attention failure are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; different bugs." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading reports, but they concern different checkpoints and failure modes." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Same audio area, but one is an unsupported center argument and the other is a _torch_extract_fbank_features issue; not clearly the same bug." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Image-preprocessor dtype mismatch is unrelated to the template-node compilation TypeError." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "Missing trust_remote_code fields and invalid group type in PagedAttentionCache are different problems." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing all_tied_weights_keys and loading cjvt/sleng-bert are unrelated." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "TrainerState token counter issue and Qwen2.5-VL iterable dataset failure are distinct." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Both are Whisper-related, but one is batch_decode skip_special_tokens handling and the other is empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Non-persistent buffer junk and custom model initialization regressions are separate v5 issues." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer special-token conflict is unrelated to Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": true, + "reason": "Both point to the same GGUF loading regression in transformers v5; 43482 looks like a specific model manifestation of the broader 43452 failure." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Different model-loading failures for different checkpoints and different triggers." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Torch dynamo grouped_mm tracing failure is unrelated to the template-node compilation error." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "FineGrainedFP8Config failure and linear_attention group-type crash are separate Qwen issues." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Output_hidden_states bug and RoPE temporal position scaling bug are different Qwen3.5/Qwen2.5-VL problems." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "gpu+cpu offload dequantized-model loading and GGUF loading regression are different code paths." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method-name conflict and notebook custom-model initialization failure do not match." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Mistral3 lm_head tying bug and device-dependent lm_head serialization are related in theme but not the same concrete bug." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Separate versioned loading failures for different models/tokenizers; not the same underlying regression." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Missing all_tied_weights_keys is unrelated to est-roberta tokenizer loading." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio processor chat-template sampling-rate default and MIMI batched-vs-single padding-mask bug are different." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template compilation TypeError and GraniteMoEHybrid invalid-method calls are different failures." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Custom model init breakage and junk-filled non-persistent buffers are separate v5 regressions." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo 640px-vs-224px preprocessing mismatch are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Notebook custom-model initialization issue does not match the non-persistent-buffer regression." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: this set is clearly heterogeneous, and the lone GGUF-loading pair (43452/43482) is a reasonable near-duplicate call based on the titles. The other soft-edge rejections are conservative and do not overclaim equivalence from theme-level similarity alone." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 9, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43296", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43421", + "issue:43425", + "issue:43531", + "issue:43540", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43742", + "issue:43819", + "issue:44062", + "issue:44079", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44462", + "issue:44479", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44610", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44877", + "issue:44987", + "issue:45081", + "issue:45084", + "issue:45127", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45542", + "issue:45663", + "issue:45812" + ], + "soft_pairs": [ + "issue:43322|issue:43540", + "issue:44403|issue:45081", + "issue:43377|issue:43819", + "issue:42915|issue:45084", + "issue:44265|issue:44610", + "issue:45216|issue:45357", + "issue:44451|issue:45081", + "issue:44987|issue:45081", + "issue:44220|issue:44683", + "issue:43404|issue:45127", + "issue:43632|issue:44756", + "issue:42915|issue:44291", + "issue:44062|issue:44987", + "issue:44479|issue:45325", + "issue:36032|issue:43742", + "issue:43296|issue:43531", + "issue:43611|issue:44534", + "issue:42915|issue:44589", + "issue:40444|issue:45325", + "issue:43531|issue:44877", + "issue:40990|issue:44811", + "issue:36032|issue:41720", + "issue:44079|issue:45663", + "issue:43425|issue:44403", + "issue:43632|issue:44534", + "issue:41553|issue:43582", + "issue:44462|issue:45812", + "issue:42175|issue:45542", + "issue:43421|issue:44568" + ], + "prepared_review_unit_hash": "e02cc04075cc0f6db45a989660e71a209e565021defc729cfcc2cf37be2ac786", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35801, + "estimated_input_tokens": 8951, + "estimated_eval_tokens": 18158 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of unrelated Transformers bugs. The only clear duplicate-like pair is the Qwen3.5 `save_pretrained` regression, where the later issue narrows the same incorrect-checkpoint problem to visual encoder keys.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45216 is the broadest, earliest report of the Qwen3.5 `save_pretrained` serialization regression; 45357 appears to be a later, narrower symptom of the same bug.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45216 is the best representative issue for the one confirmed duplicate pair in this set and has the clearest, most general description of the shared bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Different multimodal model families and different failure modes (segfault on Llava Next loading vs ValueError on Qwen3OmniMoe video processing)." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "One is a generic loading-noise complaint; the other is a specific Mistral tokenizer regex crash. Not the same code-path bug." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Both are audio-related, but one is a padding-mask batching mismatch in MIMI and the other is a DAC STE/from_latents mismatch. Different bugs." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3Moe FP8 failure and a template-node compilation TypeError are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "Torch export failure with `torch_compilable_check` is unrelated to OmDet-Turbo processor image-size mismatch." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same Qwen3.5 `save_pretrained` checkpoint regression; the later issue specifically calls out incorrect visual encoder keys, which is a narrower manifestation of the same bug." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Different models and different failures: ScandiBERT load failure vs Mistral regex patch crash." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different affected models and separate loading paths; not the same underlying bug." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Feature-extraction helper issue and compiled flex_attention torch-version incompatibility are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Weight-tying bug in Mistral3 is unrelated to LoRA merging with extended vocab collapse." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "`_is_hf_initialized` regression and mmap OOM on Strix Halo are different problems." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "Both mention loading/training regressions, but one is Qwen3Moe FP8-specific and the other is an `init_empty_weights` argument error." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "Tokenizer added-token argument collision is unrelated to loading `physical-intelligence/fast`." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Both touch Qwen VL rope/vision logic, but one is a broad video-input regression across several models and the other is a specific still-image temporal position scaling bug." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "T5 special-token conflict and MobileLLM key error are different tokenizer/model-loading issues." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding-window issue are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Base-model-prefix loading breakage and non-persistent buffer junk-filling are different v5 regressions." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and Float8 storage TypeError are distinct errors." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen2.5-VL bugs: iterable-dataset multi-image finetuning failure vs rope-index temporal-position scaling." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding-window issue and Granite speech strict-config loading are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Perplexity on GPT-OSS and Whisper batch_decode special-token handling are not the same bug." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer load conflict and Qwen3 auto device-mapping cuda assert are unrelated." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment bug and Gemma4 FSDP2 shared_kv_states KeyError are different issues." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch version compatibility and noisy loading messages are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Different v5 regressions with different symptoms and code paths." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral and Apple Silicon warmup TypeError are unrelated." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Both involve AutoTokenizer regressions, but one is about ignoring repository `tokenizer.json` and the other is a Granite-specific wrong-token-IDs regression; too broad to treat as the same bug without stronger evidence." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both concern TensorFlow/backend packaging, but one is missing tf in the torch extra while the other is a tensorboard-without-TF runtime backend error." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime special-token post-processor updates and `add_special_tokens=True` BOS/EOS behavior are distinct tokenizer bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The Qwen3.5 save_pretrained pair is the only plausible duplicate-like match from the packet, and the other soft-edge rejections stay appropriately cautious." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 10, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43454", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43646", + "issue:43701", + "issue:43749", + "issue:43792", + "issue:43818", + "issue:43856", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44387", + "issue:44403", + "issue:44466", + "issue:44534", + "issue:44589", + "issue:44617", + "issue:44683", + "issue:44829", + "issue:44898", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45491", + "issue:45538", + "issue:45800", + "issue:45834" + ], + "soft_pairs": [ + "issue:43317|issue:43856", + "issue:43818|issue:45357", + "issue:33666|issue:40444", + "issue:43792|issue:44589", + "issue:44829|issue:45479", + "issue:44403|issue:44991", + "issue:44683|issue:45020", + "issue:43482|issue:45538", + "issue:35707|issue:40444", + "issue:44079|issue:44534", + "issue:44291|issue:44589", + "issue:43482|issue:43525", + "issue:43377|issue:45491", + "issue:43749|issue:44079", + "issue:43644|issue:44079", + "issue:40444|issue:45381", + "issue:41720|issue:45081", + "issue:43531|issue:44291", + "issue:44265|issue:45072", + "issue:43632|issue:45072", + "issue:45325|issue:45381", + "issue:43611|issue:43646", + "issue:43994|issue:44898", + "issue:42175|issue:43296", + "issue:43454|issue:44466", + "issue:45325|issue:45834", + "issue:42915|issue:43482", + "issue:43792|issue:43957", + "issue:44387|issue:44617", + "issue:43296|issue:45800", + "issue:43701|issue:45325" + ], + "prepared_review_unit_hash": "0e0781c0851a795e07924fcd1f7d7639a26c4a88457fb8de10d46bb964442d9a", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36997, + "estimated_input_tokens": 9250, + "estimated_eval_tokens": 18756 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of distinct model-specific Transformer regressions, with one clear duplicate pair around the non-persistent buffer junk bug. Most soft links are only broad similarity (same model family, same subsystem, or same Transformers v5 regression theme) and should not be merged.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43644 is the earlier, clearer report of the non-persistent-buffer regression, and issue:44534 appears to be the same bug with nearly identical wording. It is the best canonical issue for that duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest canonical issue here because it is specific, concrete, and the earliest of the near-identical reports; it best represents the underlying bug among the provided items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Different failures: dequantized model loading on CPU/GPU offload vs Qwen3 MoE training memory usage." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Both are vision-model related, but they describe different bugs on different models: Video-LLaVA tower weights/attention vs Qwen3.5 save_pretrained key serialization." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Different Qwen2-VL training problems: multi-GPU training vs iterable dataset with multiple images per prompt." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Unrelated symptoms: Whisper model runtime failure vs float8 storage lookup error." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both affect training, but one is a flash_attention_3 regression and the other is a classification-loss bug for num_labels=1." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Noise on model load vs tokenizer loading failure for a specific model; not the same code-path bug." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "Different issues: compiled flex_attention on torch >=2.9 vs remote_code model breakage." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer model_max_length behavior are unrelated." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to multi-image finetuning failures." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer junk are different regressions." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized argument handling is unrelated to the float8 storage object error." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Different models and symptoms: GGUF loading vs missing pad_token_id on Llama4Config." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both are batching/shape-related, but MIMI padding-mask output mismatch is not the same as Gemma3 sliding-window NaNs." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and ModelOutput key assignment are separate bugs." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer junk regression is unrelated to ModelOutput key assignment." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both involve Qwen2.5-VL, but one is multi-image finetuning and the other is video vision_position_ids." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Auto device-map CUDA assert on Qwen3 is unrelated to Mistral tokenizer regex patch failure." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument handling are different bugs." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export with torch_compilable_check is unrelated to dtype mismatch in bfloat16 inference." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "_is_hf_initialized flag regression and dtype mismatch are different failures." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Closely related Qwen2.5-VL position-id topics, but one is still-image temporal scaling and the other is video input vision_position_ids; not clearly the same bug." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are broad Transformers v5 breakages, but they describe different aspects of model loading/initialization and are too general to merge confidently." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different vision-model failures: SigLIP2 nonsensical outputs vs Perceiver non-default resolution classification failure." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Packaging/backend availability issue vs PaddleOCR-VL loading failure in vLLM." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Tied lm_head weights causing bad generation and inconsistent serialization of lm_head.weight are related but not the same concrete bug." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Different models and failures: Qwen2.5-VL rope indexing vs Kosmos2.5 OCR index error." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Qwen3Moe FP8 config failure is unrelated to Qwen2.5-GGUF loading on Transformers v5." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper runtime failure and torch.device('meta') model loading regression are different problems." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Int4 quantization memory regression is unrelated to Sam3Video OOM." + }, + { + "left": "issue:43296", + "right": "issue:45800", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and torch 2.4.1/transformers 5.8 incompatibility are not the same bug." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint key mismatch and Qwen2.5-VL rope_index position-id bug are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both mention Transformers v5 load/init regressions, but the titles point to different concrete breakages and not one shared code-path." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Near-identical reports of the same regression: Transformers v5 fills non-persistent buffers with junk." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 11, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43819", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44008", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44466", + "issue:44479", + "issue:44488", + "issue:44534", + "issue:44610", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45357", + "issue:45584", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45759", + "issue:45823" + ], + "soft_pairs": [ + "issue:44479|issue:45823", + "issue:43881|issue:44488", + "issue:44291|issue:44877", + "issue:43262|issue:43746", + "issue:43606|issue:44265", + "issue:40990|issue:45584", + "issue:34634|issue:35707", + "issue:43873|issue:45701", + "issue:44186|issue:44265", + "issue:43317|issue:43873", + "issue:35707|issue:41720", + "issue:43606|issue:43819", + "issue:43454|issue:43550", + "issue:41720|issue:45823", + "issue:43377|issue:44610", + "issue:43454|issue:45072", + "issue:43377|issue:44898", + "issue:39692|issue:40444", + "issue:43572|issue:45759", + "issue:43526|issue:43883", + "issue:43749|issue:44534", + "issue:44466|issue:45357", + "issue:43454|issue:43720", + "issue:43881|issue:44991", + "issue:41720|issue:43322", + "issue:43296|issue:44877", + "issue:41720|issue:44977", + "issue:40990|issue:45701", + "issue:44008|issue:45823", + "issue:45632|issue:45684", + "issue:43404|issue:45357", + "issue:44186|issue:45072", + "issue:42617|issue:42915" + ], + "prepared_review_unit_hash": "c56c0f7a47573ef5e6c54dfa445deb998b25ed4f241d6eb3ae85b457dbfbaf7d", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 37164, + "estimated_input_tokens": 9291, + "estimated_eval_tokens": 18838 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is highly heterogeneous. The proposed soft edges mostly share broad subsystem words like offloading, device mapping, serialization, or multimodal models, but they do not appear to describe the same underlying bug or change. I would reject all candidate merges.", + "confidence": 0.96, + "canonical_issue_reason": "If a representative issue must be chosen, issue 43873 is the broadest and most cluster-central because it captures the recurring offloading/quantization theme; however, the cluster is too diverse for a true canonical duplicate anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the best overall representative among these mostly unrelated reports because it is broad, open, and closest to the repeated offloading/device-placement subset, though it is not a strong duplicate canonical.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Different models and failure modes: video-input regression in Qwen VL family vs Gemma4 PLE device mismatch." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Unrelated loading failures for different models; one is glm-4v-9b, the other a tokenizer load issue for a BERT model." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "Different root causes: unexpected `_is_hf_initialized` argument during init-empty-weights vs strict config rejection for granite_speech." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sampling-rate default bug is unrelated to GraniteSpeech PEFT local checkpoint loading." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch for Bark-small is unrelated to torch.export failing on torch_compilable_check." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "High perplexity on gpt-oss is a modeling/eval issue, not the Whisper empty-transcription generation bug." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "BarkProcessor voice_preset and progressive generation with inputs_embeds/past_key_values are separate code paths." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Offloading/quantization failure is unrelated to tokenizer behavior changing across transformer versions." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer padding/truncation crash is unrelated to torch.export and torch_compilable_check." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both mention offloading, but one is a specific dequantized-model load failure and the other is a broader quantization/offloading complaint; not the same bug." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with inputs_embeds is unrelated to Qwen3 auto device-map CUDA assertion." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "Bark CPU offload device mismatch and DAC.from_latents STE mismatch are different model-specific issues." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "lm_head weight tying bug is unrelated to Bamba torch.compile SDPA failure." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both involve device placement, but the errors and models differ materially: CUDA assert on Qwen3 vs Gemma4 PLE forward mismatch." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "MIMI padding-mask batching bug is unrelated to OmDet-Turbo processor/model input-size mismatch." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "Weight-tying/generation bug is unrelated to dtype mismatches in SwitchTransformers and TimmWrapperModel." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Different subsystems and causes: missing padding-mask support vs Perceiver non-default-resolution failure." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Documentation/example errors for SigLIP2 are unrelated to Qwen2.5-VL finetuning with multi-image iterable datasets." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Missing `pad_token_idx` in StableLmConfig is unrelated to `from_config` not unwrapping `text_config` for Qwen multimodal configs." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Molmo missing `all_tied_weights_keys`." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are distinct v5 regressions." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both concern serialization, but one is tied lm_head.weight behavior and the other is incorrect visual encoder keys for Qwen3.5." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "lm_head tying bug is unrelated to BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Model loading failure for glm-4v-9b is unrelated to tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Different models and failures: Qwen3 CUDA assert vs Llava Next segmentation fault." + }, + { + "left": "issue:43296", + "right": "issue:44877", + "accept": false, + "reason": "PaddleOCR-VL vLLM load failure is unrelated to granite_speech strict-config loading rejection." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Both are Qwen-family reports, but one is an auto device-map CUDA assert and the other is a flash-attention generation regression." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Gemma 3n variable-name collision and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Cache-path collision for trust_remote_code is unrelated to permission propagation in save_pretrained/register_for_auto_class." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Mistral3 lm_head tying bug is unrelated to Qwen3.5 visual encoder key serialization regression." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crash on NER/padding is unrelated to bfloat16 dtype mismatches in model inference." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "3d_parallel.py execution problem and Qwen3Moe FineGrainedFP8Config failure are different issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 12, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36010", + "issue:36331", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43450", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43708", + "issue:43720", + "issue:43782", + "issue:43873", + "issue:43881", + "issue:44186", + "issue:44361", + "issue:44387", + "issue:44423", + "issue:44479", + "issue:44492", + "issue:44509", + "issue:44610", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44918", + "issue:44991", + "issue:45059", + "issue:45081", + "issue:45161", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704", + "issue:45812" + ], + "soft_pairs": [ + "issue:43577|issue:43708", + "issue:43873|issue:44756", + "issue:44387|issue:45704", + "issue:44423|issue:45406", + "issue:44877|issue:44991", + "issue:36331|issue:42617", + "issue:43550|issue:43720", + "issue:43450|issue:45701", + "issue:36331|issue:42915", + "issue:44610|issue:44898", + "issue:44186|issue:44610", + "issue:44479|issue:44918", + "issue:43550|issue:43606", + "issue:41720|issue:42617", + "issue:42915|issue:43317", + "issue:43316|issue:44913", + "issue:43295|issue:45701", + "issue:44671|issue:44743", + "issue:45081|issue:45812", + "issue:43526|issue:43881", + "issue:41720|issue:45161", + "issue:43531|issue:43572", + "issue:43421|issue:45812", + "issue:43540|issue:43782", + "issue:36010|issue:42617", + "issue:44186|issue:44361", + "issue:42915|issue:43322", + "issue:43388|issue:45059", + "issue:45406|issue:45464", + "issue:43421|issue:45701", + "issue:44492|issue:44509" + ], + "prepared_review_unit_hash": "0c8ff351c40c42985369f6ad679f03dab3a5f727c6d49cb8c9492ab9df2caf20", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37022, + "estimated_input_tokens": 9256, + "estimated_eval_tokens": 18768 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues are mostly unrelated Transformers regressions spread across trainers, tokenizers, multimodal processors, quantization, offloading, and config loading. The soft pairs only share broad subsystems or similar symptoms, not the same concrete underlying bug, so all should be rejected. No single canonical issue clearly represents the whole set.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Different bugs: one is model dtype initialization on Blip2 load, the other is Trainer checkpoint resume step \u0917\u0923ulation." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Both mention offloading/memory, but one is quantization offload behavior and the other is mmap-related OOM on Strix Halo; different code paths." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "One is int4 quantization causing higher CUDA reserved memory; the other is a T5/apex RMSNorm memory leak. Not the same defect." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve/multimodal crashes, but they involve different models and different attribute errors in different processor paths." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Config strictness for granite_speech and tokenizer loading failure for est-roberta are unrelated failures." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer compute_loss signature regression is unrelated to running 3d_parallel.py." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Different failures: SDPA with torch.compile in Bamba versus packed-weight handling during accelerate loading for BitNet." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video batch-shape bug is unrelated to a tokenization-version regression." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "CustomTrainer compute_loss kwargs issue is unrelated to Qwen3MoE + FineGrainedFP8Config loading." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both are image-size/preprocessing issues, but they affect different models and different mismatch logic." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash is a tokenizer bug, not the OmDet-Turbo processor image-size bug." + }, + { + "left": "issue:44479", + "right": "issue:44918", + "accept": false, + "reason": "A video-input regression in Qwen VL models is not the same as Qwen3.5 embedding unpacking with TRL SFT." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Both are loading/execution failures, but one is SDPA/compile in Bamba and the other is CPU-offload device mismatch in bark-small." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping cuda assert is unrelated to failing to run 3d_parallel.py." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Qwen3MoE FP8 config failure and dequantized-model GPU+CPU offload failure are different loading/offload bugs." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "API discrepancy in Gemma3TextConfig and GPTNeoX rotary_pct reload behavior are separate configuration issues." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "Processor/tokenizer API regression and generic tokenization-version change are not the same concrete bug." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM prediction regression is unrelated to Qwen3.5 recurrent state reset with cache." + }, + { + "left": "issue:45081", + "right": "issue:45812", + "accept": false, + "reason": "Mistral regex patch crash in tokenizer loading is a different failure from Granite token-ID regression." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BEiT label reduction bug and glm-4v-9b loading failure do not share the same root cause." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both involve Qwen/GPT-OSS MoE models, but one is an A800 auto-device-map assert and the other is TP-only support not working." + }, + { + "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:43421", + "right": "issue:45812", + "accept": false, + "reason": "Special-token post-processor update behavior and Granite token ID regression are different tokenizer issues." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Video-input ValueError in Qwen3OmniMoe is unrelated to Qwen3VL weight_only loading error." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import error and 3d_parallel.py execution failure are distinct issues." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer-related, but one is LayoutLMv2 NER/batching crash and the other is MLukeTokenizer task AttributeError." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3MoE FP8 loading failure and Llava Next segmentation fault are different model-loading bugs." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Metric-gathering label truncation and SAM3 text/bbox odd behavior are unrelated." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Both hit serve/chat multimodal paths, but they are different models with different failure modes and code paths." + }, + { + "left": "issue:43421", + "right": "issue:45701", + "accept": false, + "reason": "Runtime special-token/post-processor updates are not the same as a version-induced tokenization change." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A typo in cache strategies is unrelated to stale docs mentioning removed pipeline tasks." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 13, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:43381", + "issue:43421", + "issue:43531", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:44062", + "issue:44077", + "issue:44186", + "issue:44368", + "issue:44451", + "issue:44514", + "issue:44545", + "issue:44589", + "issue:44610", + "issue:44625", + "issue:44779", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45588", + "issue:45704", + "issue:45759", + "issue:45812", + "issue:45859" + ], + "soft_pairs": [ + "issue:33357|issue:35707", + "issue:43606|issue:43720", + "issue:41720|issue:43746", + "issue:43550|issue:43653", + "issue:30064|issue:34634", + "issue:44779|issue:45812", + "issue:43873|issue:45704", + "issue:44514|issue:44545", + "issue:44625|issue:45479", + "issue:43606|issue:43653", + "issue:43531|issue:45759", + "issue:43421|issue:43927", + "issue:34634|issue:41720", + "issue:34567|issue:35707", + "issue:44368|issue:45357", + "issue:43782|issue:45381", + "issue:44898|issue:45072", + "issue:44451|issue:44991", + "issue:44077|issue:44625", + "issue:45759|issue:45859", + "issue:44987|issue:45042", + "issue:43531|issue:44589", + "issue:44918|issue:44936", + "issue:43421|issue:43653", + "issue:44062|issue:44451", + "issue:45588|issue:45759", + "issue:43381|issue:43844", + "issue:39692|issue:44610", + "issue:42886|issue:44843", + "issue:43421|issue:44186" + ], + "prepared_review_unit_hash": "9e976e45e04cad27c62deb18763837515b7bdc8cdf504b16455521de0df7246a", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36329, + "estimated_input_tokens": 9083, + "estimated_eval_tokens": 18422 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a loose set of unrelated bug reports that only share broad subsystem overlap (tokenizers, model loading, multimodal processors, offload, training). There is one clear duplicate pair: the Qwen2_5_VLProcessor.apply_chat_template batched-input crash.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44514 is the best canonical issue because it is the earlier report of the exact same batched-input crash in Qwen2_5_VLProcessor.apply_chat_template, with essentially identical reproduction conditions to issue:44545.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44514 is the strongest representative issue in this cluster since it is a clean, specific bug report with a direct duplicate and a well-scoped failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "Different failures: MacOS bus error with a CLIP model vs progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Both are CI/offload-related, but they describe different code paths and failures (device mismatch vs packed-weight unpacking during loading)." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Different models and symptoms: Qwen3 auto device map CUDA assert vs GraniteSpeech local checkpoint/PEFT loading." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is unrelated to BigBirdTokenizer mask-token decode behavior." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Different processors and bugs: void segmentation map handling vs Bark voice_preset not working." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both concern tokenizer regressions, but they affect different model families and appear to stem from different bugs." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Offloading with quantization is not the same as T5 silently using apex.FusedRMSNorm and leaking memory." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same underlying bug: identical Qwen2_5_VLProcessor.apply_chat_template crash on batched input when padding=False." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Different configuration bugs: Qwen3.5 num_labels propagation vs degenerate zero loss for single-label classification." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Device mismatch during CPU offload is unrelated to BigBirdTokenizer special-token registration and decode output." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is not the same as missing text_config unwrapping in composite Qwen 3.5/3.6 configs." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Runtime post-processor update for special tokens is a different tokenizer feature from DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping failures are unrelated." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token counting and progressive generation with cache/embeds are unrelated training/inference bugs." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "A warning about tie_word_embeddings is not the same as incorrect visual-encoder keys being saved in Qwen3.5." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Weight-only loading error for Qwen3VL is unrelated to incorrect vision_position_ids in qwen2.5-vl video input." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different failures: Perceiver non-default resolution handling vs dtype mismatches in SwitchTransformers/TimmWrapperModel bfloat16 inference." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer loading regressions, but they target different models and likely different causes." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "patchtsmixer post_init policy is unrelated to Qwen3.5 num_labels propagation." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Missing text_config unwrapping is not the same as CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding.forward." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Model loading regression for physical-intelligence/fast is unrelated to PIL backend processors requiring torchvision." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding-window issue for Qwen3-MoE is unrelated to missing Float8 storage type resolution." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "trl SFT input-embedding unpacking and trainer.evaluate() after trainer.train() are different trainer workflows." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Special-tokens runtime update behavior is a different bug than BigBirdTokenizer mask-token registration." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken constructor TypeError is unrelated to loading ScandiBERT." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Flash attention sink-less-model crash is unrelated to Qwen 3.5/3.6 config unwrapping." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Gradient checkpointing in eval mode and exploding gradients under ZeRO-3 are different training issues." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Documentation example errors for SigLIP2 are unrelated to OmDet-Turbo processor output-size mismatch." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Both are HF_HUB_OFFLINE-related, but one is cache loading for tokenizers and the other is an unconditional model_info call in mistral regex patching." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Runtime special-token/post-processor updates and LayoutLMv2 NER padding/truncation crashes are different tokenizer bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The one accepted duplicate pair, issue:44514 vs issue:44545, is well supported by the identical titles and the metadata showing references; the rest of the soft-edge verdicts are appropriately marked false without overclaiming. Minor phrasing like \"essentially identical reproduction conditions\" is a bit stronger than the visible metadata alone, but not enough to undermine the overall assessment." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 14, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:43295", + "issue:43322", + "issue:43335", + "issue:43421", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43596", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43716", + "issue:43720", + "issue:43746", + "issue:43756", + "issue:43818", + "issue:43819", + "issue:43873", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44488", + "issue:44661", + "issue:44671", + "issue:44811", + "issue:44843", + "issue:44898", + "issue:45072", + "issue:45081", + "issue:45183", + "issue:45356", + "issue:45381", + "issue:45479", + "issue:45685", + "issue:45704", + "issue:45834", + "issue:45859", + "issue:45907" + ], + "soft_pairs": [ + "issue:43994|issue:44671", + "issue:45704|issue:45859", + "issue:43525|issue:43531", + "issue:45081|issue:45356", + "issue:41720|issue:45704", + "issue:45356|issue:45907", + "issue:39692|issue:43873", + "issue:43746|issue:43873", + "issue:40444|issue:43746", + "issue:43596|issue:44220", + "issue:43746|issue:45381", + "issue:43540|issue:44008", + "issue:43720|issue:43819", + "issue:43618|issue:43697", + "issue:45183|issue:45479", + "issue:44220|issue:44661", + "issue:43653|issue:43720", + "issue:30064|issue:36010", + "issue:43335|issue:43756", + "issue:43295|issue:45834", + "issue:43716|issue:44008", + "issue:43550|issue:45072", + "issue:43550|issue:43819", + "issue:44062|issue:44488", + "issue:43322|issue:45685", + "issue:44186|issue:44898", + "issue:44843|issue:45081", + "issue:43577|issue:43818", + "issue:43421|issue:44811", + "issue:43531|issue:43643", + "issue:43653|issue:43819" + ], + "prepared_review_unit_hash": "de36c963ba6b79a9602c92128260190d83c0597bf27bad482689aca04e70663e", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37471, + "estimated_input_tokens": 9368, + "estimated_eval_tokens": 18992 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is highly heterogeneous: it spans unrelated bugs in tokenizers, multimodal processors, quantization/offload, model configs, and CI/documentation. None of the soft pairs look like the same underlying issue, so there is no duplicate cluster to collapse.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue: the items do not share one concrete code path or failure mode, so treating any one as the cluster representative would be misleading.", + "canonical_pr_reason": null, + "best_issue_reason": "No clear best issue for the cluster as a whole, because the issues are unrelated rather than variants of one bug. If forced, a broad/high-activity item could be picked, but it would not be representative.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different subsystems: model/pipeline output correctness vs tokenizer-mapping/model addition logic." + }, + { + "left": "issue:45704", + "right": "issue:45859", + "accept": false, + "reason": "Unrelated model/runtime bugs: T5 Apex RMSNorm memory leak vs Qwen3 MoE CPU offload incompatibility." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Different failures: missing pad_token_id on Llama4Config vs sliding_window handling in Qwen3-MoE." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both involve tokenizer regex handling, but they are different concrete bugs with different triggers and symptoms." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "Different model/code-path issues: Qwen3 device mapping assert vs T5 Apex memory leak." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression vs repo file listing API forwarding; no shared underlying bug." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "Documentation/model example errors are unrelated to quantization offloading behavior." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Local checkpoint/PEFT loading failure is a different path from quantization offload problems." + }, + { + "left": "issue:40444", + "right": "issue:43746", + "accept": false, + "reason": "Multimodal dataset processing failure vs checkpoint loading with PEFT adapters." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed/BertModel init failure and audio fbank feature extraction are separate bugs." + }, + { + "left": "issue:43746", + "right": "issue:45381", + "accept": false, + "reason": "GraniteSpeech loading issue is unrelated to Qwen2.5-VL video position-id errors." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Different multimodal models and different failures: video input ValueError vs Gemma3n name collision." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Packed-weight loading in BitNet is unrelated to DAC.from_latents / STE mismatch." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "Both are output regressions, but for different models and different concrete code paths." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Input validation messaging bug vs classification loss degeneracy; not the same issue." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction failure and tokenizer-mapping model registration failure are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBird special-token registration and BitNet weight unpacking are different bugs." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Image segmentation-map processing and GenerationMixin import error are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Different model architectures and failures; no shared concrete defect." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Processor/tokenizer regression and Kosmos2.5 OCR index error are different multimodal bugs." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Image preprocessor dtype mismatch is unrelated to Gemma3n audio-mask variable collision." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is different from bfloat16 dtype mismatches in other models." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Different models and failure modes: compile+SDPA vs DAC latent conversion mismatch." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken keyword collision and sleng-bert loading are not the same underlying bug." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Llava Next loading segfault and MPS histogram-on-int error are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer NER/padding crash vs Perceiver image-resolution failure are different paths." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Same helper function, but different defects: offline model_info call vs backend_tokenizer attribute crash." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "Blip2 dtype loading bug and Video-LLaVA temporal-attention/weight-sharing issue are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Runtime special-token post-processing update bug is different from Whisper batch_decode skip_special_tokens handling." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to trust_remote_code config field loss." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "BigBird tokenizer special-token registration and DAC latent conversion are unrelated bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 15, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30064", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:42175", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43643", + "issue:43644", + "issue:43720", + "issue:43749", + "issue:43782", + "issue:43825", + "issue:43827", + "issue:43994", + "issue:44060", + "issue:44186", + "issue:44291", + "issue:44368", + "issue:44442", + "issue:44448", + "issue:44479", + "issue:44485", + "issue:44509", + "issue:44561", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44821", + "issue:44877", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45381", + "issue:45406", + "issue:45584", + "issue:45701", + "issue:45834" + ], + "soft_pairs": [ + "issue:43643|issue:44877", + "issue:43540|issue:45381", + "issue:43782|issue:44977", + "issue:44448|issue:44671", + "issue:45584|issue:45834", + "issue:43644|issue:43749", + "issue:43825|issue:43994", + "issue:44485|issue:44509", + "issue:44561|issue:45020", + "issue:39692|issue:45834", + "issue:44821|issue:45084", + "issue:41720|issue:44991", + "issue:30064|issue:36331", + "issue:44186|issue:44442", + "issue:43825|issue:43827", + "issue:43577|issue:43582", + "issue:44479|issue:45834", + "issue:44805|issue:44936", + "issue:44671|issue:45701", + "issue:43606|issue:45072", + "issue:34634|issue:36010", + "issue:44610|issue:45072", + "issue:43782|issue:45325", + "issue:43720|issue:45072", + "issue:42175|issue:45081", + "issue:45081|issue:45406", + "issue:43749|issue:44291", + "issue:44060|issue:44368" + ], + "prepared_review_unit_hash": "44075519b3a63e5f0ddd475f221da71d0e3016f8e8185d7a215a705b4ac1a3e9", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35677, + "estimated_input_tokens": 8920, + "estimated_eval_tokens": 18096 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous set of Transformers bug reports. The only near-duplicate pair is the v5 pipeline-removal docs/error-message fallout; the rest are distinct model-, tokenizer-, loading-, or dtype-specific regressions. No PRs were provided.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45020 is the broadest and most representative umbrella report in the set: it covers recent-version regressions affecting `remote_code` models, has active discussion, and can anchor the cluster better than the narrower model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the best overall issue to keep as the canonical tracker because it is the most general and central report, with enough discussion/inbound references to serve as an umbrella for related breakages.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config loading, but one is missing fields from `trust_remote_code` config objects while the other is strict config loading for `granite_speech`; different failure modes." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Different models and different video-processing bugs: Qwen3OmniMoe ValueError vs Qwen2.5-VL wrong `vision_position_ids`." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated symptoms: Qwen3VL `weight_only=True` load error vs Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Different models and outputs: Pegasus v4/v5 output drift vs CamemBERT MLM prediction regression." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription failure and Kosmos2.5 OCR index error are distinct code paths." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Both touch loading, but one is non-persistent buffer initialization junk and the other is FSDP CPU RAM efficient loading; not the same bug." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Both mention pipeline-related behavior, but one is a bad error message and the other is a SigLIP2 model/processor mismatch plus quantization failure." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "Different scope: GLM-5 RoPE implementation discussion vs stale docs mentioning removed pipeline tasks." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Related broad theme around `remote_code`, but 44561 is a specific `is_torch_fx_available` removal breakage and 45020 is a wider umbrella report." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "Different models and problems: SigLIP2 example errors vs Kosmos2.5 long-OCR index error." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor-from-URL loading and template compilation errors are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Different subsystems: Qwen3 device-map CUDA assert vs tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "ImageProcessor void segmentation handling and CustomTrainer `compute_loss` signature mismatch are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2 NER/padding crash and FastSpeech2Conformer tokenizer loading failure are different tokenizer issues." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": true, + "reason": "Same underlying v5 pipeline-removal fallout: one reports the bad runtime error message, the other the stale docs referencing removed summarization/translation pipeline tasks." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "Blip2 dtype preservation bug and Apple Silicon `caching_allocator_warmup` TypeError are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Different regressions: Qwen2.5/3.5 VL video-input issue vs Kosmos2.5 OCR index error." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Mask-shape indexing error and trainer evaluate-after-train failure are separate bugs." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "CamemBERT masked-LM regression and vague tokenization change report are not the same concrete bug." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "CPU offload device-mismatch for Bark and bfloat16 dtype mismatches for SwitchTransformers/TimmWrapper are different issues." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "BarkProcessor `voice_preset` failure and `GenerationMixin` import error are unrelated." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor output size mismatch for OmDet-Turbo vs bfloat16 dtype mismatch in unrelated models." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen VL regressions: `weight_only=True` load error vs still-image temporal position-id scaling bug." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "BitNet packed-weight unpacking bug and bfloat16 dtype mismatch are different model-loading paths." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "TensorFlow backend packaging and Mistral regex patch crash are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Different tokenizer/processor bugs: Mistral regex patch vs Gemma4Processor missing `_tokenizer`." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and `init_empty_weights` unexpected `_is_hf_initialized` argument are separate loading regressions." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Different Qwen3-Next warning behaviors: incorrect tied-weights mapping vs noisy `tie_word_embeddings=False` warning." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The canonical choice is grounded in the packet: issue 45020 is the broadest remote_code umbrella and has enough activity/inbound references to justify centering the cluster. The summary stays conservative overall, and the lone accepted soft pair (43825/43827) is clearly tied to the same v5 pipeline-removal fallout. No PR-pair mergeability concerns apply because no PRs were provided." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 16, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43454", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43638", + "issue:43645", + "issue:43650", + "issue:43653", + "issue:43696", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43854", + "issue:43906", + "issue:43927", + "issue:44479", + "issue:44554", + "issue:44568", + "issue:44589", + "issue:44610", + "issue:44811", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45042", + "issue:45084", + "issue:45356", + "issue:45459", + "issue:45488", + "issue:45538", + "issue:45600", + "issue:45803" + ], + "soft_pairs": [ + "issue:43650|issue:43906", + "issue:43582|issue:44857", + "issue:43262|issue:43819", + "issue:43526|issue:43697", + "issue:43526|issue:44554", + "issue:45459|issue:45803", + "issue:34634|issue:36331", + "issue:43653|issue:44912", + "issue:43329|issue:45600", + "issue:45084|issue:45538", + "issue:43742|issue:45020", + "issue:43638|issue:44936", + "issue:43257|issue:43329", + "issue:43452|issue:44991", + "issue:43746|issue:44811", + "issue:43257|issue:43696", + "issue:44589|issue:45042", + "issue:43854|issue:44977", + "issue:38175|issue:44610", + "issue:39692|issue:44912", + "issue:43819|issue:43927", + "issue:36010|issue:43645", + "issue:43454|issue:43577", + "issue:43550|issue:44912", + "issue:43720|issue:44610", + "issue:44568|issue:45488", + "issue:43262|issue:44811", + "issue:43329|issue:44479", + "issue:43531|issue:45042", + "issue:44568|issue:45356" + ], + "prepared_review_unit_hash": "023f8ab5e89266279ca36f6eb6c1ae6d291f2274ce140feaed1e0e4c69b04363", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36932, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The candidate pairs are mostly superficial similarity matches rather than true duplicates. They span unrelated bugs across tokenizers, processors, training, quantization, and model-loading paths, so no soft edge is safe to merge.", + "confidence": 0.94, + "canonical_issue_reason": "issue:45020 is the broadest open umbrella for recent Transformers v5 model-loading regressions, but the cluster is too heterogeneous for a true duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the best global anchor because it is broad and current; the other issues are narrower, model-specific failures that do not form one coherent duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "Completely unrelated: a vague data request vs a reproduction of a separate transformers bug." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Different subsystems and failures: Apple Silicon allocator TypeError vs CUDA/AMP loss crash." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate default bug is unrelated to DAC latent decoding/STE behavior." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "Different models and code paths: BeitImageProcessorFast label reduction vs RTDetrV2 output regression." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "No overlap beyond being correctness bugs; BEiT label reduction and MPS attention issues are unrelated." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve exception handling, but in different tokenizer utilities and with different root causes/code paths." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Voice preset behavior in BarkProcessor is unrelated to CustomTrainer.compute_loss signature breakage." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBirdTokenizer special-token registration bug is unrelated to MXFP4 quantization fallback for gpt-oss." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Video multimodal token counting bug and auto_mappings CI break are unrelated issues." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Compiler template-node error is unrelated to CLIPTokenizer's model_max_length constant." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "A single model loading KeyError is not the same as the broad remote_code regression reported in the other issue." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Deepspeed ZeRO3/empty-parameter index error is unrelated to evaluate() failing after train()." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3-MoE weight conversion/loading bug is unrelated to undefined variables in multimodal video token counting." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "Both concern loading, but one is gguf_file support and the other is a tokenizer regression for a specific model." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "PEFT local checkpoint loading and Whisper batch_decode skip_special_tokens are distinct failures." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3-MoE loading bug is unrelated to a GPT-oss CUDA out-of-memory report." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Float8 storage lookup error and PIL backend torchvision dependency regression are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading issue is not the same as Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo processor/model size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "SigLIP2 documentation errors and gpt-oss quantization fallback are different problems." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent/forward mismatch and DiaConfig token-ID persistence are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import path break and custom-model notebook initialization are different v5 issues." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "AyaVision lm_head tying bug and BLIP2 dtype propagation bug affect different models and code paths." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba is unrelated to MXFP4 quantization fallback." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Packed-weight unpacking during accelerate loading is unrelated to an OmDet-Turbo input-size mismatch." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer regressions, but they hit different tokenizers and different behaviors; not the same bug." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Audio sampling-rate default in apply_chat_template and Whisper batch_decode skip_special_tokens are unrelated." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Undefined video-token helper variables and a broader v5.3.0 video-input regression are not the same concrete defect." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding window behavior and PIL backend torchvision dependency are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer regressions: special-token addition vs codec/regex warning behavior in Kimi-K2.5." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 17, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:33357", + "issue:34567", + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43329", + "issue:43450", + "issue:43452", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43643", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43824", + "issue:43927", + "issue:43931", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44560", + "issue:44756", + "issue:44821", + "issue:44849", + "issue:44912", + "issue:44938", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45137", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45538", + "issue:45701", + "issue:45759", + "issue:45859", + "issue:45907" + ], + "soft_pairs": [ + "issue:43749|issue:44756", + "issue:36010|issue:36331", + "issue:44849|issue:45701", + "issue:43643|issue:45759", + "issue:43257|issue:43931", + "issue:43329|issue:43450", + "issue:34567|issue:40990", + "issue:43606|issue:45859", + "issue:43257|issue:43756", + "issue:43697|issue:45357", + "issue:43452|issue:43582", + "issue:44291|issue:45042", + "issue:44987|issue:45907", + "issue:42915|issue:43742", + "issue:44987|issue:45538", + "issue:33357|issue:34567", + "issue:44479|issue:45381", + "issue:41720|issue:43540", + "issue:43927|issue:44912", + "issue:42915|issue:44821", + "issue:42915|issue:45538", + "issue:43577|issue:45072", + "issue:43317|issue:45538", + "issue:43540|issue:44479", + "issue:42175|issue:43824", + "issue:43540|issue:44560", + "issue:44062|issue:44991", + "issue:42175|issue:44938", + "issue:45137|issue:45405" + ], + "prepared_review_unit_hash": "20d828a995cf07e15b099110ed621c1d311899900d2fdca9910b832d2b9ab313", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35776, + "estimated_input_tokens": 8944, + "estimated_eval_tokens": 18144 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue reports clustered by superficial similarity in model loading, offload, and multimodal/video wording. I only see one likely duplicate pair around the Qwen v5.3.0 video-input regression; the rest look like distinct bugs and should not be merged.", + "confidence": 0.94, + "canonical_issue_reason": "Use issue:44479 as the canonical issue: it is the broadest and clearest description of the Qwen v5.3.0 video-input regression, and issue:45381 reads like a specific manifestation of the same problem.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the best issue representative because it captures the shared regression at the right level of abstraction while naming the affected model families and version.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Both are loading/memory-related, but one is FSDP CPU RAM efficient loading and the other is mmap OOM on Strix Halo; different mechanisms and fixes." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "An ImportError for GenerationMixin is unrelated to a Trainer.compute_loss signature break." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "One is a Qwen3.5 output_hidden_states bug; the other is a tokenizer/version-dependent tokenization change. Different layers of the stack." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Both concern config/model construction, but missing fields from trust_remote_code and not unwrapping text_config are distinct code paths." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Different models and failures: Qwen3 MOE weight conversion with accelerate/deepspeed versus Qwen3-VL weight shape mismatch." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Both mention video, but one is undefined helpers in multimodal token counting and the other is a batched video processor shape bug." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counting and high perplexity on gpt-oss are unrelated problems." + }, + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both involve CPU offload, but they affect different models and different failure points; not the same concrete bug." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MOE weight conversion and Smollm3 RoPE layer selection are unrelated." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "RTDetrV2 output drift in v5 and Qwen3.5 save_pretrained key handling are separate issues." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading failures and an Apple Silicon caching_allocator_warmup TypeError do not share a code path." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument and PIL backend requiring torchvision are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Loading physical-intelligence/fast on transformers>=5.1.0 and forwarding revision in list_repo_files are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3Moe fine-grained FP8 loading failure and MobileLLM-125M key error are different loading bugs." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Model loading regression and CLIPTokenizer max length handling are unrelated." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS bus error with CLIP and TrainerState token counting are not the same bug." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both describe the same v5.3.0 Qwen2.5-VL video-input regression; 45381 looks like a specific symptom of the broader regression in 44479." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Qwen3 auto device-mapping cudaErrorAssert and Qwen3OmniMoe video processing ValueError are different failures." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig save/load token ID loss and MXFP4 quantization fallback are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor URL loading are different problems." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Qwen3Moe FP8 loading and CLIPTokenizer model_max_length are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "BLIP2 dtype mismatch and SwitchTransformers/TimmWrapperModel bf16 mismatch are separate model-specific dtype bugs." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "device_map auto offload/dequantized model loading is unrelated to CLIPTokenizer max length." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Different video bugs in different model families; Qwen3OmniMoe ValueError is not the same as the broader Qwen2.5-VL/Qwen3.5 video regression." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Missing TensorFlow backend in a pip extra is unrelated to a missing Qwen2_5_VLForConditionalGeneration import." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Qwen3OmniMoe video processing ValueError and Qwen3-vl-embedding StopIteration are distinct video-input failures." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "AddedToken special-argument duplication and tokenizer loading for EMBEDDIA/est-roberta are different tokenizer bugs." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Missing TensorFlow in an install extra and Python 3.14 import failure are unrelated environment/package issues." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque underflow and MIN_PEFT_VERSION bump are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays conservative. The only accepted pair, issue:44479 vs issue:45381, is plausibly the same Qwen 5.3.0 video-input regression at different levels of specificity. The other soft pairs are rejected with reasons that match the issue titles and do not overclaim shared root causes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 18, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:28282", + "issue:29127", + "issue:34567", + "issue:34689", + "issue:36331", + "issue:38175", + "issue:42175", + "issue:42915", + "issue:43452", + "issue:43493", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43582", + "issue:43643", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43906", + "issue:43994", + "issue:44062", + "issue:44291", + "issue:44484", + "issue:44488", + "issue:44514", + "issue:44589", + "issue:44704", + "issue:44737", + "issue:44792", + "issue:44821", + "issue:44898", + "issue:44964", + "issue:45042", + "issue:45081", + "issue:45216", + "issue:45325", + "issue:45375", + "issue:45381", + "issue:45405", + "issue:45479", + "issue:45588", + "issue:45600", + "issue:45823" + ], + "soft_pairs": [ + "issue:38175|issue:44898", + "issue:42175|issue:44488", + "issue:44737|issue:45588", + "issue:28282|issue:34689", + "issue:43906|issue:45381", + "issue:42915|issue:43749", + "issue:43742|issue:44291", + "issue:43526|issue:45479", + "issue:28282|issue:29127", + "issue:43582|issue:43782", + "issue:43540|issue:45823", + "issue:43756|issue:45325", + "issue:43493|issue:43994", + "issue:34567|issue:36331", + "issue:44291|issue:44821", + "issue:44062|issue:45081", + "issue:43643|issue:45375", + "issue:43452|issue:44488", + "issue:44792|issue:44964", + "issue:45042|issue:45081", + "issue:43742|issue:44589", + "issue:43531|issue:44821", + "issue:45405|issue:45600", + "issue:43697|issue:45216", + "issue:43688|issue:43749", + "issue:44484|issue:44704", + "issue:44514|issue:45381", + "issue:43720|issue:44898" + ], + "prepared_review_unit_hash": "8103e143b8f9d76fb75d711ade97aeafb21ec3aa166b27715116ce68e151b893", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35777, + "estimated_input_tokens": 8945, + "estimated_eval_tokens": 18146 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are largely unrelated closed Transformers issues rather than a true duplicate cluster. None of the soft pairs look like the same concrete bug/change with enough confidence to merge.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43749 has the strongest signal in this set (highest discussion/inbound references) and is a concrete, broadly impactful failure report, so it is the best anchor issue among otherwise unrelated tickets.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43749 is the most actionable and representative issue here: it describes a specific broken loading path with substantial discussion and external references.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different models and failure modes: zero probabilities in SigLIP2 vs resolution/interpolation failure in Perceiver." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Backend packaging/dependency issue vs model loading failure for a specific checkpoint; not the same bug." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "Unrelated code paths: XLNet CPU tensor creation vs flash attention auxiliary-sink crash." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Generic PyTorch import error vs a regression loading Llama 3.2 Vision; different root causes." + }, + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "No clear evidence they target the same bug; one is a reproduction of another issue, the other is a Qwen2.5-VL video position-id problem." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "Different subsystems and symptoms: FineGrainedFP8/Qwen3Moe vs FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Different loading failures: MobileLLM key error vs init_empty_weights/_is_hf_initialized argument mismatch." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "Label reduction bug in BeitImageProcessorFast is unrelated to sequence-classification loss degeneracy." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "ImportError for missing PyTorch and a LayoutLMv3 error-message clarity issue are not the same bug." + }, + { + "left": "issue:43582", + "right": "issue:43782", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError is unrelated to Qwen3VL weight_only loading error." + }, + { + "left": "issue:43540", + "right": "issue:45823", + "accept": false, + "reason": "Different models and failure modes: Qwen3OmniMoe video processing vs Gemma4 device mismatch." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Separate RoPE/position-id bugs in different models; not a single underlying change." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": false, + "reason": "Both mention SigLIP2, but one is a HF-vs-JAX discrepancy while the other is a nonsensical AutoModel/pipeline result; insufficient evidence of the same bug." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "TrainerState token counting and CustomTrainer compute_loss signature mismatch are distinct trainer issues." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "Init-empty-weights argument handling is unrelated to loading an AutoImageProcessor from URL." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "AddedToken constructor conflict and Mistral regex patch crash are different tokenizer-loading problems." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both involve missing config fields, but one is generic trust_remote_code behavior and the other is a specific Qwen config field drop; not clearly the same code path." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "gguf_file loading breakage and a separate sleng-bert load failure are not the same issue." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Different multimodal model failures and different symptoms; no shared bug is evident." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Torchvision dependency regression in image processors is unrelated to Mistral tokenizer regex patching." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "MobileLLM load failure and missing Float8 storage object are distinct errors." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3-MoE sliding window behavior is unrelated to URL loading for AutoImageProcessor." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "PEFT version pinning and stale auto_mappings references are unrelated CI/setup problems." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "RTDetrV2 output drift and Qwen3.5 save_pretrained checkpoint corruption are different regressions." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Auxiliary-loss normalization and FSDP CPU RAM efficient loading affect different code paths." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to AutoProcessor cached_file kwargs forwarding." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL issues, but one is apply_chat_template batching and the other is video position-id computation; not the same bug." + }, + { + "left": "issue:44589", + "right": "issue:44704", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to AutoProcessor kwargs forwarding." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "Already distinct XLNet CPU placement and flash_attention s_aux handling issues." + }, + { + "left": "issue:44821", + "right": "issue:43531", + "accept": false, + "reason": "AutoImageProcessor URL loading and Qwen3-MoE sliding window are unrelated." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "These touch different subsystems and have different failure signatures." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "BitNet packed-weight loading and Perceiver interpolation are different 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": 1, + "component_count": 17, + "review_unit_index": 19, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:30064", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:35141", + "issue:38175", + "issue:41720", + "issue:41762", + "issue:43295", + "issue:43452", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44514", + "issue:44561", + "issue:44568", + "issue:44625", + "issue:44743", + "issue:44811", + "issue:44843", + "issue:44877", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45005", + "issue:45290", + "issue:45356", + "issue:45440", + "issue:45701", + "issue:45704", + "issue:45823" + ], + "soft_pairs": [ + "issue:41762|issue:44936", + "issue:43643|issue:44561", + "issue:43452|issue:43577", + "issue:24643|issue:29127", + "issue:44625|issue:44877", + "issue:43452|issue:44843", + "issue:44466|issue:45005", + "issue:44514|issue:45290", + "issue:43295|issue:45356", + "issue:43819|issue:44811", + "issue:44464|issue:44743", + "issue:43653|issue:44568", + "issue:44936|issue:45823", + "issue:43688|issue:45440", + "issue:38175|issue:43295", + "issue:44361|issue:44442", + "issue:43653|issue:44361", + "issue:43688|issue:45704", + "issue:30064|issue:33666", + "issue:41720|issue:43906", + "issue:28282|issue:32090", + "issue:43577|issue:43697", + "issue:28282|issue:33357", + "issue:43646|issue:43749", + "issue:43688|issue:45701", + "issue:28282|issue:35141", + "issue:44843|issue:44991", + "issue:43653|issue:43927", + "issue:43531|issue:43742", + "issue:43550|issue:43927", + "issue:44464|issue:44977" + ], + "prepared_review_unit_hash": "db311ba966acd53570ee842ca48b7057c3964aeee6015a1f3636b7753596174c", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37104, + "estimated_input_tokens": 9276, + "estimated_eval_tokens": 18808 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No duplicate cluster here: the items span many unrelated bugs across loading, tokenization, training, MoE, and device-map behavior, so every soft edge should be rejected.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is appropriate; the set is heterogeneous and the paired items are not the same underlying defect.", + "canonical_pr_reason": "No pull requests are present in this cluster.", + "best_issue_reason": "No global best issue stands out as a representative duplicate anchor; at most, a few items share broad subsystems, but not the same concrete bug.", + "best_pr_reason": "No pull requests are present in this cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: Gemma3 ZeRO-3 load-time index error vs trainer.evaluate() after train; no shared code-path bug." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both touch trust_remote_code, but one is missing fields in AutoConfig output and the other is v5 removal of is_torch_fx_available; distinct regressions." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "gguf_file breaks tokenizer/model loading, while the BLIP2 report is about dtype propagation; unrelated bugs." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed weight-shape training error vs LayoutLMv3 box-clarity/error-message issue; different problems." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "num_labels propagation in Qwen3.5 is unrelated to strict config loading for granite_speech." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both involve loading/tokenizer paths, but gguf_file breakage and unconditional model_info() in offline mode are different bugs." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but one is device-dependent lm_head serialization and the other is translation-model tied-weight handling; not the same fix." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both crash in apply_chat_template, but with different inputs and causes: batched padding=False vs assistant tool-call messages with no content." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "One is processor.tokenizer regression; the other is Kimi-K2.5 codec/warning behavior. Same area, different defect." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC from_latents mismatch and Whisper batch_decode skip_special_tokens are unrelated code paths." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is distinct from Qwen3-MoE recurrent-state reset with cache." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and mdeberta add_special_tokens behavior are separate tokenizer bugs." + }, + { + "left": "issue:44936", + "right": "issue:45823", + "accept": false, + "reason": "trainer.evaluate() after train is unrelated to Gemma4 PLE device mismatch under device_map=auto." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Auxiliary-loss normalization in OLMoE/GPT Oss is not the same as DeepseekV3MoE diverging from the remote implementation." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to the processor.tokenizer regression." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "MLukeTokenizer AttributeError and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are different tokenizer-loading failures." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "Special-token registration in BigBirdTokenizer is unrelated to MLukeTokenizer AttributeError." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "MoE auxiliary-loss normalization and T5's apex FusedRMSNorm memory leak are unrelated." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Void segmentation-map processing and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:43906", + "accept": false, + "reason": "Qwen3 auto device-mapping cudaErrorAssert is not the same as an isolated reproduction of a different upstream issue." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "PyTorch-missing import error and NoneType broadcast failure in Trainer are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "BLIP2 dtype state is unrelated to RTDetrV2 producing different outputs in v5." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "ImportError about missing PyTorch and Mac bus-error on CLIP are different environment/runtime problems." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Transformers v5 custom-model initialization break is distinct from FSDP_CPU_RAM_EFFICIENT_LOADING." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Auxiliary-loss normalization is unrelated to tokenization changing across versions." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "Missing-PyTorch import error and output-embedding reinitialization after resize_token_embeddings are different bugs." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Offline model_info() regression and tokenizer-loading failure for a specific model are different failure modes." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and DiaConfig losing custom token IDs after save/load are separate state/token bugs." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "torch.compile with SDPA on Bamba-9B-v2 and DiaConfig save/load token-ID loss are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Chunked generation with compiled forward and Qwen3.5 flash-attention generation issues are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 20, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:28218", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:34567", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43122", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43329", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43825", + "issue:43844", + "issue:43906", + "issue:43927", + "issue:44361", + "issue:44451", + "issue:44464", + "issue:44479", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44625", + "issue:44821", + "issue:44849", + "issue:44913", + "issue:44977", + "issue:45325", + "issue:45357", + "issue:45588", + "issue:45812" + ], + "soft_pairs": [ + "issue:43577|issue:45588", + "issue:43329|issue:45325", + "issue:43577|issue:44554", + "issue:44479|issue:44625", + "issue:43550|issue:44361", + "issue:43122|issue:43906", + "issue:32090|issue:34567", + "issue:43504|issue:43550", + "issue:43504|issue:43720", + "issue:41720|issue:44464", + "issue:43504|issue:44610", + "issue:32090|issue:33666", + "issue:44849|issue:45357", + "issue:43452|issue:44451", + "issue:43278|issue:43688", + "issue:44568|issue:45812", + "issue:43742|issue:44821", + "issue:33666|issue:36010", + "issue:43493|issue:43825", + "issue:43606|issue:43927", + "issue:43720|issue:43927", + "issue:43756|issue:44913", + "issue:43262|issue:43295", + "issue:43525|issue:43572", + "issue:28218|issue:33357", + "issue:43720|issue:44361", + "issue:43493|issue:43844", + "issue:33666|issue:36331", + "issue:39692|issue:43493", + "issue:43540|issue:44977", + "issue:43278|issue:43844" + ], + "prepared_review_unit_hash": "3e2064489a5265579473ceba9ba615fb6769f7796b3aaf87ad9d01fdf81db26e", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36622, + "estimated_input_tokens": 9156, + "estimated_eval_tokens": 18568 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: many items are tokenizer/config/training/loading bugs, but the soft pairs are only superficially similar and do not appear to be the same concrete underlying issue. No soft edge should be merged.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45812 is the broadest and most central tokenizer regression in the set, with multiple inbound refs and a clear v4\u2192v5 impact; the other items are mostly unrelated one-off model/training/loading bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45812 is the strongest representative issue because it describes a widespread, silent tokenizer regression affecting all Granite models and is more cluster-representative than the narrower, model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43577", + "right": "issue:45588", + "accept": false, + "reason": "Different code paths: BLIP2 dtype handling vs flash-attention sinkless-model crash." + }, + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Both involve video/VL models, but one is undefined variables in multimodal token counting and the other is rope-index scaling for still-image temporal positions." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Unrelated bugs: loading dtype mismatch in BLIP2 vs an MPS attention correctness issue." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is a video-input regression and the other is num_labels propagation." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Different failures: torch.compile + SDPA on Bamba vs MLukeTokenizer attribute error on tasks." + }, + { + "left": "issue:43122", + "right": "issue:43906", + "accept": false, + "reason": "One is a tokenizer output-change regression across versions; the other is an isolated repro of a different issue." + }, + { + "left": "issue:32090", + "right": "issue:34567", + "accept": false, + "reason": "Trainer errors, but different symptoms and code paths: gpu broadcast NoneType vs num_input_tokens_seen not updating." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Legacy-field model loading bug vs compile/SDPA runtime failure on a different model." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Different loading problems: legacy preset field vs packed-weight handling during accelerate loading." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "A800 auto device mapping cudaErrorAssert is not the same as inconsistent outputs from compiled forward chunked generation." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Different issues: pretrained preset load failure vs processor producing the wrong input size for the model." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Both mention training, but one is a broadcast TypeError and the other is a Qwen2-VL multi-GPU training question/bug." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 regressions, but hidden-states behavior and save_pretrained key mapping are distinct bugs." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Loading failures, but gguf_file support breakage and inability to load a specific model are not the same bug." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Different model behaviors: dtype drift in eval vs incorrect auxiliary-loss normalization." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they affect different models and symptoms (special tokens vs wrong token IDs) with no evidence of the same code-path bug." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "Both are load/processor problems, but one is a MobileLLM key error and the other is loading AutoImageProcessor from URL." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training vs GenerationMixin import error; unrelated failures." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy is unrelated to the pipeline translation-task support message regression." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "Device-mismatch during CPU offload is unrelated to loss of custom token IDs and generation IndexError." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Packed-weight loading failure is a different bug from DiaConfig losing custom token IDs after save/load." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "RoPE-layer selection bug and GPTNeoX rotary_pct reload bug are both config-related but not the same issue." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio chat-template sampling-rate default is unrelated to processor.tokenizer access / image-passing regression." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Both mention missing pad token fields, but they affect different configs and different attribute names; no shared concrete bug is shown." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer spacing bug and MacOS bus error in a CLIP model are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "Packed-weight loading and MLukeTokenizer task attribute errors are separate failures." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 model discrepancy is unrelated to DeepSpeed ZeRO-3 gradient growth." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Different training issues: multi-GPU Qwen2-VL vs CustomTrainer.compute_loss signature mismatch." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Both are SigLIP2-related, but one is a docs example mismatch/quantization failure and the other is a core implementation discrepancy; not the same bug." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Video processing ValueError in Qwen3OmniMoe and flash-attention generation failure in Qwen3.5 are separate bugs." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "BF16-to-FP32 dtype drift is unrelated to abnormal gradient growth under ZeRO-3." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 21, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:29942", + "issue:30333", + "issue:33357", + "issue:33666", + "issue:39692", + "issue:43232", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43504", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44112", + "issue:44291", + "issue:44361", + "issue:44423", + "issue:44464", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44521", + "issue:44589", + "issue:44625", + "issue:44734", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:44964", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45290", + "issue:45405", + "issue:45520", + "issue:45584", + "issue:45759", + "issue:45823" + ], + "soft_pairs": [ + "issue:44964|issue:45405", + "issue:43257|issue:43299", + "issue:43334|issue:43572", + "issue:44625|issue:45584", + "issue:43756|issue:45083", + "issue:44488|issue:44843", + "issue:45200|issue:45229", + "issue:43334|issue:45759", + "issue:43646|issue:44466", + "issue:43824|issue:44589", + "issue:39692|issue:43295", + "issue:45229|issue:45823", + "issue:33357|issue:33666", + "issue:44589|issue:45520", + "issue:43504|issue:43606", + "issue:43572|issue:44291", + "issue:43927|issue:44361", + "issue:44423|issue:44734", + "issue:43504|issue:43653", + "issue:43575|issue:45405", + "issue:44464|issue:45584", + "issue:43994|issue:44493", + "issue:43606|issue:44112", + "issue:43278|issue:43701", + "issue:44792|issue:45405", + "issue:29942|issue:30333", + "issue:44521|issue:45290", + "issue:43232|issue:44849", + "issue:43844|issue:44964" + ], + "prepared_review_unit_hash": "948bf06835ded405396d65d1e684828ebfcdc8593dee50168d20d75bda2f82c3", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36537, + "estimated_input_tokens": 9135, + "estimated_eval_tokens": 18526 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose similarity group of Transformers/model-regression issues, not true duplicates. The strongest unifying thread is v5-era model/config loading breakage, but most pairs are distinct model-specific failures. No PRs are present.", + "confidence": 0.38, + "canonical_issue_reason": "issue:43646 is the broadest and most central regression in the set: it describes Transformers 5.0.0 breaking custom model initialization, which is the closest umbrella symptom to several other v5 loading/init complaints.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best representative issue because it is general, clearly stated, and captures the broader initialization regression theme better than the more model-specific tickets.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "One is a model-loading failure for Phi-4 multimodal; the other is a dependency/version bump issue for PEFT. No shared bug." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen3, but one is MOE weight conversion with accelerate/deepspeed and the other is Qwen3VL loading. Different code paths and failures." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Both are missing config fields, but for different models and different fields (`pad_token_id` vs `pad_token_idx`). Not the same bug." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation and Whisper empty-transcription behavior are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:45083", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and qwen3_omni_moe feature-length helper behavior are different issues." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "A specific model load failure versus an offline-mode tokenizer regex patch bug; different failure modes." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Same Gemma4 family, but one is a missing default tensor input and the other is a multi-GPU OOM. Not mergeable as one fix." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Both involve Qwen multimodal configs, but one is a missing `pad_token_id` attribute and the other is incorrect unwrapping of `text_config`. Different concrete bugs." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization regression versus device-dependent `lm_head.weight` serialization inconsistency. Not the same code-path problem." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "ImportError for a missing Qwen2.5-VL class versus a Float8 storage lookup error. Unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Docs/example errors for SigLIP2 versus a runtime regression in custom model code using `processor.tokenizer` and images. Different artifacts and bugs." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4-related, but one is CUDA OOM in inference and the other is a device-mismatch forward bug. Distinct issues." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS CLIP bus error and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure and Python 3.13 flash_attn KeyError are different import/runtime problems." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT pretrained-load legacy-field issue versus Bark CPU-offload device mismatch. Different models and failures." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Missing `pad_token_idx` in StableLmConfig and unexpected `_is_hf_initialized` in `init_empty_weights` are different initialization regressions." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "DiaConfig save/load token-ID loss and MLukeTokenizer task AttributeError are separate tokenizer/config bugs." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both are serve-related, but one is a multimodal `.to` crash and the other is KV-cache continuation tensor indexing. Different endpoints and code paths." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "A BEiT model-loading legacy-field issue versus a BigBirdTokenizer special-token registration bug. Unrelated." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "Qwen2-57B-A14B-Auto tensor-parallel OOM and unreleased PEFT version bump are not the same issue." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and Whisper empty-transcription failure are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 pipeline/AutoModel nonsense outputs and unexpected position-id keys across many models do not describe the same concrete bug." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "Bark-small CPU offload device mismatch versus a GraniteSpeech CI test issue. One is runtime behavior, the other is test flakiness." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift and checkpoint resume key mismatch are unrelated training issues." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and PEFT version bump are unrelated." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures and MLFlow reporting on failed jobs are entirely different problems." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve `apply_chat_template`, but one is all-zero assistant masks and the other is a crash with tool-call messages. Different failure modes." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Generation kwarg update after `sync_gpus` and Qwen3.5 hidden-state behavior are unrelated." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading failure are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 22, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:34567", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:42915", + "issue:43295", + "issue:43334", + "issue:43388", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43526", + "issue:43540", + "issue:43572", + "issue:43582", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43957", + "issue:43975", + "issue:44315", + "issue:44464", + "issue:44466", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44792", + "issue:44849", + "issue:44898", + "issue:44936", + "issue:44977", + "issue:44987", + "issue:45072", + "issue:45084", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45325", + "issue:45356", + "issue:45412", + "issue:45446", + "issue:45520", + "issue:45563", + "issue:45701", + "issue:45823" + ], + "soft_pairs": [ + "issue:28218|issue:34567", + "issue:44554|issue:45446", + "issue:43582|issue:44977", + "issue:45084|issue:45520", + "issue:15354|issue:38175", + "issue:44568|issue:45701", + "issue:43540|issue:45200", + "issue:45072|issue:45198", + "issue:45200|issue:45823", + "issue:37428|issue:43957", + "issue:44315|issue:44849", + "issue:44554|issue:45356", + "issue:43493|issue:43701", + "issue:43295|issue:44610", + "issue:43873|issue:44464", + "issue:44466|issue:45127", + "issue:44898|issue:45412", + "issue:44792|issue:45563", + "issue:45200|issue:45325", + "issue:43572|issue:44987", + "issue:37428|issue:43824", + "issue:43844|issue:44936", + "issue:43572|issue:43957", + "issue:43295|issue:43450", + "issue:43334|issue:43525", + "issue:43824|issue:43975", + "issue:43388|issue:43526", + "issue:43295|issue:43873", + "issue:38175|issue:45412", + "issue:28218|issue:35141", + "issue:38175|issue:43493", + "issue:42915|issue:43824" + ], + "prepared_review_unit_hash": "a37dff7dd02d0ed5bce829c300ed284069afda4d19db255f083d136cc8eabe13", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37505, + "estimated_input_tokens": 9377, + "estimated_eval_tokens": 19010 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are unrelated false positives across different models/features. I only see two likely duplicates: the missing `pad_token_id` config regressions and the SigLIP2 implementation discrepancy. The rest are separate bugs. I chose the broad tokenization/version-regression report as the best overall issue umbrella.", + "confidence": 0.41, + "canonical_issue_reason": "Issue 45701 is the broadest umbrella in this set: a version-dependent tokenization regression. It is the closest high-level representative of the cluster, though many items are still distinct model-specific bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "45701 is the most general issue to keep as the cluster representative; it can absorb the tokenizer/version-regression theme better than the narrower model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer spacing after added tokens and TrainerState token counting are unrelated bugs." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness and a flex_attention import/version guard are different code paths." + }, + { + "left": "issue:43582", + "right": "issue:44977", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError is unrelated to Qwen3.5 flash-attention generation." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Compile/template-node errors and a Python 3.13 flash_attn import KeyError are separate failures." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp support and SigLIP2 zero probabilities are not the same issue." + }, + { + "left": "issue:44568", + "right": "issue:45701", + "accept": false, + "reason": "A specific mdeberta special-tokens bug is too narrow to be the same as the broader tokenization-version regression." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video-input processing and Gemma4 text-only token type defaults are different problems." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "bfloat16 dtype mismatches and Wav2Vec2 save/tokenization failures are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Missing `mm_token_type_ids` defaults and a Gemma4 device-map mismatch are distinct bugs." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "A missing flash-attention helper import and meta-device loading breakage are different regressions." + }, + { + "left": "issue:44315", + "right": "issue:44849", + "accept": false, + "reason": "Liger kernel application in `model_init` and a Qwen3.5 hidden-states bug do not share the same root cause." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "MPS attention correctness and tokenizer codec handling are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 HF/JAX discrepancy and checkpoint resume key mismatch are separate issues." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Processor.tokenizer regression with images and OmDet-Turbo resize/shape mismatch are not the same bug." + }, + { + "left": "issue:43873", + "right": "issue:44464", + "accept": false, + "reason": "Quantization offloading and compiled-forward chunked generation are different failure modes." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both involve tied weights, but one is serialization inconsistency and the other is LoRA-merge collapse; not clearly the same concrete bug." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resolution handling and RT-DETR memory release are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "A Janus generation test failure and a stale paged-generate warning are different issues." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Defaulting `mm_token_type_ids` and Qwen2.5-VL temporal rope scaling are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "Missing `pad_token_idx` in StableLmConfig may be adjacent, but the loading failure is too unspecific to confirm the same bug." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Two different missing-import symbols; same broad area, but not the same underlying defect." + }, + { + "left": "issue:43844", + "right": "issue:44936", + "accept": false, + "reason": "ZeRO-3 gradient growth and post-train evaluate failure are different training issues." + }, + { + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "A config field regression and meta-device loading breakage are not the same problem." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression and batched video-shape handling are distinct." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": true, + "reason": "Both are the same `pad_token_id` AttributeError class of config regression, just reported on different model configs." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Missing model import and incorrect detokenization are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Metrics label dropping and BeitImageProcessorFast label reduction are different code paths." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor/tokenizer regression and quantization offloading are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 output issues and RT-DETR memory release are different bugs." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Added-token spacing and output-embedding reinitialization are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": true, + "reason": "Both describe the same SigLIP2 HF implementation mismatch/incorrect outputs compared with the reference implementation." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays conservative. The two accepted soft pairs are well supported by the titles/body themes, and the rejections avoid overclaiming across clearly different bugs. The choice of issue:45701 as the umbrella representative is also reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 23, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:29942", + "issue:32090", + "issue:37428", + "issue:39692", + "issue:43262", + "issue:43278", + "issue:43335", + "issue:43425", + "issue:43450", + "issue:43493", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43856", + "issue:43901", + "issue:43931", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44360", + "issue:44442", + "issue:44448", + "issue:44462", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44743", + "issue:44779", + "issue:44857", + "issue:45083", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45381", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45823" + ], + "soft_pairs": [ + "issue:44442|issue:45198", + "issue:43278|issue:43856", + "issue:44514|issue:44521", + "issue:43653|issue:44112", + "issue:43827|issue:43901", + "issue:43450|issue:43575", + "issue:43450|issue:43493", + "issue:44206|issue:45584", + "issue:45127|issue:45245", + "issue:44360|issue:44561", + "issue:43819|issue:45083", + "issue:43262|issue:43450", + "issue:43819|issue:44112", + "issue:44610|issue:44857", + "issue:43278|issue:43825", + "issue:43335|issue:43550", + "issue:43606|issue:45083", + "issue:37428|issue:43425", + "issue:44857|issue:45593", + "issue:44448|issue:44743", + "issue:45290|issue:45381", + "issue:45200|issue:45588", + "issue:29942|issue:32090", + "issue:39692|issue:43526", + "issue:43611|issue:43931", + "issue:37428|issue:44351", + "issue:44206|issue:45823", + "issue:44462|issue:44561", + "issue:44561|issue:44779", + "issue:43335|issue:43653", + "issue:43819|issue:44493", + "issue:43550|issue:44112" + ], + "prepared_review_unit_hash": "86a59824798bcf2f445aa873cef843216448d8645270fc80e93d3f6df34eb986", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37293, + "estimated_input_tokens": 9324, + "estimated_eval_tokens": 18904 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous with no clear duplicate pair. Most items are isolated bug reports or docs regressions across different models/APIs, so I would not merge any soft-edge candidates.", + "confidence": 0.31, + "canonical_issue_reason": "issue:44521 is the strongest representative because it has the most discussion/inbound references and centers on a high-traffic API (`apply_chat_template`) with a concrete multimodal bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the best standalone issue in this set: concrete, user-impacting, and well-documented compared with the more niche or documentation-only reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizer families and different failures: AutoTokenizer loading vs save_pretrained/tokenization in Wav2Vec2." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "One is a dtype regression during evaluate; the other is MoE training memory usage. Not the same bug." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both involve `apply_chat_template`, but one is a batching crash and the other is incorrect assistant masks; likely separate fixes." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Tokenizer special-token registration vs a GraniteSpeech CI device-override failure; unrelated." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs/deprecation complaints, but they target different pipeline behaviors and different APIs." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video processor shape bug and Qwen2-57B TP OOM are unrelated." + }, + { + "left": "issue:43450", + "right": "issue:43493", + "accept": false, + "reason": "Batched video shape issue vs SigLIP2 implementation discrepancy; different code paths." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Feature-extractor arg regression vs Whisper empty-transcription failure; not the same underlying defect." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/tied-embeddings model collapse is unrelated to the 2^24 category limit error." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA indexer activation issue vs removal of `is_torch_fx_available`; no shared bug." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC latent mismatch vs Qwen3 Omni MoE helper length behavior; different model components." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate default vs batched video input shape bug; different subsystems." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "Different problems entirely: DAC latent reconstruction vs GraniteSpeech CI override." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "Processor input-size mismatch vs image loss AMP crash; not the same issue." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Training/eval dtype regression is unrelated to the pipeline translation-error messaging bug." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug vs Bamba torch.compile/SDPA failure; different code paths." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to the Qwen3 Omni helper output-length issue." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Import error around a flash-attention helper vs broad Torch 2.10 incompatibility; too general to merge." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Loss crash under AMP vs D-FINE auxiliary-loss omission; separate training bugs." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Pegasus v4/v5 output drift vs Qwen3 recurrent-state reset; different model behaviors." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call crash vs video vision position-id bug; unrelated multimodal issues." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "Gemma4 mm_token_type_ids defaulting vs flash_attention sinkless-model AttributeError; different failures." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention test failure vs Trainer `_gpu_broadcast_one` NoneType error; no common bug." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 docs/model mismatch vs BEiT Fast image processor label reduction bug; unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Base-model-prefix loading breakage vs Qwen3-VL weight-shape mismatch; different model-loading problems." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing symbols and different root causes." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Both mention device mismatch, but one is a feature extractor crash and the other is a Gemma4 forward-pass placement issue." + }, + { + "left": "issue:44462", + "right": "issue:44561", + "accept": false, + "reason": "Tokenizer repository resolution bug vs removal of `is_torch_fx_available`; unrelated." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Trust-remote-code breakage from API removal vs Deepseek tokenizer regression; different causes." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Sparse-layer config bug vs BigBird tokenizer special-token registration; no overlap." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC latent reconstruction bug vs widespread position-id key warnings; unrelated." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Bamba SDPA/compile failure vs GraniteSpeech CI stale override test; different failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 24, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:28218", + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:43335", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43701", + "issue:43720", + "issue:43723", + "issue:43873", + "issue:43927", + "issue:44112", + "issue:44292", + "issue:44448", + "issue:44493", + "issue:44554", + "issue:44568", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45198", + "issue:45229", + "issue:45310", + "issue:45372", + "issue:45459", + "issue:45478", + "issue:45491", + "issue:45520", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45698", + "issue:45712", + "issue:45803", + "issue:45859" + ], + "soft_pairs": [ + "issue:45310|issue:45478", + "issue:43335|issue:43504", + "issue:45588|issue:45859", + "issue:29127|issue:32090", + "issue:43335|issue:45083", + "issue:44855|issue:45083", + "issue:39692|issue:44493", + "issue:44292|issue:45520", + "issue:44448|issue:44779", + "issue:38175|issue:44493", + "issue:43335|issue:45072", + "issue:43335|issue:45593", + "issue:43720|issue:44112", + "issue:44977|issue:45229", + "issue:40444|issue:43526", + "issue:43723|issue:44554", + "issue:34689|issue:35141", + "issue:43606|issue:44857", + "issue:44448|issue:44977", + "issue:44857|issue:45072", + "issue:44568|issue:45491", + "issue:28218|issue:32090", + "issue:43927|issue:45698", + "issue:45198|issue:45803", + "issue:44977|issue:45563", + "issue:43701|issue:45083", + "issue:29942|issue:35141", + "issue:45459|issue:45712", + "issue:43550|issue:45083", + "issue:43873|issue:44493", + "issue:45372|issue:45588", + "issue:32090|issue:33357" + ], + "prepared_review_unit_hash": "459d6727d4e679a87b44677636080abf2e73ea25d105ea020b08ca808ba9a631", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36930, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The Qwen3.5 MoE pair (#45310/#45478) still looks like a true duplicate: same title, same model family, and same from_pretrained failure. The earlier issue #45310 is the better anchor, but not because it is open; it is simply the earlier report and the more suitable representative.", + "confidence": 0.9, + "canonical_issue_reason": "#45310 is the earlier report for the same Qwen3.5 MoE from_pretrained failure, so it is the best issue anchor without relying on issue state.", + "canonical_pr_reason": null, + "best_issue_reason": "#45310 best represents the duplicate pair because it is the first report of the same concrete loading bug and has the broader discussion trail.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same underlying bug: both report transformers>=5.4.0 Qwen3.5 MoE from_pretrained failures for the same model family and code path." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate judgment for #45310/#45478 is plausible, but the analyst overstates key facts from the packet. #45310 is not open in the provided data, so the rationale that it is the \u201cearlier, open report\u201d and the \u201cbest representative\u201d is not grounded. Because the canonical-selection reasoning relies on that incorrect state claim, the response should be rejected." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 25, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:25251", + "issue:28218", + "issue:29127", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:37428", + "issue:41720", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43066", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43412", + "issue:43653", + "issue:43873", + "issue:43976", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44361", + "issue:44448", + "issue:44451", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44743", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:44987", + "issue:45198", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45520", + "issue:45563", + "issue:45588", + "issue:45823" + ], + "soft_pairs": [ + "issue:37428|issue:44987", + "issue:43412|issue:45588", + "issue:43873|issue:45229", + "issue:45245|issue:45563", + "issue:43278|issue:43329", + "issue:41720|issue:45229", + "issue:10000|issue:15354", + "issue:43976|issue:45440", + "issue:28218|issue:29127", + "issue:45372|issue:45406", + "issue:45372|issue:45459", + "issue:43065|issue:43994", + "issue:43066|issue:44462", + "issue:44857|issue:45588", + "issue:44448|issue:44466", + "issue:43278|issue:43381", + "issue:43065|issue:45412", + "issue:43653|issue:44743", + "issue:45588|issue:45823", + "issue:25251|issue:32090", + "issue:44186|issue:45198", + "issue:44206|issue:44479", + "issue:43994|issue:45412", + "issue:44451|issue:45520", + "issue:44361|issue:45198", + "issue:44871|issue:44977", + "issue:42915|issue:42994", + "issue:44265|issue:45198", + "issue:29942|issue:30064" + ], + "prepared_review_unit_hash": "8d78d47bab9b9743fc316086ec9a5485a218973e891e16ebf51d2e137fd2bd7a", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36377, + "estimated_input_tokens": 9095, + "estimated_eval_tokens": 18446 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft-similarity pairs span unrelated tokenizer, export, quantization, flash-attention, model-loading, and training issues. None look like the same underlying bug/change, so all soft edges should be rejected and there is no meaningful duplicate canonical for the cluster.", + "confidence": 0.96, + "canonical_issue_reason": "No strong canonical issue: the items do not form a single duplicate set, so there is no sensible anchor issue for this cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global representative because the cluster is a mix of unrelated regressions across different subsystems and models.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "One is a flash-attention import error; the other is a model loading failure. Different failure modes and code paths." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n export with erfinv/training-guard issues is unrelated to flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading behavior and Gemma4 multi-GPU CUDA OOM are different problems, not the same bug." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count limits in image handling are unrelated to a stale generate warning about num_return_sequences." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "A dtype regression between train/eval is unrelated to a multimodal token-counting bug in video branches." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both involve GPU memory/runtime failures, but one is Qwen3 device-mapping assert and the other is Gemma4 OOM; not the same code path." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel issue and torch.jit generator-exp export failure are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python-version compatibility and DeepSeekV3MoE divergence are distinct bugs." + }, + { + "left": "issue:28218", + "right": "issue:29127", + "accept": false, + "reason": "Tokenizer spacing after added tokens is unrelated to a LayoutLMv3 error-message clarity issue." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Gemma4 processor import failure and Gemma4Processor missing _tokenizer during serve are different failures." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "A mistral_common import breakage is unrelated to tokenizer error masking when protobuf is absent." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and SigLIP2 nonsensical outputs are unrelated model-specific bugs." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type and AutoTokenizer ignoring tokenizer.json are both tokenizer-related but clearly different defects." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "LwDetrImageLoss AMP crash and flash_attention s_aux=None crash are different code paths and symptoms." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Pegasus output drift across versions is unrelated to lm_head.weight serialization inconsistency." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype regression and gradient-checkpointing-in-eval failure are separate behaviors." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "A placeholder Conv2d implementation is unrelated to RT-DETR memory not being released after deletion." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "flash_attention s_aux=None crash and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting behavior and Trainer _gpu_broadcast_one NoneType error are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes and Wav2Vec2 save/tokenization failures are different tokenizer/model issues." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Unsupported center arg in LasrFeatureExtractor and Qwen2.5/3 VL video regression do not share the same root cause." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 incorrect outputs and RT-DETR memory leak are unrelated." + }, + { + "left": "issue:44451", + "right": "issue:45520", + "accept": false, + "reason": "Loading ScandiBERT and a flash_attn import KeyError on Python 3.13 are different issues." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure are unrelated." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id config mismatch is not the same as Qwen3.5 flash-attention generation regression." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8 failure and quantized model saving failure are different quantization-related bugs." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failing on torch_compilable_check and Wav2Vec2 save/tokenization failure are unrelated." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Failing Flash Attention 2 tests and void segmentation-map processing are unrelated test/product bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 26, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:35141", + "issue:36010", + "issue:39692", + "issue:42915", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43278", + "issue:43334", + "issue:43381", + "issue:43388", + "issue:43421", + "issue:43441", + "issue:43526", + "issue:43531", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43931", + "issue:44062", + "issue:44186", + "issue:44206", + "issue:44279", + "issue:44442", + "issue:44464", + "issue:44610", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45103", + "issue:45245", + "issue:45372", + "issue:45412", + "issue:45440", + "issue:45491", + "issue:45588", + "issue:45593", + "issue:45834", + "issue:45859", + "issue:45907" + ], + "soft_pairs": [ + "issue:43756|issue:45440", + "issue:43257|issue:43278", + "issue:43421|issue:45491", + "issue:33357|issue:39692", + "issue:29127|issue:34567", + "issue:32090|issue:33290", + "issue:44610|issue:45412", + "issue:45103|issue:45372", + "issue:45440|issue:45593", + "issue:43824|issue:44279", + "issue:43653|issue:44442", + "issue:43381|issue:43701", + "issue:43761|issue:45859", + "issue:43526|issue:45834", + "issue:29942|issue:36010", + "issue:43526|issue:44186", + "issue:29127|issue:33357", + "issue:43334|issue:43531", + "issue:29127|issue:35141", + "issue:44464|issue:45491", + "issue:44977|issue:45245", + "issue:42915|issue:43931", + "issue:44442|issue:45072", + "issue:43262|issue:44206", + "issue:43526|issue:44857", + "issue:43232|issue:43388", + "issue:43441|issue:45588", + "issue:44062|issue:45907" + ], + "prepared_review_unit_hash": "018ca9623e7ca17f1c06409133fb7a8607d7430f76edf3a7beda930d4cc88baa", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36212, + "estimated_input_tokens": 9053, + "estimated_eval_tokens": 18362 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: the paired items share broad transformer-adjacent vocabulary but do not describe the same underlying bug or change. I would not collapse these into one duplicate set; each soft pair is a false positive.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 43441 is the clearest self-contained bug report in the set, with a specific model, version context, and runtime failure path. It is a reasonable anchor if one issue must be chosen, but it does not represent the rest of the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43441 is the best single issue artifact here because it is concrete and actionable, but only for its own flash-attention failure; the cluster as a whole is too mixed for a broader canonical issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Both mention MoE/DeepSeek, but one is about dropped RoPE layers while the other is about native-vs-remote implementation divergence; not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43278", + "accept": false, + "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed is unrelated to an embedding dtype change between training and evaluation." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime post-processor updates for special tokens are unrelated to Gemma3 NaNs from sliding-window attention and all-padding windows." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "A MacOS bus error in CLIP loading is not the same as documentation example errors for SigLIP2." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and TrainerState token-count updates are different problems in different code paths." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "A NoneType broadcast crash in Trainer is not the same as an Adafactor/deepspeed OOM issue." + }, + { + "left": "issue:44610", + "right": "issue:45412", + "accept": false, + "reason": "Processor/model input-size mismatch is unrelated to RT-DETR memory not being released on deletion." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotation handling and Gemma 4 processor loading/import errors are different failures." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepSeekV3 implementation divergence does not match D-FINE's missing auxiliary losses when denoising is off." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "A missing Qwen2.5-VL import and a generic dependency issue are too different to be duplicates." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird mask-token special registration and FastSpeech2Conformer tokenizer loading are different tokenizer bugs." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate trainer behaviors." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "CLIPVision hidden_states regression and Qwen3_5Moe CPU-offload incompatibility affect different models and code paths." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels behavior is unrelated to Kosmos2.5 OCR long-input index errors." + }, + { + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "FlashAttention 2 test failures and a GenerationMixin import error are distinct regressions." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Both are tokenizer/processor-adjacent, but one is a label-reduction bug and the other is a LayoutLMv2Tokenizer crash on NER/padding." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "An error-message issue in LayoutLMv3 is not the same as a CLIP MacOS bus error." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id on Qwen3-VL text config is unrelated to sliding_window behavior in Qwen3-MoE." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error messaging and token-embedding reinitialization after resize are different bugs." + }, + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Chunked generation inconsistencies with compiled forward are not the same as Gemma3 NaNs from sliding-window attention." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation issues are unrelated to the category-count runtime limit." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "FineGrainedFP8Config failures for Qwen3Moe and weight-shape mismatches for Qwen3-VL are different model-loading bugs." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer loading failure and bfloat16 dtype mismatches affect different subsystems and failure modes." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and a feature-extractor unsupported-argument crash are not the same bug." + }, + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and LwDetrImageLoss AMP/CUDA crashes are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs update after sync_gpus and metric-gathering label truncation are separate distributed-training issues." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both involve flash attention, but one is a Ministral-3 failure in Transformers v5 RC while the other is an s_aux=None crash for sink-less models; not the same concrete bug." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "An AddedToken special-argument conflict is unrelated to list_repo_files failing to forward revision." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 27, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:33290", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:39692", + "issue:41628", + "issue:42907", + "issue:43316", + "issue:43381", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43575", + "issue:43611", + "issue:43643", + "issue:43646", + "issue:43701", + "issue:43720", + "issue:43723", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44387", + "issue:44442", + "issue:44479", + "issue:44496", + "issue:44743", + "issue:44811", + "issue:44871", + "issue:44898", + "issue:44913", + "issue:45003", + "issue:45216", + "issue:45290", + "issue:45372", + "issue:45440", + "issue:45479", + "issue:45491", + "issue:45698", + "issue:45759", + "issue:45803" + ], + "soft_pairs": [ + "issue:44743|issue:44811", + "issue:43575|issue:43701", + "issue:43646|issue:44077", + "issue:43575|issue:44387", + "issue:45003|issue:45440", + "issue:43761|issue:44079", + "issue:43723|issue:45479", + "issue:44442|issue:45803", + "issue:43761|issue:44188", + "issue:43316|issue:43525", + "issue:44479|issue:45290", + "issue:42907|issue:45698", + "issue:44442|issue:44898", + "issue:33290|issue:34567", + "issue:34689|issue:43611", + "issue:39692|issue:45491", + "issue:33290|issue:35141", + "issue:43381|issue:43873", + "issue:33290|issue:34689", + "issue:43931|issue:45216", + "issue:43493|issue:43827", + "issue:43550|issue:44743", + "issue:44077|issue:44479", + "issue:43720|issue:44442", + "issue:43316|issue:45759", + "issue:41628|issue:45372", + "issue:43643|issue:44913", + "issue:43749|issue:43931", + "issue:33290|issue:36331", + "issue:44190|issue:44496", + "issue:43873|issue:44871" + ], + "prepared_review_unit_hash": "0b13ed7c4c6876dd90cbbcff021a75a3c0d63f06522e49722a5c036e46f89f75", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37215, + "estimated_input_tokens": 9304, + "estimated_eval_tokens": 18864 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of Transformers v5 regressions around loading, config round-tripping, and serialization, but the proposed soft pairs are not true duplicates of the same underlying bug/change.", + "confidence": 0.77, + "canonical_issue_reason": "issue:43646 is the broadest representative bug here: a v5 regression breaking custom model initialization, which aligns best with the cluster\u2019s general loading/config-compatibility theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best single issue to anchor the cluster because it captures a broad, high-impact initialization regression rather than a narrow model-specific failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different bugs: recurrent-state reset with cache vs Whisper decode parameter handling." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Model-load OOM with tensor parallelism is unrelated to resume_from_checkpoint key mismatches." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Different components and failures: custom model init regression vs patchtsmixer post_init policy change." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both mention OOM, but one is TP model loading and the other is int4 quantization memory inflation; not the same bug." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "sys.modules access safety bug is unrelated to DeepseekV3MoE implementation divergence." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "Specific CLIP hidden_states regression is not the same as generic ModelOutput key assignment behavior." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "Tokenizer loading in v5 and degenerate classification loss are separate issues." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure is unrelated to Fuyu tensor conversion swallowing exceptions." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "CLIP hidden_states bug and torch.compile attention-kernel divergence are different code paths." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both are config-related, but they concern different models and different missing/discrepant fields." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "Video-input regression and chat-template tool-call crash are not the same underlying bug." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load flows, but one is dequantized Ministral/Devstral saving and the other is wrong custom-module loading after save_pretrained." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer loading failure is unrelated to Perceiver image classification with non-default resolution." + }, + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "AdaFactor/deepspeed OOM and num_input_tokens_seen not updating are unrelated trainer bugs." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Both are model-loading regressions, but they affect different mechanisms and failure modes." + }, + { + "left": "issue:39692", + "right": "issue:45491", + "accept": false, + "reason": "Documentation/example errors for SigLIP2 are unrelated to Gemma3 NaN embeddings." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "Optimizer OOM and output embedding reinitialization in post_init are distinct problems." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Eval-mode gradient-checkpointing restriction is not the same as quantization offloading failure." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Different bugs: optimizer memory blowup vs Llama model-loading regression." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Weight-shape mismatch at load time is separate from a save_pretrained regression." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation drift is unrelated to docs still referencing removed pipelines." + }, + { + "left": "issue:43550", + "right": "issue:44743", + "accept": false, + "reason": "torch.compile SDPA failure and recurrent-state reset with cache are different code-path bugs." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet accelerated-loading unpacking bug is unrelated to tokenizer loading failure." + }, + { + "left": "issue:41628", + "right": "issue:45372", + "accept": false, + "reason": "Both are import errors, but they involve different symbols and different processor-loading paths." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Config missing-fields on trust_remote_code load is not the same as rotary_pct not persisting on reload." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU-RAM-efficient loading and Qwen3-VL weight mismatch are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "Implementation discrepancy and docs pipeline removal are unrelated." + }, + { + "left": "issue:43749", + "right": "issue:44496", + "accept": false, + "reason": "Broken FSDP loading and unrecognized model config are different issues." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Example script dataset loading failure is unrelated to config/model-type recognition." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization offloading behavior and Gemma eos_token_id config inconsistency are separate bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 28, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:28282", + "issue:34689", + "issue:36010", + "issue:41950", + "issue:42175", + "issue:42617", + "issue:43065", + "issue:43295", + "issue:43316", + "issue:43381", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43452", + "issue:43475", + "issue:43540", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43673", + "issue:43688", + "issue:43701", + "issue:43742", + "issue:43761", + "issue:43825", + "issue:43856", + "issue:44077", + "issue:44162", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44617", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44877", + "issue:44913", + "issue:45127", + "issue:45137", + "issue:45200", + "issue:45584" + ], + "soft_pairs": [ + "issue:43381|issue:43856", + "issue:34689|issue:36010", + "issue:44855|issue:45137", + "issue:43381|issue:43742", + "issue:43540|issue:44514", + "issue:43316|issue:44877", + "issue:43673|issue:44743", + "issue:42175|issue:43316", + "issue:43404|issue:43441", + "issue:25251|issue:28282", + "issue:43065|issue:44617", + "issue:44162|issue:44792", + "issue:43388|issue:43688", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:44077|issue:44534", + "issue:41950|issue:43475", + "issue:45127|issue:45200", + "issue:44871|issue:45584", + "issue:43550|issue:43701", + "issue:43761|issue:44479", + "issue:43606|issue:43701", + "issue:43295|issue:44871", + "issue:43575|issue:43742", + "issue:43452|issue:44496", + "issue:42617|issue:43575", + "issue:43646|issue:44913" + ], + "prepared_review_unit_hash": "de38021d5b862389defef0917faba0f0ba4bad3cbe25b097c44ed0f146113c21", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35060, + "estimated_input_tokens": 8765, + "estimated_eval_tokens": 17786 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: mostly unrelated Transformers issues spanning model loading, config quirks, generation regressions, and pipeline/model-specific bugs. The soft matches are mostly broad topical overlap, not the same underlying defect or change.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43646 is the broadest representative of the cluster\u2019s v5 regression / model-initialization failures, though the overall set is too mixed to form a strong duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best single issue to anchor this cluster because it most generally reflects the recurring v5 breakage theme.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Different problems: gradient checkpointing in eval mode vs Qwen3 MoE training memory usage." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Unrelated failures: Llama 3.2 Vision model loading vs missing GenerationMixin import." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Different code paths: Python 3.13 Deberta import parsing vs DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:43381", + "right": "issue:43742", + "accept": false, + "reason": "Checkpointing/eval-mode behavior does not match a MobileLLM loading key error." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both mention multimodal/video processing, but one is a video-input validation error and the other is a batched chat-template padding crash." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Different config bugs in different models: Gemma3TextConfig API mismatch vs granite_speech config strictness." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both involve caching/state, but one is missing generation cache in v5 and the other is recurrent state reset in a specific Qwen module." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "Package/backend installation issue vs model config API discrepancy; no shared bug." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Weight tying bug in Mistral3ForConditionalGeneration is not the same as a FlashAttention failure in Ministral-3." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline output nesting bug vs missing PyTorch import; entirely different failures." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "SAM3 pixel decoder dummy layer issue is not the same as a SAM3Video OOM." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage and janus image-generation test failure are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Training/evaluation utility bug vs auxiliary-loss normalization bug; different mechanisms." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Both are v5 regressions, but one is processor/tokenizer behavior and the other is a pipeline error message." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "DeBERTa model_max_length question and ONNX import error are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "PatchTSMixer post_init policy is not the same as non-persistent buffer initialization corruption." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor selection and Sam3Video encoder attribute access are different bugs." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge / tied embeddings issue is unrelated to Gemma 4 token-type defaults." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma eos_token_id config mismatch is not the same as Whisper empty-transcription generation failure." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile + SDPA failure is unrelated to resume_from_checkpoint key mismatch." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "CLIPVision hidden_states regression and Qwen VL video-input regression affect different models and paths." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch is unrelated to checkpoint resume key mismatch." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Different v5-era config regressions with different symptoms and models." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Both are loading problems, but one is tensor-parallel OOM and the other is a MobileLLM key error." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "GGUF loading breakage is not the same as an unrecognized config/model_type error." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py runtime issue and Qwen2 load-time OOM are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage is not the same as GPTNeoXConfig rotary_pct not persisting on reload." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 29, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:41720", + "issue:43065", + "issue:43335", + "issue:43388", + "issue:43504", + "issue:43575", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44079", + "issue:44190", + "issue:44242", + "issue:44462", + "issue:44464", + "issue:44479", + "issue:44521", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44933", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45245", + "issue:45290", + "issue:45357", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45698", + "issue:45803", + "issue:45907" + ], + "soft_pairs": [ + "issue:44734|issue:45406", + "issue:43504|issue:45698", + "issue:43575|issue:43873", + "issue:44617|issue:45584", + "issue:43931|issue:44060", + "issue:43761|issue:44242", + "issue:32090|issue:35141", + "issue:45563|issue:45584", + "issue:44464|issue:44871", + "issue:44792|issue:44977", + "issue:44479|issue:44734", + "issue:43065|issue:44610", + "issue:44521|issue:45381", + "issue:43827|issue:43994", + "issue:43906|issue:45290", + "issue:44464|issue:45563", + "issue:44534|issue:44913", + "issue:44933|issue:45362", + "issue:44190|issue:44964", + "issue:41720|issue:45245", + "issue:44792|issue:45600", + "issue:44855|issue:45588", + "issue:44737|issue:45803", + "issue:43388|issue:44079", + "issue:45127|issue:45357", + "issue:44462|issue:45907", + "issue:43335|issue:44855" + ], + "prepared_review_unit_hash": "3375ef60bbe5e5295191826186b4790a8338d015895d74246414147d1058f449", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35993, + "estimated_input_tokens": 8999, + "estimated_eval_tokens": 18254 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated issues with only superficial title overlap; I don\u2019t see any soft pair that is confidently the same underlying bug. The clearest standalone issue here is the ModelOutput None-key bug.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44079 is a precise, self-contained framework bug with clear failure semantics and broad utility as a representative issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44079 is the most canonical-looking issue: specific, actionable, and not tied to a model-specific edge case or a flaky environment report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Both touch `transformers serve`, but they describe different failures: KV-cache tensor indexing vs a missing `_tokenizer` attribute." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "One is a legacy preset-loading field problem; the other is loading the wrong custom module after save/load. Different code paths." + }, + { + "left": "issue:43575", + "right": "issue:43873", + "accept": false, + "reason": "OOM under tensor parallelism and offloading failures with quantization are different resource-management problems." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video CUDA OOM and Whisper empty-transcription failure are unrelated model-specific bugs." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Weight-shape mismatch when loading Qwen3-VL is not the same as an incorrect tied-weights warning in Qwen3-Next." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "CLIP hidden-states regression and missing load-balancing loss are different forward-pass behaviors in different model families." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Trainer GPU broadcast `NoneType` error and embedding reinitialization after resize are separate mechanisms." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale generation warning is unrelated to Whisper failing on empty transcription after token alignment." + }, + { + "left": "issue:44464", + "right": "issue:44871", + "accept": false, + "reason": "Chunked generation inconsistency with compiled forward and Gemma EOS config mismatch are different bugs." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "A Janus image-generation test failure and Qwen3.5 flash-attention generation issues are not the same defect." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video-input regression in Qwen multimodal models and a serve KV-cache crash are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "A dummy Conv2d implementation and a processor/model input-size mismatch are different issues." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Zero assistant masks in multimodal chat templates and wrong video vision_position_ids are distinct multimodal bugs." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still mentioning `pipeline()` and SigLIP2 nonsense outputs are different problems." + }, + { + "left": "issue:43906", + "right": "issue:45290", + "accept": false, + "reason": "`43906` is only a reproduction marker for another issue; the titles alone do not establish the same concrete bug as the chat-template/tool-call crash." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Generation inconsistency under compilation and a stale warning about `num_return_sequences` are not the same code-path problem." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk values and GPTNeoX config reload behavior are unrelated regressions." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Missing `image_utils` import and a Qwen3.5 chat crash do not describe the same failure." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading failure in a training script and model loading failure for Phi-4 multimodal are different subsystems." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "CUDA assert on auto device mapping and a category-cardinality runtime limit are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Janus test failure and stale auto-mapping references to removed configs are different CI/model issues." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python parsing failure in DebertaV2 import and flash-attention `s_aux=None` crash are unrelated." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "CPU-bound rotary-position computation and a bare-except bug in tensor conversion are different defects." + }, + { + "left": "issue:43388", + "right": "issue:44079", + "accept": false, + "reason": "Metric gathering drops tuple labels in the last batch; ModelOutput key assignment with prior `None` is a separate data-structure bug." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA/vocabulary merge collapse and incorrect visual encoder keys during save_pretrained are different persistence bugs." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "Ignoring `tokenizer.json` and failing to forward `revision` are separate Hub-loading issues." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Sparse-layer construction in SwitchTransformers and a Python 3.13 import parse error are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 30, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36246", + "issue:38175", + "issue:38617", + "issue:39692", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:43065", + "issue:43295", + "issue:43335", + "issue:43404", + "issue:43479", + "issue:43540", + "issue:43550", + "issue:43643", + "issue:43644", + "issue:43650", + "issue:43976", + "issue:44016", + "issue:44077", + "issue:44079", + "issue:44112", + "issue:44186", + "issue:44190", + "issue:44220", + "issue:44246", + "issue:44279", + "issue:44466", + "issue:44496", + "issue:44561", + "issue:44683", + "issue:44734", + "issue:44855", + "issue:44913", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45216", + "issue:45245", + "issue:45276", + "issue:45375", + "issue:45406", + "issue:45446", + "issue:45542", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803" + ], + "soft_pairs": [ + "issue:44938|issue:45542", + "issue:44077|issue:44079", + "issue:43976|issue:44855", + "issue:43644|issue:44913", + "issue:43550|issue:44190", + "issue:43540|issue:44734", + "issue:44112|issue:45600", + "issue:44496|issue:45803", + "issue:45276|issue:45584", + "issue:43335|issue:44190", + "issue:43650|issue:44220", + "issue:39692|issue:44190", + "issue:45245|issue:45584", + "issue:43295|issue:45276", + "issue:43479|issue:45375", + "issue:41950|issue:44016", + "issue:36246|issue:41628", + "issue:36246|issue:38617", + "issue:44186|issue:44734", + "issue:38175|issue:43065", + "issue:44683|issue:45446", + "issue:44466|issue:44913", + "issue:44734|issue:45632", + "issue:42175|issue:44279", + "issue:44016|issue:44220", + "issue:43404|issue:45216", + "issue:44561|issue:45632", + "issue:44246|issue:45003", + "issue:43643|issue:44496", + "issue:45406|issue:45632", + "issue:44964|issue:45600", + "issue:44112|issue:44964" + ], + "prepared_review_unit_hash": "170330efc71c1dd7d5bdc9ba9e466b581c7b122e298b5aa7acd1e653e789c034", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37420, + "estimated_input_tokens": 9355, + "estimated_eval_tokens": 18966 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly unrelated issues across different Transformers features and regressions. I don\u2019t see any soft pair that is strong enough to treat as the same underlying bug/change, so I\u2019d keep the items separate. The broad v5 serialization regression is the best cluster representative among the issues listed.", + "confidence": 0.41, + "canonical_issue_reason": "issue:43644 is the broadest, most central regression here: it targets a core v5 save/load behavior with wider model impact than the narrower model-, pipeline-, or environment-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest representative issue in the set because it describes a concrete, high-impact core regression rather than a niche model-specific failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Different failure classes: Python 3.14 import/load issue versus a missing TensorFlow backend dependency when only tensorboard is installed." + }, + { + "left": "issue:44077", + "right": "issue:44079", + "accept": false, + "reason": "Unrelated bugs: one is about an optional post_init constraint in patchtsmixer, the other about ModelOutput key assignment semantics." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Both are version-related import problems, but they affect different code paths and root causes (package compatibility vs. JIT parsing of a model file)." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "One is a v5 buffer serialization regression; the other is GPTNeoXConfig losing rotary_pct on reload. Different model/config paths." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Different subsystems and symptoms: Bamba torch.compile/SDPA failure versus dataset loading in an image-classification example script." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Different products and code paths: Qwen3OmniMoe video input validation versus transformers serve KV-cache indexing in /v1/responses." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "Both touch CI, but one is a stale GraniteSpeech test and the other is auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:44496", + "right": "issue:45803", + "accept": false, + "reason": "An unrecognized-model/config issue is not the same bug as a bare-except swallowing interrupts/errors in FuyuBatchFeature." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Different model behaviors: Gemma4 embedding resize bug versus Whisper empty-transcription generation failure." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers config-layer construction and a local dataset loading failure are unrelated." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "The first is a non-informative placeholder issue; the second is an fbank feature extraction bug. No shared underlying change." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "Documentation/example errors in SigLIP2 are not the same as a dataset-loading script failure." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Category-count overflow and Whisper special-token generation are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Processor/tokenizer regression and Gemma4 embedding-resize behavior are different problems." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Different config classes and fields; these are separate multimodal config initialization bugs." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "A video-classification pipeline processor lookup bug is unrelated to a notebook syntax error." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and likely different fixes: Qwen2_5_VLImageProcessor versus AutoImageProcessor." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different missing imports in different modules; no evidence of a shared root cause." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "LayoutLMv2 tokenizer padding/NER crashes are unrelated to serve KV-cache continuation indexing." + }, + { + "left": "issue:38175", + "right": "issue:43065", + "accept": false, + "reason": "A SigLIP2 probability anomaly and a dummy Conv2d in Sam3PixelDecoder are different model-level bugs." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both mention flex_attention, but one is a compiled flex_attention failure on torch>=2.9 while the other is a version-check/import bug for AuxRequest; not enough to merge." + }, + { + "left": "issue:44466", + "right": "issue:44913", + "accept": false, + "reason": "Serialization of tied weights depending on device is a different issue from GPTNeoXConfig rotary_pct not persisting on reload." + }, + { + "left": "issue:44734", + "right": "issue:45632", + "accept": false, + "reason": "Serve KV-cache tensor indexing and trust_remote_code cache-path collisions are unrelated code paths." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "A TensorFlow backend packaging problem is not the same as a generic dependency issue." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and audio feature extraction bug are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "lm_head weight tying in Mistral3ForConditionalGeneration is a different bug from Qwen3.5 checkpoint save_pretrained serialization." + }, + { + "left": "issue:44561", + "right": "issue:45632", + "accept": false, + "reason": "Removal of is_torch_fx_available breaking trust_remote_code models is separate from local-model cache-path collisions." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Slow import and unsafe sys.modules access are different issues, even if both concern import-time behavior." + }, + { + "left": "issue:45406", + "right": "issue:45632", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to trust_remote_code cache path collisions." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Loading Phi-4-multimodal-instruct and auto_mappings referencing removed Sam3LiteText configs are different breakages." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "A stale CI test failure and a model-loading regression are not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 31, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:38175", + "issue:39692", + "issue:41950", + "issue:42907", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43577", + "issue:43673", + "issue:43824", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44242", + "issue:44279", + "issue:44336", + "issue:44351", + "issue:44368", + "issue:44496", + "issue:44521", + "issue:44617", + "issue:44683", + "issue:44933", + "issue:44964", + "issue:44987", + "issue:45020", + "issue:45042", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45325", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45542", + "issue:45584", + "issue:45593", + "issue:45600", + "issue:45759", + "issue:45823" + ], + "soft_pairs": [ + "issue:42907|issue:43441", + "issue:43854|issue:44964", + "issue:43295|issue:45335", + "issue:44016|issue:44062", + "issue:43976|issue:44987", + "issue:43673|issue:44336", + "issue:43531|issue:45600", + "issue:44496|issue:45042", + "issue:43526|issue:43901", + "issue:43824|issue:44933", + "issue:43299|issue:45216", + "issue:45335|issue:45584", + "issue:44075|issue:44368", + "issue:43976|issue:44279", + "issue:43901|issue:44496", + "issue:43381|issue:45593", + "issue:43299|issue:45357", + "issue:44351|issue:45278", + "issue:44496|issue:45759", + "issue:44521|issue:45325", + "issue:44242|issue:45200", + "issue:43976|issue:45405", + "issue:41950|issue:45020", + "issue:43479|issue:43525", + "issue:43577|issue:45216", + "issue:41950|issue:45823", + "issue:38175|issue:44617", + "issue:44279|issue:45405", + "issue:39692|issue:43901", + "issue:43299|issue:43854", + "issue:44683|issue:45542" + ], + "prepared_review_unit_hash": "38275d3f886c76e2757f0075006a8870a86accd1dfbe02d9b188ecf4c8bb379c", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36903, + "estimated_input_tokens": 9226, + "estimated_eval_tokens": 18708 + }, + "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 docs/import issues. The only strong thematic center is recent-version breakage around model loading and custom/remote-code models, so I\u2019d use that as the representative issue, but I would not deduplicate any of the soft pairs on titles alone.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45020 is the broadest and most representative report in the set: it frames the common theme of recent Transformers versions breaking model loading for custom/remote_code models. It is the best umbrella issue for this mixed regression cluster, even though it is not a strict duplicate of most others.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 best captures the shared cluster theme of recent-version compatibility regressions and model-loading breakages, and it is broad enough to serve as the main triage anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Same model family, but one is about saving dequantized Ministral/Devstral models and the other is a FlashAttention loading failure. Different bugs and code paths." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different root causes. Not the same underlying bug." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor/tokenizer regression versus resize_token_embeddings affecting decoder.embed_tokens. Different subsystems and failure modes." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken keyword conflict are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python-version support issue versus a specific model loading failure. Too different to merge." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Generation cache regression and ANSI escape handling in loading_report are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior versus auto_mappings cleanup for Sam3LiteText configs. Different problems." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized-model config error versus PIL backend torchvision dependency regression. Different code paths." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug versus stale pipeline documentation. Not the same bug." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols and different modules; no clear shared underlying defect." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3VL MoE loading regression versus Qwen3.5 save_pretrained checkpoint corruption. Related model family, but different concrete bugs." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize bug and Whisper empty-transcription generation failure are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Unused SGD args and a warning about tie_word_embeddings are not the same issue." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python compatibility regression and a vague dependency issue do not point to the same concrete defect." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Pipeline docs mismatch versus unrecognized model/config error. Different artifacts and different fixes." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing in eval mode and missing auxiliary losses are unrelated training behaviors." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3VL loading regression versus save_pretrained visual-encoder-key regression. Same family, but not the same underlying bug." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "A specific missing import and a broad collection of import errors are too weakly linked without evidence of the same root cause." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized model config versus AutoModelForCausalLM.from_config not unwrapping text_config. Different API paths." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "All-zero assistant masks in multimodal apply_chat_template is unrelated to Qwen2.5-VL rope position scaling." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "Missing load-balancing loss when router logits are off versus mm_token_type_ids defaulting behavior. Different model-training bugs." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python-version support regression and MIN_PEFT_VERSION bump are unrelated." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification pipeline processor selection and remote_code model loading are different problems." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Phi4MultimodalConfig default-initialization bug and missing pad_token_id on Llama4Config are distinct configuration issues." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "BLIP2 dtype propagation and Qwen3.5 save_pretrained corruption are unrelated." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Video pipeline image-processor lookup and Gemma4 PLE device mismatch are different failure modes." + }, + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 and Sam3Video CUDA OOM are unrelated." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Generic dependency issue and unreleased PEFT minimum version are not the same bug." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "SigLIP2 docs example errors and stale pipeline docs mention are both documentation-related, but they are different documents and different errors." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Qwen3VL loading regression and GLM-4.7 flash-loading test failure are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and tensorboard/TensorFlow backend detection are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 32, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:33290", + "issue:41950", + "issue:42175", + "issue:42994", + "issue:43065", + "issue:43262", + "issue:43295", + "issue:43317", + "issue:43441", + "issue:43450", + "issue:43650", + "issue:43653", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43901", + "issue:43976", + "issue:43994", + "issue:44016", + "issue:44162", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44291", + "issue:44297", + "issue:44361", + "issue:44462", + "issue:44466", + "issue:44521", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:44898", + "issue:45081", + "issue:45276", + "issue:45290", + "issue:45397", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45823", + "issue:45834", + "issue:45907" + ], + "soft_pairs": [ + "issue:43295|issue:44779", + "issue:44462|issue:45081", + "issue:41950|issue:43650", + "issue:33290|issue:45656", + "issue:44297|issue:44462", + "issue:45081|issue:45907", + "issue:44016|issue:44291", + "issue:45600|issue:45712", + "issue:43441|issue:45834", + "issue:43450|issue:44279", + "issue:44261|issue:44263", + "issue:43065|issue:44898", + "issue:42994|issue:43317", + "issue:44466|issue:45276", + "issue:43441|issue:44683", + "issue:43650|issue:45290", + "issue:44623|issue:45397", + "issue:44623|issue:44749", + "issue:43653|issue:45593", + "issue:43723|issue:44162", + "issue:43262|issue:44521", + "issue:43673|issue:43761", + "issue:43784|issue:45542", + "issue:44361|issue:45593", + "issue:44671|issue:44779", + "issue:44016|issue:45823", + "issue:43901|issue:43994", + "issue:42175|issue:43976" + ], + "prepared_review_unit_hash": "19a395c2956d0ff4f86ae84dae2bc8e87d6b42e143617334c6bffd19e38a7d65", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36089, + "estimated_input_tokens": 9023, + "estimated_eval_tokens": 18302 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: the soft pairs mostly share a subsystem label (tokenizers, processors, deepspeed, CI, model loading) but not the same concrete bug. None of the candidate pairs look mergeable as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 tokenizer regressions, but they affect different models and failure modes; not the same bug." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "One is AutoTokenizer ignoring tokenizer.json; the other is a Mistral regex patch crash. Different load paths." + }, + { + "left": "issue:41950", + "right": "issue:43650", + "accept": false, + "reason": "Unrelated topics: pipeline processor lookup vs a vague data request." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention deepspeed/optimizer, but one is OOM and the other is double-stepping. Different defects." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Both concern tokenizer persistence/loading, but one is save_pretrained metadata mismatch and the other is repository tokenizer.json being ignored." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Different bugs: Mistral tokenizer regex crash vs list_repo_files not forwarding revision." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error and init_empty_weights argument error are unrelated." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both are repo-maintenance/CI issues, but they concern different files and failures." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Different model-specific failures: FlashAttention on Ministral vs OCR index error in Kosmos2.5." + }, + { + "left": "issue:43450", + "right": "issue:44279", + "accept": false, + "reason": "Incorrect batched video shape vs generic dependency issue; no shared code-path bug." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different model internals: layernorm epsilon mismatch vs torch.split return handling." + }, + { + "left": "issue:43065", + "right": "issue:44898", + "accept": false, + "reason": "Different model components and symptoms; one is a dummy Conv2d, the other is Perceiver resolution handling." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both involve quantization/loading, but save failure and offload loading failure are distinct." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Serialization of tied weights and resize_token_embeddings propagation are different bugs." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Different attention backends and environments; not the same concrete failure." + }, + { + "left": "issue:43650", + "right": "issue:45290", + "accept": false, + "reason": "Vague data request vs apply_chat_template crash on tool-call assistant messages." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Processor save_pretrained missing files and zero3 from_pretrained are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Missing processor save artifacts vs tokenizer throughput regression; different bug classes." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "Tokenizer special-token registration and auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading problem and ESM2 model breakage are not the same defect." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both touch apply_chat_template, but one is audio sampling-rate defaulting and the other is multimodal assistant masks." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Generation cache handling and CLIP hidden_states output are unrelated code paths." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Both are optional-backend import issues, but one is nn missing and the other is tf backend detection." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "Tokenizer AttributeError and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 model regressions, but CamemBERT masked LM outputs and Deepseek tokenizer behavior are different bugs." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error vs Gemma4 device mismatch; no shared underlying issue." + }, + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Documentation mismatch in pipeline docs and wrong SigLIP2 outputs are not the same bug." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "TensorFlow dependency packaging issue and Python version compatibility issue are distinct." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 33, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42491", + "issue:42757", + "issue:43122", + "issue:43299", + "issue:43352", + "issue:43450", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43596", + "issue:43643", + "issue:43650", + "issue:43673", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:44162", + "issue:44188", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44351", + "issue:44373", + "issue:44485", + "issue:44496", + "issue:44561", + "issue:44589", + "issue:44617", + "issue:44749", + "issue:44857", + "issue:44871", + "issue:44987", + "issue:45003", + "issue:45200", + "issue:45250", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45593" + ], + "soft_pairs": [ + "issue:43122|issue:43650", + "issue:44987|issue:45542", + "issue:43824|issue:43976", + "issue:44297|issue:45397", + "issue:43525|issue:43881", + "issue:44188|issue:44336", + "issue:43525|issue:43643", + "issue:43450|issue:44617", + "issue:42491|issue:43299", + "issue:43901|issue:44246", + "issue:44589|issue:45362", + "issue:45561|issue:45563", + "issue:41628|issue:43976", + "issue:44336|issue:44373", + "issue:44242|issue:45593", + "issue:43352|issue:45250", + "issue:44295|issue:45397", + "issue:41950|issue:44220", + "issue:44246|issue:44496", + "issue:43531|issue:45362", + "issue:36246|issue:42757", + "issue:43526|issue:45003", + "issue:36246|issue:44351", + "issue:44336|issue:44485", + "issue:45003|issue:45561", + "issue:43673|issue:44561", + "issue:44246|issue:44871", + "issue:44242|issue:44857", + "issue:44162|issue:45200", + "issue:44295|issue:44749", + "issue:36246|issue:43596" + ], + "prepared_review_unit_hash": "8c282ab5eaea7574b317f7ee6cdc1fde24cb3e030d104d68a5c3965f6811aa9d", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36619, + "estimated_input_tokens": 9155, + "estimated_eval_tokens": 18566 + }, + "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. A small subset concerns import regressions, but the rest span tokenizer behavior, model-loading failures, docs, performance, and model-specific training/runtime bugs. None of the soft pairs look like true duplicates.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 41628 is the best single representative because it is a clear, user-facing top-level import regression and is closest to the recurring import/error-failure theme in the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "41628 is the most suitable cluster anchor: concrete, reproducible, and broadly understandable, while still being representative of the import-regression subset better than the model-specific or docs-only reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43122", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer behavior regression versus a placeholder/low-signal issue; no shared bug or code path." + }, + { + "left": "issue:44987", + "right": "issue:45542", + "accept": false, + "reason": "Different failure modes: model loading on transformers>=5.1.0 versus TensorFlow backend discovery when only tensorboard is installed." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Specific Qwen2_5_VL import failure versus a Python version compatibility problem; unrelated root causes." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer serialization mismatch and Gemma-4 ZeRO3 loading are different bugs in different paths." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Llama4Config missing pad_token_id is unrelated to glm-4v-9b loading failure." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Torch.compile attention-kernel divergence and ANSI output in loading_report are completely unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Both mention config fields, but one is a missing model attribute and the other is a trust_remote_code config hydration issue." + }, + { + "left": "issue:43450", + "right": "issue:44617", + "accept": false, + "reason": "Batched video processor shape bug versus CUDA OOM in Sam3Video; not the same underlying issue." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen/moe models, but one is a HF4.x to HF5.x LoRA incompatibility and the other is a sliding-window loading regression." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch for return_all_scores versus import-time slowness; different categories and symptoms." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage type error and Qwen3.5 chat crash are unrelated runtime failures." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "Pytest-xdist file race and a stale generation warning are different parts of the codebase and different bugs." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Top-level import error versus Python 3.9+/3.10+ support matrix issue; not the same defect." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "Terminal ANSI formatting bug and a wrong docstring for position_ids do not share a concrete code path." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Both concern missing losses, but they affect different models and gating conditions; too broad to treat as one bug." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "Nemotron Flash Attention 2 support gap is not the same as a generic Flash Attention 2 topic issue." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Position_ids buffer access bug versus Gemma-4 ZeRO3 loading failure; unrelated." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification processor selection bug versus audio fbank extraction issue." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Import-time performance issue and unrecognized model/config loading are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5 chat crash are different problems." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they concern different symbols from different packages and different fixes." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports for different APIs; same symptom class, not the same bug." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "ANSI code emission and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Unsafe sys.modules access and a pytest worker race do not overlap in code path or fix." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Chunked-prefill cache handling and removal of is_torch_fx_available are different v5 regressions." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowness and eos_token_id configuration inconsistency are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Load-balancing loss gating and a float16 AMP crash in LwDetrImageLoss are different bugs." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma-4 mm_token_type_ids defaults are unrelated model-specific issues." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Position_ids buffer access and slower filtering after upgrade are unrelated performance/runtime issues." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "Qwen2_5_VL image-processor import error and BertModel zero3 index error are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 34, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:42886", + "issue:43065", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43408", + "issue:43475", + "issue:43540", + "issue:43653", + "issue:43673", + "issue:43901", + "issue:43931", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44246", + "issue:44297", + "issue:44336", + "issue:44368", + "issue:44466", + "issue:44496", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44877", + "issue:44933", + "issue:44977", + "issue:44987", + "issue:44995", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45278", + "issue:45292", + "issue:45335", + "issue:45356", + "issue:45375", + "issue:45397", + "issue:45459", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45656", + "issue:45701", + "issue:45759" + ], + "soft_pairs": [ + "issue:44779|issue:45292", + "issue:44877|issue:45375", + "issue:45292|issue:45584", + "issue:45292|issue:45701", + "issue:43901|issue:45003", + "issue:42886|issue:45229", + "issue:43232|issue:44336", + "issue:44496|issue:45003", + "issue:44987|issue:45459", + "issue:43065|issue:43408", + "issue:45375|issue:45759", + "issue:43257|issue:45137", + "issue:43262|issue:44246", + "issue:44995|issue:45200", + "issue:44060|issue:44077", + "issue:43475|issue:43540", + "issue:44466|issue:45335", + "issue:43653|issue:44162", + "issue:44995|issue:45588", + "issue:44368|issue:45563", + "issue:44368|issue:44977", + "issue:45397|issue:45656", + "issue:43673|issue:44734", + "issue:45137|issue:45600", + "issue:43931|issue:44368", + "issue:44933|issue:45459", + "issue:44869|issue:45356", + "issue:43232|issue:43673", + "issue:44297|issue:44623", + "issue:43673|issue:45278" + ], + "prepared_review_unit_hash": "f2016cf240b0e95c9506908e1f173f567e9e077d2c0874a217e207c7c55321ef", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36678, + "estimated_input_tokens": 9170, + "estimated_eval_tokens": 18596 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is not a true duplicate set: it mixes unrelated tokenizer, config, generation, Deepspeed, and SAM/Qwen/Gemma issues. None of the soft pairs appear to be the same underlying bug or change.", + "confidence": 0.88, + "canonical_issue_reason": "issue:45701 is the broadest, most cluster-like issue: a version-dependent tokenization regression. It is the closest representative of the tokenizer/version-regression subset, even though it is not a duplicate of the other items.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45701 is the best single issue to represent the loose theme of version-induced tokenizer behavior changes across the set. It is still only a thematic anchor, not a duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Deepseek tokenizer correctness regression vs resize_token_embeddings behavior; different code paths and symptoms." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both involve config handling/strictness, but they target different models and different failure modes." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize bug and Whisper empty-transcription generation crash are unrelated." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "Resize-token-embeddings behavior is not the same as a tokenization change across versions." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs mismatch for a pipeline option is unrelated to unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are different bugs." + }, + { + "left": "issue:43232", + "right": "issue:44336", + "accept": false, + "reason": "Generation kwargs sync issue is unrelated to ANSI output in loading_report." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config loading error and sys.modules access bug do not share the same code defect." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Model loading failure for a specific repo and protobuf-related tokenizer error masking are distinct." + }, + { + "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." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Missing config field under strict handling and text_config unwrapping for composite configs are different loading bugs." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Deepspeed weight conversion issue and ZeRO3 deque underflow are unrelated runtime failures." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio sampling-rate default mismatch is unrelated to slow import performance." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Stale indexer cache on second forward pass is different from missing mm_token_type_ids defaulting." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "Incorrect tied-weights warning and patchtsmixer post_init validation are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM3 video encoder output attribute error and Qwen3OmniMoe video-input processing failure are different model bugs." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Serialization inconsistency for lm_head.weight is not the same as resize_token_embeddings not updating decoder.embed_tokens." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration bug and ESM2 breakage are not the same defect." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Second-forward stale cache crash is unrelated to flash_attention handling of s_aux=None." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Both mention warnings, but they concern different generation behaviors and different warning conditions." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Tie-word-embeddings warning is not the same as flash-attention generation failure." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma-4 ZeRO3 loading and double optimizer step under deepspeed are different issues." + }, + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Chunked-prefill cache missing and KV-cache continuation tensor indexing are both generation/cache related but not the same bug." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "ZeRO3 deque pop error and stale Sam3LiteText references in auto_mappings are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and tie_word_embeddings warning are different loading/model-config problems." + }, + { + "left": "issue:44933", + "right": "issue:45459", + "accept": false, + "reason": "Missing import from image_utils is unrelated to tokenizer error masking when protobuf is absent." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper trailing-replacement decode crash and Kimi tokenizer codec regression are separate tokenizer bugs." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both are generation-related, but one is kwargs syncing and the other is cache availability in chunked prefill." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "tokenizer.save_pretrained metadata mismatch and processor.save_pretrained missing files are different save-path bugs." + }, + { + "left": "issue:43673", + "right": "issue:45278", + "accept": false, + "reason": "Generation cache missing in chunked prefill and broad import failures after upgrade are not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 35, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42831", + "issue:42994", + "issue:43232", + "issue:43377", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43704", + "issue:43873", + "issue:44038", + "issue:44162", + "issue:44164", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44351", + "issue:44361", + "issue:44623", + "issue:44661", + "issue:44749", + "issue:44792", + "issue:44933", + "issue:44945", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45200", + "issue:45230", + "issue:45310", + "issue:45375", + "issue:45381", + "issue:45397", + "issue:45478", + "issue:45656", + "issue:45712", + "issue:45753" + ], + "soft_pairs": [ + "issue:43646|issue:44162", + "issue:44361|issue:44749", + "issue:44623|issue:45656", + "issue:44749|issue:45656", + "issue:28282|issue:30990", + "issue:44749|issue:45310", + "issue:42994|issue:43482", + "issue:43475|issue:43643", + "issue:42371|issue:45092", + "issue:43475|issue:45381", + "issue:45230|issue:45397", + "issue:44038|issue:44623", + "issue:41628|issue:44933", + "issue:45081|issue:45103", + "issue:42994|issue:43525", + "issue:43475|issue:44661", + "issue:45375|issue:45478", + "issue:42673|issue:43704", + "issue:43232|issue:44188", + "issue:45310|issue:45375", + "issue:43873|issue:44945", + "issue:42831|issue:44222", + "issue:41628|issue:44351", + "issue:44945|issue:45753", + "issue:44222|issue:45310", + "issue:42222|issue:45200", + "issue:43673|issue:44188", + "issue:42994|issue:44164", + "issue:36010|issue:41628", + "issue:42175|issue:45712", + "issue:43232|issue:44792", + "issue:44038|issue:45310", + "issue:43377|issue:44242" + ], + "prepared_review_unit_hash": "75fcb2a0ff93d32c98527c4950fddebcc1afea3f916f3479f84cada4193c160a", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 36602, + "estimated_input_tokens": 9151, + "estimated_eval_tokens": 18558 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous set of Transformers-v5 regression reports across imports, model init/load, quantization, and generation. Only one soft edge looks like a true duplicate: the repeated Qwen3ForCausalLM VRAM-leak report. The Qwen3.5 MoE-related items are similar in theme but still appear to be distinct bugs rather than the same concrete failure.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the broadest umbrella report in the set: a Transformers 5.0 custom-model initialization regression. It is the best single anchor for the overall theme, even though the cluster contains several distinct sub-bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the most representative issue because it describes the highest-level regression pattern rather than one model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Both are v5-init regressions, but one is generic custom-model init breakage and the other is ESM2-specific; not clearly the same bug." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "Tokenizer AttributeError vs upgrade-induced slowdown; different failure modes." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "save_pretrained missing files vs deepspeed optimizer stepping twice; different code paths." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance regression in data filtering is unrelated to a duplicated optimizer step." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "PyTorch ImportError vs Sentence Transformers loading hang; separate runtime issues." + }, + { + "left": "issue:44749", + "right": "issue:45310", + "accept": false, + "reason": "Slow filtering after upgrade is not the same as Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized saving failure vs GGUF loading failure; save/load bugs, not one issue." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM3 attribute error vs missing fields from trust_remote_code config; different objects and causes." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API warning is unrelated to InternVL2 meta-initialization incompatibility." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "Different models and different attribute/position-id failures." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "Generic bug report vs gemma-4 zero3 from_pretrained issue; no clear overlap." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe from_pretrained bug and processor.save_pretrained missing files are different components." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Both involve imports, but AutoImageProcessor and image_utils failures are different symbols/code paths." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Tokenizer regex crash vs auto_docstring annotations crash; unrelated." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized save failure vs missing pad_token_id on Llama4Config; distinct regressions." + }, + { + "left": "issue:43475", + "right": "issue:44661", + "accept": false, + "reason": "SAM3 output attribute error vs tokenizer-mapping add-new-model-like failure; unrelated." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Missing config field in Qwen3_5MoeVisionConfig vs Qwen3.5 MoE from_pretrained error; same family, different bug." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Exact same title and same Qwen3ForCausalLM VRAM-leak symptom in multiple dataloader threads; this is the same report." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Generation kwargs update issue vs torch.compile attention-kernel branching issue; different code paths." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Qwen3.5 MoE from_pretrained error and missing vision config field are related but not the same concrete failure." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization/offloading behavior vs incorrect pipeline-parallel output; different problems." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FP8 accuracy regression vs FP8 save_pretrained failure; different symptoms." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "AutoImageProcessor import error vs HybridCache import error; separate missing exports." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output bug vs infinite loop on a specific image; unrelated." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained on MoE vs Qwen3.5 MoE from_pretrained error; different operation and failure mode." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "VitPose breakage vs Gemma 4 mm_token_type_ids defaulting; unrelated model families." + }, + { + "left": "issue:43673", + "right": "issue:44188", + "accept": false, + "reason": "Chunked_prefill cache regression vs torch.compile bidirectional-skip branching issue; distinct generation bugs." + }, + { + "left": "issue:42994", + "right": "issue:44164", + "accept": false, + "reason": "Quantized model saving failure vs extra_state save/from_pretrained handling; serialization-related but different defects." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "GenerationMixin import error vs AutoImageProcessor import error; different missing symbols." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "TensorFlow backend packaging issue vs leftover dummy PyTorch classes; different packaging problems." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation kwargs bug vs janus generate test failure; not the same underlying issue." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Different Qwen MoE variants and different failure modes; not a clear duplicate." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "MIMI batching/padding-mask bug vs router-logits load-balancing-loss bug; unrelated architecture issues." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster-level summary is slightly inconsistent with the structured verdicts: it says only two soft edges look like true duplicates, but only one pair is actually accepted. More importantly, the Qwen3.5 MoE pair is described as a duplicate-like case in the summary even though the verdict marks it false, so the final writeup is not fully grounded in the packet. The rest of the conservative rejections look reasonable." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 36, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:42175", + "issue:42831", + "issue:42898", + "issue:43116", + "issue:43278", + "issue:43479", + "issue:43504", + "issue:43526", + "issue:43644", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:43976", + "issue:44060", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44206", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44373", + "issue:44448", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44933", + "issue:44964", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45137", + "issue:45310", + "issue:45356", + "issue:45405", + "issue:45561", + "issue:45656", + "issue:45712", + "issue:45812" + ], + "soft_pairs": [ + "issue:43526|issue:45561", + "issue:44496|issue:45561", + "issue:43504|issue:45561", + "issue:44964|issue:45137", + "issue:42898|issue:44568", + "issue:42175|issue:44933", + "issue:30990|issue:35141", + "issue:43479|issue:44869", + "issue:43479|issue:44079", + "issue:43976|issue:44933", + "issue:42898|issue:45812", + "issue:43784|issue:43824", + "issue:44188|issue:44373", + "issue:44623|issue:45310", + "issue:43278|issue:45137", + "issue:44734|issue:45103", + "issue:30333|issue:33290", + "issue:43723|issue:44568", + "issue:42898|issue:45356", + "issue:44060|issue:45125", + "issue:43874|issue:45092", + "issue:42898|issue:44779", + "issue:45405|issue:45712", + "issue:42898|issue:44448", + "issue:44295|issue:45656", + "issue:44933|issue:45712", + "issue:44222|issue:44297", + "issue:43116|issue:44190", + "issue:44297|issue:45656", + "issue:43479|issue:43644", + "issue:44206|issue:45103", + "issue:44222|issue:44623", + "issue:42831|issue:44623" + ], + "prepared_review_unit_hash": "9ef9e86958d1a1a7427bda638dd6080cf321840f7ca15836accd358d611f2a68", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 33, + "serialized_chars": 37179, + "estimated_input_tokens": 9295, + "estimated_eval_tokens": 18846 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: model loading, tokenizer regressions, training/runtime errors, and tooling issues. None of the soft pairs look like the same underlying bug, so all are rejected.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43824 is the strongest canonical candidate: it has high discussion/inbound-reference activity and a clear, user-facing, concrete failure mode with enough detail to anchor duplicate triage.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43824 is the best standalone issue in this set because it is well-scoped, highly referenced, and describes a specific reproducible import failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "Different subsystems and failures: BEiT label reduction vs a pytest-xdist file race." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Unrecognized-model/config loading error vs test harness race on captured_info.txt." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Pretrained preset loading/legacy field issue is unrelated to xdist test-worker file contention." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Model loading failure for Phi-4 multimodal is unrelated to DeepSpeed ZeRO3 deque underflow." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both involve tokenization, but one is a v5 behavior change while the other is tokenizer-specific BOS/EOS insertion regression." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Backend install packaging issue vs missing/nonexistent image_utils import; different failure classes." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-Transformers loading hang is unrelated to token embedding reinitialization in post_init." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Multimodal config default initialization bug vs Whisper word-timestamp decode crash; different code paths." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Config initialization with None is unrelated to ModelOutput key assignment semantics." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility regression vs missing import from image_utils." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; not the same concrete bug." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Sentence-transformers import NameError is unrelated to Qwen2.5-VL missing export/import error." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "torch.compile attention-kernel divergence and a docstring typo are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Missing processor-save files vs Qwen3.5-Moe from_pretrained failure; different loading/saving paths." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift during eval is unrelated to DeepSpeed ZeRO3 deque errors." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "KV-cache continuation crash in serve is unrelated to auto_docstring crashing on future annotations." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLFlow reporting on failed training jobs is unrelated to adafactor/deepspeed OOM." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "General AutoTokenizer loading failure is not the same bug as mdeberta BOS/EOS handling regression." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but the affected models and codec/warning behavior differ; no same underlying bug evidence." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Tied-weights warning bug in Qwen3-Next is unrelated to missing _tp_plan in Qwen3.5-Moe." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Missing image-patch method in GLM46V is unrelated to old InternVL2 remote-code meta-init incompatibility." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer regressions affecting different models; no indication they are the same fix." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "PEFT version pin mismatch is unrelated to dummy class cleanup/check_repo failures." + }, + { + "left": "issue:42898", + "right": "issue:44448", + "accept": false, + "reason": "Pegasus output drift and clean_up_tokenization_spaces behavior change are different regression reports." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Buffer read error for position_ids is unrelated to DeepSpeed double optimizer-step behavior." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Missing image_utils import and leftover dummy classes are unrelated packaging/codegen issues." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "FP8 MoE save_pretrained bug is unrelated to tokenizer_class metadata mismatch on save_pretrained." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Multi-label example returning empty results is unrelated to local dataset loading in another example script." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata bug is unrelated to DeepSpeed optimizer-step duplication." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Multimodal config defaults and junk-filled non-persistent buffers are different initialization problems." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature extractor center-arg crash is unrelated to auto_docstring annotation handling." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "FP8 MoE save_pretrained issue is not the same as processor.save_pretrained missing files." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is unrelated to processor save_pretrained file omissions." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 37, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:30990", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:38175", + "issue:39401", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42994", + "issue:43232", + "issue:43479", + "issue:43723", + "issue:43756", + "issue:43824", + "issue:44038", + "issue:44060", + "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:44568", + "issue:44623", + "issue:44871", + "issue:44913", + "issue:44977", + "issue:44991", + "issue:45092", + "issue:45103", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45406", + "issue:45593", + "issue:45834" + ], + "soft_pairs": [ + "issue:41628|issue:42757", + "issue:22355|issue:28282", + "issue:30333|issue:36331", + "issue:43232|issue:45834", + "issue:30333|issue:34689", + "issue:44222|issue:44484", + "issue:44297|issue:45310", + "issue:43723|issue:44977", + "issue:28282|issue:30333", + "issue:30333|issue:35141", + "issue:44263|issue:45092", + "issue:44263|issue:44360", + "issue:44483|issue:45406", + "issue:44373|issue:44485", + "issue:44514|issue:45103", + "issue:44295|issue:44623", + "issue:43723|issue:44871", + "issue:44261|issue:45092", + "issue:45092|issue:45357", + "issue:43723|issue:44991", + "issue:30990|issue:38175", + "issue:43723|issue:44442", + "issue:41628|issue:43824", + "issue:43479|issue:44913", + "issue:42831|issue:45310", + "issue:39401|issue:44568", + "issue:42886|issue:42994", + "issue:44263|issue:44373", + "issue:42757|issue:45278", + "issue:42831|issue:44038", + "issue:43756|issue:44060", + "issue:43232|issue:45593" + ], + "prepared_review_unit_hash": "887ba8a34e842c77239a2b062eee7b302cf3b78b7d5aaab97487d7847dcc112e", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36440, + "estimated_input_tokens": 9110, + "estimated_eval_tokens": 18476 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is not a true duplicate cluster: it mixes import/load regressions, tokenizer/model-specific bugs, save_pretrained issues, and unrelated training/runtime problems. Most soft pairs are only superficially similar and should be kept separate.", + "confidence": 0.37, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella here, covering multiple import regressions after a Transformers upgrade, which matches the largest visible theme in the cluster better than the narrower model-specific bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best single representative because it is the most general and open-ended report in the set; the rest are too specific or unrelated to serve as a canonical duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Different failing imports from different packages/components: AutoImageProcessor in transformers vs is_offline_mode from huggingface_hub." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "One is missing transformers.onnx; the other is missing PyTorch for AutoModel. Different dependency failures." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "Unrelated problems: MLflow reporting on failed training vs CustomTrainer.compute_loss signature breakage." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Different generation/pathology issues: sync_gpus model kwargs handling vs Kosmos2.5 long OCR index error." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training-job reporting bug vs model loading regression for Llama 3.2 vision are not the same bug." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "Both touch save_pretrained, but one is FP8/MoE saving correctness and the other is shard-size behavior." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer config mismatch on save_pretrained is unrelated to Qwen3.5-Moe from_pretrained loading errors." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Generic tokenizer loading in v5 vs flash-attention generation failure are different code paths and symptoms." + }, + { + "left": "issue:28282", + "right": "issue:30333", + "accept": false, + "reason": "Missing PyTorch import dependency and MLflow reporting on failed training are unrelated." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "Training failure reporting vs token-embedding reinitialization in post_init are different bugs." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Different model internals: GLM DSA indexer split behavior vs InternVL2 meta-initialization incompatibility." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Same broad DSA area, but one is torch.split return values and the other is a missing ReLU; not the same concrete bug." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "API request handling in chat/completions is unrelated to Gemma4Processor crashing in serve due to missing _tokenizer." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Wrong docstring for position_ids is documentation-only; GLM-5 RoPE implementation is a code behavior issue." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Batched chat-template crash with padding=False is unrelated to AttributeError in auto_docstring with future annotations." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Reading position_ids after buffer registration and processor.save_pretrained missing files are different problems." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Tokenizer loading regression and eos_token_id inconsistency are different classes of tokenizer/model config issues." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "MLA layernorm epsilon precision issue is unrelated to remote-code checkpoint/meta-init compatibility." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Meta-initialization incompatibility and incorrect visual encoder keys on save_pretrained are different bugs." + }, + { + "left": "issue:43723", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading related, but they involve different repositories/models and likely different root causes." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence Transformers hanging on load and SigLIP2 zero probabilities are not the same underlying issue." + }, + { + "left": "issue:43723", + "right": "issue:44442", + "accept": false, + "reason": "AutoTokenizer loading failure for one tokenizer is not the same as FastSpeech2ConformerTokenizer import/load failure." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Different import errors for different symbols/models: AutoImageProcessor vs Qwen2_5_VLForConditionalGeneration." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Config default-initialization bug and GPTNeoX rotary_pct not persisting on reload are different config issues." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is unrelated to Qwen3.5-Moe from_pretrained loading failure." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Different tokenizer bugs: wrong offset_mapping vs missing BOS/EOS insertion." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache loading and quantized model saving failure are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "DSA indexer split behavior and a docstring issue for position_ids are not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "45278 is a broad import-regression report, but 42757 is a specific missing symbol in huggingface_hub; not the same concrete failure." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FP8 accuracy issue and Qwen3-VL-Moe transformers v5 bug are different model/runtime failures." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "RoPE layer-dropping in Smollm3 and tied-weights warning in Qwen3-Next are different model implementation bugs." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation kwargs update after sync_gpus is unrelated to D-FINE auxiliary-loss behavior when denoising is off." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 38, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:22355", + "issue:28218", + "issue:29127", + "issue:30990", + "issue:33357", + "issue:36296", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43576", + "issue:43784", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44117", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44485", + "issue:44488", + "issue:44514", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44743", + "issue:44779", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:45200", + "issue:45278", + "issue:45290", + "issue:45600", + "issue:45753" + ], + "soft_pairs": [ + "issue:43576|issue:43784", + "issue:42831|issue:45753", + "issue:42886|issue:43066", + "issue:22355|issue:28218", + "issue:42371|issue:44263", + "issue:43122|issue:44521", + "issue:44117|issue:44704", + "issue:30990|issue:33357", + "issue:43232|issue:44743", + "issue:42907|issue:44945", + "issue:44295|issue:44297", + "issue:44704|issue:45290", + "issue:41628|issue:45278", + "issue:43976|issue:44945", + "issue:43122|issue:44514", + "issue:39401|issue:44488", + "issue:43784|issue:45600", + "issue:42907|issue:43937", + "issue:43408|issue:43475", + "issue:44261|issue:44373", + "issue:44360|issue:44485", + "issue:44625|issue:44829", + "issue:39401|issue:44521", + "issue:43116|issue:45200", + "issue:29127|issue:30990", + "issue:42907|issue:44964", + "issue:44038|issue:44297", + "issue:43122|issue:44779", + "issue:44038|issue:44295", + "issue:42907|issue:43452", + "issue:36296|issue:43116", + "issue:44038|issue:44222" + ], + "prepared_review_unit_hash": "d20d62990132357af9e9e324e1ef5ca53ebdf7f55dedc987b445b84f1ee5a801", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36782, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous issue pile rather than a true duplicate cluster. There is a loose subgroup around Transformers v5 import/loading regressions, but none of the soft pairs look like the same underlying bug or change.", + "confidence": 0.94, + "canonical_issue_reason": "No true canonical issue: the set spans unrelated import, tokenizer, generation, multimodal, FP8, and training bugs; at most, issue:45278 is a loose umbrella for post-upgrade import regressions.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the broadest representative candidate because it captures the only clear mini-theme (many import errors after upgrading), but it still does not unify the overall set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "`transformers env` CLI breakage vs a sentence-transformers import-time `nn` NameError; different code paths." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy regression vs Qwen3_5 image infinite loop; unrelated bugs." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading failure vs wrong tokenizer decoder type in v5; both tokenizer-related but not the same defect." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "Missing `transformers.onnx` module import vs extra space after added token; unrelated." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 warning/settings issue vs `torch.split` return handling in DSA indexer; different subsystems." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Cross-version tokenization drift vs multimodal assistant mask bug; not the same bug." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "Tokenizer mapping `None` handling vs AutoProcessor kwargs forwarding; different failure points." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Sentence-Transformers loading hang vs MacOS bus error loading CLIP; unrelated symptoms." + }, + { + "left": "issue:43232", + "right": "issue:44743", + "accept": false, + "reason": "Generation kwargs update after `sync_gpus` vs Qwen3.5 recurrent-state reset with cache; different code paths." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Dequantized model save failure vs incorrect pipeline-parallel outputs; unrelated." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "`position_ids` buffer access error vs tokenizer class mismatch on save; different save/load bugs." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor cached-file kwargs bug vs chat-template crash on tool-call messages; different APIs." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Specific `AutoImageProcessor` import failure vs broad post-upgrade import errors; overlap is too general to be a duplicate." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python-version compatibility issue vs pipeline-parallel output bug; unrelated." + }, + { + "left": "issue:43122", + "right": "issue:44514", + "accept": false, + "reason": "Tokenizer behavior drift vs `apply_chat_template` crash on batched multimodal input; different bugs." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 offset-mapping bug vs failure to load a specific model; not the same issue." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Import-time `nn` NameError in sentence-transformers vs stale `auto_mappings` CI break; both import-ish but not the same bug." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Model save failure vs invalid `GenerationConfig`; unrelated." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "SAM3 type warning vs missing `fpn_position_embeddings` attribute; related family, but different concrete failures." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "Layernorm epsilon precision bug vs wrong docstring; functional bug and documentation issue are not duplicates." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "DSA indexer missing ReLU vs GLM-5 RoPE implementation; unrelated model logic." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation bug vs degenerate training with flash_attention_3; different problems." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset mapping vs all-zero assistant masks for multimodal inputs; different processing stages." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Example script returning empty labels vs Gemma 4 token-type defaulting issue; unrelated." + }, + { + "left": "issue:29127", + "right": "issue:30990", + "accept": false, + "reason": "LayoutLMv3 error-message clarity vs Sentence-Transformers loading stall; no overlap." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Dequantized save bug vs failure to load Phi-4 multimodal; different model-loading/saving paths." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe bug vs tokenizer config class mismatch; unrelated." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "General tokenization drift vs Deepseek-specific incorrect tokenization; not the same concrete bug." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Qwen3-VL-Moe issue vs `position_ids` buffer access error; unrelated." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Dequantized save failure vs `gguf_file` breakage in from_pretrained; different code paths." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug vs example-script multi-label output bug; unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Qwen3-VL-Moe bug vs FP8 `save_pretrained` bug; not the same defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 39, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:33453", + "issue:38617", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42757", + "issue:42907", + "issue:43064", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43441", + "issue:43493", + "issue:43673", + "issue:43784", + "issue:43824", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43906", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44117", + "issue:44222", + "issue:44351", + "issue:44485", + "issue:44704", + "issue:44779", + "issue:44936", + "issue:44945", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45216", + "issue:45278", + "issue:45446", + "issue:45478", + "issue:45563", + "issue:45588", + "issue:45753" + ], + "soft_pairs": [ + "issue:43824|issue:44351", + "issue:45216|issue:45478", + "issue:42757|issue:44351", + "issue:43278|issue:44945", + "issue:43066|issue:44779", + "issue:43493|issue:45753", + "issue:42907|issue:44222", + "issue:43122|issue:44977", + "issue:43992|issue:44704", + "issue:43784|issue:45446", + "issue:38617|issue:45278", + "issue:42673|issue:45072", + "issue:43937|issue:44485", + "issue:22355|issue:29127", + "issue:42371|issue:43874", + "issue:43064|issue:43673", + "issue:42222|issue:43441", + "issue:43976|issue:45446", + "issue:44075|issue:44936", + "issue:43116|issue:43493", + "issue:43116|issue:43825", + "issue:44038|issue:45478", + "issue:44222|issue:45478", + "issue:45083|issue:45446", + "issue:43937|issue:45563", + "issue:33453|issue:39401", + "issue:44117|issue:45588", + "issue:39401|issue:44779", + "issue:43827|issue:45446", + "issue:43937|issue:44977", + "issue:42673|issue:43906" + ], + "prepared_review_unit_hash": "51a44ef657dd386e77a684959b9e8ec9e4663f3c1d4188aacba3d23548b52122", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 35975, + "estimated_input_tokens": 8994, + "estimated_eval_tokens": 18244 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mix of Transformers v5 regressions and model-specific bugs, not a single duplicate thread. Issue 45278 is the broadest umbrella for the import-error subset, but the rest are too heterogeneous for strong deduping. No PRs were provided.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45278 is the broadest, most cluster-like report: it frames a general upgrade regression with multiple import errors after moving to 5.5.0, which best matches the shared v5 breakage theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best representative because it is the most general symptom statement and can plausibly cover several import-related reports in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they name different missing symbols and likely different code paths." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "One is about save_pretrained checkpoint correctness, the other about from_pretrained loading; related family, but not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports from different modules; no evidence of one underlying defect." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "BF16/FP32 dtype drift during eval is unrelated to incorrect output under pipeline parallelism." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both tokenizer regressions, but they affect different models and different failure modes." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Different model families and symptoms; no sign of the same underlying bug." + }, + { + "left": "issue:42907", + "right": "issue:44222", + "accept": false, + "reason": "Both involve saving models, but one is dequantized Ministral/Devstral and the other is FP8 MoE; too different to merge." + }, + { + "left": "issue:43122", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer-version difference vs flash-attention generation failure are separate problems." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "A missing weight in from_pretrained is not the same as AutoProcessor failing to forward kwargs." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Import-time NameError in sentence-transformers is unrelated to a PyTorch version check in flex_attention." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "Generic import failure vs broad upgrade import regressions; similar theme but not demonstrably the same bug." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in dataloader threads is unrelated to dtype mismatch in bfloat16 inference." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "GenerationConfig validation and RoPE implementation are different GLM-5 issues." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "A missing module import and a LayoutLMv3 error message issue are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API guidance and a missing image-patch method are unrelated." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Optimizer-state corruption under FSDP2/PEFT is unrelated to GenerationMixin cache handling." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Broken VitPose imports and Ministral flash-attention failures are different model-specific regressions." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version support issue and flex_attention version check bug are not the same underlying defect." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Unused SGD args and evaluate-after-train failure are separate training bugs." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script multi-label output bug is unrelated to SigLIP2 implementation discrepancy." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example-script classification output bug and pipeline() error messaging are different issues." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Related Qwen3 MoE area, but one is a broad v5/moe bug and the other is a specific from_pretrained error; not enough to merge." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Feature-extract length helper behavior is unrelated to an import version check." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "GLM-5 generation config validation and a stale warning about num_return_sequences are different code paths." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Tokenizer loading regression and offset-mapping regression are distinct tokenizer issues." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Tokenizer mapping fallback issue and flash_attention crash on s_aux=None are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both tokenizer regressions, but for different models and different symptoms." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() is unrelated to a runtime import/version-check bug." + }, + { + "left": "issue:43906", + "right": "issue:42673", + "accept": false, + "reason": "A repro of an older issue and a VRAM leak in dataloader threads are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 40, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:42673", + "issue:42831", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43408", + "issue:43421", + "issue:43450", + "issue:43577", + "issue:43582", + "issue:43673", + "issue:43742", + "issue:43874", + "issue:43937", + "issue:43986", + "issue:44075", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44291", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44493", + "issue:44496", + "issue:44829", + "issue:44933", + "issue:45071", + "issue:45081", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45405", + "issue:45446", + "issue:45464", + "issue:45478", + "issue:45542", + "issue:45561", + "issue:45656" + ], + "soft_pairs": [ + "issue:44291|issue:45071", + "issue:45405|issue:45446", + "issue:43450|issue:43986", + "issue:45341|issue:45561", + "issue:44117|issue:45081", + "issue:38617|issue:41628", + "issue:39401|issue:43421", + "issue:44410|issue:45310", + "issue:44829|issue:45216", + "issue:44484|issue:45310", + "issue:42673|issue:43582", + "issue:42371|issue:44261", + "issue:43408|issue:43742", + "issue:42617|issue:45341", + "issue:42222|issue:45542", + "issue:44933|issue:45278", + "issue:43937|issue:44496", + "issue:42673|issue:43577", + "issue:42175|issue:42222", + "issue:44075|issue:45656", + "issue:38617|issue:44220", + "issue:44483|issue:45464", + "issue:42371|issue:42831", + "issue:43066|issue:43122", + "issue:43012|issue:43577", + "issue:43673|issue:45341", + "issue:44484|issue:45478", + "issue:44410|issue:45478", + "issue:43874|issue:44261", + "issue:43097|issue:44493" + ], + "prepared_review_unit_hash": "cb3c2f4cb71d2d6916846567d0f0cbe9f8f52c5f9d5e5338dcae274059c491c3", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36472, + "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 is mostly a loose v5-regression theme (imports, tokenizers, model loading, backend/config behavior), not a tight duplicate set. The best representative issue is the broad import-regression tracker.", + "confidence": 0.72, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report ('Many import errors after update from 4.57.0 to 5.5.0') and best matches the cluster's overall migration/import-breakage theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the most representative top-level tracker: broad scope, current, and explicitly about multiple import errors after upgrade.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Both are v5 breakages, but they describe different failures in different code paths (_is_hf_initialized vs PreTrainedConfig type checking)." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "Unrelated problems: PEFT version pinning vs a PyTorch version check in flex_attention." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Both touch video processing, but one is a batched-shape bug and the other is an AutoProcessor/torchvision import crash." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Same file area, but the reports are different: a generic testing_utils bug vs an xdist race on captured_info.txt." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Both are tokenizer-loading issues, but they target different mechanisms (mapping fallback vs Mistral regex patch crash)." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different regression causes." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Tokenizer-related, but one is wrong offset mapping and the other is runtime post-processor update behavior." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Different Qwen-family model issues with different failing paths; not the same bug." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Training instability with flash_attention_3 is unrelated to the Qwen3.5 save_pretrained checkpoint regression." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "Chat/completions request rejection and from_pretrained model-loading failure are different code paths." + }, + { + "left": "issue:42673", + "right": "issue:43582", + "accept": false, + "reason": "VRAM leak in multi-threaded dataloader use is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API guidance and MLA precision/config epsilon issues are unrelated." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "Different model-loading warnings/errors for different model families." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for a missing symbol and an audio feature-extraction issue are not the same underlying bug." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is request validation/acceptance and the other is streaming inference failure." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 backend behavior and FineGrainedFP8 accuracy are separate numeric/configuration issues." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both concern tokenizer behavior in v5, but the symptoms are different and there is no clear same-root-cause evidence." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "PyTorch precision warning on compile and BLIP2 dtype propagation are distinct bugs." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache behavior during chunked_prefill is unrelated to a small testing_utils bug." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen chat/completions or from_pretrained, but they are different failures and likely different fixes." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen model variants and different failure modes; not mergeable as one bug." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing image-patch helper and MLA norm epsilon precision error are unrelated." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "A removed config option and unexpected position-id keys are separate v5 regressions." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "A missing import symbol and a feature-extraction helper issue do not share the same code path." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Endpoint request acceptance and streaming inference behavior are different problems." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "No overlap beyond general numerical/backend concerns." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "The tokenizer decoder-type issue is not clearly the same as the broader tokenization-difference report." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Different precision/dtype bugs in different model flows." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache bug and testing utility bug are unrelated." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "Similar surface area, but different model-loading/API failures." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen regressions with different root causes." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Different multimodal/image-patch vs norm-epsilon precision issues." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Different regression symptoms 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": 1, + "component_count": 17, + "review_unit_index": 41, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:38617", + "issue:41628", + "issue:42222", + "issue:42757", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43335", + "issue:43408", + "issue:43441", + "issue:43452", + "issue:43502", + "issue:43531", + "issue:43824", + "issue:43873", + "issue:43874", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44351", + "issue:44368", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44704", + "issue:44829", + "issue:44987", + "issue:45003", + "issue:45071", + "issue:45103", + "issue:45245", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45459", + "issue:45542", + "issue:45588" + ], + "soft_pairs": [ + "issue:44263|issue:44485", + "issue:43012|issue:43408", + "issue:42222|issue:45459", + "issue:41628|issue:45071", + "issue:43441|issue:44829", + "issue:45071|issue:45542", + "issue:43931|issue:44410", + "issue:43931|issue:44038", + "issue:45003|issue:45341", + "issue:43335|issue:45071", + "issue:43097|issue:43452", + "issue:43986|issue:45341", + "issue:43986|issue:44484", + "issue:44117|issue:44987", + "issue:43408|issue:44368", + "issue:43874|issue:44263", + "issue:43976|issue:45071", + "issue:44829|issue:45245", + "issue:43986|issue:45372", + "issue:43531|issue:45341", + "issue:38617|issue:42757", + "issue:44261|issue:44485", + "issue:43064|issue:44315", + "issue:44410|issue:45375", + "issue:38617|issue:44351", + "issue:43097|issue:45588", + "issue:44062|issue:44117", + "issue:44038|issue:44410", + "issue:44704|issue:45103", + "issue:43502|issue:43992", + "issue:43986|issue:44117", + "issue:43097|issue:43873", + "issue:43824|issue:43986", + "issue:43066|issue:43906", + "issue:43097|issue:43901" + ], + "prepared_review_unit_hash": "cfefcf720ef82fdd46256997845345596f37e44d6d0caeff1e1bd0a03e94fec4", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 35, + "serialized_chars": 37656, + "estimated_input_tokens": 9414, + "estimated_eval_tokens": 19084 + }, + "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 issues that share broad keywords like v5, import errors, Qwen/GLM, or flash attention. None of the soft pairs looks like the same underlying bug/change strongly enough to merge.", + "confidence": 0.86, + "canonical_issue_reason": "No single canonical issue stands out: the items span unrelated import failures, model-specific regressions, training bugs, and documentation/warning reports, with only superficial title similarity.", + "canonical_pr_reason": "No PRs are present in the cluster.", + "best_issue_reason": "No strong global best issue for deduping; any representative would be misleading because the cluster is not centered on one concrete code path.", + "best_pr_reason": "No PRs are present in the cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "Different bugs: torch.split return handling in a GLM MoE indexer vs RoPE implementation details." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "Both mention warnings, but one is a bfloat16 precision warning and the other is a model-type mismatch warning." + }, + { + "left": "issue:42222", + "right": "issue:45459", + "accept": false, + "reason": "Unrelated subsystems: vitpose model breakage vs tokenizer protobuf error masking." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "Import error for AutoImageProcessor is unrelated to PretrainedConfig type-checking breakage." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both involve flash attention, but one is a load-time failure and the other is degenerate training behavior." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "PretrainedConfig typing regression is unrelated to TensorBoard/TensorFlow backend detection." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Different Qwen3-family failures: weight-shape mismatch vs missing attention projections in a specific layer." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Both are Qwen model issues, but one is a loading shape mismatch and the other is a Qwen3-VL-Moe regression." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "sys.modules access bug is unrelated to a testing_utils bug." + }, + { + "left": "issue:43335", + "right": "issue:45071", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction is not the same issue as PretrainedConfig typing." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed config flag warning and gguf_file loading breakage are different code paths." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Video AutoProcessor crash and a testing_utils bug do not match." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "AutoProcessor/torchvision crash is unrelated to save_pretrained shard size behavior." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Tokenizer mapping None-handling is a different failure than loading physical-intelligence/fast." + }, + { + "left": "issue:43408", + "right": "issue:44368", + "accept": false, + "reason": "Different warnings in different contexts: model-type mismatch vs tie_word_embeddings advice." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Different GLM bugs: missing image-patch helper vs torch.split return values." + }, + { + "left": "issue:43976", + "right": "issue:45071", + "accept": false, + "reason": "Python version support issue is unrelated to a config typing regression." + }, + { + "left": "issue:44829", + "right": "issue:45245", + "accept": false, + "reason": "Flash-attention training degeneration is unrelated to category cardinality limits." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Video AutoProcessor/torchvision crash is unrelated to ReasoningEffort import failure in Gemma 4 loading." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug is unrelated to a testing_utils bug." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but for different missing symbols from different packages." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "RMS norm epsilon precision bug does not match the GLM-5 RoPE implementation issue." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "Different training-time problems: optimizer state/rank mismatch vs Liger Kernel application during model init." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Both are Qwen config/model issues, but one is missing projections and the other is a missing vision config field." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Distinct import failures for different symbols: layer_type_validation vs HybridCache." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "Removed tie_embeddings warning is unrelated to flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer AddedToken argument collision is unrelated to TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Both mention Qwen3 variants, but the concrete failures and code paths differ." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor cached_file kwargs forwarding is unrelated to auto_docstring annotation handling." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "Local-files-only network leak and missing embed_tokens.weight are unrelated loading issues." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video AutoProcessor crash and tokenizer mapping None-handling do not share the same bug." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Removal of a config flag is unrelated to offloading/quantization behavior." + }, + { + "left": "issue:43824", + "right": "issue:43986", + "accept": false, + "reason": "Qwen2.5-VL import error and video-model AutoProcessor crash are different model/loading problems." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type is not the same as the separate reproduced issue linked by 43906." + }, + { + "left": "issue:43097", + "right": "issue:43901", + "accept": false, + "reason": "A removed config API warning is unrelated to a documentation mismatch in TextClassificationPipeline." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 42, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:38617", + "issue:42757", + "issue:42886", + "issue:42898", + "issue:42907", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43381", + "issue:43475", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "issue:43611", + "issue:43618", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43825", + "issue:43867", + "issue:43901", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44077", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44655", + "issue:44683", + "issue:44938", + "issue:45216", + "issue:45290", + "issue:45381", + "issue:45479", + "issue:45753" + ], + "soft_pairs": [ + "issue:43825|issue:44655", + "issue:43992|issue:44484", + "issue:43352|issue:44380", + "issue:43208|issue:43550", + "issue:43012|issue:43550", + "issue:43012|issue:44368", + "issue:42886|issue:43381", + "issue:42907|issue:45753", + "issue:42907|issue:44655", + "issue:44380|issue:45479", + "issue:43352|issue:43976", + "issue:43064|issue:44380", + "issue:43208|issue:43526", + "issue:36683|issue:43582", + "issue:44483|issue:44514", + "issue:43618|issue:44380", + "issue:43618|issue:43761", + "issue:43012|issue:44683", + "issue:42757|issue:43824", + "issue:38617|issue:43502", + "issue:44038|issue:44230", + "issue:44483|issue:45290", + "issue:43299|issue:43867", + "issue:44483|issue:45381", + "issue:43756|issue:44485", + "issue:43299|issue:45753", + "issue:43901|issue:44655", + "issue:36683|issue:43475", + "issue:42994|issue:44655", + "issue:43352|issue:43937", + "issue:43867|issue:45216", + "issue:44077|issue:44380", + "issue:42898|issue:43066", + "issue:36010|issue:42757", + "issue:36683|issue:44077", + "issue:43611|issue:44655", + "issue:43976|issue:44938" + ], + "prepared_review_unit_hash": "0a2daaa890d35229510251a3943cd2fb0d66d01b574ce0f8d2830e59018cc1e3", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 37, + "serialized_chars": 38515, + "estimated_input_tokens": 9629, + "estimated_eval_tokens": 19514 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is not a true duplicate cluster: it mixes unrelated import errors, model-loading regressions, tokenizer/pipeline behavior changes, save_pretrained issues, and model-specific bugs. The closest thematic hub is pipeline/model saving, but none of the soft pairs look like the same concrete fix.", + "confidence": 0.48, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:44655 is the best representative hub because several soft similarities point toward save_pretrained/pipeline serialization, even though the overall set is too heterogeneous to treat as one duplicate cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Translation-task help text vs pipeline save_pretrained failure; unrelated." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens.weight vs shard-size question; unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Both are attention/backend bugs, but they affect different models and different code paths." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs vs Bamba torch.compile/SDPA issue; unrelated." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "PyTorch precision warning vs a compile-time SDPA failure; not the same bug." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Different warning symptoms in different feature paths; unrelated." + }, + { + "left": "issue:42886", + "right": "issue:43381", + "accept": false, + "reason": "Offline cache loading vs gradient checkpointing in eval mode; unrelated." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Saving dequantized models vs Qwen3_5 image looping; unrelated." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Both are saving-related, but one is quantized model serialization and the other is pipeline save_pretrained; separate fixes." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "GPT2 attention-scaling backend bug vs sequence-classification zero-loss bug; unrelated." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model-specific attention support issue vs Python-version compatibility problem; unrelated." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state bug vs attention-scaling backend bug; unrelated." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training issues vs BeitImageProcessorFast label reduction bug; unrelated." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Missing Gemma3Config attribute vs Apple Silicon allocator TypeError; unrelated." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Chat completions endpoint rejection vs batched chat_template crash; different layers and failures." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIP output regression vs GPT2 attention-scaling bug; unrelated." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both are CLIP regressions, but one drops attentions and the other hidden_states; not clearly the same fix." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "A precision warning is not the same as a flex_attention compilation failure." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Two import errors, but for different symbols from different modules; unrelated." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "ImportError in configuration utils vs network calls despite local_files_only=True; unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Qwen3-VL-Moe loading regression vs fp8 inference support request; not the same bug." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Chat API acceptance regression vs assistant tool-call template crash; different code paths." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL-Moe loading breakage vs sorted state_dict load error; unrelated." + }, + { + "left": "issue:44483", + "right": "issue:45381", + "accept": false, + "reason": "Chat completions API regression vs Qwen2.5-VL video position_ids bug; unrelated." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Both mention RoPE, but they concern different models and likely different implementations." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for return_all_scores vs pipeline save_pretrained failure; unrelated." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Gemma3 config missing vocab_size vs SAM3 encoder output missing an attribute; unrelated." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Both saving-related, but quantized-model save failure and pipeline save_pretrained are separate code paths." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "FlashAttention support bug vs GLM-5 GenerationConfig validation error; unrelated." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "State_dict ordering load error vs checkpoint save_pretrained regression; different directions and symptoms." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "PatchTSMixer post_init policy issue vs GPT2 attention scaling; unrelated." + }, + { + "left": "issue:42898", + "right": "issue:43066", + "accept": false, + "reason": "Tokenizer cleanup behavior change vs wrong tokenizer decoder type; related area but distinct defects." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Import errors from different packages and symbols; unrelated." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Gemma3Config attribute bug vs PatchTSMixer post_init policy change; unrelated." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "base_model_prefix load regression vs pipeline save_pretrained issue; unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python compatibility reports, but they involve different versions and different failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 43, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:28282", + "issue:33453", + "issue:36296", + "issue:36683", + "issue:41669", + "issue:42890", + "issue:43010", + "issue:43064", + "issue:43352", + "issue:43502", + "issue:43525", + "issue:43575", + "issue:43618", + "issue:43867", + "issue:43931", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44075", + "issue:44077", + "issue:44112", + "issue:44188", + "issue:44242", + "issue:44246", + "issue:44263", + "issue:44355", + "issue:44360", + "issue:44380", + "issue:44484", + "issue:44492", + "issue:44855", + "issue:44857", + "issue:44938", + "issue:44964", + "issue:44977", + "issue:45071", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45381", + "issue:45440", + "issue:45478", + "issue:45561", + "issue:45593", + "issue:45600", + "issue:45850" + ], + "soft_pairs": [ + "issue:44263|issue:45440", + "issue:36683|issue:43525", + "issue:43352|issue:45593", + "issue:43618|issue:45216", + "issue:44188|issue:44380", + "issue:36683|issue:44242", + "issue:43064|issue:44492", + "issue:43352|issue:43931", + "issue:43010|issue:45278", + "issue:43992|issue:45310", + "issue:43867|issue:45375", + "issue:41669|issue:44246", + "issue:44355|issue:44855", + "issue:36683|issue:43064", + "issue:42890|issue:45561", + "issue:44938|issue:45071", + "issue:43867|issue:45357", + "issue:43010|issue:45850", + "issue:43992|issue:45478", + "issue:43010|issue:43976", + "issue:44938|issue:45381", + "issue:43618|issue:44077", + "issue:44360|issue:45440", + "issue:36296|issue:43575", + "issue:28282|issue:33453", + "issue:43867|issue:44484", + "issue:43618|issue:44977", + "issue:43618|issue:44857", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43010|issue:44075" + ], + "prepared_review_unit_hash": "b8a9f17214c8e31e093982cc53633f778e4da6be9e73b06d5b2c337079c957b5", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37119, + "estimated_input_tokens": 9280, + "estimated_eval_tokens": 18816 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issues spanning different models, config attributes, training paths, import regressions, and test flakes. I don\u2019t see any soft pair that looks like the same underlying bug/change, so this cluster does not collapse cleanly into duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:45278 is the broadest, most umbrella-like report here: multiple import regressions after upgrading versions. It\u2019s the closest thing to a representative issue for the set, even though the rest are not true duplicates of it.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 best represents the cluster because it captures the broad upgrade/regression theme, rather than a narrow model-specific or test-only failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44263", + "right": "issue:45440", + "accept": false, + "reason": "Both concern model internals, but one is a GlmMoe DSA indexer return issue and the other is DeepSeekV3 MoE divergence; different code paths and symptoms." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Both are missing config attributes, but they affect different configs (`Gemma3Config.vocab_size` vs `Llama4Config.pad_token_id`) and different failure points." + }, + { + "left": "issue:43352", + "right": "issue:45593", + "accept": false, + "reason": "Nemotron FlashAttention support gating is unrelated to D-FINE auxiliary-loss behavior when denoising is off." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIPOutput attentions assignment and Qwen3.5 checkpoint key saving are separate bugs in different parts of the stack." + }, + { + "left": "issue:44188", + "right": "issue:44380", + "accept": false, + "reason": "Both touch attention, but one is a torch.compile branching divergence bug and the other is backend-specific GPT2 attention scaling being ignored." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Missing `vocab_size` on Gemma3Config is not the same as load-balancing loss being omitted when `output_router_logits=False`." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "A distributed training optimizer-state bug is unrelated to a documentation/typing typo in cache strategies." + }, + { + "left": "issue:43352", + "right": "issue:43931", + "accept": false, + "reason": "One is unsupported FlashAttention 2.0 for Nemotron; the other is a Qwen3-VL weight-shape mismatch during model loading." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "`no_grad` decoration on cache updates is a code-path fix; widespread import errors after upgrading versions are a different regression class." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both are from loading models, but they are model-specific missing-key/from_pretrained failures with different root causes." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "A sorted-state_dict loading error and a missing `deepstack_visual_indexes` config field are distinct failure modes." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Import-time performance from `import *` usage is not the same bug as intermittent slow `import transformers`." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Compiled Python file errors and a Python 3.13 `@torch.jit.script` parsing/IndentationError are different import-time issues." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "A missing config attribute on Gemma3 has nothing to do with FSDP2/PEFT optimizer-state corruption in training." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Missing `set_seed()` in a test file is unrelated to an xdist race writing `captured_info.txt`." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 import/load compatibility and PretrainedConfig type-checking breakage are separate regressions." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "State-dict sorting/load failure is not the same as saving incorrect visual encoder keys in Qwen3.5." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "A specific `@torch.no_grad` cache update bug is not the same as a broad complaint that v5 breaks too many models." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5-related from_pretrained issues, but the reported missing key/error context does not match closely enough to be the same bug." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache update decoration and Python version support for v5.1.0 are unrelated." + }, + { + "left": "issue:44938", + "right": "issue:45381", + "accept": false, + "reason": "A Python 3.14 load failure is not the same as a Qwen2.5-VL video position-id bug." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "CLIPOutput attentions missing and patchtsmixer `post_init` policy are unrelated model API issues." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "A missing ReLU in the DSA indexer is not the same as DeepSeekV3 MoE divergence from the remote implementation." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Tensor-parallel training bug and Qwen2-57B-A14B-Instruct OOM are both TP-related but not the same underlying defect." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "PyTorch import-missing errors and tokenizer-loading regression are different startup problems with different triggers." + }, + { + "left": "issue:43867", + "right": "issue:44484", + "accept": false, + "reason": "A load-model state_dict ordering error is unrelated to the `max_shard_size` default in save_pretrained()." + }, + { + "left": "issue:43618", + "right": "issue:44977", + "accept": false, + "reason": "CLIP output attentions assignment and Qwen3.5 flash-attention generation issues are separate code paths." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "Different failures: missing CLIP attentions vs LwDetr float16 AMP CUDA crash." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Test seeding flakiness is unrelated to a Phi-4 multimodal model-loading regression." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "A flaky integration test due to missing seeds is not the same as a GraniteSpeech CI device-override issue." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Test flakiness and stale auto_mappings references are different classes of failures." + }, + { + "left": "issue:43010", + "right": "issue:44075", + "accept": false, + "reason": "Cache update decoration and unused SGD args are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 44, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:42548", + "issue:42757", + "issue:42831", + "issue:42890", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43299", + "issue:43502", + "issue:43526", + "issue:43576", + "issue:43673", + "issue:43723", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:44222", + "issue:44230", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44655", + "issue:44749", + "issue:44908", + "issue:44938", + "issue:44961", + "issue:45071", + "issue:45125", + "issue:45341", + "issue:45362", + "issue:45375", + "issue:45468", + "issue:45478", + "issue:45850" + ], + "soft_pairs": [ + "issue:43723|issue:44556", + "issue:44222|issue:45125", + "issue:42890|issue:45341", + "issue:44387|issue:44556", + "issue:42548|issue:44351", + "issue:43010|issue:43065", + "issue:43502|issue:44485", + "issue:44355|issue:44749", + "issue:30990|issue:31795", + "issue:43010|issue:43232", + "issue:45125|issue:45478", + "issue:45125|issue:45375", + "issue:30333|issue:33453", + "issue:43502|issue:44222", + "issue:43299|issue:44556", + "issue:44315|issue:44360", + "issue:42831|issue:43874", + "issue:43502|issue:44961", + "issue:30333|issue:36296", + "issue:44355|issue:45071", + "issue:44222|issue:44230", + "issue:43827|issue:44961", + "issue:44355|issue:44655", + "issue:33453|issue:34689", + "issue:42757|issue:43576", + "issue:33453|issue:36296", + "issue:43576|issue:45071", + "issue:43576|issue:43824", + "issue:31795|issue:38617", + "issue:44961|issue:45468", + "issue:41628|issue:42548", + "issue:38617|issue:43576", + "issue:43673|issue:44492", + "issue:43576|issue:44938", + "issue:44908|issue:44961", + "issue:44485|issue:44961", + "issue:36296|issue:43901", + "issue:41628|issue:44355", + "issue:36296|issue:43526", + "issue:45362|issue:45850" + ], + "prepared_review_unit_hash": "10a8091e81d7185058e30d9d3ae5dfb90e1c988302afb71dc981bcc60b235dab", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 40, + "serialized_chars": 38987, + "estimated_input_tokens": 9747, + "estimated_eval_tokens": 19750 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is dominated by heterogeneous Transformers v5 regressions, import/load failures, and a few unrelated docs/test bugs. The soft pairs are mostly not duplicates of the same underlying defect; they just share broad subsystem or version-related similarity.", + "confidence": 0.62, + "canonical_issue_reason": "issue:44556 is the best concrete umbrella among the set: it clearly captures a version-upgrade loading regression and is specific enough to be actionable, while many other issues are narrower variants around v5 breakage.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44556 is the strongest representative issue because it describes a concrete, user-facing reload failure after upgrading versions, which aligns with the broader v5 regression theme in this cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both are loading-related, but one is tokenizer-from_pretrained failure and the other is checkpoint reload after upgrade; different bugs." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained/MoE bug vs missing _tp_plan for tensor parallelism; related area, but not the same defect." + }, + { + "left": "issue:42890", + "right": "issue:45341", + "accept": false, + "reason": "Both mention tests, but one is flakiness from missing set_seed and the other is a separate testing_utils bug." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Quantization OOM from reserved memory is a different failure mode than checkpoint reload regression." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they target different missing symbols and likely different code paths." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache no_grad decoration and dummy Conv2d in Sam3PixelDecoder are unrelated bugs." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "Offline-mode network leakage and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled-file import/runtime errors vs a performance regression after upgrading; not the same issue." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Sentence Transformers loading hang vs documentation confusion in model.forward; different problem class." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Both are generation/cache adjacent, but they describe different cache-update bugs in different paths." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 MoE loading/tensor-parallel concerns, but one is missing TP plan and the other is a from_pretrained error; too distinct to merge." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Tensor parallel plan issue vs missing config field silently dropped; not the same bug." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLflow status reporting after failed training vs tokenizer loading regression; unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Offline network requests vs FP8 save_pretrained bug; same broad area but not the same concrete change." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 loading regressions, but one is Qwen3VL MoE-specific and the other is generic checkpoint reload failure." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "model_init Liger-kernel application and DSA indexer ReLU are unrelated." + }, + { + "left": "issue:42831", + "right": "issue:43874", + "accept": false, + "reason": "FP8 accuracy issue and missing image-patch method are different bugs." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "The second item is effectively unrelated/too vague to match the local_files_only issue." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow reporting after failed training is unrelated to tensor-parallel training behavior." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled Python file failures vs PretrainedConfig type-checking breakage are different issues." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Both are FP8/MoE related, but one is save_pretrained and the other is inference support; not the same code-path bug." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still referencing pipeline() is unrelated to the unrelated/placeholder item." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Compiled-file errors and inability to save Pipeline objects are different defects." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Tokenizer-loading regression vs Llama 3.2 model-loading regression; similar surface area, different underlying bug." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "ImportError for is_offline_mode and the transformers env command are unrelated." + }, + { + "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": "Broken env command and PretrainedConfig type-checking breakage are separate v5 issues." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Env command failure and Qwen2_5_VL import error are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation ambiguity in forward args vs missing layer_type_validation import; different problems." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "Placeholder title versus Gemma-4 audio positional encoding bug; no duplicate evidence." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "AutoImageProcessor import error and PreTrainedModel import error are different missing exports." + }, + { + "left": "issue:38617", + "right": "issue:43576", + "accept": false, + "reason": "Missing layer_type_validation import and env command breakage do not share the same bug." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Chunked-prefill cache bug vs a typo in cache strategies; not the same issue." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Broken env command and Python 3.14 load failure are different regressions." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwargs bug and unrelated placeholder item are not duplicates." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation bug and unrelated placeholder item are not duplicates." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug vs pipeline docs mention return_all_scores; unrelated." + }, + { + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Import error for AutoImageProcessor and compiled Python file errors are unrelated." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor-parallel training bug and BeitImageProcessorFast reduce_labels bug are different code paths." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Specific Qwen3.5 chat crash vs broad complaint that v5 breaks too many models; too general to treat as the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 45, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43827", + "issue:43874", + "issue:43976", + "issue:44038", + "issue:44077", + "issue:44246", + "issue:44261", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44393", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44655", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:44945", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45478", + "issue:45542", + "issue:45753", + "issue:45850" + ], + "soft_pairs": [ + "issue:44261|issue:44360", + "issue:44623|issue:45230", + "issue:44273|issue:45542", + "issue:44273|issue:45003", + "issue:44393|issue:45753", + "issue:43576|issue:44556", + "issue:44038|issue:45125", + "issue:44393|issue:45071", + "issue:43525|issue:45850", + "issue:41669|issue:43673", + "issue:44556|issue:45092", + "issue:44492|issue:45850", + "issue:42371|issue:44492", + "issue:43572|issue:45850", + "issue:44704|issue:44908", + "issue:42890|issue:44908", + "issue:45278|issue:45850", + "issue:43408|issue:44315", + "issue:44246|issue:44273", + "issue:43976|issue:44393", + "issue:43352|issue:44393", + "issue:43576|issue:44908", + "issue:44297|issue:45230", + "issue:44908|issue:45071", + "issue:44295|issue:45230", + "issue:41669|issue:44351", + "issue:41669|issue:45341", + "issue:44492|issue:44945", + "issue:43874|issue:45850", + "issue:44393|issue:45478", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "issue:44315|issue:44829", + "issue:31795|issue:33453", + "issue:41669|issue:44908", + "issue:43576|issue:44273", + "issue:44655|issue:44861", + "issue:44360|issue:45468", + "issue:45230|issue:45310", + "issue:44273|issue:44908" + ], + "prepared_review_unit_hash": "bf502ef058a435bada82af1187f28ba1057ccec0f5d1e79f7d628cd3b7dee6e6", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 40, + "serialized_chars": 38524, + "estimated_input_tokens": 9631, + "estimated_eval_tokens": 19518 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mix of unrelated Transformers v5-era regression reports (imports, config fields, model loading, schedulers, docs, tests). The soft edges do not look like true duplicates, so I rejected them all.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45278 is the best representative issue: it is broad, concrete, and matches the recurring theme of upgrade/import breakage better than the more vague umbrella complaints.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 has the clearest actionable scope among the cluster\u2019s broad regression reports and best captures the common v5 compatibility/import problem pattern.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44261", + "right": "issue:44360", + "accept": false, + "reason": "Different bugs: RMS norm epsilon precision/config concern vs DSA indexer ReLU implementation issue." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "One is a specific processor save_pretrained file-saving bug; the other is just a placeholder title with no matching concrete failure." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy-loading/import behavior and a TensorFlow backend dependency error are different failure modes." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Possible thematic overlap around imports, but not enough to treat lazy loading symptoms as the same bug as a specific sys.modules access issue." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Same model family, but the reported failures differ: 2D bbox hallucination/error vs infinite loop on a specific image." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Both involve v5 regressions, but one breaks the env command and the other breaks checkpoint reload compatibility; not the same bug." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both mention Qwen3/Qwen3-VL-Moe, but one is a general bug report and the other is a missing tensor-parallel plan; different code-path issues." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Model output/vision bug versus PretrainedConfig type-checking regression; unrelated code paths." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "A specific missing pad_token_id attribute is not the same as a broad complaint that v5 breaks many models." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-time slowdown from import * removal is distinct from a missing GenerationMixin cache during chunked_prefill." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both are checkpoint-loading regressions after v5, but they affect different model families and likely different root causes." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo in cache strategy naming is not the same as a vague broad regression report." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API migration guidance and a cache-strategy typo are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is a concrete config regression, not the same as a generic v5 breakage complaint." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "Processor.from_pretrained kwargs plumbing and inverse_sqrt scheduler kwargs handling are different APIs and bugs." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test seeding flakiness and scheduler kwargs propagation are unrelated." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Both are broad v5 complaints, but one is a concrete import-error regression and the other is too vague to be the same issue." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Model type mismatch warning and Liger Kernel application with model_init are different problems." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "General import slowness and lazy loading not working are adjacent topics, but not clearly the same underlying bug." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version incompatibility is unrelated to a Qwen3-VL output/hallucination bug." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention 2 support gating and a Qwen3-VL bbox output issue are different failures." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "The env command regression and scheduler kwargs bug touch different subsystems." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to an empty placeholder bug report." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwarg handling and PretrainedConfig type-checking are different issues." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "position_ids buffer access error is specific and unrelated to the placeholder report." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import * performance regression and cannot-import-HybridCache are different import-time bugs." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Model import performance regression is unrelated to a testing_utils bug." + }, + { + "left": "issue:44492", + "right": "issue:44945", + "accept": false, + "reason": "Cache strategy typo and incorrect pipeline-parallel output are not the same code-path." + }, + { + "left": "issue:43874", + "right": "issue:45850", + "accept": false, + "reason": "A missing image-patch helper in GLM46V is far more specific than the generic v5 breakage complaint." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen3.5/VL-related, but one is a bbox output issue and the other is a from_pretrained error; not clearly the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "PatchTSMixer post_init policy and tied-weight-key AttributeError are unrelated internals." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still referencing pipeline() and scheduler kwargs are unrelated." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied with model_init and degenerate training under flash_attention_3 are different concrete problems." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion in model.forward and a tokenizer loading regression are not the same bug." + }, + { + "left": "issue:41669", + "right": "issue:44908", + "accept": false, + "reason": "Import-time performance regression and scheduler kwarg handling are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command and lazy loading problems are distinct upgrade regressions." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving Pipeline objects and tied-weight key handling are separate features/issues." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer ReLU concern and Gemma4AudioRelPositionalEncoding are unrelated model internals." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Placeholder bug report versus Qwen3.5 Moe from_pretrained regression; no shared concrete bug." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading failure and inverse_sqrt scheduler kwargs are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 17, + "review_unit_index": 46, + "review_unit_count": 46, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-397", + "nodes": [ + "issue:30333", + "issue:31515", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:37428", + "issue:38617", + "issue:41669", + "issue:42371", + "issue:42913", + "issue:43066", + "issue:43298", + "issue:43502", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43606", + "issue:43618", + "issue:43704", + "issue:44230", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44485", + "issue:44492", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44998", + "issue:45092", + "issue:45290", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45850" + ], + "soft_pairs": [ + "issue:43519|issue:44485", + "issue:43519|issue:44908", + "issue:31795|issue:34689", + "issue:41669|issue:43704", + "issue:31795|issue:37428", + "issue:44315|issue:45092", + "issue:44485|issue:45468", + "issue:43704|issue:44485", + "issue:42371|issue:43704", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:44263|issue:45468", + "issue:43606|issue:45468", + "issue:44261|issue:45468", + "issue:42913|issue:43066", + "issue:44908|issue:44998", + "issue:44861|issue:45290", + "issue:44492|issue:44998", + "issue:44998|issue:45850", + "issue:43618|issue:44998", + "issue:44998|issue:45341", + "issue:43525|issue:44861", + "issue:43572|issue:44861", + "issue:31515|issue:38617", + "issue:44230|issue:44623", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "issue:31515|issue:33453", + "issue:43298|issue:44230" + ], + "prepared_review_unit_hash": "b437dc525c33d54a474984b6ae324656e000bba886af10470b1afd7598cb5fb0", + "component_budget": { + "node_count": 397, + "item_count": 397, + "soft_pair_count": 1488, + "serialized_chars": 767144, + "estimated_input_tokens": 191786, + "estimated_eval_tokens": 383828 + }, + "budget": { + "node_count": 36, + "item_count": 36, + "soft_pair_count": 32, + "serialized_chars": 29800, + "estimated_input_tokens": 7450, + "estimated_eval_tokens": 15156 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: a loose subgroup concerns Transformers v5 loading/tokenizer/config regressions, but the soft pairs are not the same underlying bug/change. I would not merge any of the candidate pairs.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 33453 is the most central concrete loading-regression report in this set and the best umbrella anchor for the tokenizer/model-loading subgroup.", + "canonical_pr_reason": null, + "best_issue_reason": "33453 is the strongest cluster anchor: it is specific, actionable, and closest to the recurring v5 loading/tokenizer complaints without being too vague.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Different domains: Qwen3VL timestamp math vs GLM-5 RoPE implementation." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp bug vs scheduler kwargs handling; unrelated code paths." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Documentation confusion vs a model-loading regression; not the same issue." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star cleanup/perf issue is unrelated to a Qwen3ForCausalLM VRAM leak." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "Docs ambiguity vs flash-attention import error; different symptoms and fixes." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Liger Kernel application in model_init is not the same as meta-init incompatibility for old remote-code checkpoints." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Both are model-specific, but one is RoPE logic and the other is audio positional encoding; distinct bugs." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak in dataloader threads vs RoPE implementation error; no shared code-path." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 settings warning is unrelated to a memory leak." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "Unexpected network access vs incorrect timestamp calculation; unrelated behavior." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Processor timestamp bug vs missing save_pretrained files; different API surfaces." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Processor timestamp bug vs VRAM leak; not the same failure mode." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split handling in GLM MoE indexing is unrelated to Gemma audio positional encoding." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch is a runtime placement bug, not an audio encoding issue." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "RMS norm epsilon precision issue does not match Gemma audio positional encoding." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both involve tokenizer v5 changes, but one is broad behavior drift and the other is a specific decoder-type bug." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler kwargs bug and an unrelated unemployment issue." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Tied-weight key handling is unrelated to chat template crashes on tool-call assistant messages." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Cache-strategy typo is unrelated to an off-topic unemployment issue." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "One is nonsensical/off-topic; the other is a broad v5 complaint, but they are not the same bug." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the off-topic unemployment issue." + }, + { + "left": "issue:44998", + "right": "issue:45341", + "accept": false, + "reason": "Unrelated off-topic issue versus a testing_utils bug." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is unrelated to tied-weight key list handling." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression is unrelated to tied-weight key extraction." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Slow checkpoint loading is not the same as an ImportError for layer_type_validation." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support for Qwen3-VL/Qwen3.5 MoE is unrelated to processor save_pretrained file emission." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow failure reporting and slow from_pretrained checkpoint loading are different problems." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Checkpoint loading slowdown vs tensor-parallel training bug; different code paths." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Loading performance issue vs documentation confusion." + }, + { + "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 regression." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Both concern loading, but one is checkpoint load speed and the other is a tokenizer-loading regression." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching audio-model support and fp8 Qwen3-VL/Qwen3.5 MoE inference 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": 17, + "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": "79543dcba06d67c60dc8ac378e42e419e6e9c38926a34e1fdd57c5d6d292fa04", + "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": "This cluster appears to contain three unrelated issue themes: tokenization mismatch, private model download/auth error messaging, and a generic model loading failure. None of the proposed soft pairs look like the same underlying bug.", + "confidence": 0.93, + "canonical_issue_reason": "Issue #16618 is the broadest and most representative of a core model-loading failure; the others are separate concerns about tokenization behavior and auth/error messaging.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 is the best single issue to anchor the cluster because it matches the most general user-facing failure mode and is the most likely duplicate target among these candidates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tokenization mismatch vs. model loading failure are different bugs affecting different code paths; not the same underlying issue." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "One is an auth/download error message problem for private models, the other is a generic inability to load a public model; related area but not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 3, + "component_count": 17, + "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": "8caa4d8a8429a71fdcb0167054a661205e24526ea34b0b016480cdd5ebbb57f7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They target different features and code paths.", + "confidence": 0.98, + "canonical_issue_reason": "Issue #26350 is the stronger representative only because it has much higher discussion activity, but it is not a duplicate of #33260.", + "canonical_pr_reason": null, + "best_issue_reason": "#26350 is the best issue to anchor the cluster if one must be chosen, due to higher activity and clearer community interest; however, the pair should not be merged.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different underlying requests: Flash Attention 2 vs GGUF support. They concern separate model/runtime integration 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": 4, + "component_count": 17, + "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": "1aaeeb0c9aa25514d5a4730a6293b3715b238718db942ab0432f38966e129d8c", + "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": "All four issues are highly similar translation-request templates, but the concrete language target differs or is unspecified, so they do not look like the same underlying issue. The Turkish issue is the most specific and best representative of the cluster.", + "confidence": 0.73, + "canonical_issue_reason": "issue:27088 is the only concrete, fully specified request ('Turkish'); the others are generic placeholder/template variants for some language and are less suitable as the canonical representative.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:27088 is the best issue candidate because it is concrete and actionable, while the others are generic template-style duplicates with unspecified language targets.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Same translation-request pattern, but one is explicitly Turkish and the other is an unspecified language placeholder; different underlying request." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "The titles indicate different language targets (Turkish vs placeholder generic language); not the same issue." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Concrete Turkish request does not match a generic '/' placeholder issue; likely separate translation requests." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "Both are template-style translation issues, but the available titles do not show the same language target or same concrete request." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "The two issues appear to be separate instances of the same template, not evidence of one identical underlying request." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "Very similar template text, but no concrete shared language target or identical request is visible; keep them separate." + } + ] + }, + "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": 17, + "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-36817-15", + "nodes": [ + "issue:36817", + "issue:36879", + "issue:37015", + "issue:41115", + "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:36817|issue:36879", + "issue:36879|issue:41211", + "issue:36879|issue:43909", + "issue:36879|issue:44121", + "issue:36879|issue:43837", + "issue:36879|issue:43671", + "issue:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:36879|issue:37015", + "issue:37015|issue:42503", + "issue:41115|issue:43837", + "issue:41115|issue:44121", + "issue:41115|issue:42503", + "issue:41115|issue:45027", + "issue:41211|issue:43671", + "issue:42503|issue:43837", + "issue:42503|issue:43671", + "issue:42503|issue:42738", + "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:37015|issue:45027", + "issue:43837|issue:45177" + ], + "prepared_review_unit_hash": "077d6d0e6e6b78f775764d673a72560220f2dc066bb262bbdfcf60c2435a49e1", + "component_budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 35, + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 + }, + "budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 35, + "serialized_chars": 20705, + "estimated_input_tokens": 5177, + "estimated_eval_tokens": 10610 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are all separate model-request issues; the title similarity is mostly generic \"add/support model\" wording, not the same underlying bug or change. None of the soft pairs look mergeable as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "issue:43837 is the strongest hub in this cluster by soft-similarity and is a clear, open request; if one issue must represent the group, this is the best fit, though it is still a distinct Qwen3-ASR request rather than a duplicate of the others.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43837 is the most representative single issue: it is specific, open, and centrally connected to several of the audio/model-request items, but it does not actually subsume the other model additions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBERT config vs DEIMv2: different model families and different requests; only the generic \"add model\" framing matches." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBERT config vs Qwen3-TTS: unrelated model families and tasks." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBERT config vs RF-DETR: one is a config entry request, the other is a specific model addition." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR vs DEIMv2: different architectures and unrelated model additions." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR vs LFM2.5 Audio 1.5B: object-detection model request vs audio model request." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR vs OpenAI Weight-Sparse Transformer: unrelated model families and change scopes." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR vs Qwen3-ASR: different model families and tasks." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR vs Qwen3-TTS: different model families and tasks." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT vs Qwen3-ASR: BERT-style text model request vs ASR support request." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT vs OpenAI Weight-Sparse Transformer: different model families and different requests." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT vs LFM2.5 Audio 1.5B: unrelated model families and tasks." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR vs NeoBERT: unrelated model additions with no shared concrete change." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT vs ModernVBERT models: similar naming, but different BERT variants and separate requests." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 vs Qwen3-ASR: different model families and different tasks." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 vs OpenAI Weight-Sparse Transformer: unrelated model requests." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 vs ModernVBERT models: different model families and no shared concrete code change." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 vs Voxtral-4B-TTS-2603: unrelated model families and modalities." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 vs Qwen3-TTS: different model families and tasks." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT models vs Qwen3-ASR: unrelated model families and tasks." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT models vs Qwen3-TTS: different model families and different requests." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT models vs BERT-like models with RoPE: related wording, but one is a specific model request and the other is a general architectural feature request." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like models with RoPE vs Voxtral TTS: unrelated model families and scopes." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS vs Qwen3-ASR: same product line, but different modalities/tasks and not the same change." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS vs Voxtral-4B-TTS-2603: both TTS-related, but different model families and separate support requests." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS vs DeepSeek-OCR2: different modalities and unrelated requests." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS vs OpenAI Weight-Sparse Transformer: unrelated model families." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS vs LFM2.5 Audio 1.5B: both audio-related, but different model families and separate additions." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR vs LFM2.5 Audio 1.5B: same broad audio domain, but different models and tasks." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR vs EUPE: ASR support request vs a perception-encoder model request; not the same change." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR vs OpenAI Weight-Sparse Transformer: unrelated model families and scopes." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio 1.5B vs SAM3-LiteText: audio model request vs multimodal perception/text model request." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "OpenAI Weight-Sparse Transformer vs EUPE: different model families and separate requests." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR vs Voxtral-4B-TTS-2603: both audio-related, but ASR and TTS are different concrete changes." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT vs Voxtral-4B-TTS-2603: unrelated model families and modalities." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR vs DeepSeek-OCR2: ASR support vs OCR model support; different underlying changes." + } + ] + }, + "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": 17, + "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": "20d02a86283c3c6a7671907080253a8ffea65c6938bf8f1df54a64934e8a2309", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6288, + "estimated_input_tokens": 1572, + "estimated_eval_tokens": 3400 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6288, + "estimated_input_tokens": 1572, + "estimated_eval_tokens": 3400 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These issues are only loosely similar at a surface level (model/config/loss/tokenizer bugs), but they describe different code paths and distinct defects. No soft-edge pair looks like the same underlying bug.", + "confidence": 0.82, + "canonical_issue_reason": "No clear canonical duplicate emerges; the issues target different components and symptoms rather than one shared bug.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43668 is the cleanest self-contained report, but it is not a duplicate representative of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "Is Llama4TextL2Norm meant to be RMS norm?", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", + "accept": false, + "reason": "Different models and problems: one questions Llama4 norm semantics, the other is a ModernBERT typing issue." + }, + { + "left": "Mask2former model's ignore_value not used after definition", + "right": "kwargs are not passed to loss calculation function.", + "accept": false, + "reason": "Different bug locations and failure modes: Mask2former ignore_value handling vs loss kwargs plumbing." + }, + { + "left": "Mask2former model's ignore_value not used after definition", + "right": "All tokenizers raise incorrect regex pattern warning after version 4.57.3?", + "accept": false, + "reason": "Different subsystems entirely: model loss configuration vs tokenizer warning regression." + }, + { + "left": "Mask2former model's ignore_value not used after definition", + "right": "Typo - should be \"orig_conversion.quantization_operation\"", + "accept": false, + "reason": "Unrelated issues: Mask2former ignore_value usage vs a typo in a quantization-related attribute name." + }, + { + "left": "Mask2former model's ignore_value not used after definition", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", + "accept": false, + "reason": "Different models and defect types; one is unused ignore_value logic, the other is a config type hint problem." + }, + { + "left": "kwargs are not passed to loss calculation function.", + "right": "ModernBERTConfig `norm_eps` type hint is incorrect", + "accept": false, + "reason": "Different concerns: missing kwargs propagation into loss computation vs incorrect type annotation." + }, + { + "left": "ModernBERTConfig `norm_eps` type hint is incorrect", + "right": "All tokenizers raise incorrect regex pattern warning after version 4.57.3?", + "accept": false, + "reason": "Different areas and symptoms: ModernBERT config typing vs tokenizer warning regression." + }, + { + "left": "ModernBERTConfig `norm_eps` type hint is incorrect", + "right": "Typo - should be \"orig_conversion.quantization_operation\"", + "accept": false, + "reason": "A type-hint fix and a string typo in another code path are not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 17, + "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": "f998e09d4a92338d7511caa5151185e8978a0cb5c117200e0998db1e961d1db0", + "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": "These are unrelated open issues: one asks for block decoding, the other asks for DeepSeek v3.2 support. They do not look like the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical duplicate issue here; the requests are about different concerns (decoding behavior vs model/version support).", + "canonical_pr_reason": null, + "best_issue_reason": "43037 is the more concrete feature request and is better suited as a standalone representative issue, but it is not a duplicate of 41084.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different topics and code paths: block decoding vs DeepSeek v3.2 support. Shared similarity is only broad ML/runtime context, not the same underlying 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": 17, + "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": "f2e08bb507e354cd34a781faf7f8ffa5c7881bf3be5b65d6cbb4e33dcb447b1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are related at a high level to custom trainers and Trainer behavior, but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", + "confidence": 0.76, + "canonical_issue_reason": "Issue 43598 is the broader, more direct Trainer customization concern and better represents the cluster theme.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43598 is the best issue candidate because it addresses the underlying Trainer customization behavior rather than a narrower derived symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "They share a general custom-trainer context, but the concrete problems differ: one concerns `remove_unused_column` customizability, the other concerns FLOPS accounting. 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": 9, + "component_count": 17, + "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": "595cd49557777a2a8b20649268c7cdbb735ed142430ac16dbd4f5f337d7afe00", + "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 issues are related only at a broad Transformers/model-registration level, but they target different model families and different APIs: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs. Qwen3.5 sequence-classification support. They should not be merged as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 43684 is the cleaner representative because it is a concrete registration request with explicit target classes and a narrowly defined model family.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43684 is the best issue to keep as the cluster representative since it is specific and actionable, while the other issue is a separate feature request for a different config/class combination.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying requests: one asks for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration, the other asks for AutoModelForSequenceClassification support for Qwen3.5 (Qwen3_5Config). They do not describe 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": 17, + "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": "733719b45928aa04790a1997019c14cad10173f1e38498632c66b76d6cd6afc6", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4267, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4267, + "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 exact same vague title \"huggingface \" and were opened close together, so they likely represent one duplicate cluster; \"Fett\" looks unrelated.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface \" reports, so it is the best representative for the duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43845 is the cleanest canonical choice because it matches the other duplicate titles and was created first.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different title/content signals; \"Fett\" does not appear to describe the same issue as the \"huggingface \" reports." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different title/content signals; no evidence these refer to the same underlying bug." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different title/content signals; no evidence these refer to the same underlying bug." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same exact title and very close creation time, with no distinguishing details, strongly suggesting duplicates of the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title and same vague report pattern; likely a duplicate of the same underlying issue." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title and same vague report pattern; likely a duplicate of the same underlying issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: 43845/43846/43859 all share the exact title \"huggingface \" and appear close enough to plausibly form one duplicate cluster, while 43835 titled \"Fett\" is distinct. The canonical choice of 43845 is conservative and supported by it being the earliest of the three similar reports. The soft-edge verdicts are also appropriately cautious." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 17, + "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": "2665af50fba4dcd2142dc08cace44ce1e4c4a3c568306408839281d473edf44c", + "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": "This cluster is mostly low-information spam/generic issues with no bodies or actionable details. The titles overlap superficially, but there is not enough evidence that they describe the same underlying problem, so I would not deduplicate them as true duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "If one issue must serve as canonical, issue #44371 is the earliest and has the same placeholder spam-style title as the closest matches, making it the most representative of the group\u2019s noise pattern.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #44371 is the best representative issue in this set because it is earliest and shares the same generic spam title form as the other placeholder issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are just '' placeholders with no substantive body; identical titles alone are not enough to prove the same underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same as above: generic spam titles and no issue details or shared targets/file context." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is spam-placeholder text and the other is a vague documentation suggestion; these do not appear to be the same concrete issue." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "'' and 'Check out \"Google LLC\"' are different low-information titles with no shared technical context." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "A generic documentation suggestion is not the same underlying problem as a spam placeholder issue." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "No concrete overlap beyond both being vague/low-detail issues." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Although both are '' placeholders, there is no body, target, or context showing they report the same underlying bug rather than separate spam reports." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Different titles and no supporting details tie them to the same issue." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "Different low-information titles with no evidence of a shared root cause." + } + ] + }, + "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": 17, + "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": "76b0893a14957d42661b96f2c88b34d9c229dc0e7949298e49ee196b611d8dbc", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "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 have identical titles and similar size, so they are very likely related and may be duplicates, but the packet does not expose body content, so the evidence is not strong enough to confirm they are the same request. #45145 remains the best canonical issue because it has the stronger reference signal.", + "confidence": 0.82, + "canonical_issue_reason": "#45145 is the best canonical issue because it shares the exact title with #45151 and has the only inbound reference, making it the likelier primary thread.", + "canonical_pr_reason": null, + "best_issue_reason": "#45145 is the strongest single issue representative due to the identical title and slightly higher linkage signal.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "The titles are identical and the issues are probably related, but the packet does not show the issue bodies or concrete content overlap. That is enough for a likely same-topic pair, not enough to confidently accept a duplicate edge." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The canonical choice is grounded, but the duplicate verdict is too strong for the evidence provided. The packet only shows identical titles, similar body lengths, and a soft-similarity signal; it does not expose the issue bodies, so phrases like \"clearly the same proposed energy-optimization work\" and \"straightforward duplicate pair\" overstate certainty. This should be framed more conservatively as a likely duplicate/same-topic pair, not a definite one." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 17, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45295-2", + "nodes": [ + "issue:45295", + "issue:45373" + ], + "soft_pairs": [], + "prepared_review_unit_hash": "f5fd95fada2f530ab0999e343c17c16266cde0c7182bb22509972ca44827cd49", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 0, + "serialized_chars": 1286, + "estimated_input_tokens": 322, + "estimated_eval_tokens": 900 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues describe the same missing Gemma 4 sequence-classification support; the later, more specific open issue is the better cluster representative.", + "confidence": 0.94, + "canonical_issue_reason": "issue #45373 is the clearest canonical tracker: it is open, explicitly names the missing class `Gemma4ForSequenceClassification`, and states the module-level gap relative to Gemma 2/3.", + "canonical_pr_reason": null, + "best_issue_reason": "issue #45373 best captures the concrete request and likely remaining work, while #45295 is a shorter earlier duplicate with less detail and is already closed.", + "best_pr_reason": null, + "soft_edge_verdicts": [] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 14, + "component_count": 17, + "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-45672-2", + "nodes": [ + "issue:45672", + "issue:45674" + ], + "soft_pairs": [ + "issue:45672|issue:45674" + ], + "prepared_review_unit_hash": "b998c2ed9383e81edb09276e441c622e520fa75a1328595f94461ba314a79ddd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1669, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1669, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both items are 4-bit inference bugs, but they describe different root causes and different failing paths: Gemma4 audio-module uint8 weights hitting `torch.finfo()` versus BitsAndBytesConfig skip-module handling removing the default `lm_head` exclusion and triggering an AssertionError. They should not be deduplicated.", + "confidence": 0.96, + "canonical_issue_reason": "Issue #45674 is the broader, more central 4-bit inference regression and has stronger discussion/inbound activity, so it is the better representative if one issue must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45674 is the better cluster representative because it is more general and appears to be the more reusable description of a 4-bit BitsAndBytesConfig failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Same high-level area (4-bit inference) but different bug mechanisms, code paths, and symptoms; not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 17, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45762-2", + "nodes": [ + "issue:45762", + "issue:45763" + ], + "soft_pairs": [ + "issue:45762|issue:45763" + ], + "prepared_review_unit_hash": "ec75034adf5686b15c4624bc5ad53fad7ad051a1ffb665e8999bbc9c7536e78c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are not duplicates; they are unrelated name/date entries with no shared technical problem or code path.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 45762 is the slightly better canonical placeholder only because it was created first, but there is no real duplicate relationship between the two issues.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a strong canonical duplicate target; if one must be chosen, 45762 is marginally better due to earlier creation, not because of content overlap.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "The titles are different personal-name/date strings and there is no evidence of a shared bug, request, or code-path change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 16, + "component_count": 17, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45764-2", + "nodes": [ + "issue:45764", + "issue:45765" + ], + "soft_pairs": [ + "issue:45764|issue:45765" + ], + "prepared_review_unit_hash": "8c24f8a175f8ca44266f737603101c473f42ed32fea0ee9d29cec92c3f965a37", + "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 closed issues appear unrelated and lack any substantive body or references; their titles are just different personal-name/date strings, so the soft similarity is not enough to treat them 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:45764", + "right": "issue:45765", + "accept": false, + "reason": "Reject: the titles only share a superficial pattern (a name plus date-like text). There is no body, discussion, target, or other evidence that they describe the same underlying bug or request." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 17, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45869-5", + "nodes": [ + "issue:45869", + "issue:45870", + "issue:45871", + "issue:45872", + "issue:45873" + ], + "soft_pairs": [ + "issue:45869|issue:45870", + "issue:45869|issue:45871", + "issue:45870|issue:45871", + "issue:45870|issue:45872", + "issue:45870|issue:45873", + "issue:45871|issue:45872", + "issue:45871|issue:45873", + "issue:45872|issue:45873" + ], + "prepared_review_unit_hash": "73474d02abf376ca2abc414ae7451c1732bf1dfbe32cdbca6347acc65b3e7b01", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "There are two clear duplicate issue pairs: 45870\u219445871 and 45872\u219445873. The dataset-sales post (45869) is unrelated, and the two financial-time-series items (45870/71 vs 45872/73) are similar in theme but not the same underlying issue.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 45872 is the strongest representative: it is a substantive, product-specific request and has an exact duplicate in 45873, while the other similar items are either a different topic or less clearly central.", + "canonical_pr_reason": null, + "best_issue_reason": "45872 is the best issue candidate because it is the clearest, most concrete example of the underlying request and has the strongest duplicate match.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Different subjects: a financial dataset listing versus an example about time series prediction. Similar finance wording is not enough to treat them as the same issue." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Different subjects: a financial dataset listing versus an example about time series prediction. Not the same underlying bug or request." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Exact duplicate title and essentially identical content; these are the same issue." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Related finance/DataForge theme, but the request types differ: one is a prediction example, the other adds a scraping service. Too different to merge as duplicates." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Same as 45870\u219445872: related domain, but not the same concrete issue/request." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Related theme only; the concrete asks differ, so they should not be merged as duplicates." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Related theme only; the concrete asks differ, so they should not be merged as duplicates." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Exact duplicate title and essentially identical content; these are the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The conclusions are grounded in the packet: 45870/45871 and 45872/45873 have identical titles and matching body lengths, while 45869 is clearly a different topic. The non-duplicate soft-edge verdicts are conservative given only title-level evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-39895-14", + "nodes": [ + "pull_request:39895", + "pull_request:41855", + "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:45655", + "pull_request:45795", + "pull_request:45899" + ], + "soft_pairs": [ + "pull_request:39895|pull_request:43424", + "pull_request:41855|pull_request:43823", + "pull_request:41855|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:45899", + "pull_request:43424|pull_request:44827", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43995|pull_request:44827", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:45899", + "pull_request:44827|pull_request:45899", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:45795", + "pull_request:44827|pull_request:45655", + "pull_request:45170|pull_request:45899", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:45213|pull_request:45899", + "pull_request:45795|pull_request:45899", + "pull_request:43823|pull_request:45795", + "pull_request:43823|pull_request:45899", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "7127ece962ac4784c8f66b9c46c4b4b547789748fd32e492b10660522cd939bf", + "component_budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 26, + "serialized_chars": 39789, + "estimated_input_tokens": 9948, + "estimated_eval_tokens": 20152 + }, + "budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 26, + "serialized_chars": 39789, + "estimated_input_tokens": 9948, + "estimated_eval_tokens": 20152 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Most items here are unrelated PRs that only share broad Transformers themes (new models, docs, refactors). The only true duplicate-style match is the pair of `fixed_cross_entropy` fixes, which target the same helper and same bug in the loss path.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43251 is the best canonical representative: it explicitly links issue 43240 and implements the fuller `cross_entropy` kwargs pass-through in the shared loss helper; 43254 is just a narrower variant of the same change.", + "best_issue_reason": null, + "best_pr_reason": "PR 43251 is the clearest triage anchor for the only real duplicate pair, with the explicit issue reference and the more complete fix in the exact same code path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "Videoprism model addition vs executorch dynamic-shape test; unrelated code paths." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Tokenizer API additions vs adding a new MobileLLM model; not the same change." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Tokenizer methods vs executorch export test; different subsystems and goals." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor auto-loading robustness vs new model integration; only broad transformer overlap." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Loss-function kwargs fix vs model addition; unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Cross-entropy kwargs fix vs Mistral4 test fix; not the same underlying bug." + }, + { + "left": "pull_request:43424", + "right": "pull_request:45899", + "accept": false, + "reason": "Executorch export test vs docs for continuous batching decode path; unrelated." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "Executorch dynamic-shape test vs Mistral4 test fix; different changes." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Executorch test vs model addition; no shared concrete bug." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "Adding MobileLLM vs fixing `cls.config` resolution order; unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "MobileLLM model addition vs Falcon output-collection refactor; different code paths." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon refactor vs Mistral4 test fixes; not mergeable as one fix." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon model refactor vs model-creation skill PR; unrelated." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45899", + "accept": false, + "reason": "Falcon refactor vs docs-only decode fast path update; unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45899", + "accept": false, + "reason": "Mistral4 test fixes vs continuous batching docs; unrelated despite both touching Transformers." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 tests vs `layrnorm` renaming across CLIP-like models; different bugs." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45795", + "accept": false, + "reason": "Mistral4 test fix vs audio/video processor docs; unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 test fix vs config-resolution order fix; unrelated." + }, + { + "left": "pull_request:45170", + "right": "pull_request:45899", + "accept": false, + "reason": "Layernorm typo cleanup vs docs for decode fast path; unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 test fix vs model-creation skill PR; unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM addition vs model-creation skill PR; not the same change." + }, + { + "left": "pull_request:45213", + "right": "pull_request:45899", + "accept": false, + "reason": "Model-creation skill tooling/docs vs continuous batching docs; unrelated." + }, + { + "left": "pull_request:45795", + "right": "pull_request:45899", + "accept": false, + "reason": "Docs for audio/video processors vs docs for decode fast path; different documentation topics." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45795", + "accept": false, + "reason": "MobileLLM model addition vs audio/video processor docs; unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45899", + "accept": false, + "reason": "MobileLLM addition vs continuous batching docs; unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both patches change `fixed_cross_entropy` to pass supported kwargs to `nn.functional.cross_entropy`; same underlying bug and same code path." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the only accepted soft pair is the one with shared issue target and the same `fixed_cross_entropy` helper. The other verdicts are conservative and do not overclaim mergeability for clearly unrelated PRs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-2", + "nodes": [ + "pull_request:40546", + "pull_request:42781" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781" + ], + "prepared_review_unit_hash": "6f8915a46745bd9085d5fd2ceaea267c1b3974c9f5ecfa0e495674df4b3db01d", + "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": "Both PRs relate to VibeVoice, but they target different scopes: #40546 is the core VibeVoice model integration, while #42781 is a draft for VibeVoice Realtime with additional acoustic-tokenizer and generation work. Shared files reflect the same model family, not the same concrete change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "#40546 is the broader, original VibeVoice integration and is the better anchor for the cluster; it has the fuller model hookup and more review activity.", + "best_issue_reason": null, + "best_pr_reason": "#40546 is the best representative because it implements the main VibeVoice support, whereas #42781 adds a separate realtime extension and related tokenizer components.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Reject: these are related VibeVoice PRs but not the same change. #40546 adds the base model integration; #42781 extends VibeVoice for realtime usage with acoustic tokenizer and extra generation/model files." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40755-2", + "nodes": [ + "pull_request:40755", + "pull_request:41224" + ], + "soft_pairs": [ + "pull_request:40755|pull_request:41224" + ], + "prepared_review_unit_hash": "a514a5a070afe4faef212a4a71147d0e4dc902b0a38b0ffb0859931f558284cf", + "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 two pull requests 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.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single best PR for this cluster: the PRs address different features in different model families, so neither can serve as a canonical duplicate representative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different underlying changes: TimesFM forecasting with covariates vs DINOv3 image-classification support. No shared bug or mergeable common 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": 4, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "aead1308c167b0bae6ff11bf9358340baa5e76f816618383418b505a10a9cd01", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two DEIMv2 pull requests target the same model family, but they are not the same change: PR 41356 is an early/basic model + image processor add, while PR 44339 is a much broader follow-up that adds loss code, auto-mappings, conversion tooling, modular files, and more integration work.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44339 is the more complete and representative DEIMv2 implementation; PR 41356 looks like an earlier partial version of the same model bring-up.", + "best_issue_reason": null, + "best_pr_reason": "PR 44339 is the strongest single artifact for the cluster because it subsumes the basic model work and includes the broader integration needed for a full Transformers addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": false, + "reason": "Same model area, but not the same concrete change: 44339 expands the implementation substantially and adds new code paths beyond the basic PR 41356. They look like related iterations, not mergeable duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41797-3", + "nodes": [ + "pull_request:41797", + "pull_request:45186", + "pull_request:45630" + ], + "soft_pairs": [ + "pull_request:41797|pull_request:45186", + "pull_request:45186|pull_request:45630" + ], + "prepared_review_unit_hash": "5dbac1d316db05553a244f2278e144606cc21daf6adf0efac42ae62fbb5c851a", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10409, + "estimated_input_tokens": 2603, + "estimated_eval_tokens": 5462 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10409, + "estimated_input_tokens": 2603, + "estimated_eval_tokens": 5462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-addition pull requests (DeepSeek OCR, Isaac, and Kimi2-6). They share Transformers model-registration plumbing, but they target different models and different implementations, so they are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45630 is the strongest representative of the cluster because it has the highest discussion/review activity and is a full model-addition change touching the expected registration, docs, tests, and model code.", + "best_issue_reason": null, + "best_pr_reason": "PR 45630 is the best single representative for triage purposes, but only as an example of the broader model-addition pattern; it is not the same underlying change as the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41797", + "right": "pull_request:45186", + "accept": false, + "reason": "Different models and code paths: DeepSeek OCR vs Isaac. The overlap is only shared Transformers integration files and docs, which is normal for separate model additions." + }, + { + "left": "pull_request:45186", + "right": "pull_request:45630", + "accept": false, + "reason": "Different model families: Isaac vs Kimi2-6. They share registry/auto-mapping updates, but they are distinct additions with different implementations and should not 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": 6, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42975-2", + "nodes": [ + "pull_request:42975", + "pull_request:42976" + ], + "soft_pairs": [ + "pull_request:42975|pull_request:42976" + ], + "prepared_review_unit_hash": "a2e880e2466dbce3e7b644939dc578c92dff6eca9823626cdf7c58a71ffb6c0a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs both upgrade GitHub Actions, but they target different action families and maintenance goals: one is a Node 24 compatibility bump for workflow helper actions, the other updates Docker build actions. They are related in theme only, not the same underlying change.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the broader, more central workflow-actions upgrade focused on Node 24 compatibility across many files, so it is the better canonical representative if this cluster were treated as one change.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the better representative because it covers the wider GitHub Actions runtime-compatibility update, while PR 42976 is a separate Docker-action version bump.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Different underlying changes: 42975 updates checkout/cache/upload-artifact and related workflow actions for Node 24 compatibility, while 42976 upgrades docker/build-push-action. They touch different action code paths and would not naturally 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": 7, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-7", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:43973", + "pull_request:44601", + "pull_request:45490", + "pull_request:45838" + ], + "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", + "pull_request:43973|pull_request:45838", + "pull_request:44601|pull_request:45838", + "pull_request:45490|pull_request:45838" + ], + "prepared_review_unit_hash": "affbe9c991d165666bd7eec81676c5605f9fba6c402b2d0c649a2b0647891f32", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 9, + "serialized_chars": 23541, + "estimated_input_tokens": 5886, + "estimated_eval_tokens": 12028 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 9, + "serialized_chars": 23541, + "estimated_input_tokens": 5886, + "estimated_eval_tokens": 12028 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: several unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM, LFM2 Audio, PP-OCRv6/PP-LCNetV4) plus one distributed pipeline-parallel infrastructure PR. The overlap is mostly boilerplate docs/auto-registration, not the same underlying change or bug, so none of the soft edges should be merged.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the items are not duplicates of one another; they span different model families and even a separate distributed-loading feature. Shared files are generic documentation/auto-mapping boilerplate.", + "best_issue_reason": null, + "best_pr_reason": "No single PR is a good global representative for dedupe triage, because there is no common concrete code-path change across the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo); shared docs/auto-registration files are just boilerplate." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "Different model families (ViT NEPA vs Molmo2); not the same underlying feature or bug." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "Different model domains (ViT NEPA vs CTSM); only generic model-registration/docs overlap." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are distinct model additions, not one concrete change; they could not be merged as a single fix." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo vs CTSM are unrelated model integrations; shared auto-mapping/docs are not enough." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM address different model families and code paths; not duplicates." + }, + { + "left": "pull_request:43973", + "right": "pull_request:45838", + "accept": false, + "reason": "LFM2 audio support is an audio model add, while PP-OCRv6/PP-LCNetV4 is OCR/text recognition; different changes." + }, + { + "left": "pull_request:44601", + "right": "pull_request:45838", + "accept": false, + "reason": "Pipeline-parallel loading infrastructure is unrelated to OCR model support; not the same code-path problem." + }, + { + "left": "pull_request:45490", + "right": "pull_request:45838", + "accept": false, + "reason": "CTSM and PP-OCRv6/PP-LCNetV4 are different model additions with only generic docs/registration overlap." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43291-3", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43665|pull_request:44815" + ], + "prepared_review_unit_hash": "54f3a77fde9e5098062f333a58ac8ff13904abb19e12a183c70aa1e65a6c6632", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7093, + "estimated_input_tokens": 1774, + "estimated_eval_tokens": 3804 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs do not look like the same underlying change. #43291 is a Whisper tokenizer/test fix, #43665 is an unrelated test-only edit set across CLIP/ViT, and #44815 is a substantive dequant/FP8/model-loading fix. No duplicate pairs should be merged.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "#44815 is the best representative PR in this cluster because it contains the clearest concrete code-path fix (core model loading plus FP8 quantization logic) and is not just a test-only or obviously unrelated change.", + "best_issue_reason": null, + "best_pr_reason": "#44815 is the strongest standalone PR: it addresses a specific implementation bug with meaningful code changes and supporting tests, whereas #43291 is narrower and #43665 does not resemble a real fix for the same problem.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Whisper tokenizer/test changes are unrelated to CLIP/ViT test edits; same broad \"fix tests\" framing is not enough to imply the same bug or mergeable PR." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "These touch different subsystems and problems: Whisper decoding vs FP8/core loading/dequant behavior. No concrete shared code-path issue." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "A test-only CLIP/ViT patch is not the same underlying bug as FP8 dequant/model-loading fixes, and the overlap is only superficial title similarity." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-2", + "nodes": [ + "pull_request:43488", + "pull_request:43532" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:43532" + ], + "prepared_review_unit_hash": "daefc180cb6afa6032a04c8957d63d045dd936dfb0765775ebea0c1c459d8978", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated: one is a setup/version/dependency tweak with a vit test file, while the other is a CI/workflow and diff-display change with broader tooling edits. The shared test filename looks incidental, not a duplicate signal.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No clear canonical PR: PR #43488 is about repo setup/version/dependency changes, while PR #43532 is about workflow/diff-display behavior and related tooling. They do not represent the same change.", + "best_issue_reason": null, + "best_pr_reason": "Neither PR is a good global best candidate for deduplication because they target different code paths and only overlap on an unrelated test filename.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Reject: the overlap is superficial. #43488 changes setup/version/deps and a vit test, while #43532 changes workflow/comment handling and diff display. They are not the same concrete fix and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 10, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43612-2", + "nodes": [ + "pull_request:43612", + "pull_request:43613" + ], + "soft_pairs": [ + "pull_request:43612|pull_request:43613" + ], + "prepared_review_unit_hash": "09117cbc12278132adbf6cdd50cb5a572422cd11d088f9499725c9ede487bcba", + "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 two related but distinct PRs adding separate pipelines for different segmentation tasks. They share registration/docs/test plumbing, but the underlying features and code paths are not the same, so they should not be deduplicated.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: #43612 adds Promptable Concept Segmentation, while #43613 adds Promptable Visual Segmentation; they target different tasks and model integrations.", + "best_issue_reason": null, + "best_pr_reason": "#43613 is the broader of the two and touches more models/tests, but it is still a separate feature rather than a duplicate of #43612.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different pipeline features with different task semantics and model support; overlap is limited to shared docs/registry scaffolding, 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": 11, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43651-2", + "nodes": [ + "pull_request:43651", + "pull_request:43663" + ], + "soft_pairs": [ + "pull_request:43651|pull_request:43663" + ], + "prepared_review_unit_hash": "609df94f1371bd8448d72f7f653be613c22ca5d9b4a5c17a3509787a40e8d0d8", + "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": "The two PRs are both trainer.py changes, but they target different code paths: one adjusts gradient-accumulation loss scaling, the other adds a hook for signature-column filtering. They are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "43651 is the more canonical representative of the cluster because it changes core training_step loss-scaling behavior and adds a dedicated control hook for GA scaling.", + "best_issue_reason": null, + "best_pr_reason": "43651 is the stronger representative: it is a concrete runtime behavior fix, while 43663 is a separate extensibility change for column filtering.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different underlying changes in trainer.py: 43651 fixes/controls gradient-accumulation loss scaling, while 43663 adds a separate signature-column override mechanism. They do not appear 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": 12, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "4ecc12039bdc0838076b2005210b677faaf64372bc7369be80dd252b167d3b16", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5992, + "estimated_input_tokens": 1498, + "estimated_eval_tokens": 3252 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5992, + "estimated_input_tokens": 1498, + "estimated_eval_tokens": 3252 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three PRs target TypeAdapter-related typing/runtime issues in `src/transformers/cli/serve.py`, but only one is the clean, narrowly scoped fix for the specific missing-pydantic NameError. The other two are broader or address a different annotation-evaluation path, so they should not be merged as duplicates.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43842 is the best canonical PR: it directly targets the explicit issue about `TypeAdapter` NameError when pydantic is not installed, with the smallest and most focused code change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43842 is the strongest candidate overall because it is narrowly scoped, directly matches its linked issue, and fixes the concrete runtime failure without bundling unrelated typing/refactor changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "Both touch `serve.py` and mention `TypeAdapter`, but #43656 is a broader annotation/`from __future__` typing cleanup while #43842 specifically guards `TypeAdapter` when pydantic is absent. Different issue targets and different concrete failure modes." + }, + { + "left": "pull_request:43836", + "right": "pull_request:43842", + "accept": false, + "reason": "They share the same file and issue target, but #43836 is a larger refactor with a generic string-literals workaround, whereas #43842 is a small targeted fix for the missing-pydantic NameError. They are not the same concrete change and would not be a clean single 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": 13, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": 14, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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: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:44018|pull_request:44116", + "pull_request:44071|pull_request:44722", + "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:44059|pull_request:44086", + "pull_request:44085|pull_request:44086", + "pull_request:44098|pull_request:44101", + "pull_request:44059|pull_request:44722", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44722", + "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:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44059|pull_request:44071", + "pull_request:44073|pull_request:44161", + "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", + "pull_request:44059|pull_request:44068", + "pull_request:44101|pull_request:44161", + "pull_request:43997|pull_request:44017", + "pull_request:43997|pull_request:44056", + "pull_request:44071|pull_request:44076", + "pull_request:44074|pull_request:44101", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44018", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44018|pull_request:44129", + "pull_request:44010|pull_request:44073", + "pull_request:43997|pull_request:44074", + "pull_request:43997|pull_request:44129", + "pull_request:44071|pull_request:44129", + "pull_request:44010|pull_request:44018", + "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" + ], + "prepared_review_unit_hash": "9554032547e5cc50580aaf9f14a041a20da4d10f1355123b86d41af7387fbd8a", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84895, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 + }, + "budget": { + "node_count": 27, + "item_count": 27, + "soft_pair_count": 72, + "serialized_chars": 75093, + "estimated_input_tokens": 18774, + "estimated_eval_tokens": 37804 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a set of separate per-model output-tracing refactors around the same umbrella issue (#43979). The only clear duplicate clusters are the GPT-J/CodeGen PRs and the GPT-Neo PR pair; the rest are same-theme but different code paths and should stay separate.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44722 is the best canonical representative: it explicitly targets #43979 and covers the same concrete GPT-J/CodeGen output-tracing refactor as the earlier GPT-J PRs, making it the most consolidated version of that change.", + "best_issue_reason": null, + "best_pr_reason": "PR 44722 is the strongest single representative of the cluster because it addresses the same concrete GPT-J/CodeGen tracing path, has the explicit issue target, and subsumes the earlier overlapping GPT-J-only PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model/code paths; sharing the same umbrella issue target is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different model/code paths; both are output-tracing refactors but not the same bug/change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Same GPT-J tracing refactor on the same concrete code path; these could plausibly be merged into one PR." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Different vision backbones and code paths; same refactor theme only." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families and code paths; only the tracing pattern overlaps." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "Different models/code paths; not the same underlying fix." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "Different models and concrete code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "Different models; same issue umbrella is too broad to treat as a duplicate." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Same GPT-Neo output-tracing refactor on the same code path; these look mergeable into one PR." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen versus GPT-Neo; different concrete changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "Different models and code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet versus GPT-J/CodeGen; same refactor family only." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "Different models/code paths; not the same bug/change." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "Different models and different concrete fixes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo versus Flaubert; same generic refactor pattern only." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo versus GPT-J/CodeGen; different concrete changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "Different model families; only shared output-handling style." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 versus MGP-STR; not the same underlying change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44161", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 versus MPT; same pattern only." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "Segformer versus LongT5; different concrete changes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo versus MPT; not the same fix." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet versus Flaubert/XLM; same refactor theme only." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet versus LongT5; different concrete changes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert versus MPT; not the same bug/change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo versus SpeechT5; only the refactor style overlaps." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 versus MPT; same pattern only." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44059", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44101", + "right": "pull_request:44161", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen tracing refactor; 44722 is the consolidated version of the same concrete change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J output-tracing refactor on the same code path; mergeable as one PR." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "Different models/code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "Different models/code paths; same refactor pattern only." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analyst summary is conservative and grounded in the packet. The accepted soft-edge pairs are limited to closely related PRs on the same model/code path, especially the GPT-J/CodeGen cluster and the GPT-Neo pair. The rejected pairs are consistently justified by different model families or different concrete fixes. I did not see an overstatement beyond the evidence provided." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 28, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44024", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44024|pull_request:44076", + "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": "f72c58a76fe68b47871b44c5183c08f227bbd49c02bebf9c5806ff2729febb9f", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84895, + "estimated_input_tokens": 21224, + "estimated_eval_tokens": 42704 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 6, + "serialized_chars": 17123, + "estimated_input_tokens": 4281, + "estimated_eval_tokens": 8818 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are all model-specific PRs in the same broad standardized-output-tracing theme, but they do not appear to be duplicates of one another. Each changes a different model implementation, so none of the soft edges should be merged.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44129 is the strongest representative of the cluster because it has an explicit issue target and a clear, concrete refactor in a single model path, but it is still specific to SpeechT5 rather than a duplicate of the other PRs.", + "best_issue_reason": null, + "best_pr_reason": "PR 44129 is the best single representative: it is well-scoped, explicitly linked to an issue, and clearly applies the standardized output-capture refactor to one concrete code path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch different model code paths (FocalNet vs ImageGPT) and are not the same underlying bug/change." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "These are separate refactors for different architectures (VisualBert vs TextNet); same theme, but not a duplicate fix." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet and VisualBert are distinct model implementations with different tracing setup; not the same concrete change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Different model families and different forward/tracing code paths; too broad to treat as one change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated model-specific output refactors, not a shared bug fix that could be merged into one PR." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "Both concern standardized output capture, but they modify different models and different implementation details, so they are not duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 15, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43998-13", + "nodes": [ + "pull_request:43998", + "pull_request:43999", + "pull_request:44000", + "pull_request:44001", + "pull_request:44002", + "pull_request:44003", + "pull_request:44004", + "pull_request:44025", + "pull_request:44026", + "pull_request:44027", + "pull_request:44028", + "pull_request:44029", + "pull_request:44030" + ], + "soft_pairs": [ + "pull_request:43998|pull_request:44002", + "pull_request:43998|pull_request:44000", + "pull_request:43998|pull_request:44001", + "pull_request:43998|pull_request:44003", + "pull_request:43998|pull_request:44029", + "pull_request:43999|pull_request:44001", + "pull_request:43999|pull_request:44002", + "pull_request:43999|pull_request:44025", + "pull_request:43999|pull_request:44026", + "pull_request:43999|pull_request:44027", + "pull_request:44000|pull_request:44001", + "pull_request:44000|pull_request:44002", + "pull_request:44000|pull_request:44003", + "pull_request:44000|pull_request:44025", + "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", + "pull_request:44001|pull_request:44025", + "pull_request:44002|pull_request:44029", + "pull_request:44002|pull_request:44003", + "pull_request:44003|pull_request:44028", + "pull_request:44003|pull_request:44004", + "pull_request:43998|pull_request:44004", + "pull_request:44001|pull_request:44004", + "pull_request:44002|pull_request:44004", + "pull_request:44000|pull_request:44004", + "pull_request:44025|pull_request:44026", + "pull_request:44025|pull_request:44027", + "pull_request:44025|pull_request:44028", + "pull_request:44025|pull_request:44029", + "pull_request:44025|pull_request:44030", + "pull_request:44026|pull_request:44029", + "pull_request:44026|pull_request:44027", + "pull_request:44026|pull_request:44028", + "pull_request:44026|pull_request:44030", + "pull_request:44027|pull_request:44028", + "pull_request:44027|pull_request:44029", + "pull_request:44027|pull_request:44030", + "pull_request:44028|pull_request:44029", + "pull_request:44028|pull_request:44030", + "pull_request:44029|pull_request:44030" + ], + "prepared_review_unit_hash": "77a4822b251bc6e9b710f0696c123e332e3f4fe9cefe11f527d2d1349f1119e4", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40168, + "estimated_input_tokens": 10042, + "estimated_eval_tokens": 20340 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40168, + "estimated_input_tokens": 10042, + "estimated_eval_tokens": 20340 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are mostly parallel, model-specific output-tracing refactors across unrelated architectures. I don\u2019t see true duplicate PRs here; the similarities are mostly from a shared refactor pattern rather than the same concrete fix. The Mamba/FalconMamba PR is the best representative of the cluster\u2019s actual tracing plumbing change.", + "confidence": 0.81, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44003 is the strongest representative: it contains the most substantive tracing-plumbing changes, spans two closely related model implementations, and is broader/more concrete than the other single-model refactors.", + "best_issue_reason": null, + "best_pr_reason": "PR 44003 best captures the underlying theme of the cluster with real code-path changes (can_return_tuple/output recording) rather than just a model-specific mechanical refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model implementations and code paths; both are output-tracing refactors, but not the same underlying bug/change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Unrelated architectures (timm_backbone vs vision_text_dual_encoder); similar refactor theme only." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Different models and runtime paths; same tracing motif, not a duplicate fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "These touch different model families; the shared output-tracing pattern is too broad to treat as one fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Separate architectures and implementations; not the same concrete change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated code paths; similarity is only the generic tracing refactor." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model families and different forward paths; not mergeable as one duplicate change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "Separate architectures; both are model-specific output-tracing updates, not one underlying fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "Different wrappers/models; same boilerplate refactor pattern only." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "Different code paths in different encoder-decoder wrappers; not the same concrete bug/change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Vision-text dual encoder and UnivNet are unrelated; shared refactor language is not enough." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different architectures and model internals; not a duplicate PR." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "These affect distinct model families and cannot plausibly be one merged fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Separate model implementations; same tracing refactor pattern only." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are unrelated changes; not the same code-path problem." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families and forward logic; not a duplicate." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Different architectures and code paths; shared output-tracing theme is too broad." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are unrelated implementations; not one underlying fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families and different tracing code paths; not duplicates." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "FalconMamba/Mamba vs SuperPoint are unrelated architectures; similar refactor pattern only." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Mamba/FalconMamba and CodeGen touch different model internals; not the same fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "Timm backbone and CodeGen are unrelated model code paths; not a duplicate." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different architectures; same refactor family only." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen are separate implementations; no shared concrete bug." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder and CodeGen are unrelated changes; not mergeable as one fix." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "DepthAnything and VisionEncoderDecoder are different model families; only the tracing refactor pattern matches." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Different code paths in different model families; not the same underlying issue." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "DepthAnything and SuperPoint are unrelated models; shared output-tracing refactor is too generic." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model families and implementation details; not a duplicate." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "DepthAnything and DPR touch different model code paths; not one concrete fix." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "VisionEncoderDecoder and RWKV are unrelated implementations; no duplicate code-path problem." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "These are two different encoder-decoder wrapper models; similar refactor style, but not the same underlying change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "VisionEncoderDecoder and SuperPoint are unrelated; the similarity is only the tracing refactor pattern." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model families and different runtime paths; not duplicates." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "SpeechEncoderDecoder and SuperPoint are unrelated code paths; same broad theme only." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "SpeechEncoderDecoder and RWKV are different architectures; not the same fix." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "Different model implementations and code paths; not a duplicate." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are unrelated changes; shared refactor pattern is not enough." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint and DPR are different model families; not the same underlying bug/change." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are unrelated architectures; 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": 16, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "eecb7edf0ff49fdef2945977cfdca42d16bf7b37972e9c9ee9be591572f4dcf2", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10913, + "estimated_input_tokens": 2729, + "estimated_eval_tokens": 5714 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10913, + "estimated_input_tokens": 2729, + "estimated_eval_tokens": 5714 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three pull requests are separate model-addition efforts (PenguinVL, DeepSeek-OCR-2, and Voxtral TTS). They share Transformers registration/docs scaffolding, but they do not appear to fix the same bug or implement the same concrete change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45075 is the strongest representative: it is a complete, non-draft model addition with substantial implementation, conversion, processing, and test coverage, and it has the highest review activity.", + "best_issue_reason": null, + "best_pr_reason": "PR #45075 is the best standalone PR in the cluster because it looks the most mature and complete, with broader model infrastructure and test coverage than the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Different models and code paths: PenguinVL vs Voxtral TTS. Shared auto-registration and docs files are generic model-addition scaffolding, not evidence of the same underlying change." + }, + { + "left": "pull_request:44662", + "right": "pull_request:45075", + "accept": false, + "reason": "Both are new model integrations, but they target different architectures and implementations (PenguinVL vs DeepSeek-OCR-2). Similar repo plumbing is not enough to treat them as duplicates." + }, + { + "left": "pull_request:45075", + "right": "pull_request:45401", + "accept": false, + "reason": "Distinct model families and functionality: DeepSeek-OCR-2 versus Voxtral TTS. They share only generic Transformers boilerplate, so they should not be merged as the same 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": 17, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-2", + "nodes": [ + "pull_request:44979", + "pull_request:45363" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363" + ], + "prepared_review_unit_hash": "11d37c6c037fa8f13f0a51f4e23d07f66ffbd972c13c30af7c74ea7ab0cc2cd1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4408, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Related fusion-feature PRs, but they target different layers: #44979 adds a generic Module Fusion API, while #45363 adds KernelConfig-driven n-to-1 kernel fusion integration and an example. They are not the same concrete change.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "#44979 is the more canonical representative of the cluster because it introduces the core reusable fusion API, whereas #45363 is a narrower KernelConfig-based integration on top of that theme.", + "best_issue_reason": null, + "best_pr_reason": "#44979 best represents the cluster\u2019s main subject since it defines the foundational module-fusion mechanism rather than a specific application of fusion.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "They are conceptually related, but not the same bug or change: one adds a general module-fusion API, the other implements n-to-1 kernel fusion via KernelConfig with different integration points and files." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45086-2", + "nodes": [ + "pull_request:45086", + "pull_request:45317" + ], + "soft_pairs": [ + "pull_request:45086|pull_request:45317" + ], + "prepared_review_unit_hash": "fd0bb7c38737398e30df6dc5011667ee7f088c4b19d195c39cd7bf609eed4ec6", + "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 implement the same Mistral regex fix in the same code path; the second PR appears to be a more complete version with tests, so they should be treated as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the stronger canonical PR because it contains the same fix as 45086 plus test coverage, making it the more complete candidate for the underlying bug.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the best overall PR candidate: it fixes the same AttributeError in `_patch_mistral_regex` and adds tests, so it better represents the intended change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying AttributeError fix in the same file and code path, with identical line changes; 45317 is effectively a superset of 45086 with added tests." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-2", + "nodes": [ + "pull_request:45115", + "pull_request:45569" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569" + ], + "prepared_review_unit_hash": "2db7f01b30d8df753ea503456d57b83df6d8e5fdfb705e761048f07cfb8e188a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs overlap on NemotronH internals, but they are not the same change: #45115 is a narrow inheritance/refactor and conversion-mapping update, while #45569 is a broader model-split PR adding dense/sparse variants, docs, and tests. They should not be merged as duplicates.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45569 is the better representative of the cluster because it covers the broader NemotronH family restructuring and adds the user-facing dense/sparse model support, while #45115 is a narrower internal refactor.", + "best_issue_reason": null, + "best_pr_reason": "PR #45569 is the strongest standalone representative: it is the larger, more complete change set and subsumes the main NemotronH model-organization work more than #45115 does.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Related NemotronH work, but different concrete changes and scopes: #45115 focuses on GraniteMoeHybrid-style inheritance and conversion key renames, while #45569 introduces separate nemotron_h_dense/sparse variants plus docs and tests. They are not the same fix and would not plausibly merge into one PR without losing scope clarity." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45332-2", + "nodes": [ + "pull_request:45332", + "pull_request:45333" + ], + "soft_pairs": [ + "pull_request:45332|pull_request:45333" + ], + "prepared_review_unit_hash": "bb445e52ebbba691fe11f6bd42ab53d3533101258835049478fecc0166e68f73", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5397, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5397, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two related PRs in the same heterogeneity initiative: one adds per-layer heterogeneous config support, the other adds heterogeneous modeling/cache handling. They overlap in helper files, but they are not the same concrete change and should not be deduplicated.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45332 is the broader umbrella change: it extends heterogeneous support into modeling and cache handling, while also relying on the config-layer helpers. It is the better representative PR for the cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 45332 is the best single representative of the cluster because it spans the largest portion of the new heterogeneous-model feature and subsumes the config work more than PR 45333 does.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Related feature work, but not the same bug/change: 45333 is per-layer config support, while 45332 is broader modeling/cache integration. Shared filenames alone are not enough to merge these PRs." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45415-2", + "nodes": [ + "pull_request:45415", + "pull_request:45425" + ], + "soft_pairs": [ + "pull_request:45415|pull_request:45425" + ], + "prepared_review_unit_hash": "142753e821cc7c25478f0cd24e8b1bb5777f8d82101ef1830c6f16ebb2b3aae7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are typing-related, but they are not duplicates: 45415 is a broad repo-wide type-checking sweep, while 45425 adds specific typing protocols/helpers for modeling and PEFT integration. They touch some of the same files, but not the same underlying change.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "45425 is the more focused, concrete typing-compatibility PR; 45415 is a much broader type-checking pass and should not be collapsed into the same change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Shared typing files are incidental; 45415 is a wide type-checking sweep, while 45425 adds targeted typing infrastructure and compatibility helpers. Different underlying changes." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45487-3", + "nodes": [ + "pull_request:45487", + "pull_request:45817", + "pull_request:45843" + ], + "soft_pairs": [ + "pull_request:45487|pull_request:45843", + "pull_request:45817|pull_request:45843" + ], + "prepared_review_unit_hash": "e6595a1048cc9f2501e2b5b1f5564f3d5ed38e57d2dba26a91f7b1af6ea22952", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 7561, + "estimated_input_tokens": 1891, + "estimated_eval_tokens": 4038 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 7561, + "estimated_input_tokens": 1891, + "estimated_eval_tokens": 4038 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-parallel fix PRs for different architectures (AltCLIP/ChineseCLIP, Gemma4, Deimv2). They share a broad theme but do not appear to fix the same concrete bug or be mergeable into one PR.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45817 is the strongest representative of the cluster theme: it has a clear, self-contained model-parallel fix, touches the relevant device-placement and no-split logic directly, and has the most discussion activity. The other PRs are separate model-specific fixes, not duplicates of it.", + "best_issue_reason": null, + "best_pr_reason": "PR #45817 is the best standalone PR in this set because it most directly addresses a concrete model-parallel failure mode with precise code-path changes. #45487 and #45843 are distinct fixes for different models and should not be conflated.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45487", + "right": "pull_request:45843", + "accept": false, + "reason": "Both are model-parallel themed, but they target different model families and different code paths (AltCLIP/ChineseCLIP vs Deimv2). They are not the same underlying bug and would not plausibly merge into one PR." + }, + { + "left": "pull_request:45817", + "right": "pull_request:45843", + "accept": false, + "reason": "Gemma4-specific parallelization fixes are unrelated to Deimv2-specific backbone/module-splitting changes. Same broad subsystem, but different concrete bugs and implementations." + } + ] + }, + "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": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45548-2", + "nodes": [ + "pull_request:45548", + "pull_request:45662" + ], + "soft_pairs": [ + "pull_request:45548|pull_request:45662" + ], + "prepared_review_unit_hash": "9d0009d8c186bcb04337a0cd054fe0e0ab0d54e0c5ff5a248a4ac51a22449c84", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related only at a broad EP/distributed-loading theme, but they fix different bugs and code paths: 45548 handles EP + DeepSpeed ZeRO-3 config/loading serialization, while 45662 fixes EP + FSDP2 expert tensors being overwritten during rank-0 broadcast.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": "No exact duplicate PR here; if one PR must represent the cluster, 45662 is the better anchor because it fixes the more concrete and severe expert-overwrite bug in the loading path.", + "best_issue_reason": null, + "best_pr_reason": "45662 is the stronger representative: it targets a specific corruption issue in distributed loading and spans the relevant code path more directly than 45548\u2019s narrower config/has_ep adjustment.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Same broad EP/distributed-loading area, but different underlying failures and fix paths; not plausible as one merged duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45639-2", + "nodes": [ + "pull_request:45639", + "pull_request:45645" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645" + ], + "prepared_review_unit_hash": "be0ad7f1b338599725a4d7c461ba48093cca7e72e0d95b7e31941d2ab7c36681", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4590, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4590, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same xdist collision problem around captured_info output, with 45645 appearing to subsume 45639 and add the CI-side file handling needed to preserve logs.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the more complete change: it includes the xdist-safe testing_utils fix from 45639 and also updates CI/reporting code to consume multiple captured_info files, making it the better canonical representative.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the strongest standalone candidate because it covers the full end-to-end fix, not just the helper change in testing_utils.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Same underlying bug: xdist-safe handling of captured_info artifacts. 45645 extends the exact same code-path fix in testing_utils and adds the related CI output changes, so these 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": 25, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45683-2", + "nodes": [ + "pull_request:45683", + "pull_request:45734" + ], + "soft_pairs": [ + "pull_request:45683|pull_request:45734" + ], + "prepared_review_unit_hash": "eba4ca0f26b2d3be0c00e316270a9341d31af2208b44dbab31b1d3a8e81e7538", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4469, + "estimated_input_tokens": 1118, + "estimated_eval_tokens": 2492 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4469, + "estimated_input_tokens": 1118, + "estimated_eval_tokens": 2492 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs touch the same quantization helper, but they fix different bugs: one special-cases audio submodules, the other changes how default skip lists are combined with user-supplied skips. They should not be clustered as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "45734 is the better representative PR because it fixes the core skip-module composition logic in `get_modules_to_not_convert`, which is broader and more central than the narrow audio-module special case in 45683.", + "best_issue_reason": null, + "best_pr_reason": "45734 is the strongest standalone PR in the set: it addresses a concrete quantization failure mode in the main code path and its fix is directly about the helper\u2019s behavior rather than a model-specific workaround.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45683", + "right": "pull_request:45734", + "accept": false, + "reason": "Both touch quantization skip logic, but 45683 adds audio-tower/embed_audio exclusions while 45734 changes additive handling of `skip_modules`; these are different underlying bugs and would not naturally merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 26, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45737-2", + "nodes": [ + "pull_request:45737", + "pull_request:45738" + ], + "soft_pairs": [ + "pull_request:45737|pull_request:45738" + ], + "prepared_review_unit_hash": "cc2ade73a49ec6a8883a7a24c433988645588bcd049e8aad2bce56b1b24daa1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same MusicgenMelody cache initialization bug in `forward()`: replacing `EncoderDecoderCache(DynamicCache, DynamicCache)` with a single `DynamicCache`. PR #45737 is the more complete duplicate because it bundles the fix with a regression test, while #45738 is the fix-only variant.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45737 is the best canonical PR because it contains the same code fix as #45738 plus a regression test, making it the most complete representation of the bug fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45737 is the strongest overall candidate: it fixes the concrete cache bug and adds a targeted regression test, whereas #45738 only changes the implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45737", + "right": "pull_request:45738", + "accept": true, + "reason": "Same underlying code-path bug in `musicgen_melody` cache setup; the diffs match on the core fix and the test-only addition in #45737 could plausibly live in the same PR as #45738." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: both PRs make the same `musicgen_melody` cache initialization change, and #45737 only adds a regression test on top of that fix. The soft-edge accept is conservative enough because the diffs are mergeable into one PR for the same bugfix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 27, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45858-2", + "nodes": [ + "pull_request:45858", + "pull_request:45890" + ], + "soft_pairs": [ + "pull_request:45858|pull_request:45890" + ], + "prepared_review_unit_hash": "43b423c313e03329c4798c8ce8a5e302c962ff893fc90d6de511f33af4986768", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4139, + "estimated_input_tokens": 1035, + "estimated_eval_tokens": 2326 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4139, + "estimated_input_tokens": 1035, + "estimated_eval_tokens": 2326 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated model-specific test PRs that both add XPU expectations, but they touch different model test files and different code paths. The soft similarity is superficial rather than duplicate-level.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:45890 is the stronger representative of the cluster because it is a complete XPU test update with deterministic-for-XPU gating, while pull_request:45858 is a separate model's expectation-only patch.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:45890 is the best PR candidate overall for this cluster: it is self-contained, clearly scoped, and the added XPU-specific guard makes the test change more robust.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45858", + "right": "pull_request:45890", + "accept": false, + "reason": "Different models and different test files; both are just adding XPU expectations, not fixing the same concrete bug or change. This is too broad to treat as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 28, + "component_count": 28, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45861-2", + "nodes": [ + "pull_request:45861", + "pull_request:45903" + ], + "soft_pairs": [ + "pull_request:45861|pull_request:45903" + ], + "prepared_review_unit_hash": "1c8f486fe1121354bc015c0bf642b856e06835e917aeccec7f6cc5156af36b18", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5529, + "estimated_input_tokens": 1383, + "estimated_eval_tokens": 3022 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5529, + "estimated_input_tokens": 1383, + "estimated_eval_tokens": 3022 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs target the same Qwen3-VL family M-RoPE/FSDP2 CPU-offload device mismatch, with 45903 expanding the earlier 45861 fix to also handle `meta` \u2192 `to_empty` re-init.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45903 is the better canonical PR because it includes the same `inv_freq.to(x.device)` fix as 45861 and additionally covers the `meta`/`to_empty` re-initialization path, making it the more complete version of the same fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 45903 is the best representative because it addresses the full concrete failure mode in the same code path and subsumes 45861's narrower patch.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45861", + "right": "pull_request:45903", + "accept": true, + "reason": "Same underlying M-RoPE device/offload bug in the Qwen3-VL family; 45903 contains the 45861 change and extends it with the additional `meta`/`to_empty` fix, so they are overlapping versions of one fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/manifest.json b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..e8f5701a2f4612342db9440fc0f54a3963c1ca41 --- /dev/null +++ b/snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260512t180027z", + "artifacts": { + "hybrid": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-05-12T18:10:17Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260512T180027Z", + "variant": "hybrid" +} diff --git a/snapshots/20260512T180027Z/manifest.json b/snapshots/20260512T180027Z/manifest.json index 73d6b9b0f865e769cc09da810c206fc1bae05caa..891d2f0b036b5846f4447a75a64bb122687c7e3d 100644 --- a/snapshots/20260512T180027Z/manifest.json +++ b/snapshots/20260512T180027Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 1253 }, "extracted_at": "2026-05-12T18:00:27Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260512t180027z", + "runs": { + "hybrid-model-20260512t180027z": { + "analysis_id": "hybrid-model-20260512t180027z", + "artifacts": { + "hybrid": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260512T180027Z/analysis-runs/hybrid-model-20260512t180027z/manifest.json", + "model": null, + "published_at": "2026-05-12T18:10:17Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260512T180027Z", "watermark": {