diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 245cc71d535fa164c0a085a1ef36871a30b53320..1ed87795637b20a556e5dae03eba1632c04a9117 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": "20260505T060044Z", - "generated_at": "2026-05-05T06:09:27Z", + "snapshot_id": "20260505T120036Z", + "generated_at": "2026-05-05T12:09:07Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 0dd584408f8186d1cb732c474be4fe725b9ffaeb..0e8ab030133f0a0f7c553615df23ebeeb71a5faf 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": "20260505T060044Z", - "generated_at": "2026-05-05T06:09:27Z", + "snapshot_id": "20260505T120036Z", + "generated_at": "2026-05-05T12:09:07Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -26,7 +26,7 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "d56c5d5a3f11345f515d7f21dbf2c83ed36190b41540e76cc58fa67ffd9340ba", + "prepared_review_unit_hash": "1fdeafe9f49fea0220d1477225d72dc9a7221709cfe69fa8056ae6f21684cb04", "component_budget": { "node_count": 3, "item_count": 3, @@ -48,24 +48,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues do not appear to be duplicates: one is about TAPAS tokenization parity, one about an unauthenticated private-model download error, and one about failing to load bert-base-uncased. The cluster is heterogeneous rather than a single underlying bug.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "The three issues do not look like duplicates: one is about Tapas tokenization, one about a bad auth/download error message, and one about a generic model-loading failure for bert-base-uncased. The soft similarities are too broad to merge them.", + "confidence": 0.88, + "canonical_issue_reason": "Issue #16618 is the broadest and most central of the set, describing a concrete model-loading failure, but it does not appear to subsume the Tapas tokenization or private-download error reports.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "#16618 is the best standalone representative because it is the most general user-facing failure report and the other two issues are clearly about different code paths.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Different problems: TAPAS tokenization mismatch vs. model loading failure for bert-base-uncased." + "reason": "Tapas tokenization differences and a bert-base-uncased load failure are different bugs in different code paths; only superficial NLP/model similarity." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Different code paths: private-model auth/download error vs. a general model-load failure." + "reason": "#15980 is specifically about a misleading error when downloading a private model while logged out, while #16618 is a generic inability to load a public model; related subsystem, but not the same underlying issue." } ] }, @@ -169,8 +169,8 @@ "issue:39290|issue:45081", "issue:43828|issue:45161", "issue:43425|issue:44292", - "issue:44918|issue:45161", "issue:34928|issue:41720", + "issue:44918|issue:45161", "issue:43716|issue:45685", "issue:43541|issue:43716", "issue:43828|issue:44292", @@ -187,8 +187,8 @@ "issue:30277|issue:36331", "issue:43039|issue:44805", "issue:34928|issue:36010", - "issue:43901|issue:44512", "issue:41762|issue:43596", + "issue:43901|issue:44512", "issue:39290|issue:41720", "issue:43039|issue:44560", "issue:43866|issue:44863", @@ -206,415 +206,361 @@ "issue:44841|issue:45084", "issue:41093|issue:41762" ], - "prepared_review_unit_hash": "afb0273794e45991c23982a6b35e3342e60e0aa4b344c67203d5ae6ad741fae4", + "prepared_review_unit_hash": "3f26833fc075bb4cba37195a72705af19d0575681da582ba6dae0ca3d30d34a4", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 67, - "serialized_chars": 47120, - "estimated_input_tokens": 11780, - "estimated_eval_tokens": 23816 + "serialized_chars": 47115, + "estimated_input_tokens": 11779, + "estimated_eval_tokens": 23814 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a grab bag of unrelated model/runtime/docs issues. The only clear duplicate-looking pair is the ZeRO-3/BERT size-0 IndexError reports.", - "confidence": 0.79, - "canonical_issue_reason": "issue:43596 is the clearest representative of the one plausible duplicate thread here (ZeRO-3 + BERT + size-0 IndexError), and it is the more general/earlier report.", + "summary": "Mostly unrelated issue reports, with one clear duplicate pair around DeepSpeed ZeRO-3 + Bert index-out-of-bounds. I would reject the docs-only pair as too different from the titles alone, and leave the rest unmerged as separate bugs.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43596 is the best canonical issue anchor because its title gives the clearest generic label for the DeepSpeed ZeRO-3/Bert index-out-of-bounds bug, and it has the strongest near-duplicate in issue:43638.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 best captures the only strong duplicate cluster in this batch and is the cleanest canonical issue candidate.", + "best_issue_reason": "issue:43596 is the strongest overall issue candidate: it is specific enough to anchor the DeepSpeed/Bert out-of-bounds cluster and is the one side of the only accepted duplicate-like pair.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different Qwen3.5 failures: config attribute error vs streaming chat/completions failure." + "reason": "Same model family name, but one is a config attribute error and the other is a streaming chat/completions failure." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Different hardware/backend problems on different models; not the same bug." + "reason": "Different backends and different failures: AMD GPU loading versus MPS histogram op support." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Mistral-3 dtype mismatch with image preprocessing is unrelated to GPT-OSS on AMD GPUs." + "reason": "Image-preprocessor dtype mismatch is not the same bug as AMD GPU model loading failure." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Phi-tiny-MoE autocast dtype mismatch is a different code path from GPT-OSS AMD GPU failures." + "reason": "Autocast dtype mismatch and AMD GPU incompatibility are different issues." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "AMD GPU runtime issue vs GraniteMoEHybrid invalid method; no shared underlying defect." + "reason": "Only loosely related by MoE; the concrete failures are different." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Qwen-3-8B-NVFP4 error and GPT-OSS AMD GPU error are unrelated model/runtime issues." + "reason": "Qwen NVFP4 runtime error is a different problem from AMD GPU support." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "DeepSpeed collective mismatch during training vs ModernBERT ONNX export are unrelated." + "reason": "DeepSpeed collective mismatch and ModernBERT ONNX export are unrelated bugs." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "ONNX export failure and Qwen3Moe FP8 failure do not look like the same bug." + "reason": "ONNX export failure and FineGrainedFP8Config failure are different model/runtime issues." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ModernBERT export problem and Qwen3 auto-device-map cuda assert are different failures." + "reason": "ModernBERT export and Qwen3 device-mapping CUDA assert do not share the same code-path problem." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are v5 docs cleanup items, but they reference different removed docs content and are not the same change." + "reason": "Both are v5 docs cleanups, but the removed topics differ: pipeline tasks versus the transformers run command." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "ModernBERT export and GenerationMixin import breakage are unrelated." + "reason": "ONNX export and GenerationMixin import error are distinct regressions." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Both are IndexError reports, but the titles point to different concrete failures and models." + "reason": "Both are index errors, but the titles point to different shapes and different model/setup contexts." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Same broad ZeRO-3/indexing area, but the reported error and symptom differ too much to merge." + "reason": "Same broad error class, but the titles indicate different bugs and different setups." }, { "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 model/trainer failures; no evidence of a shared code path." + "reason": "Mask-shape mismatch and GPT-OSS tensor-parallel failure are different issues." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "Tokenizer-mapping setup bug is unrelated to the mask/indexing error." + "reason": "Tokenizer-mapping utility failure is unrelated to the mask/tensor shape indexing bug." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset issue is unrelated to ModernBERT ONNX export." + "reason": "BarkProcessor voice preset handling and ModernBERT ONNX export are unrelated." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both involve DeepSpeed, but one is collective mismatch and the other is Bert zero-3 indexing; different bugs." + "reason": "Both involve DeepSpeed, but the titles indicate different concrete failures." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Near-identical ZeRO-3 + BERT + size-0 IndexError reports; same underlying bug and plausible single fix." + "reason": "The titles describe the same DeepSpeed ZeRO-3/Bert index-out-of-bounds failure, with the second report only narrowing the Bert setup." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "DeepSeek detokenization bug and LoRA save/reload inconsistency are different problems." + "reason": "Different symptoms and models: detokenization bug versus LoRA save/reload mismatch." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are unrelated." + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are not the same issue." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Qwen NVFP4 runtime issue and LoRA merge/save mismatch are not the same defect." + "reason": "Qwen NVFP4 runtime failure and LoRA merge/save inconsistency are unrelated." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Activation checkpointing/FSDP recompute mismatch is unrelated to the mask shape IndexError." + "reason": "Activation-checkpointing/FSDP recomputation mismatch is a different bug from the mask/tensor index error." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "DeepSpeed weight-dimension error and collective mismatch are different training failures." + "reason": "Both are DeepSpeed-related, but the titles point to different failures." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "ZeRO-3/BERT index error is unrelated to GPT-OSS tensor-parallel not working." - }, - { - "left": "issue:43716", - "right": "issue:45507", - "accept": false, - "reason": "Different model families and different failure modes." - }, - { - "left": "issue:39290", - "right": "issue:45081", - "accept": false, - "reason": "Gemma3 sliding_window_pattern and Mistral tokenizer regex patch are unrelated." - }, - { - "left": "issue:43828", - "right": "issue:45161", - "accept": false, - "reason": "Autocast dtype mismatch is not the same as GPT-OSS TP failure." + "reason": "Bert ZeRO-3 index error and GPT-OSS tensor-parallel failure do not match." }, { - "left": "issue:43425", - "right": "issue:44292", + "left": "issue:34928", + "right": "issue:36331", "accept": false, - "reason": "Torch version compatibility and Qwen NVFP4 runtime error are different issues." + "reason": "FSDP checkpointing mismatch and custom trainer signature change are unrelated." }, { - "left": "issue:44918", + "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Qwen3.5 embedding unpacking under TRL is unrelated to GPT-OSS TP behavior." - }, - { - "left": "issue:34928", - "right": "issue:41720", - "accept": false, - "reason": "FSDP checkpointing mismatch and Qwen3 auto-device-map assert are unrelated." + "reason": "DeepSpeed+Bert index error and GPT-OSS tensor-parallel issue are different code paths." }, { "left": "issue:43716", - "right": "issue:45685", + "right": "issue:45507", "accept": false, - "reason": "Mistral-3 dtype mismatch and MPS histogram kernel error are different backend bugs." + "reason": "Image-preprocessor dtype mismatch is unrelated to the GraniteMoE invalid-method failure." }, { - "left": "issue:43541", - "right": "issue:43716", + "left": "issue:39290", + "right": "issue:45081", "accept": false, - "reason": "Mixtral grouped_mm tracing error is unrelated to Mistral-3 image preprocessing dtype mismatch." + "reason": "Gemma3 config attribute issue and Mistral tokenizer regex patch crash are different bugs." }, { "left": "issue:43828", - "right": "issue:44292", - "accept": false, - "reason": "Both are model runtime errors, but the model and failure mechanism differ." - }, - { - "left": "issue:43039", - "right": "issue:44918", - "accept": false, - "reason": "Liger Kernel cross-entropy routing and Qwen3.5 embedding unpacking are unrelated." - }, - { - "left": "issue:41762", - "right": "issue:43638", - "accept": false, - "reason": "Both involve ZeRO-3, but Gemma3 loading and Bert loading failures are not the same concrete bug." - }, - { - "left": "issue:43827", - "right": "issue:44512", - "accept": false, - "reason": "Both are docs issues, but they concern different removed v5 references and should stay separate." - }, - { - "left": "issue:43716", - "right": "issue:44560", + "right": "issue:45161", "accept": false, - "reason": "Mistral-3 dtype mismatch and Qwen3-vl StopIteration are unrelated." + "reason": "Autocast dtype mismatch and MoE tensor-parallel failure are not the same underlying defect." }, { - "left": "issue:43828", - "right": "issue:45685", + "left": "issue:43425", + "right": "issue:44292", "accept": false, - "reason": "Autocast dtype mismatch and MPS histogram kernel error are different backend issues." + "reason": "Torch version incompatibility and Qwen NVFP4 runtime failure are different problems." }, { - "left": "issue:44560", - "right": "issue:45685", + "left": "issue:44292", + "right": "issue:44918", "accept": false, - "reason": "Video embedding StopIteration and MPS Int histogram support are unrelated." + "reason": "Different model families and different failure modes; no shared concrete bug." }, { "left": "issue:43039", "right": "issue:43828", "accept": false, - "reason": "Different model/backend failures with no shared code-path evidence." + "reason": "Liger Kernel cross-entropy routing and autocast dtype mismatch are not duplicates." }, { "left": "issue:30277", "right": "issue:36331", "accept": false, - "reason": "DeepSpeed collective mismatch and CustomTrainer signature breakage are unrelated." + "reason": "DeepSpeed collective mismatch and trainer kwarg mismatch are unrelated." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Liger Kernel cross-entropy issue is unrelated to the mask/indexing IndexError." + "reason": "Kernel selection issue and indexing shape mismatch are unrelated." }, { "left": "issue:34928", "right": "issue:36010", "accept": false, - "reason": "Activation checkpointing/FSDP mismatch and GenerationMixin import error are unrelated." + "reason": "Checkpointing/FSDP issue and GenerationMixin import error are unrelated." }, { - "left": "issue:43901", - "right": "issue:44512", + "left": "issue:41762", + "right": "issue:43638", "accept": false, - "reason": "Different documentation topics and different removed behaviors." + "reason": "Both are zero-3/loading errors, but Gemma3 and Bert failures are different concrete bugs." }, { - "left": "issue:41762", - "right": "issue:43596", + "left": "issue:43901", + "right": "issue:44512", "accept": false, - "reason": "Same DeepSpeed/size-0 theme, but Gemma3 zero-3 loading and Bert zero-3 loading are not the same bug." + "reason": "Different docs topics: return_all_scores versus the removed run command." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Gemma3 sliding_window_pattern and Qwen3 auto device mapping are different defects." + "reason": "Gemma3 config attribute issue and Qwen3 device-mapping CUDA assert are unrelated." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Liger Kernel loss routing and Qwen3-vl video StopIteration are unrelated." + "reason": "Liger Kernel cross-entropy routing and Qwen3-vl video StopIteration are different issues." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Checkpoint corruption and NemotronH checkpoint loading are not the same underlying issue." + "reason": "Checkpoint corruption and NemotronH checkpoint loading failure are not the same bug." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Unit-test load failure and checkpoint corruption are different problems." + "reason": "Unit-test model loading failure and corrupted checkpoint report different root causes." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Both mention Qwen3-related sliding-window behavior, but the titles point to different concrete failures." + "reason": "Both mention sliding-window-like behavior, but the model/config contexts are different." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Mask/tensor shape mismatch and Bert zero-3 size-0 index error are not clearly the same bug." + "reason": "Mask/index shape mismatch and Bert ZeRO-3 index error are different failures." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch version incompatibility and pydantic PretrainedConfig field breakage are unrelated." + "reason": "Torch compatibility and pydantic PretrainedConfig field regression are unrelated." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Qwen3.5 chat crash and Qwen3.5 streaming inference failure are different symptoms with no evidence of one root cause." + "reason": "Both are Qwen3.5 chat-related, but the crash and streaming API failure are different bugs." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper load failure and pydantic config breakage are unrelated." + "reason": "Whisper runtime failure and pydantic model-field breakage are unrelated." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Mistral3 untied lm_head weight and Ovis2 checkpoint corruption are different defects." + "reason": "lm_head tying bug and checkpoint corruption are different issues." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Qwen3.5 streaming inference failure is unrelated to the MPS histogram kernel issue." + "reason": "Qwen3.5 streaming failure and MPS histogram unsupported error are unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Mixtral dynamo-tracing grouped_mm error and MPS Int histogram error are unrelated." + "reason": "Torch Dynamo grouped_mm tracing error and MPS histogram issue are different code paths." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Grouped_mm tracing failure and autocast dtype mismatch are different bugs." + "reason": "Grouped_mm tracing failure and autocast dtype mismatch are different runtime bugs." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "GenerationMixin import breakage and Gemma3 sliding_window_pattern issue are unrelated." + "reason": "GenerationMixin import error and Gemma3 sliding-window config error are unrelated." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Processor load failure and template compilation error are different code paths." + "reason": "Processor failure for Voxtral and non-template-node compile error are different problems." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are indexing/loading errors, but the specific symptoms and models differ too much." + "reason": "Mask-shape mismatch and Gemma3 ZeRO-3 index error are not the same bug." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The assessment is grounded in the packet and stays conservative. The one accepted duplicate pair, issue:43596 and issue:43638, is well supported by the near-identical ZeRO-3/BERT size-0 IndexError titles and timestamps, and the rejection of the rest avoids overclaiming from broad thematic similarity alone. The summary is appropriately cautious and does not overstate evidence." + "accept": false, + "feedback": "Mostly conservative, but not fully grounded in the packet. The canonical/best-issue reasons rely on body-level claims like \u201cconcrete, reproducible\u201d and \u201cnarrower reproduction,\u201d which are not supported by the metadata shown here. The accepted 44509/44512 docs pair is also only weakly justified from the titles alone, since they describe different removed docs topics rather than an obviously identical change." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -632,11 +578,13 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:36331", "issue:39186", "issue:39290", "issue:39692", "issue:40990", "issue:41093", + "issue:41762", "issue:42915", "issue:43054", "issue:43296", @@ -677,7 +625,6 @@ "issue:45464", "issue:45507", "issue:45632", - "issue:45685", "issue:45698" ], "soft_pairs": [ @@ -693,18 +640,17 @@ "issue:43541|issue:44589", "issue:41093|issue:43596", "issue:43054|issue:43994", + "issue:40990|issue:44960", "issue:43716|issue:44918", - "issue:44918|issue:45685", + "issue:39186|issue:44918", "issue:45632|issue:45698", "issue:43404|issue:45005", "issue:43645|issue:45399", - "issue:40990|issue:44960", "issue:44530|issue:45005", - "issue:42915|issue:45399", - "issue:44360|issue:44512", "issue:44863|issue:45005", + "issue:42915|issue:45399", "issue:43873|issue:45005", - "issue:39186|issue:44918", + "issue:44360|issue:44512", "issue:44492|issue:44512", "issue:44918|issue:45507", "issue:43975|issue:44292", @@ -713,216 +659,241 @@ "issue:44292|issue:45507", "issue:39692|issue:43054", "issue:43828|issue:44928", - "issue:43866|issue:44403" + "issue:43866|issue:44403", + "issue:36331|issue:41762" ], - "prepared_review_unit_hash": "b1b1d3bd3df7e53a75860fbf5af49e93061f9dce33bf3cfd6aa2bc4aa83daf90", + "prepared_review_unit_hash": "62f08e0d122c65f4a427d1f9d66f92a3ff0221169ff87ec37ca974fb80ba22f9", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 36618, - "estimated_input_tokens": 9155, - "estimated_eval_tokens": 18566 + "serialized_chars": 37278, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed-topic cluster rather than one duplicate set. Two clear duplicate pockets stand out: SigLIP2 bad outputs, and the v5 non-persistent-buffer corruption regression. Most other soft pairs are only broadly related by subsystem or symptom and should stay separate.", - "confidence": 0.88, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a grab-bag of unrelated Transformers/model-specific bugs, but there are a few real duplicate pairs around the same root regressions: non-persistent buffer corruption, SigLIP2 output degradation, and trust_remote_code cache/module mixups.", + "confidence": 0.85, + "canonical_issue_reason": "issue:43644 is the clearest root report for the non-persistent-buffer regression and sits at the center of the strongest duplicate chain with issue:43950 and issue:44534.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43950 is the strongest standalone representative: it describes the concrete from_pretrained/non-persistent-buffer corruption regression clearly and is corroborated by near-duplicate wording in issue:43644 and issue:44534.", + "best_issue_reason": "issue:43644 is the best single canonical issue because it describes the underlying bug most generally and has direct duplicate support from later reports with near-identical symptoms.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "Different models and different failures: PaddleOCR-VL/vLLM load error vs. gpt-oss GGUF support." + "reason": "Different problems: PaddleOCR-VL/vLLM load failure vs GGUF support for gpt-oss architecture." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "One is a missing StableLmConfig field after v5 update; the other is a broader PretrainedConfig/pydantic-field regression." + "reason": "Both are config-related v5 regressions, but they affect different fields/code paths and are not the same bug." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Int4 quantization OOM and tied-weights bugs are different code paths and fixes." + "reason": "Unrelated failures: int4 CUDA memory/OOM vs tied-weights translation-model regression." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Mixtral torch-dynamo grouped_mm tracing failure is unrelated to GPT-OSS tensor-parallel behavior." + "reason": "Both involve MoE/MoE-like models, but one is a torch-dynamo grouped_mm tracing error and the other is a tensor-parallel limitation." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is not the same as Qwen3.5 streaming chat/completions failure." + "reason": "Different symptoms and code paths: autocast dtype mismatch vs streaming chat/completions failure." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a different bug than the pydantic PretrainedConfig field issue." + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a distinct initialization bug from the pydantic PretrainedConfig break." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "Both mention modern models, but GPT-OSS GGUF support and Qwen3.5 streaming inference are separate problems." + "reason": "No meaningful overlap beyond being model-specific issues; one is GGUF architecture support, the other is streaming inference." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding_window handling is not the same as a PretrainedConfig field regression." + "reason": "Sliding-window handling for Qwen3-MoE is a different regression from the PretrainedConfig/pydantic field issue." }, { "left": "issue:39290", "right": "issue:43742", "accept": false, - "reason": "Gemma3TextConfig missing attribute in vLLM and MobileLLM key-error loading are distinct model/load bugs." + "reason": "Both are loading-time errors, but the missing sliding_window_pattern on Gemma3TextConfig is unrelated to the MobileLLM key error." }, { "left": "issue:43541", "right": "issue:44589", "accept": false, - "reason": "Torch-dynamo grouped_mm tracing failure is unrelated to the missing Float8 storage type error." + "reason": "Tracing/grouped_mm failure is unrelated to the missing Float8 storage serialization issue." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Mask length mismatch during indexing is a different failure from the zero-size tensor index error with ZeRO-3/BERT." + "reason": "Both are IndexErrors, but one is a mask/tensor shape mismatch and the other is a zero-sized dimension during DeepSpeed ZeRO-3 init." }, { "left": "issue:43054", "right": "issue:43994", "accept": true, - "reason": "Both describe the same SigLIP2 regression: degraded/nonsensical text embeddings or outputs from the same model family, likely the same processor/model mismatch." + "reason": "Both report the same SigLIP2 regression: text embeddings / AutoModel outputs are badly degraded or nonsensical." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Different model families and failure modes: high perplexity on gpt-oss-20b vs a GLM5 issue." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Mistral-3 image-preprocessor dtype mismatch and Qwen3.5 SFT embedding unpacking are unrelated bugs." + "reason": "Different bugs: Mistral-3 image preprocessor dtype mismatch vs Qwen3.5 embedding unpacking failure in TRL SFT trainer." }, { - "left": "issue:44918", - "right": "issue:45685", + "left": "issue:39186", + "right": "issue:44918", "accept": false, - "reason": "TRL SFT embedding unpacking and the MPS histogram_mps Int error are different failures in different stacks." + "reason": "FSDP 'weight must be 2-D' is unrelated to the SFT trainer embedding unpacking error." }, { "left": "issue:45632", "right": "issue:45698", - "accept": false, - "reason": "Both involve custom loading, but one is a trust_remote_code cache-path collision and the other is wrong module selection after save_pretrained; not the same concrete bug." + "accept": true, + "reason": "Both point to the same trust_remote_code/custom-module cache collision, where the wrong local module is loaded after saving or when models share a leaf directory name." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights, but one is a Mistral3 lm_head tying bug and the other is a translation-models v5 regression." + "reason": "Mistral3 tied-weight loading bug is different from the translation-model tied-weights regression described in issue:45005." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Notebook custom-model initialization and flash-attn fallback gating are unrelated." + "reason": "Jupyter custom-model initialization and flash-attn fallback blocking are unrelated regressions." }, { - "left": "issue:40990", - "right": "issue:44960", + "left": "issue:44530", + "right": "issue:45005", "accept": false, - "reason": "High perplexity on gpt-oss-20b and the GLM5 issue do not share a concrete fix path." + "reason": "PagedAttentionCache on Qwen3.5 is a different runtime bug from tied weights in translation models." }, { - "left": "issue:44530", + "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "Qwen3.5 PagedAttentionCache linear_attention crash is not the same as tied-weights behavior in translation models." + "reason": "NemotronH checkpoint loading failure is unrelated to translation tied-weights behavior." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "FineGrainedFP8Config failure and flash-attn fallback checks are separate issues." - }, - { - "left": "issue:44360", - "right": "issue:44512", - "accept": false, - "reason": "DSA indexer ReLU discussion and a docs typo about the removed run command are unrelated." - }, - { - "left": "issue:44863", - "right": "issue:45005", - "accept": false, - "reason": "NemotronH checkpoint-loading failure is different from the tied-weights translation-model regression." + "reason": "Qwen3 MoE + FineGrainedFP8 failure is unrelated to flash-attn fallback gating." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization/offloading behavior and tied-weights handling are not the same bug." + "reason": "Quantization offloading behavior is a different problem from tied weights in translation models." }, { - "left": "issue:39186", - "right": "issue:44918", + "left": "issue:44360", + "right": "issue:44512", "accept": false, - "reason": "FSDP weight dimensionality error and Qwen3.5 embedding unpacking are unrelated." + "reason": "DSA indexer/ReLU discussion is unrelated to the removed transformers run command docs issue." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "A cache-strategy typo is not the same as outdated docs about the removed transformers run command." + "reason": "A typo in cache strategies and outdated CLI docs are separate documentation issues." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "TRL SFT embedding unpacking and GraniteMoEHybrid invalid method calls are different code paths." + "reason": "Qwen3.5 embedding unpacking in TRL is unrelated to GraniteMoEHybrid invalid method calls." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Incorrect detokenization for deepseek-coder is unrelated to the Qwen-3-8B-NVFP4 runtime error." + "reason": "Incorrect detokenization and Qwen-3-8B-NVFP4 runtime errors are different bugs." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Same v5 regression: non-persistent buffers are being overwritten/corrupted on load, just phrased differently." + "reason": "Same underlying non-persistent-buffer corruption regression in Transformers 5.x, described with very similar wording." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Same underlying non-persistent-buffer corruption bug in Transformers 5.x." + "reason": "Same bug: from_pretrained corrupts non-persistent buffers / fills them with junk in Transformers 5.x." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime error and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "SigLIP2 doc example errors are not the same as the broader SigLIP2 embedding quality regression." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different Qwen/Phi model training/runtime failures with different mechanisms." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Corrupted Ovis2 checkpoint is unrelated to the loading-noise behavior issue." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "CustomTrainer unexpected argument and Gemma3 ZeRO-3 out-of-bounds are unrelated failures despite both involving training/loading." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted soft edges are well supported by the issue titles, especially the non-persistent-buffer regression trio and the SigLIP2 output pair. The rejected pairs are mostly appropriately kept separate based on distinct models, failure modes, or subsystems, with no clear overstatement of evidence." + "feedback": "The canonical issue choice is well grounded: issue:43644 is clearly supported by the duplicate chain to 43950 and 44534, and the cluster summary stays appropriately cautious about the rest being mostly unrelated model-specific bugs. The accepted soft-edge pairs are also conservative given the packet, and I don't see any overstatement that would warrant rejection." }, "evaluator_used": true, "retried": false, @@ -943,7 +914,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:36331", "issue:38175", "issue:41553", "issue:41762", @@ -985,6 +955,7 @@ "issue:45020", "issue:45081", "issue:45237", + "issue:45305", "issue:45313", "issue:45356", "issue:45399", @@ -992,31 +963,30 @@ "issue:45701" ], "soft_pairs": [ - "issue:36331|issue:41762", "issue:41553|issue:43883", "issue:43872|issue:44589", + "issue:43054|issue:45701", "issue:38175|issue:43054", "issue:41553|issue:43881", - "issue:43054|issue:45701", "issue:42947|issue:44387", + "issue:43054|issue:43493", "issue:44403|issue:44451", "issue:43643|issue:45020", "issue:43881|issue:43883", "issue:43975|issue:45520", - "issue:43054|issue:43493", "issue:43975|issue:45237", "issue:44403|issue:44488", "issue:42947|issue:43856", - "issue:42491|issue:44368", "issue:41762|issue:42915", + "issue:42491|issue:44368", + "issue:43782|issue:44960", "issue:43975|issue:44451", - "issue:44661|issue:45020", "issue:43824|issue:43872", + "issue:44661|issue:45020", "issue:44991|issue:45399", "issue:45020|issue:45356", - "issue:43782|issue:44960", - "issue:43827|issue:44509", "issue:45081|issue:45399", + "issue:43827|issue:44509", "issue:43792|issue:44220", "issue:43404|issue:43950", "issue:43742|issue:45399", @@ -1024,264 +994,271 @@ "issue:43257|issue:43866", "issue:43742|issue:45020", "issue:41553|issue:44991", - "issue:43940|issue:45313", "issue:43872|issue:44291", + "issue:43940|issue:45313", + "issue:42915|issue:43872", "issue:43854|issue:44863", - "issue:42915|issue:43872" + "issue:43257|issue:45305" ], - "prepared_review_unit_hash": "5e37c8dfa8b5574eace33cd366420468aeab63c6c0c49ff005fcd56b808a5210", + "prepared_review_unit_hash": "3a45aa316acf5c5e5d54d3730c425dc93bbf57944a5da7632eb90c389a130a5b", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 36, - "serialized_chars": 37866, - "estimated_input_tokens": 9467, - "estimated_eval_tokens": 19190 + "serialized_chars": 37873, + "estimated_input_tokens": 9469, + "estimated_eval_tokens": 19194 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a mixed cluster of mostly unrelated model-loading, tokenizer, training, and docs issues. Only two soft edges look like true duplicates: the v5 docs pipeline cleanup pair and the `_is_hf_initialized` loading regression pair.", - "confidence": 0.77, - "canonical_issue_reason": "issue:43872 is the clearest exact-duplicate anchor here: issue:44291 describes the same `_is_hf_initialized` kwarg failure in the init_empty_weights / quantized-loading path.", + "summary": "Mostly a heterogeneous set of Transformers issues with only a few clear duplicate pairs: SigLIP2 implementation mismatch, remote_code/config loading regressions, docs about removed pipelines, and a bitsandbytes/_is_hf_initialized loading regression. Most other soft edges are different model-specific failures and should stay separate.", + "confidence": 0.58, + "canonical_issue_reason": "issue:45020 is the broadest open umbrella for the recent remote_code / config-loading regression subset, even though the cluster overall is mixed.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the broadest cluster-level umbrella for recent remote_code/model-loading regressions, even though most items are not exact duplicates of it.", + "best_issue_reason": "issue:45020 is the best global issue candidate because it is open and captures the widest shared regression theme among the closest matches.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:36331", - "right": "issue:41762", - "accept": false, - "reason": "Different failure modes: custom trainer signature mismatch vs Gemma3 ZeRO-3 index error." - }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Unrelated models and code paths: Voxtral AutoTokenizer error message vs Molmo tied-weights attribute error." + "reason": "Different models and failures: Voxtral tokenizer error vs Molmo tied-weights attribute error." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are loading-related TypeErrors, but they point to different incompatibilities (_is_hf_initialized vs Float8 storage lookup)." + "reason": "Both are loading/runtime errors, but the root causes differ: bitsandbytes init arg vs Float8 storage lookup." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "One is a SigLIP2 quality complaint; the other is a tokenizer/version regression. Not the same bug." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Both mention SigLIP2, but one is zero probabilities and the other is degraded text embeddings; not clearly the same bug." + "reason": "Both concern SigLIP2, but one is zero probabilities and the other is lower embedding quality; too broad to merge." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Different models and unrelated loading failures." - }, - { - "left": "issue:43054", - "right": "issue:45701", - "accept": false, - "reason": "SigLIP2 embedding quality issue vs generic tokenization change regression; not the same underlying bug." + "reason": "Unrelated model-loading failures for different models and error paths." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Both involve memory, but one is LoRA gradient checkpointing and the other is int4 reserved-memory growth; different regressions." + "reason": "Different areas: LoRA gradient checkpointing vs int4 quantization memory/OOM behavior." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": true, + "reason": "Both point to the same SigLIP2 implementation discrepancy causing degraded text embeddings." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "Generic loading noise vs a concrete ScandiBERT loading failure; not a duplicate." + "reason": "Generic loading-noise report vs a specific ScandiBERT load failure; not the same underlying bug." }, { "left": "issue:43643", "right": "issue:45020", - "accept": false, - "reason": "401remote_code issues are broad, but this is specifically missing fields from AutoConfig; not enough evidence of the same bug." + "accept": true, + "reason": "Both describe recent remote_code/config-loading breakage; 43643 looks like a specific manifestation of the broader regression." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different model families and different failures." + "reason": "Different models and different failure modes." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Tokenizer detokenization regression vs flash_attn import key error on Python 3.13." - }, - { - "left": "issue:43054", - "right": "issue:43493", - "accept": false, - "reason": "Related SigLIP2 area, but this is a performance/embedding-quality report and the other is a broader implementation discrepancy claim." + "reason": "Tokenizer detokenization regression vs flash_attn import failure; unrelated code paths." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Tokenizer regression vs AMD GPU runtime support issue." + "reason": "Different problems: tokenizer behavior vs AMD GPU incompatibility." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "Noise/warnings on load vs a specific model load failure; different problems." + "reason": "Both are load-related, but they are model-specific complaints with no clear shared root cause." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Both are training/memory reports, but they describe different symptoms and likely different code paths." - }, - { - "left": "issue:42491", - "right": "issue:44368", - "accept": false, - "reason": "Qwen3 MoE version-compatibility/load issue vs a tie_word_embeddings warning in Qwen3.5 LoRA fine-tuning." + "reason": "Training memory issues, but one is gradient checkpointing with LoRA and the other is general Qwen3 MoE memory usage." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Gemma3 ZeRO-3 index error vs Qwen3 MoE FP8 failure; unrelated." + "reason": "Different models and configurations: Gemma3 ZeRO-3 indexing error vs Qwen3 MoE FP8 load failure." }, { - "left": "issue:43975", - "right": "issue:44451", + "left": "issue:42491", + "right": "issue:44368", "accept": false, - "reason": "Different model/tokenizer regressions with no clear shared root cause." + "reason": "Both mention Qwen3 and LoRA, but one is an hf4.x/hf5.x compatibility break and the other is a tie_word_embeddings warning." }, { - "left": "issue:44661", - "right": "issue:45020", + "left": "issue:43782", + "right": "issue:44960", "accept": false, - "reason": "add-new-model-like mapping bug is unrelated to the broader remote_code loading regressions." + "reason": "Different model families and unrelated failure descriptions." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Qwen2.5-VL import error vs bitsandbytes _is_hf_initialized TypeError; unrelated." + "reason": "Import error for a missing class vs bitsandbytes parameter incompatibility; different underlying issues." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "A tokenizer mapping utility bug is not the same as the remote_code loading regression." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading regression vs flash-attn fallback gating; different bugs." + "reason": "Tokenizer loading failure vs flash-attn fallback logic; unrelated." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Broad remote_code breakage vs a specific Kimi-K2.5 tokenizer regression; not the same concrete defect." + "reason": "Remote_code regression and Kimi-K2.5 tokenizer codec regression are different bugs." }, { - "left": "issue:43782", - "right": "issue:44960", + "left": "issue:45081", + "right": "issue:45399", "accept": false, - "reason": "Different model families and distinct loading errors." + "reason": "Mistral regex patch crash vs flash-attn fallback gating; separate code paths." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Both report the same stale docs problem: pipeline task names removed in v5 are still referenced." - }, - { - "left": "issue:45081", - "right": "issue:45399", - "accept": false, - "reason": "Tokenizer regex patch crash vs flash-attn fallback gating; unrelated." + "reason": "Same docs regression: removed pipeline tasks still referenced in v5 documentation." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Whisper runtime failure and a feature-extraction-function issue may be related, but the evidence is not strong enough to call them the same bug." + "reason": "Whisper cannot run is too broad to equate confidently with the specific fbank feature-extraction issue." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "Weight tying bug in Mistral3 vs silent non-persistent buffer corruption; different regressions." + "reason": "Tied lm_head weights vs non-persistent buffer corruption; distinct model state bugs." }, { "left": "issue:43742", "right": "issue:45399", "accept": false, - "reason": "MobileLLM loading key error vs flash-attn fallback behavior; unrelated." + "reason": "MobileLLM loading failure and flash-attn fallback gating are unrelated." }, { "left": "issue:42915", "right": "issue:45020", "accept": false, - "reason": "Qwen3 MoE FP8 failure is much narrower than the broad remote_code regression bucket." + "reason": "Qwen3 MoE FP8 loading failure is model/config specific, not the remote_code regression." }, { "left": "issue:43257", "right": "issue:43866", "accept": false, - "reason": "Qwen3 MoE weight-conversion/load issue vs Ovis2 checkpoint corruption; unrelated." + "reason": "Qwen3 MoE accelerate/deepspeed weight conversion vs corrupted Ovis2 checkpoint; different causes." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "Specific MobileLLM key error does not match the broad remote_code regression." + "reason": "Different model-loading bugs with no shared concrete failure path." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Voxtral tokenizer error message vs tokenizer-loading failure for a different model; unrelated." - }, - { - "left": "issue:43940", - "right": "issue:45313", - "accept": false, - "reason": "Both are ZeRO-3 weight-loading issues, but they affect different Qwen variants and describe different missing-weight patterns." + "reason": "Different tokenizer/loading failures for different models." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same underlying `_is_hf_initialized` TypeError during init_empty_weights / quantized model loading." + "reason": "Both center on the unexpected `_is_hf_initialized` argument during model init / init_empty_weights loading." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 weight-loading failure pattern across Qwen3 variants." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "GLM-4.7-Flash test loading problem vs NemotronH checkpoint-loading failure; unrelated." + "reason": "Different model families and different load failures." + }, + { + "left": "issue:44403", + "right": "issue:45020", + "accept": false, + "reason": "Generic loading noise and remote_code breakage are not the same bug." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "Different failures: FP8 Qwen3 MoE loading vs `_is_hf_initialized` quantized-loading TypeError." - } - ] - }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet, and the two accepted soft edges are well-supported by the issue titles. The remaining soft-edge rejections stay conservative and do not overclaim duplicate status." - }, - "evaluator_used": true, - "retried": false, + "reason": "FP8 config loading failure vs bitsandbytes Int8Params argument incompatibility." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-specific checkpoint/load problems." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Different Qwen3 training/deepspeed issues: weight conversion vs gradient averaging." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative overall. The accepted duplicate pairs are plausible from the issue titles and themes, and the rejections generally avoid overclaiming shared root cause. The summary correctly characterizes the cluster as mostly heterogeneous with a few clear duplicate themes. One small omission is the Qwen3 ZeRO-3 pair, but that does not undermine the overall evaluation." + }, + "evaluator_used": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1309,13 +1286,13 @@ "issue:41553", "issue:42617", "issue:42915", - "issue:43257", "issue:43317", "issue:43366", "issue:43531", "issue:43550", "issue:43596", "issue:43638", + "issue:43644", "issue:43645", "issue:43646", "issue:43824", @@ -1328,6 +1305,7 @@ "issue:44479", "issue:44488", "issue:44530", + "issue:44534", "issue:44560", "issue:44589", "issue:44661", @@ -1339,7 +1317,6 @@ "issue:44918", "issue:44928", "issue:45030", - "issue:45305", "issue:45313", "issue:45325", "issue:45357", @@ -1349,250 +1326,250 @@ ], "soft_pairs": [ "issue:44928|issue:45588", - "issue:41553|issue:44488", - "issue:43257|issue:45305", "issue:36331|issue:41093", + "issue:41553|issue:44488", "issue:43872|issue:45538", - "issue:44530|issue:44863", "issue:41093|issue:44805", + "issue:44530|issue:44863", "issue:44479|issue:44560", "issue:24643|issue:30064", "issue:43366|issue:43828", + "issue:44155|issue:45584", "issue:43366|issue:43531", "issue:43550|issue:44910", + "issue:43824|issue:43957", "issue:43645|issue:43646", "issue:43366|issue:44589", + "issue:42915|issue:43957", "issue:43950|issue:45325", "issue:44560|issue:44918", "issue:42617|issue:43366", - "issue:42915|issue:43957", - "issue:43824|issue:43957", "issue:36010|issue:42915", "issue:38175|issue:43994", "issue:43646|issue:43950", - "issue:44155|issue:45584", - "issue:44877|issue:45030", "issue:44560|issue:44805", + "issue:44877|issue:45030", "issue:43638|issue:44661", "issue:43596|issue:44661", "issue:43317|issue:44756", + "issue:43644|issue:44534", "issue:38175|issue:39692", "issue:45313|issue:45357" ], - "prepared_review_unit_hash": "8bf8f40d4c157d33c86d4fe87cd01e8ca5272531b6a58533cdc480a1bd89acd5", + "prepared_review_unit_hash": "f49ee4c7b46a9b62ed03917a01b7498a973f7f8322ab87ff1f013b1e1fa3baaf", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36297, - "estimated_input_tokens": 9075, - "estimated_eval_tokens": 18406 + "serialized_chars": 36228, + "estimated_input_tokens": 9057, + "estimated_eval_tokens": 18370 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous cluster of unrelated issues, with only a few clear duplicate pockets. The strongest overlaps are the mask/tensor shape IndexError reports, the Transformers 5.0 custom model initialization regression, and the SigLIP2 output regression; the rest look distinct.", - "confidence": 0.82, - "canonical_issue_reason": "No single canonical issue for the whole cluster; it splits into multiple unrelated bug families. The closest duplicate anchor is the mask-shape IndexError pair.", + "summary": "This cluster is mostly a mix of unrelated issue reports across different models and subsystems. Only a few soft pairs look like true duplicates: the Transformers 5 custom-model initialization regressions, the non-persistent buffer regression, and the SigLIP2 output regression.", + "confidence": 0.89, + "canonical_issue_reason": "No single canonical issue fits the whole cluster: the items span unrelated bugs and regressions, with only a handful of local duplicate pairs.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue for the entire cluster for the same reason; if forced to anchor one duplicate pocket, the earliest mask-shape IndexError report is the cleanest representative.", + "best_issue_reason": "No global best issue: there is no coherent umbrella bug in this set, so any single issue would misrepresent the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "Different Qwen3.5 bugs: SDPA/3D position_ids NaN vs flash_attention s_aux=None crash." + "reason": "Different failures: Qwen3.5 RLHF/SDPA BF16 gradient instability vs flash_attention crash on sink-less models." }, { - "left": "issue:41553", - "right": "issue:44488", + "left": "issue:36331", + "right": "issue:41093", "accept": false, - "reason": "Different models and failures; Voxtral tokenizer error message vs sleng-bert load failure." + "reason": "Unrelated symptoms and code paths: trainer API keyword mismatch vs mask/tensor shape IndexError." }, { - "left": "issue:43257", - "right": "issue:45305", + "left": "issue:41553", + "right": "issue:44488", "accept": false, - "reason": "Both DeepSpeed-related, but one is weight conversion and the other gradient averaging; not the same bug." + "reason": "Both are loading problems, but for different models and different failure modes; not the same bug." }, { - "left": "issue:36331", - "right": "issue:41093", + "left": "issue:43872", + "right": "issue:45538", "accept": false, - "reason": "Different code paths: custom trainer signature mismatch vs mask/tensor shape IndexError." + "reason": "Bitsandbytes constructor incompatibility is unrelated to the CLIPTokenizer model_max_length issue." }, { - "left": "issue:43872", - "right": "issue:45538", + "left": "issue:41093", + "right": "issue:44805", "accept": false, - "reason": "bitsandbytes init incompatibility vs CLIPTokenizer max_length bug." + "reason": "Same error family, but no evidence they come from the same concrete bug or code path." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Qwen3.5 PagedAttention cache crash vs NemotronH checkpoint loading." - }, - { - "left": "issue:41093", - "right": "issue:44805", - "accept": true, - "reason": "Same generic mask/tensor shape mismatch IndexError; 44805 looks like a later duplicate report." + "reason": "Distinct model-specific problems: PagedAttentionCache linear_attention crash vs NemotronH checkpoint loading failure." }, { "left": "issue:44479", "right": "issue:44560", "accept": false, - "reason": "Both are video/Qwen issues, but the failures differ: broad video-input regression vs StopIteration in Qwen3-vl-embedding." + "reason": "Both involve Qwen video flows, but one is a broad regression and the other is a specific embedding StopIteration failure." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Training/DeepSpeed weight shape error vs image processor void-segmentation issue." + "reason": "Completely different subsystems: DeepSpeed training RuntimeError vs image processor void segmentation map handling." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "GGUF/gpt-oss support vs Phi-tiny-MoE autocast dtype mismatch." + "reason": "GGUF gpt-oss support and torch.autocast dtype mismatch are unrelated." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 batched leakage and Whisper empty-transcription failure are different bugs." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "Unrelated Qwen3-MoE sliding_window bug vs GGUF gpt-oss support." + "reason": "Different Qwen/GGUF topics with no shared concrete failure path." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Different models and kernels: Bamba torch.compile/SDPA vs Qwen3.5 flash-attention position_ids crash." + "reason": "Different model families and different bugs: torch.compile/SDPA vs FlashAttention 3D position_ids handling." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Importing a specific class and meta-device loading regressions are separate issues." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Same Transformers 5.0 custom model initialization regression; one is just a notebook-specific reproduction." + "reason": "Same custom-model initialization regression in Transformers 5.0.0; one report is just more specific (Jupyter notebooks)." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "GGUF architecture support vs missing Float8 storage type." + "reason": "Unrelated: GGUF architecture support vs missing Float8 storage object." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Different failures: FineGrainedFP8Config on Qwen3Moe vs meta-device model loading regression." }, { "left": "issue:43950", "right": "issue:45325", "accept": false, - "reason": "Buffer loading regression vs Qwen2.5-VL rope/temporal position scaling." + "reason": "Non-persistent buffer corruption is unrelated to Qwen2.5-VL rope/index scaling." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Video StopIteration in Qwen3-vl-embedding vs Qwen3.5 TRL unpacking failure." + "reason": "Different bugs: video StopIteration in Qwen3-vl-embedding vs TRL SFT unpacking failure." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel.py runtime failure unrelated to GGUF support." - }, - { - "left": "issue:42915", - "right": "issue:43957", - "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure vs meta-device model loading." - }, - { - "left": "issue:43824", - "right": "issue:43957", - "accept": false, - "reason": "Import/export of Qwen2_5_VL class vs meta-device loading regression." + "reason": "3d_parallel.py runtime failure is unrelated to GGUF gpt-oss support." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "GenerationMixin import error vs Qwen3Moe FP8 config." + "reason": "GenerationMixin import breakage and Qwen3Moe FP8 loading are unrelated." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Same SigLIP2 model/output regression; both report broken predictions for google/siglip2-base-patch16-224." + "reason": "Same SigLIP2 model regression causing incorrect/zero probabilities and nonsensical outputs; the second is the broader symptom report." }, { "left": "issue:43646", "right": "issue:43950", "accept": false, - "reason": "Custom model init regression vs non-persistent buffer corruption." + "reason": "Custom model init regression and non-persistent buffer corruption are separate Transformers 5.x bugs." }, { - "left": "issue:44155", - "right": "issue:45584", + "left": "issue:44560", + "right": "issue:44805", "accept": false, - "reason": "AudioFlamingo3 track leak vs Whisper empty transcription bug." + "reason": "Video-generation StopIteration and generic mask shape mismatch are unrelated." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Different config-validation failures on different models." - }, - { - "left": "issue:44560", - "right": "issue:44805", - "accept": false, - "reason": "Different error classes: video StopIteration vs generic mask shape mismatch." + "reason": "Both are config validation issues, but they affect different models and are not the same bug." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "Zero3/Bert index error vs tokenizer-mapping restriction in add-new-model-like." + "reason": "DeepSpeed ZeRO-3 BertModel initialization failure is unrelated to tokenizer-mapping add-new-model-like behavior." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Zero3/Bert index error vs tokenizer-mapping restriction in add-new-model-like." + "reason": "Different code paths and symptoms: ZeRO-3 index error vs tokenizer mapping failure." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Dequantized model offload failure vs Strix Halo mmap OOM." + "reason": "device_map/offload loading failure is unrelated to disabling mmap for Strix Halo OOM." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same regression about non-persistent buffers being filled with junk in Transformers v5; the titles are effectively the same bug." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Same model family but different issues: doc-example validation/quantization vs output probabilities." + "reason": "Same model family, but the documented example errors are not clearly the same root cause as the zero-probabilities regression." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Load-time ZeRO-3 weight issue vs save_pretrained visual encoder key regression." + "reason": "Qwen3.5 ZeRO-3 weight loading and incorrect visual encoder save keys are different bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the cluster is mostly heterogeneous, and the three accepted duplicate pockets are tightly supported by the titles. The rejected soft edges are conservative and do not overclaim evidence." + "feedback": "The cluster-level summary is grounded in the packet: the items span many unrelated model/subsystem failures, and the three accepted soft pairs are plausible near-duplicates based on their titles. The rejected pairs are phrased conservatively and do not overclaim equivalence." }, "evaluator_used": true, "retried": false, @@ -1636,13 +1613,13 @@ "issue:43957", "issue:43994", "issue:44075", + "issue:44155", "issue:44164", "issue:44291", "issue:44387", "issue:44402", "issue:44451", "issue:44488", - "issue:44534", "issue:44661", "issue:44683", "issue:44756", @@ -1662,279 +1639,279 @@ "issue:45759" ], "soft_pairs": [ - "issue:43644|issue:44534", "issue:44960|issue:44977", "issue:44451|issue:44488", "issue:42491|issue:43720", + "issue:41720|issue:42915", "issue:43708|issue:44075", "issue:43854|issue:43931", "issue:42491|issue:45276", - "issue:41720|issue:42915", "issue:43856|issue:43873", "issue:43257|issue:43854", - "issue:39692|issue:43994", "issue:42915|issue:44756", + "issue:39692|issue:43994", + "issue:44488|issue:44991", "issue:42491|issue:43931", "issue:43856|issue:44387", "issue:44661|issue:44683", "issue:43644|issue:43645", - "issue:44910|issue:45588", "issue:45362|issue:45406", + "issue:44910|issue:45588", "issue:42491|issue:45292", - "issue:44488|issue:44991", "issue:43404|issue:43454", "issue:36010|issue:39404", "issue:45276|issue:45335", "issue:44164|issue:44291", + "issue:41720|issue:44155", "issue:44402|issue:44488", "issue:39692|issue:44960", "issue:43425|issue:43957", "issue:42947|issue:45704", + "issue:43873|issue:44387", "issue:43335|issue:45663", "issue:45663|issue:45759", - "issue:43873|issue:44387", "issue:44387|issue:44756", - "issue:43854|issue:45216", - "issue:44291|issue:44756" + "issue:44291|issue:44756", + "issue:43854|issue:45216" ], - "prepared_review_unit_hash": "8637dd97acadfb7b439285db8292b5acf7620bdea49ad9e3fa2adf2155f78a6e", + "prepared_review_unit_hash": "24566ff4e2209aa9feed2401135e79bd80b14eea04a6dff752a1750ac0004a9f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 34, - "serialized_chars": 37316, - "estimated_input_tokens": 9329, - "estimated_eval_tokens": 18914 + "serialized_chars": 37373, + "estimated_input_tokens": 9344, + "estimated_eval_tokens": 18944 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a grab bag of distinct Transformers bugs; only two soft edges look like true duplicates: the Transformers 5 non-persistent-buffer junk issue and the resize_token_embeddings propagation bug. The rest are too model- or code-path-specific to merge.", - "confidence": 0.81, - "canonical_issue_reason": "issue:43644 is the strongest representative because it has an almost exact duplicate in 44534 and the bug statement is very specific and reusable as a canonical report.", + "summary": "The set is mostly heterogeneous; there is one likely duplicate subgroup around `resize_token_embeddings` not propagating to nested embeddings in Gemma/T5Gemma variants, but the rest of the items look unrelated.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45276 is the broadest description of the `resize_token_embeddings` regression, covering nested embedding/output weight updates rather than a single narrow symptom.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the best overall issue candidate: clear repro class, strong duplicate evidence, and a concrete Transformers 5 regression with broad impact.", + "best_issue_reason": "issue:45276 is the best representative of the only plausible duplicate cluster: it states the underlying embedding-resize bug more generally than the paired report.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43644", - "right": "issue:44534", - "accept": true, - "reason": "Same bug: Transformers v5 filling non-persistent buffers with junk; titles describe the same regression nearly verbatim." - }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Both involve generation/flash-attention, but they target different models and failure modes; not the same concrete bug." + "reason": "Both are model-specific generation failures, but one is a generic GLM5 report and the other is a Qwen3.5 flash-attention crash; not the same bug." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are model-loading complaints, but the titles point to different models and no shared root cause is evident." + "reason": "Both are tokenizer/model loading complaints, but they target different checkpoints and likely different model metadata problems." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Unrelated: Qwen3 MoE LoRA compatibility vs BitNet packed-weight loading in CI." + "reason": "Different code paths: one is LoRA compatibility across HF versions, the other is BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Same broad family (Qwen3 MoE) but different failures: device-mapping CUDA assert vs FineGrainedFP8Config loading." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Different problems: checkpoint resume step counting vs SGD optimizer arguments not being used." + "reason": "Trainer checkpoint math and optimizer argument handling are unrelated bugs." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different model-loading failures for different models and different symptoms." + "reason": "Different model-loading failures on different architectures; no shared code-path is evident." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Different subsystems: Qwen3 MoE LoRA loading vs token embedding resize propagation." - }, - { - "left": "issue:41720", - "right": "issue:42915", - "accept": false, - "reason": "Both are Qwen3-related, but one is a device-mapping CUDA assert and the other is a FineGrainedFP8Config failure." + "reason": "LoRA/Qwen3-MoE compatibility is unrelated to token embedding resizing in Gemma4." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both mention memory/quantization themes, but the concrete bugs are different." + "reason": "Both mention memory/resource issues, but one is Qwen3 MoE training memory usage and the other is quantization offloading behavior." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different model families and unrelated loading paths." + "reason": "Qwen3 MoE weight conversion with accelerate+deepspeed is unrelated to GLM-4.7-Flash loading." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3Moe FP8 loading and Strix Halo mmap OOM are different subsystems and failure modes." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Both are SigLIP2-related, but one is a docs example with model/processor mismatch and the other is a nonsensical-output complaint." + "reason": "Both are SigLIP2-related, but one is a docs example with model/processor mismatch and the other is a specific checkpoint producing bad outputs; not clearly the same bug." }, { - "left": "issue:42915", - "right": "issue:44756", + "left": "issue:44488", + "right": "issue:44991", "accept": false, - "reason": "Unrelated: Qwen3 MoE FP8 failure vs Strix Halo mmap/OOM behavior." + "reason": "Both involve tokenizer loading, but they affect different models and no common underlying regression is shown." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "Different issues: Qwen3 MoE LoRA loading vs Qwen3-VL weight-shape mismatch." + "reason": "Qwen3-MoE LoRA incompatibility is unrelated to Qwen3-VL weight-shape mismatch." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both are memory-related, but one is training inefficiency and the other is int4 quantization OOM; not the same bug." + "reason": "Different quantization issues: training memory inefficiency vs int4 reserved-memory/OOM regression." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Different areas: tokenizer-mapping registration vs compiled flex_attention on new torch." + "reason": "Tokenizer-mapping registration and compiled flex_attention are unrelated." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Different regressions: junk non-persistent buffers vs custom-model initialization in notebooks." - }, - { - "left": "issue:44910", - "right": "issue:45588", - "accept": false, - "reason": "Both are flash-attention crashes, but one is a packed-sequence position_ids bug and the other is an s_aux=None AttributeError." + "reason": "One is junk-filled non-persistent buffers, the other is notebook custom-model initialization; both may be v5 regressions but not the same bug." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different model/application paths: chat crash in Qwen3.5 vs serve crash in Gemma4Processor." + "reason": "Qwen3.5 chat crashes and Gemma4Processor missing `_tokenizer` are different model-specific failures." }, { - "left": "issue:42491", - "right": "issue:45292", + "left": "issue:44910", + "right": "issue:45588", "accept": false, - "reason": "Unrelated: Qwen3 MoE LoRA loading vs resize_token_embeddings not updating output embeddings." + "reason": "Both touch flash-attention, but one is a packed-sequence/position_ids crash and the other is an `s_aux=None` AttributeError in sink-less models." }, { - "left": "issue:44488", - "right": "issue:44991", + "left": "issue:42491", + "right": "issue:45292", "accept": false, - "reason": "Both are tokenizer/loading problems, but they concern different models and different failure symptoms." + "reason": "LoRA/Qwen3-MoE compatibility is not the same bug as generic `resize_token_embeddings` not updating output embeddings." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Similar wording about tied lm_head weights, but these are separate model implementations and not clearly one duplicate bug." + "reason": "Both involve untied `lm_head`, but they are different model classes with different generation paths; too model-specific to merge as one duplicate." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "Unrelated: missing GenerationMixin import vs Whisper pipeline return_language regression." + "reason": "A GenerationMixin import error and Whisper `return_language` pipeline regression are unrelated." }, { "left": "issue:45276", "right": "issue:45335", "accept": true, - "reason": "Same underlying bug: resize_token_embeddings is not propagating to the relevant tied output/decoder embeddings in related model variants." + "reason": "Both report the same underlying regression: `resize_token_embeddings` does not propagate to nested embeddings/output weights in composite Gemma/T5Gemma models." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Different serialization/loading failures: extra_state handling vs init_empty_weights/_is_hf_initialized TypeError." + "reason": "Saving/loading extra_state and `init_empty_weights` `_is_hf_initialized` are distinct serialization/init bugs." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Qwen3 device-mapping CUDA assert and AudioFlamingo3 batched inference token leak are unrelated." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Both are model loading issues, but they involve different models and different failure descriptions." + "reason": "One is a tokenizer vocab-size mismatch for a specific model, the other is a separate model loading failure." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "Different model families and unrelated failure modes." + "reason": "SigLIP2 docs errors and a GLM5 report are unrelated model families." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch version incompatibility is unrelated to meta-device loading breakage." + "reason": "Torch 2.10 compatibility and meta-device loading failures are different regressions." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "Different concerns: LoRA gradient checkpointing vs an apex RMSNorm memory leak." + "reason": "Gradient checkpointing with PEFT LoRA and T5 using apex FusedRMSNorm are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Quantization offloading expectations and int4 reserved-memory OOM are not the same concrete issue." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Unrelated model-specific bugs: sparse-layer config creation vs FSDP2 shared_kv_states handling." + "reason": "Sparse-layer config bug and Gemma-4 FSDP2 shared_kv_states error are different model internals." }, { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Different bugs in different model/config paths; no shared concrete code-path problem." + "reason": "Gemma-4 FSDP2 attention state rebuilding and Qwen 3.5/3.6 `from_config` text_config unwrapping are unrelated." }, { - "left": "issue:43873", - "right": "issue:44387", + "left": "issue:44387", + "right": "issue:44756", "accept": false, - "reason": "Both are memory-related, but one is quantization offloading behavior and the other is int4 CUDA reserved-memory growth." + "reason": "Int4 quantization OOM and Strix Halo mmap OOM are different memory problems with different triggers." }, { - "left": "issue:44387", + "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Different memory/OOM issues with no shared root cause." + "reason": "`init_empty_weights` loading error and Strix Halo mmap OOM are unrelated." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "Different model-loading/save regressions for different models." - }, - { - "left": "issue:44291", - "right": "issue:44756", - "accept": false, - "reason": "Different issues: init_empty_weights argument mismatch vs mmap-induced OOM." + "reason": "GLM-4.7-Flash unit-test loading and Qwen3.5 save_pretrained regression are different model/version problems." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays grounded in the packet and the duplicate calls are conservative. The two accepted soft edges are strongly supported by the titles/body-level similarity, and the rejected pairs are appropriately cautious about model- or failure-specific differences. The only mild stretch is the phrasing about \"broad impact\" for issue:43644, but it does not materially overstate the evidence." + "feedback": "The analysis is grounded in the packet and stays appropriately conservative. The one accepted duplicate pair (45276/45335) is well supported by the titles and the shared `resize_token_embeddings` propagation bug. The rest of the soft-edge rejections avoid overclaiming and do not appear to merge unrelated issues." }, "evaluator_used": true, "retried": false, @@ -1956,7 +1933,6 @@ "cluster_id": "cluster-15354-384", "nodes": [ "issue:39404", - "issue:41720", "issue:42947", "issue:43257", "issue:43299", @@ -1968,13 +1944,12 @@ "issue:43708", "issue:43716", "issue:43828", - "issue:43854", - "issue:43856", "issue:43901", "issue:43906", "issue:43931", "issue:43957", - "issue:44155", + "issue:43994", + "issue:44079", "issue:44164", "issue:44265", "issue:44292", @@ -1984,6 +1959,8 @@ "issue:44509", "issue:44530", "issue:44589", + "issue:44671", + "issue:44743", "issue:44756", "issue:44805", "issue:44863", @@ -1993,6 +1970,7 @@ "issue:44918", "issue:44928", "issue:44960", + "issue:44991", "issue:45084", "issue:45276", "issue:45292", @@ -2005,279 +1983,285 @@ ], "soft_pairs": [ "issue:44292|issue:44912", + "issue:44743|issue:44960", + "issue:43957|issue:44756", "issue:42947|issue:43381", "issue:45684|issue:45698", "issue:45276|issue:45292", "issue:43381|issue:44928", "issue:44164|issue:44756", - "issue:41720|issue:44155", "issue:45588|issue:45663", - "issue:43646|issue:43708", "issue:44479|issue:44877", - "issue:43854|issue:44960", - "issue:43957|issue:44756", + "issue:43646|issue:43708", "issue:45292|issue:45335", "issue:44402|issue:44451", "issue:43645|issue:44164", - "issue:43828|issue:43957", "issue:43257|issue:44863", + "issue:43828|issue:43957", "issue:43425|issue:43828", + "issue:39404|issue:45701", "issue:44589|issue:44756", "issue:44479|issue:45663", + "issue:43931|issue:44863", "issue:43906|issue:44530", - "issue:39404|issue:45701", "issue:43611|issue:43645", "issue:44530|issue:45084", "issue:43425|issue:43716", - "issue:43931|issue:44863", - "issue:43901|issue:44509", "issue:39404|issue:44912", + "issue:43901|issue:44509", "issue:44863|issue:45663", "issue:44265|issue:44898", "issue:43299|issue:43931", "issue:44479|issue:44805", - "issue:43425|issue:44918", - "issue:43856|issue:44756", - "issue:44589|issue:45084" + "issue:43994|issue:44079", + "issue:44671|issue:44991", + "issue:39404|issue:44671", + "issue:43425|issue:44918" ], - "prepared_review_unit_hash": "1b38e4851ea3336280727f64fbf49915264e4aa012bb64d2496b1de1c7b7ce10", + "prepared_review_unit_hash": "cecc7395e8c2b83be9c11081938997f6b4564fd76e2f057caa650afd76bedb99", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 37526, - "estimated_input_tokens": 9382, - "estimated_eval_tokens": 19020 + "serialized_chars": 38104, + "estimated_input_tokens": 9526, + "estimated_eval_tokens": 19308 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: most items are separate regressions across loading, quantization, training, and docs. The only clearly duplicate-worthy subcluster is the resize_token_embeddings regression around output embeddings/decoder embeddings (45276, 45292, 45335).", - "confidence": 0.69, - "canonical_issue_reason": "issue:45292 is the broadest, most central description of the resize_token_embeddings regression affecting output embeddings; issue:45276 and issue:45335 look like model-specific manifestations of the same bug.", + "summary": "Most pairs are distinct regressions affecting different models, code paths, or behaviors. The only clear duplicate cluster here is the resize_token_embeddings propagation bug around output_embeddings / decoder.embed_tokens, where the Gemma4 and T5Gemma reports describe the same underlying failure pattern.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45292 is the cleanest general representative of the resize_token_embeddings regression because it states the core symptom without model-specific noise, and it matches the shared output_embeddings failure described in the nearby model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45292 is the best anchor for deduping the resize_token_embeddings bug because it is generic rather than model-specific and matches the shared underlying failure mode.", + "best_issue_reason": "issue:45292 is the best standalone anchor for this cluster: it is broad, concise, and captures the shared API-level bug that the other resize_token_embeddings reports appear to instantiate.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are quantization-related load failures, but they involve different models and symptoms (NVFP4 run error vs MXFP4 fallback to bf16), so they do not look like the same bug." + "reason": "Same broad quantization/loading theme, but different models and different failure modes; not the same underlying bug." + }, + { + "left": "issue:44743", + "right": "issue:44960", + "accept": false, + "reason": "Different model families and different runtime behavior; no shared concrete code-path bug." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "One is a meta-device loading regression, the other is an mmap/OOM workaround issue; unrelated." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Both mention gradient checkpointing, but one is about PEFT/LoRA effectiveness and the other is an eval-mode restriction; different code paths and failure modes." + "reason": "Both mention gradient checkpointing, but one is about PEFT LoRA ineffectiveness and the other about eval-mode prohibition; different bugs." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "One is about read-only permissions during save_pretrained, the other about loading the wrong custom module after save_pretrained; related area, but distinct bugs." + "reason": "Both involve custom-model save/load flows, but one is a permissions propagation issue and the other is the wrong module being loaded." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Both report resize_token_embeddings not propagating to output embeddings; issue:45276 is a Gemma4-specific manifestation of the broader bug in issue:45292." + "reason": "Both describe the same resize_token_embeddings propagation failure; 45276 is the Gemma4-specific manifestation of the broader output_embeddings bug." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "These are unrelated: eval-mode gradient checkpointing restriction versus RLHF NaN/gradient explosion from 3D position_ids and SDPA fallback." + "reason": "Different problems: eval-mode checkpointing restriction versus RLHF NaN/SDPA fallback instability." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "Save/from_pretrained extra_state handling and mmap OOM on Strix Halo are unrelated problems." - }, - { - "left": "issue:41720", - "right": "issue:44155", - "accept": false, - "reason": "Different subsystems and symptoms: device-mapping cuda assert on Qwen3 versus batched AudioFlamingo3 inference leakage." + "reason": "Extra_state serialization failure is unrelated to the mmap/OOM issue." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash-attention s_aux=None crash and Gemma4 FSDP2 KeyError are different concrete bugs despite both touching model internals." - }, - { - "left": "issue:43646", - "right": "issue:43708", - "accept": false, - "reason": "Custom model initialization breakage and Trainer max_steps miscalculation on checkpoint resume are unrelated." + "reason": "Different subsystems and failure points: flash attention aux handling versus Gemma4 FSDP2 shared_kv_states construction." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Video-input regression for Qwen models and strict config blocking granite_speech loading are different failures." - }, - { - "left": "issue:43854", - "right": "issue:44960", - "accept": false, - "reason": "GLM-4.7-Flash loading issues and GLM5 are related only by family name; the reports describe different model/version problems." + "reason": "One is a video-input regression in Qwen models; the other is strict config handling for granite_speech." }, { - "left": "issue:43957", - "right": "issue:44756", + "left": "issue:43646", + "right": "issue:43708", "accept": false, - "reason": "Meta-device loading failures and Strix Halo mmap OOM are unrelated." + "reason": "Custom model initialization in notebooks and Trainer resume/max_steps calculation are unrelated." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Both describe the same resize_token_embeddings propagation bug: issue:45292 mentions output_embeddings generally, while issue:45335 is the decoder.embed_tokens manifestation in t5gemma." + "reason": "Both report the same resize_token_embeddings bug on model embeddings; 45335 is the T5Gemma-specific variant of the broader output_embeddings regression." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "A tokenizer vocab mismatch and a generic load failure for ScandiBERT are not clearly the same underlying bug." + "reason": "Tokenizer vocab-size mismatch and inability to load a specific model are different issues." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Notebook custom-model initialization and extra_state save/load handling are different issues." + "reason": "Notebook-defined custom model initialization and extra_state save/load handling are distinct regressions." }, { - "left": "issue:43828", - "right": "issue:43957", + "left": "issue:43257", + "right": "issue:44863", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE and meta-device loading breakage are separate bugs." + "reason": "Different model-loading failures for different architectures; not the same code-path bug." }, { - "left": "issue:43257", - "right": "issue:44863", + "left": "issue:43828", + "right": "issue:43957", "accept": false, - "reason": "Both are loading/checkpoint problems, but one is Qwen3 MoE conversion with accelerate+deepspeed and the other is NemotronH checkpoint compatibility; not the same bug." + "reason": "Autocast dtype mismatch and meta-device loading failure are separate bugs." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch version incompatibility and autocast dtype mismatch are different failure modes." + "reason": "Torch version compatibility and an autocast dtype mismatch are not the same issue." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to tokenizer version-dependent behavior." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage deserialization and mmap OOM are unrelated." + "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap/OOM problem." }, { "left": "issue:44479", "right": "issue:45663", "accept": false, - "reason": "Qwen video regression and Gemma4 FSDP2 shared_kv_states KeyError are unrelated." + "reason": "Video-input regression in Qwen models and Gemma4 FSDP2 state handling are different bugs." }, { - "left": "issue:43906", - "right": "issue:44530", + "left": "issue:43931", + "right": "issue:44863", "accept": false, - "reason": "An isolated reproduction of another issue and a PagedAttentionCache linear_attention crash are not clearly the same bug." + "reason": "Different model families and distinct checkpoint-loading failures." }, { - "left": "issue:39404", - "right": "issue:45701", + "left": "issue:43906", + "right": "issue:44530", "accept": false, - "reason": "Whisper pipeline return_language and version-dependent tokenization changes are unrelated." + "reason": "Insufficient evidence of a shared underlying bug; one is only an isolated reproduction reference." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both are Transformers 5.0 custom-model regressions, but one is about base_model_prefix loading and the other about notebook-based initialization; too different to merge as duplicates." + "reason": "Both are Transformers 5 custom-model regressions, but they affect different mechanisms and should not be merged." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and compile non-template-nodes TypeError are unrelated." + "reason": "PagedAttentionCache linear_attention crash and template-node compilation TypeError are unrelated." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch 2.10 incompatibility and Mistral-3 image-preprocessor/model dtype mismatch are different issues." + "reason": "Torch compatibility and a model/preprocessor dtype mismatch are different problems." }, { - "left": "issue:43931", - "right": "issue:44863", + "left": "issue:39404", + "right": "issue:44912", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and NemotronH checkpoint loading are different model-specific load failures." + "reason": "Whisper pipeline behavior and MXFP4 quantization loading are unrelated." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs issues, but they cover different pipeline/task deprecations and do not describe the same change." - }, - { - "left": "issue:39404", - "right": "issue:44912", - "accept": false, - "reason": "Whisper pipeline language return and quantization loading/fallback issues are unrelated." + "reason": "Both are docs-related, but they cover different pipeline task deprecations and different pages." }, { "left": "issue:44863", "right": "issue:45663", "accept": false, - "reason": "NemotronH checkpoint loading and Gemma4 FSDP2 KeyError are distinct model-loading/training bugs." + "reason": "NemotronH checkpoint loading and Gemma4 FSDP2 shared_kv_states rebuilding are unrelated." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export.export with torch_compilable_check and Perceiver resize/interpolation failure are different problems." + "reason": "torch.export failure with torch_compilable_check and Perceiver non-default-resolution failure are different code paths." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Both involve Qwen3-VL, but one is a dev-version MoE loading regression and the other is a checkpoint shape mismatch; not clearly the same underlying bug." + "reason": "Qwen3VL MoE loading regression and Qwen3-VL weight-shape mismatch are distinct loading issues." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Video-input regression and mask/index shape mismatch are not the same issue." + "reason": "Video-input regression and mask shape IndexError are different failures." }, { - "left": "issue:43856", - "right": "issue:44756", + "left": "issue:43994", + "right": "issue:44079", "accept": false, - "reason": "Qwen3 MoE training memory inefficiency and Strix Halo mmap OOM are unrelated performance/memory issues." + "reason": "Incorrect SigLIP2 outputs and ModelOutput key assignment are unrelated." }, { - "left": "issue:44589", - "right": "issue:45084", + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT MLM prediction regression and tokenizer loading failure are separate issues." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline return_language and CamemBERT MLM predictions are unrelated bugs." + }, + { + "left": "issue:43425", + "right": "issue:44918", "accept": false, - "reason": "Float8 storage lookup failure and compile-on-template-nodes error are unrelated." + "reason": "Torch version incompatibility and Qwen3.5 embedding unpacking in TRL are different regressions." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays mostly conservative. The resize_token_embeddings trio (45276, 45292, 45335) is the only clearly duplicate-looking subcluster from the titles, and the other pairwise rejections are cautious rather than overstated." + "feedback": "The cluster summary is grounded in the packet and stays appropriately conservative. The only soft-edge accepts (45276\u219445292 and 45292\u219445335) are well-supported by the near-identical resize_token_embeddings/output_embeddings wording and the high similarity scores. The non-duplicates are mostly correctly rejected, and the choice of 45292 as the canonical anchor is reasonable because it is the broadest formulation of the shared bug." }, "evaluator_used": true, "retried": false, @@ -2322,19 +2306,19 @@ "issue:43645", "issue:43646", "issue:43716", + "issue:43856", "issue:43883", - "issue:43994", - "issue:44079", + "issue:43957", "issue:44164", "issue:44206", "issue:44220", "issue:44451", "issue:44466", - "issue:44479", "issue:44488", "issue:44530", - "issue:44671", + "issue:44589", "issue:44683", + "issue:44756", "issue:44811", "issue:44849", "issue:44987", @@ -2348,283 +2332,280 @@ "issue:45759" ], "soft_pairs": [ + "issue:43856|issue:44756", + "issue:44589|issue:45084", "issue:43596|issue:44683", "issue:40990|issue:42947", "issue:43425|issue:45685", - "issue:39404|issue:44671", - "issue:43994|issue:44079", - "issue:44671|issue:44991", "issue:44164|issue:45759", "issue:36032|issue:43452", - "issue:44991|issue:45081", "issue:32090|issue:39404", - "issue:44164|issue:44479", + "issue:44991|issue:45081", + "issue:43957|issue:44164", + "issue:43582|issue:43632", "issue:44488|issue:45081", "issue:40990|issue:43421", - "issue:43582|issue:43632", "issue:43883|issue:44488", "issue:43611|issue:43644", + "issue:44488|issue:44987", "issue:44206|issue:44220", "issue:43716|issue:45084", - "issue:44811|issue:45584", - "issue:44488|issue:44987", "issue:34567|issue:40444", "issue:43643|issue:44530", + "issue:44811|issue:45584", "issue:43644|issue:43646", + "issue:36032|issue:42915", "issue:43883|issue:44991", "issue:44451|issue:44987", - "issue:36032|issue:42915", "issue:43541|issue:45084", "issue:43452|issue:43482", "issue:42915|issue:44530", "issue:43317|issue:43482", "issue:44849|issue:45325", - "issue:43404|issue:44466", "issue:36032|issue:43645", + "issue:43404|issue:44466", + "issue:44987|issue:44991", "issue:45084|issue:45507" ], - "prepared_review_unit_hash": "9e21ceda20d06098cbb57ef658b06518e23060997fc7ba19d0709546217658a3", + "prepared_review_unit_hash": "e255bb86267d6131dedc9646957504987dc0c2696dc169b2b386156846c5cbbf", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37891, - "estimated_input_tokens": 9473, - "estimated_eval_tokens": 19202 + "serialized_chars": 37846, + "estimated_input_tokens": 9462, + "estimated_eval_tokens": 19180 }, "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 regression reports, with one clear duplicate-like GGUF loading pair. The most representative issue is the generic v5 model-loading regression around `base_model_prefix`/core loading behavior.", - "confidence": 0.69, - "canonical_issue_reason": "Issue 43611 is the broadest, most central regression report in the set: a Transformers v5 model-loading failure affecting the core loading path, which matches the overall cluster theme better than the more model-specific reports.", + "summary": "Highly heterogeneous issue-only cluster with many unrelated Transformers regressions and model-specific bugs. None of the proposed soft edges look like true duplicates; they mostly share broad themes (v5 loading, Whisper, Qwen, tokenizer/runtime issues) without the same concrete failure mode.", + "confidence": 0.88, + "canonical_issue_reason": "No single canonical issue: the set spans unrelated bugs across training, tokenizers, multimodal loading, serialization, and version regressions.", "canonical_pr_reason": null, - "best_issue_reason": "43611 is the best single issue to anchor this cluster because it is generic, high-signal, and closest to the shared v5 loading-regression pattern seen across several nearby issues.", + "best_issue_reason": "issue:43611 is the broadest representative of the Transformers v5 model-loading regression reports, so it is the closest anchor despite the cluster not being a true duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43596", - "right": "issue:44683", - "accept": false, - "reason": "Different failures in different stacks: DeepSpeed ZeRO-3 initialization vs compiled flex_attention on torch>=2.9." - }, - { - "left": "issue:40990", - "right": "issue:42947", + "left": "issue:43856", + "right": "issue:44756", "accept": false, - "reason": "Unrelated bugs: perplexity on a specific model/dataset vs gradient-checkpointing behavior with PEFT LoRA." + "reason": "Different problems: Qwen3 MoE memory usage during training vs an mmap/OOM workaround for Strix Halo." }, { - "left": "issue:43425", - "right": "issue:45685", + "left": "issue:44589", + "right": "issue:45084", "accept": false, - "reason": "No shared code path; one is a Torch version incompatibility, the other is an MPS-specific unsupported op in a MoE path." + "reason": "Unrelated failures: missing Float8 storage object vs a template-compilation error." }, { - "left": "issue:39404", - "right": "issue:44671", + "left": "issue:43596", + "right": "issue:44683", "accept": false, - "reason": "Different model families and symptoms: Whisper pipeline return_language vs CamemBERT masked-LM predictions." + "reason": "Different code paths: DeepSpeed ZeRO3/BERT init error vs compiled flex_attention on newer torch." }, { - "left": "issue:43994", - "right": "issue:44079", + "left": "issue:40990", + "right": "issue:42947", "accept": false, - "reason": "Both mention incorrect outputs, but the concrete bugs differ and there is no clear shared code-path evidence." + "reason": "Different symptoms and subsystems: high perplexity on gpt-oss vs gradient checkpointing with PEFT LoRA." }, { - "left": "issue:44671", - "right": "issue:44991", + "left": "issue:43425", + "right": "issue:45685", "accept": false, - "reason": "Separate tokenizer/model loading issues for different models; no indication they share the same underlying regression." + "reason": "Torch version compatibility issue vs an MPS MoE histogram kernel limitation; not the same bug." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "save/from_pretrained extra_state handling is unrelated to `from_config` unwrapping of `text_config` in composite configs." + "reason": "save/from_pretrained extra_state handling is unrelated to unwrapping text_config in composite configs." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "T5 tokenizer special-token method conflict is unrelated to GGUF loading failures." + "reason": "Tokenizer special-token method conflict vs gguf_file loading regression; different failure modes." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error vs Whisper pipeline return_language regression." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-loading problems, but for different models and different failure mechanisms." + "reason": "Tokenizer loading failure for EMBEDDIA/est-roberta vs a Mistral regex patch crash; not the same defect." }, { - "left": "issue:32090", - "right": "issue:39404", + "left": "issue:43957", + "right": "issue:44164", "accept": false, - "reason": "Trainer GPU broadcast TypeError is unrelated to Whisper pipeline return_language behavior." + "reason": "Meta-device loading breakage is different from extra_state serialization handling." }, { - "left": "issue:44164", - "right": "issue:44479", + "left": "issue:43582", + "right": "issue:43632", "accept": false, - "reason": "Save/load extra_state handling and Qwen VL video input regression are unrelated." + "reason": "Apple Silicon allocator warmup TypeError vs the _is_hf_initialized flag regression." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Different model loaders and different root causes; only a broad tokenizer-loading theme is shared." + "reason": "Two different Mistral/tokenizer loading issues with different root causes." }, { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "Perplexity quality issue vs runtime post-processor update bug; not the same bug or fix." - }, - { - "left": "issue:43582", - "right": "issue:43632", - "accept": false, - "reason": "Apple Silicon allocator TypeError and `_is_hf_initialized` regression are unrelated." + "reason": "Perplexity regression on a model/data setup vs runtime special-token post-processor updates." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Missing tied-weights key attribute in Molmo is unrelated to a BERT model loading failure." + "reason": "Missing all_tied_weights_keys on Molmo is unrelated to loading sleng-bert." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are v5 regressions, but one is model-loading/base_model_prefix and the other is non-persistent buffer initialization; different bugs." + "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer serialization; different bugs." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Different model-loading regressions affecting different checkpoints and failure points." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "The titles suggest the same audio area, but the concrete failures are not clearly the same code-path problem." + "reason": "Unsupported center argument in LasrFeatureExtractor is not the same as _torch_extract_fbank_features behavior." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Image preprocessor dtype mismatch is unrelated to a template-node compilation error." - }, - { - "left": "issue:44811", - "right": "issue:45584", - "accept": false, - "reason": "Both involve Whisper and special tokens, but one is batch_decode skipping tokens and the other is empty-transcription generation after alignment; too different to merge." - }, - { - "left": "issue:44488", - "right": "issue:44987", - "accept": false, - "reason": "Different model-loading failures with no shared concrete regression evidence." + "reason": "Mistral image-preprocessor dtype mismatch vs a template-node compilation error." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Trainer token-count tracking is unrelated to multi-image IterableDataset fine-tuning failure." + "reason": "Trainer token-count tracking bug vs Qwen2.5-VL iterable dataset with multiple images." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "AutoConfig trust_remote_code field loss is unrelated to PagedAttentionCache crashing on Qwen3.5 linear_attention." + "reason": "trust_remote_code missing fields in AutoConfig is unrelated to PagedAttentionCache linear_attention failures." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Whisper batch_decode skip_special_tokens handling is a different issue from empty-transcription generation after align_special_tokens." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "A non-persistent buffer initialization regression is not the same as custom model initialization in Jupyter notebooks." + "reason": "Both involve v5 custom-model behavior, but one is non-persistent buffer serialization and the other is custom model initialization; not the same concrete bug." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer add_special_tokens conflict is unrelated to Qwen3 MoE FineGrainedFP8Config failure." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Molmo tied-weights attribute error and tokenizer loading for EMBEDDIA/est-roberta are unrelated." + "reason": "Molmo tied-weights attribute error vs tokenizer loading failure for a specific model." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Different model-specific loading failures with no clear shared implementation bug." - }, - { - "left": "issue:36032", - "right": "issue:42915", - "accept": false, - "reason": "Tokenizer method-name conflict is unrelated to Qwen3Moe FineGrainedFP8Config failure." + "reason": "ScandiBERT loading failure and physical-intelligence/fast loading failure are separate model-specific regressions." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "TorchDynamo grouped_mm tracing error is unrelated to template-node compilation." + "reason": "Mixtral torch-dynamo tracing error at grouped_mm is unrelated to a template compilation error." }, { "left": "issue:43452", "right": "issue:43482", - "accept": true, - "reason": "Both are GGUF loading regressions in Transformers v5 affecting `from_pretrained`; they plausibly come from the same GGUF loader-path bug." + "accept": false, + "reason": "gguf_file breaking from_pretrained calls is different from Qwen2.5-GGUF failing under transformers v5." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "Different issues in different subsystems: FineGrainedFP8Config vs PagedAttentionCache linear_attention crash." + "reason": "Qwen3 MoE FP8 training issue vs PagedAttentionCache linear_attention crash; different concrete problems." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "GPU+CPU offload dequantized-model loading and GGUF loading in v5 are distinct failures." + "reason": "device_map=auto dequantized offload failure is unrelated to Qwen2.5-GGUF loading under v5." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Qwen hidden-states behavior and Qwen2.5-VL rope-index scaling are different bugs." + "reason": "Qwen3.5 output_hidden_states bug is separate from Qwen2.5-VL rope-index scaling." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method conflict and notebook custom-model initialization are different regressions." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both mention tied weights, but one is a Mistral3 model definition bug and the other is device-dependent serialization; not clearly the same fix." + "reason": "Untied lm_head in Mistral3 is not the same as device-dependent inconsistent serialization of lm_head.weight." }, { - "left": "issue:36032", - "right": "issue:43645", + "left": "issue:44987", + "right": "issue:44991", "accept": false, - "reason": "Tokenizer special-token conflict is unrelated to custom model initialization in notebooks." + "reason": "Two different model-loading failures with different affected checkpoints and failure points." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Template compilation failure and GraniteMoEHybrid invalid method call are unrelated." + "reason": "Template-node compilation error is unrelated to GraniteMoEHybrid calling an invalid method." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The cluster read is mostly grounded in the packet: the chosen anchor issue 43611 is a reasonable broad v5 loading-regression representative, and the rejected soft pairs are generally conservative and tied to the issue titles. The only accepted GGUF pair is phrased as plausible rather than certain, which keeps it within bounds. No mergeability concerns apply here because these are issues, not PR pairs." - }, - "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 }, @@ -2642,8 +2623,10 @@ "cluster_id": "cluster-15354-384", "nodes": [ "issue:36032", + "issue:40444", "issue:40990", "issue:41553", + "issue:41720", "issue:42175", "issue:42915", "issue:43262", @@ -2652,7 +2635,6 @@ "issue:43322", "issue:43377", "issue:43404", - "issue:43425", "issue:43531", "issue:43540", "issue:43582", @@ -2661,11 +2643,11 @@ "issue:43645", "issue:43646", "issue:43742", - "issue:43818", "issue:43819", "issue:43856", "issue:43994", "issue:44062", + "issue:44079", "issue:44220", "issue:44265", "issue:44291", @@ -2680,264 +2662,261 @@ "issue:44811", "issue:44877", "issue:44987", - "issue:44991", "issue:45081", "issue:45084", "issue:45127", "issue:45216", "issue:45325", "issue:45357", - "issue:45542" + "issue:45542", + "issue:45663" ], "soft_pairs": [ "issue:43262|issue:43377", "issue:43646|issue:44534", "issue:43994|issue:44610", - "issue:44987|issue:44991", + "issue:43645|issue:44534", "issue:44403|issue:45081", "issue:43322|issue:43540", "issue:43377|issue:43819", "issue:44220|issue:44683", - "issue:45216|issue:45357", "issue:44451|issue:45081", + "issue:45216|issue:45357", "issue:42915|issue:45084", - "issue:43645|issue:44534", "issue:44265|issue:44610", "issue:44987|issue:45081", "issue:43404|issue:45127", - "issue:41553|issue:43582", "issue:43632|issue:44756", - "issue:43296|issue:43531", + "issue:41553|issue:43582", "issue:44062|issue:44987", "issue:42915|issue:44291", + "issue:44479|issue:45325", + "issue:43296|issue:43531", "issue:36032|issue:43742", "issue:43531|issue:44877", "issue:43611|issue:44534", "issue:42915|issue:44589", + "issue:44079|issue:45663", + "issue:40444|issue:45325", "issue:40990|issue:44811", - "issue:44479|issue:45325", "issue:43317|issue:43856", "issue:42175|issue:45542", - "issue:43425|issue:44403", - "issue:43818|issue:45357" + "issue:36032|issue:41720" ], - "prepared_review_unit_hash": "4e64e49d7f90b6fe2920451b49ddb20f97a5039dd2df1390722f78b44cef57d1", + "prepared_review_unit_hash": "bd83c715d17aa61f4c0a0468d316c3010607f05318b731b727f1f4f9e6a1c396", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36043, - "estimated_input_tokens": 9011, - "estimated_eval_tokens": 18278 + "serialized_chars": 36679, + "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 mostly unrelated issue reports with broad Transformers/v5 overlap. The Qwen3.5 save_pretrained pair is the only close match, but it should be treated as likely related rather than definitively the same bug from titles alone.", - "confidence": 0.69, - "canonical_issue_reason": "issue:45216 is the broadest Qwen3.5 save_pretrained regression report and the best anchor for that mini-cluster, but 45357 should only be considered likely related without stronger body-level evidence.", + "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer-loading, multimodal, quantization, device-mapping, and config-regression reports. None of the soft pairs look like the same concrete bug or change, so no duplicates should be merged here.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue emerges; the items cover many unrelated failure modes across different models and subsystems.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45216 is the best issue anchor because it names the core Qwen3.5 save_pretrained regression and is broader than the later, more specific report.", + "best_issue_reason": "No issue is a strong global representative because the cluster is thematically fragmented rather than one bug family.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Different audio bugs: sampling-rate defaulting in apply_chat_template vs batched/single mismatch from missing padding-mask support." + "reason": "Audio chat-template sampling-rate defaulting is unrelated to MIMI batched-vs-single padding-mask behavior." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Different v5 regressions: custom model initialization vs non-persistent buffers being filled with junk." + "reason": "Custom model initialization breakage is a different issue from non-persistent buffer corruption." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different models and failure modes: SigLIP2 bad outputs vs OmDet-Turbo processor/model input-size mismatch." + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo input-size mismatch are separate model-specific bugs." }, { - "left": "issue:44987", - "right": "issue:44991", + "left": "issue:43645", + "right": "issue:44534", "accept": false, - "reason": "Different affected models and code paths: loading physical-intelligence/fast vs tokenizer loading for EMBEDDIA/est-roberta." + "reason": "Jupyter/custom-model init problems do not match the non-persistent buffer junk-data regression." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Unrelated problems: generic loading noise vs a Mistral tokenizer regex patch crash." + "reason": "Loading noise vs tokenizer patch crash are distinct symptoms and code paths." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Different multimodal models and bugs: Llava Next segfault vs Qwen3OmniMoe video-processing ValueError." + "reason": "Llava Next loading segfault and Qwen3OmniMoe video ValueError are different multimodal failures." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Both are audio/model correctness issues, but they affect different models and distinct mechanisms: padding masks vs missing STE in DAC." + "reason": "MIMI padding-mask mismatch is unrelated to DAC forward/from_latents STE behavior." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Different subsystems: audio fbank extraction vs compiled flex_attention on newer torch." - }, - { - "left": "issue:45216", - "right": "issue:45357", - "accept": false, - "reason": "Likely related Qwen3.5 save_pretrained regressions, but the titles alone do not prove the same root cause or that 45357 is a direct manifestation of 45216." + "reason": "Feature-extraction issues in audio helpers are not the same as compiled flex_attention failures." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "Different models and failures: ScandiBERT load failure vs Mistral regex-patching crash." + "reason": "ScandiBERT loading failure and Mistral regex patch crash are separate tokenizer/model-loading bugs." }, { - "left": "issue:42915", - "right": "issue:45084", + "left": "issue:45216", + "right": "issue:45357", "accept": false, - "reason": "Different bug classes: FineGrainedFP8Config load failure vs template-node compilation TypeError." + "reason": "Both are Qwen3.5 regressions, but one is checkpoint saving correctness and the other is visual-encoder key naming; not the same fix." }, { - "left": "issue:43645", - "right": "issue:44534", + "left": "issue:42915", + "right": "issue:45084", "accept": false, - "reason": "Different regressions: notebook/custom-model initialization issues vs corrupted non-persistent buffers." + "reason": "Qwen3Moe FineGrainedFP8Config failure is unrelated to template compilation errors." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "Different code paths: torch.export with torch_compilable_check vs processor output size mismatch." + "reason": "torch.export with torch_compilable_check and OmDet-Turbo processor size mismatch are unrelated." }, { "left": "issue:44987", "right": "issue:45081", "accept": false, - "reason": "Different affected components: model loading for physical-intelligence/fast vs Mistral tokenizer regex patching." + "reason": "Loading physical-intelligence/fast and Mistral tokenizer backend access are different loading paths." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Related to weight tying in broad terms, but one is Mistral3 lm_head tying and the other is LoRA merge with extended vocab." - }, - { - "left": "issue:41553", - "right": "issue:43582", - "accept": false, - "reason": "Different domains: AutoTokenizer error messaging for Voxtral vs Apple Silicon warmup TypeError." + "reason": "Untied lm_head weights in Mistral3 are not the same bug as LoRA merge collapse with extended vocab." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "Different regressions: _is_hf_initialized handling vs mmap OOM behavior on Strix Halo." + "reason": "_is_hf_initialized flag regression and Strix Halo mmap OOM are unrelated runtime issues." }, { - "left": "issue:43296", - "right": "issue:43531", + "left": "issue:41553", + "right": "issue:43582", "accept": false, - "reason": "Different models and bugs: PaddleOCR-VL load failure in vLLM vs Qwen3-MoE sliding_window issue." + "reason": "Bad tokenizer error messaging for Voxtral is unrelated to Apple Silicon TypeError in caching_allocator_warmup." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "Tokenizer AddedToken keyword duplication is unrelated to loading physical-intelligence/fast." + "reason": "AddedToken special-argument duplication and physical-intelligence/fast loading failure are different bugs." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Different failures in different paths: Qwen3Moe FP8 config vs unexpected init_empty_weights argument." + "reason": "Qwen3Moe FP8 configuration failure is not the same as init_empty_weights/_is_hf_initialized argument breakage." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Video-input regression in Qwen3.5/Qwen2.5-VL is different from still-image rope-index scaling." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window issue are unrelated model-loading versus attention-window bugs." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "Different model/tokenizer failures: T5Tokenizer special-token conflict vs MobileLLM key error." + "reason": "T5Tokenizer add_special_tokens conflict and MobileLLM key error are separate tokenizer/model init problems." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Different areas and symptoms: Qwen3-MoE sliding-window behavior vs granite_speech strict-config loading." + "reason": "Qwen3-MoE sliding_window behavior and granite_speech strict-config loading are not the same underlying issue." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "Different v5 regressions: base_model_prefix loading breakage vs junk values in non-persistent buffers." + "reason": "base_model_prefix loading regression and junk non-persistent buffers are different Transformers v5 breakages." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "Different failures in different paths: Qwen3Moe FP8 config vs missing Float8 storage type." + "reason": "Qwen3Moe FP8 config failure is unrelated to missing Float8 storage handling." }, { - "left": "issue:40990", - "right": "issue:44811", + "left": "issue:44079", + "right": "issue:45663", "accept": false, - "reason": "Unrelated issues: perplexity regression on gpt-oss-20b vs Whisper batch_decode special-token handling." + "reason": "ModelOutput key assignment bug and Gemma4 shared_kv_states FSDP2 KeyError are unrelated internal state issues." }, { - "left": "issue:44479", + "left": "issue:40444", "right": "issue:45325", "accept": false, - "reason": "Both involve video/VL behavior, but the affected models and concrete bugs differ: video-input regression vs rope-index scaling." + "reason": "IterableDataset multi-image finetuning failure is not the same as Qwen2.5-VL rope-index scaling." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and Whisper batch_decode skip_special_tokens are unrelated evaluation/decoding bugs." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Different problems: device_map/offload loading failure vs Qwen3 MoE training memory usage." + "reason": "device_map auto dequantized offload failure and Qwen3 MoE memory usage are different memory-management problems." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Related TensorFlow packaging themes, but these are different packaging/backend failures and not the same bug." - }, - { - "left": "issue:43425", - "right": "issue:44403", - "accept": false, - "reason": "Torch version incompatibility is unrelated to generic transformer-loading noise." + "reason": "pip extra dependency resolution and tensorboard-without-TF backend error are not the same packaging bug." }, { - "left": "issue:43818", - "right": "issue:45357", + "left": "issue:36032", + "right": "issue:41720", "accept": false, - "reason": "Different models and code paths: Video-LLaVA weight/temporal-attention issues vs Qwen3.5 save_pretrained serialization." + "reason": "T5 tokenizer loading conflict and Qwen3 auto-device-map cudaErrorAssert are unrelated failures." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The cluster summary is mostly grounded, but the accepted 45216\u219445357 linkage is stated too strongly for the packet. The titles support that both are Qwen3.5 save_pretrained regressions, but not that they are definitively the same underlying bug or that 45357 is proven to be a more specific manifestation of 45216. Please soften that verdict to something like 'likely related' unless you have body-level evidence." - }, - "evaluator_used": true, - "retried": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -2957,7 +2936,6 @@ "nodes": [ "issue:33666", "issue:35707", - "issue:36032", "issue:40444", "issue:41720", "issue:42175", @@ -2965,6 +2943,7 @@ "issue:43296", "issue:43377", "issue:43421", + "issue:43425", "issue:43454", "issue:43482", "issue:43525", @@ -2973,7 +2952,9 @@ "issue:43632", "issue:43644", "issue:43646", + "issue:43749", "issue:43792", + "issue:43818", "issue:43828", "issue:43881", "issue:43957", @@ -2981,297 +2962,273 @@ "issue:44079", "issue:44265", "issue:44291", - "issue:44387", "issue:44403", "issue:44466", "issue:44488", "issue:44534", "issue:44568", "issue:44589", - "issue:44617", "issue:44683", "issue:44829", - "issue:44877", "issue:44898", "issue:44991", "issue:45020", "issue:45072", "issue:45081", "issue:45325", + "issue:45357", "issue:45381", "issue:45479", "issue:45491", - "issue:45538", - "issue:45663" + "issue:45538" ], "soft_pairs": [ + "issue:43425|issue:44403", + "issue:43632|issue:44534", + "issue:43818|issue:45357", "issue:43421|issue:44568", - "issue:44079|issue:45663", "issue:44683|issue:45020", - "issue:36032|issue:41720", - "issue:43632|issue:44534", "issue:43792|issue:44589", "issue:44403|issue:44991", "issue:43482|issue:45538", "issue:33666|issue:40444", "issue:44829|issue:45479", - "issue:40444|issue:45325", + "issue:44079|issue:44534", + "issue:35707|issue:40444", "issue:43482|issue:43525", "issue:44291|issue:44589", "issue:43377|issue:45491", + "issue:43749|issue:44079", + "issue:43644|issue:44079", "issue:41720|issue:45081", "issue:44265|issue:45072", - "issue:42175|issue:43296", + "issue:43531|issue:44291", + "issue:40444|issue:45381", "issue:43632|issue:45072", + "issue:42175|issue:43296", + "issue:45325|issue:45381", "issue:43611|issue:43646", - "issue:44079|issue:44534", - "issue:43531|issue:44291", "issue:43994|issue:44898", "issue:43454|issue:44466", "issue:43881|issue:44488", "issue:43828|issue:44683", "issue:42915|issue:43482", - "issue:44079|issue:45072", - "issue:43792|issue:43957", - "issue:43644|issue:44079", - "issue:40444|issue:45381", - "issue:45325|issue:45381", - "issue:44387|issue:44617", - "issue:35707|issue:40444", - "issue:44291|issue:44877" + "issue:43792|issue:43957" ], - "prepared_review_unit_hash": "d9e3f7ba65062e2ca027a88be24c1586969e49e92cc5f3e2ac8ee32fa5d53156", + "prepared_review_unit_hash": "d76ec877ff01877395ffcef9444d017fb713fb3258ce7dad8ab225ab28460442", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 38070, - "estimated_input_tokens": 9518, - "estimated_eval_tokens": 19292 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36533, + "estimated_input_tokens": 9134, + "estimated_eval_tokens": 18524 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of separate Transformers v5 / model-loading regressions, not one duplicate set. The closest center is a broad initialization/loading regression, but most pairs are only thematically similar.", - "confidence": 0.38, - "canonical_issue_reason": "issue:43632 is the strongest concrete anchor for the cluster: a core Transformers v5 initialization regression (`_is_hf_initialized`) that fits the many nearby model-loading breakages better than the narrower, model-specific reports.", + "summary": "This cluster is mostly a loose collection of transformers v5 regressions and model-specific loading/training bugs, not a tight duplicate set. A few items are thematically adjacent, but none of the soft-edge pairs look like the same concrete defect.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43632 is the best concrete representative of the recurring transformers v5 model-initialization/loading regressions in this set, and it is narrower/more actionable than the broad umbrella reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the best cluster representative overall because it is the broadest umbrella for the shared theme of recent Transformers versions breaking model loading, especially for remote-code models.", + "best_issue_reason": "issue:45020 is the broadest open umbrella for recent transformers-version breakages, but it is too generic to be a true duplicate canonical; if forced to pick one issue as the cluster's overall anchor, it best reflects the cross-cutting regression theme.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43421", - "right": "issue:44568", + "left": "issue:43425", + "right": "issue:44403", "accept": false, - "reason": "Both are tokenizer/special-token regressions, but they describe different failure modes and code paths: runtime post-processor updates vs BOS/EOS insertion." + "reason": "One is a Torch 2.10 compatibility complaint; the other is a loading-noise report. Different bugs and different failure modes." }, { - "left": "issue:44079", - "right": "issue:45663", + "left": "issue:43632", + "right": "issue:44534", "accept": false, - "reason": "One is a ModelOutput key assignment bug; the other is a Gemma-4 FSDP2 attention state KeyError. They are unrelated." + "reason": "Both involve transformers v5 initialization behavior, but one is about `_is_hf_initialized` and the other about non-persistent buffers; these are distinct code-path regressions." }, { - "left": "issue:44683", - "right": "issue:45020", + "left": "issue:43818", + "right": "issue:45357", "accept": false, - "reason": "torch >= 2.9 flex_attention compilation failure is a different problem from remote_code model loading regressions." + "reason": "Different models and different symptoms: Video-LLaVA temporal attention/weight-sharing vs Qwen3.5 visual encoder key serialization." }, { - "left": "issue:36032", - "right": "issue:41720", + "left": "issue:43421", + "right": "issue:44568", "accept": false, - "reason": "Tokenizer method-name conflict and Qwen3 auto device mapping CUDA assert are unrelated." + "reason": "Tokenizer post-processor update logic at runtime is not the same as missing BOS/EOS insertion for a specific tokenizer." }, { - "left": "issue:43632", - "right": "issue:44534", + "left": "issue:44683", + "right": "issue:45020", "accept": false, - "reason": "Both are v5 regressions, but one concerns `_is_hf_initialized` during init and the other concerns non-persistent buffer contents; different bugs." + "reason": "A torch>=2.9 flex_attention compile failure is unrelated to the broad remote_code loading regressions." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper loading failure and missing Float8 storage are distinct load/runtime issues." + "reason": "Different runtime errors on different models; one is a Whisper load failure, the other is a missing Float8 storage type." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "Generic load noise and a specific tokenizer load failure for EMBEDDIA/est-roberta are not the same bug." + "reason": "Generic loading-noise issue vs a specific tokenizer-loading failure for EMBEDDIA/est-roberta." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "GGUF loading under v5 and CLIPTokenizer max length behavior are unrelated." + "reason": "GGUF loading failure for Qwen2.5 is unrelated to CLIPTokenizer's model_max_length sentinel value." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both involve Qwen VL training, but one is multi-GPU training and the other is iterable-dataset/multi-image prompt handling; not the same issue." + "reason": "Both are Qwen VL training issues, but one is multi-GPU training and the other is IterableDataset multi-image finetuning; not the same bug." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "Degenerate training with flash_attention_3 is a different defect from the single-label-classification zero-loss bug." + "reason": "Both mention degenerate classification behavior, but the root causes differ: flash_attention_3 training pathology vs single-label/num_labels=1 loss handling." }, { - "left": "issue:40444", - "right": "issue:45325", + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer serialization are unrelated." + }, + { + "left": "issue:35707", + "right": "issue:40444", "accept": false, - "reason": "Both are Qwen2.5-VL problems, but they affect different paths: dataset/multi-image finetuning vs still-image temporal position-id scaling." + "reason": "Progressive generation with inputs_embeds/past_key_values is a different generation path than VL finetuning with multiple images." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and missing pad_token_id on Llama4Config are unrelated loading errors." + "reason": "Different loading errors for different models: GGUF/transformers v5 vs missing pad_token_id on Llama4Config." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Unexpected `_is_hf_initialized` during init-empty-weights is not the same as the Float8 storage lookup failure." + "reason": "init_empty_weights/_is_hf_initialized argument handling is distinct from the Float8 storage lookup failure." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Missing padding-mask support in MIMI encoder and Gemma3 sliding-window NaNs are separate batching/attention bugs." + "reason": "Both are batching-related, but MIMI padding masks and Gemma3 sliding-window/all-padding NaNs are separate model-specific bugs." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "FSDP CPU RAM efficient loading is unrelated to ModelOutput key bookkeeping." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer serialization junk is not the same defect as ModelOutput key assignment." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Qwen3 device-mapping CUDA asserts and a Mistral tokenizer regex crash are unrelated." + "reason": "CUDA assert on Qwen3 auto device mapping is unrelated to tokenizer patching crashing on missing backend_tokenizer." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export/torch_compilable_check failure is distinct from dtype mismatches in SwitchTransformers/TimmWrapperModel." + "reason": "torch.export with torch_compilable_check and bfloat16 dtype mismatches are different failure modes." }, { - "left": "issue:42175", - "right": "issue:43296", + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is not the same as init_empty_weights argument incompatibility." + }, + { + "left": "issue:40444", + "right": "issue:45381", "accept": false, - "reason": "Packaging/backend inclusion issue and PaddleOCR-VL loading in vLLM are unrelated." + "reason": "Multi-image finetuning failure and Qwen2.5-VL video position_ids bug are separate VL preprocessing issues." }, { "left": "issue:43632", "right": "issue:45072", "accept": false, - "reason": "Initialization-flag regression and bf16 dtype mismatch are different code-path problems." + "reason": "_is_hf_initialized regression and bfloat16 dtype mismatches are different transformers v5 issues." }, { - "left": "issue:43611", - "right": "issue:43646", + "left": "issue:42175", + "right": "issue:43296", "accept": false, - "reason": "Both are broad v5 loading regressions, but one is about `base_model_prefix` and the other about custom model initialization; not the same concrete bug." + "reason": "Packaging/backend inclusion for pip extras is unrelated to PaddleOCR-VL loading in vLLM." }, { - "left": "issue:44079", - "right": "issue:44534", + "left": "issue:45325", + "right": "issue:45381", "accept": false, - "reason": "ModelOutput key handling and junk-filled non-persistent buffers are unrelated." + "reason": "Both are Qwen2.5-VL position-id bugs, but one is still-image temporal scaling and the other is video vision_position_ids; not the same bug." }, { - "left": "issue:43531", - "right": "issue:44291", + "left": "issue:43611", + "right": "issue:43646", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and init-empty-weights `_is_hf_initialized` are separate issues." + "reason": "Both are v5 model-loading regressions, but one is about base_model_prefix handling and the other about custom model initialization." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "SigLIP2 wrong outputs and Perceiver resolution/interpolation failure are different model-specific bugs." + "reason": "Different vision-model failures with different inputs and symptoms." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Tied lm_head weights not being set and serialization differences of lm_head.weight are related only thematically, not the same failure." + "reason": "AyaVision lm_head tying causing garbage generation is not the same as device-dependent lm_head.weight serialization inconsistency." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "glm-4v-9b loading failure and cjvt/sleng-bert tokenizer loading failure are unrelated." + "reason": "Different model-loading failures with no shared concrete code path." }, { "left": "issue:43828", "right": "issue:44683", "accept": false, - "reason": "Autocast dtype mismatch and compiled flex_attention failure are different problems." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to compiled flex_attention failures on newer torch." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "FineGrainedFP8Config failure and GGUF loading under v5 are unrelated." - }, - { - "left": "issue:44079", - "right": "issue:45072", - "accept": false, - "reason": "ModelOutput key assignment bug and bf16 dtype mismatch are not the same underlying issue." + "reason": "Qwen3Moe FineGrainedFP8Config failure and Qwen2.5-GGUF loading failure are distinct model/config issues." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "Whisper load failure and meta-device loading regression are distinct." - }, - { - "left": "issue:43644", - "right": "issue:44079", - "accept": false, - "reason": "Non-persistent buffer serialization junk and ModelOutput key assignment are different bugs." - }, - { - "left": "issue:40444", - "right": "issue:45381", - "accept": false, - "reason": "Qwen2.5-VL multi-image iterable training failure and video vision_position_ids bug affect different paths." - }, - { - "left": "issue:45325", - "right": "issue:45381", - "accept": false, - "reason": "Both are Qwen2.5-VL position-id complaints, but one is still-image temporal scaling and the other is video vision_position_ids; close, but not clearly the same concrete bug." - }, - { - "left": "issue:44387", - "right": "issue:44617", - "accept": false, - "reason": "v5 int4 memory growth/OOM and Sam3Video OOM are different resource issues." - }, - { - "left": "issue:35707", - "right": "issue:40444", - "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image iterable finetuning." - }, - { - "left": "issue:44291", - "right": "issue:44877", - "accept": false, - "reason": "Unexpected `_is_hf_initialized` argument and strict config blocking granite_speech are different loading failures." + "reason": "Whisper-large-v2 load failure and meta-device loading breakage are separate problems." } ] }, @@ -3304,19 +3261,18 @@ "issue:43262", "issue:43296", "issue:43317", + "issue:43322", "issue:43377", - "issue:43404", "issue:43454", "issue:43526", "issue:43550", "issue:43572", - "issue:43577", "issue:43606", "issue:43632", "issue:43701", - "issue:43708", "issue:43720", "issue:43746", + "issue:43749", "issue:43792", "issue:43819", "issue:43873", @@ -3324,11 +3280,13 @@ "issue:43883", "issue:44186", "issue:44265", - "issue:44423", + "issue:44291", + "issue:44387", "issue:44466", + "issue:44534", "issue:44610", + "issue:44617", "issue:44683", - "issue:44756", "issue:44877", "issue:44898", "issue:44977", @@ -3336,7 +3294,6 @@ "issue:45072", "issue:45325", "issue:45357", - "issue:45406", "issue:45563", "issue:45584", "issue:45632", @@ -3345,259 +3302,252 @@ "issue:45759" ], "soft_pairs": [ + "issue:44387|issue:44617", + "issue:43701|issue:45325", + "issue:44291|issue:44877", "issue:43526|issue:43883", - "issue:43317|issue:43873", + "issue:40990|issue:45584", + "issue:43262|issue:43746", "issue:43606|issue:44265", - "issue:43881|issue:44991", - "issue:43701|issue:45325", - "issue:44186|issue:44265", + "issue:43317|issue:43873", + "issue:34634|issue:35707", "issue:43632|issue:45563", - "issue:43262|issue:43746", + "issue:43873|issue:45701", + "issue:44186|issue:44265", + "issue:43881|issue:44991", + "issue:35707|issue:41720", "issue:43606|issue:43819", "issue:44610|issue:45325", "issue:43454|issue:43550", "issue:43377|issue:44610", "issue:43454|issue:45072", - "issue:40990|issue:45584", "issue:43572|issue:45759", "issue:43377|issue:44898", "issue:39692|issue:40444", + "issue:43749|issue:44534", "issue:44466|issue:45357", "issue:43792|issue:44683", "issue:43296|issue:44991", "issue:43296|issue:44877", "issue:43454|issue:43720", - "issue:43873|issue:45701", + "issue:41720|issue:43322", "issue:41720|issue:44977", - "issue:45632|issue:45684", - "issue:43404|issue:45357", - "issue:44423|issue:45406", - "issue:44186|issue:45072", - "issue:43577|issue:43708", - "issue:34634|issue:35707", - "issue:35707|issue:41720", - "issue:43873|issue:44756" + "issue:45632|issue:45684" ], - "prepared_review_unit_hash": "83a9dcad506494e60c7e5ecedfa7d64dda1259d04ebb768fdcd1a5015bc50775", + "prepared_review_unit_hash": "ca84ac56a22ffb4ed9b48ff6885e79d5da8949fffd6a1ab081734e031800413f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37539, - "estimated_input_tokens": 9385, - "estimated_eval_tokens": 19026 + "soft_pair_count": 31, + "serialized_chars": 36985, + "estimated_input_tokens": 9247, + "estimated_eval_tokens": 18750 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated closed issues with only a loose shared theme around loading/offloading, config, and multimodal runtime regressions. None of the soft pairs look like true duplicates, so I reject them all and center the cluster on the broad offloading/quantization report.", - "confidence": 0.71, - "canonical_issue_reason": "issue:43873 is the broadest and most central report in the small offloading/quantization subset, with additional discussion and inbound references.", + "summary": "This is a noisy cluster of mostly unrelated Transformers issues, with a loose theme around v5-era loading/init/offload regressions but no safe duplicate pairs.", + "confidence": 0.42, + "canonical_issue_reason": "issue:43632 is the broadest platform-level regression here: a core v5 `_is_hf_initialized` breakage that can affect multiple loading paths, making it a better umbrella than the model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best representative issue because it captures the widest symptom area among the related loading/offloading reports.", + "best_issue_reason": "issue:43632 is the best global representative because it captures the widest shared loading/init regression theme in the set; the other issues are mostly narrower, model-specific variants or unrelated bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both mention memory, but one is an int4 quantization regression and the other is a video-model OOM; different failure modes and code paths." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL rope/position-id scaling." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "`_is_hf_initialized` constructor incompatibility and strict config loading are separate regression surfaces." + }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "Different components and failures: BeitImageProcessorFast label reduction vs Molmo missing attribute during loading." + "reason": "A Fast image processor label-reduction bug is unrelated to a model attribute missing during tied-weight handling." }, { - "left": "issue:43317", - "right": "issue:43873", + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "Extremely high perplexity and Whisper empty-transcription generation failure are different model/runtime problems." + }, + { + "left": "issue:43262", + "right": "issue:43746", "accept": false, - "reason": "Both mention offloading/quantization, but one is a dequantized-model load failure and the other is a generic offloading complaint; not the same bug." + "reason": "Audio chat-template sample-rate defaulting is not the same bug as PEFT local checkpoint loading." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU offload device mismatch is unrelated to torch.export failures around torch_compilable_check." + "reason": "CPU offload device mismatch and torch.export/export failure via `torch_compilable_check` are distinct issues." }, { - "left": "issue:43881", - "right": "issue:44991", + "left": "issue:43317", + "right": "issue:43873", "accept": false, - "reason": "PaddleOCR-VL model loading in vLLM and tokenizer loading for est-roberta are different failure modes." + "reason": "Dequantized-model loading with auto device map is related to offloading only broadly; the concrete bug and path differ from the quantization offload report." }, { - "left": "issue:43701", - "right": "issue:45325", + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice preset handling and progressive generation with `inputs_embeds`/`past_key_values` are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:45563", "accept": false, - "reason": "Checkpoint-resume key mismatch is unrelated to Qwen2.5-VL rope-index scaling." + "reason": "Core init-flag regression is unrelated to a stale warning around `num_return_sequences`." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offloading behavior and tokenization changes are not the same bug." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "Tokenizer crashes on NER/padding are unrelated to torch.export compatibility issues." + "reason": "LayoutLMv2Tokenizer NER/padding crashes are unrelated to export-time `torch_compilable_check` failures." }, { - "left": "issue:43632", - "right": "issue:45563", + "left": "issue:43881", + "right": "issue:44991", "accept": false, - "reason": "_is_hf_initialized flag breakage is unrelated to a stale generate warning." + "reason": "A model loading failure for glm-4v-9b is not the same as tokenizer loading for EMBEDDIA/est-roberta." }, { - "left": "issue:43262", - "right": "issue:43746", + "left": "issue:35707", + "right": "issue:41720", "accept": false, - "reason": "Audio sampling-rate defaulting and PEFT adapter local-checkpoint loading are different code paths." + "reason": "Progressive generation state handling and Qwen3 device-map CUDA asserts are different bugs." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "CPU offload device mismatch and DAC.from_latents/STE mismatch are unrelated model bugs." + "reason": "CPU offload device mismatch and DAC latent reconstruction mismatch have no shared code-path." }, { "left": "issue:44610", "right": "issue:45325", "accept": false, - "reason": "Processor image size mismatch is unrelated to temporal position-id scaling in Qwen2.5-VL." + "reason": "Processor image-size mismatch and rope-index temporal scaling are unrelated model-preprocessing bugs." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "lm_head weight tying and torch.compile SDPA failures are not the same underlying issue." + "reason": "Weight tying failure and torch.compile SDPA failure are different problems." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "MIMI batching/padding-mask behavior and OmDet processor resolution mismatch are unrelated." + "reason": "MIMI batching/padding-mask output drift is unrelated to OmDet processor resolution mismatch." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "Weight-tying bugs are unrelated to dtype mismatch issues in bfloat16 inference." - }, - { - "left": "issue:40990", - "right": "issue:45584", - "accept": false, - "reason": "High perplexity on GPT-OSS and Whisper empty-transcription failure are unrelated generation problems." + "reason": "AyaVision tied-weight serialization and dtype mismatch in inference are not the same defect." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is unrelated to unwrapping text_config for Qwen multimodal configs." + "reason": "Missing `pad_token_idx` and failure to unwrap `text_config` are separate config-loading regressions." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "Missing padding-mask support in MIMI encoder is unrelated to Perceiver positional encoding interpolation." + "reason": "Batched-vs-single encoder output mismatch is not the same as Perceiver interpolation failing at non-default resolution." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 doc/example errors and Qwen2.5-VL iterable-dataset finetuning failures are different multimodal issues." + "reason": "Documentation/model-processor mismatch for SigLIP2 and Qwen2.5-VL iterable-dataset multi-image finetuning are different issues." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are separate v5 regressions." }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "lm_head.weight serialization inconsistency and incorrect visual encoder keys are both save_pretrained-related but not the same bug." + "reason": "Device-dependent lm_head serialization and incorrect visual encoder keys are different serialization bugs." }, { "left": "issue:43792", "right": "issue:44683", "accept": false, - "reason": "Whisper runtime failure and compiled flex_attention failure are unrelated." + "reason": "Whisper inference failure and compiled flex_attention failure on newer torch are unrelated." }, { "left": "issue:43296", "right": "issue:44991", "accept": false, - "reason": "PaddleOCR-VL load failure in vLLM is unrelated to tokenizer loading for est-roberta." + "reason": "PaddleOCR-VL load failure in vLLM and tokenizer loading failure for est-roberta are different artifact types and paths." }, { "left": "issue:43296", "right": "issue:44877", "accept": false, - "reason": "Strict config rejection for granite_speech is a different loading/config bug than PaddleOCR-VL model loading." + "reason": "PaddleOCR-VL model loading and strict config rejection of granite_speech are not the same underlying bug." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "lm_head weight tying is unrelated to BitNet packed-weight unpacking during accelerate loading." + "reason": "AyaVision tied weights and BitNet packed-weight unpacking during accelerate loading are distinct model-loading issues." }, { - "left": "issue:43873", - "right": "issue:45701", + "left": "issue:41720", + "right": "issue:43322", "accept": false, - "reason": "Generic offloading/quantization issues do not match a tokenization-version regression." + "reason": "Qwen3 auto device-map CUDA assert and Llava Next segmentation fault are different crashes." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Both are Qwen-family runtime bugs, but auto device mapping cuda asserts and flash-attention generation failures are different code paths." + "reason": "CUDA assert on device mapping and flash-attention generation issues for Qwen3.5 are not the same code path." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache-path collisions for trust_remote_code are unrelated to permission propagation in save_pretrained." - }, - { - "left": "issue:43404", - "right": "issue:45357", - "accept": false, - "reason": "A missing tie_weights call and incorrect visual encoder keys are different serialization/model-init bugs." - }, - { - "left": "issue:44423", - "right": "issue:45406", - "accept": false, - "reason": "Continuous-batching multimodal crash and Gemma4Processor missing _tokenizer are related to serve, but not the same concrete defect." - }, - { - "left": "issue:44186", - "right": "issue:45072", - "accept": false, - "reason": "Tokenizer padding/truncation crashes are unrelated to bfloat16 dtype mismatches." - }, - { - "left": "issue:43577", - "right": "issue:43708", - "accept": false, - "reason": "dtype propagation for BLIP2 is unrelated to Trainer resume max_steps calculation." - }, - { - "left": "issue:34634", - "right": "issue:35707", - "accept": false, - "reason": "Bark voice_preset failure and progressive generation with inputs_embeds/past_key_values are unrelated." - }, - { - "left": "issue:35707", - "right": "issue:41720", - "accept": false, - "reason": "Progressive generation state handling is unrelated to Qwen3 auto-device-map cuda asserts." - }, - { - "left": "issue:43873", - "right": "issue:44756", - "accept": false, - "reason": "Generic offloading/quantization problems are not the same as disabling mmap to avoid OOM on Strix Halo." + "reason": "Cache-path collisions for `trust_remote_code` and permission propagation in `save_pretrained` are unrelated filesystem behaviors." } ] }, @@ -3621,8 +3571,6 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:30064", - "issue:34634", "issue:36010", "issue:36331", "issue:40990", @@ -3634,309 +3582,306 @@ "issue:43317", "issue:43322", "issue:43388", - "issue:43421", + "issue:43404", "issue:43450", "issue:43526", "issue:43531", "issue:43540", "issue:43550", "issue:43572", + "issue:43577", "issue:43606", + "issue:43708", "issue:43720", - "issue:43749", "issue:43782", + "issue:43873", "issue:43881", "issue:44186", "issue:44292", "issue:44361", "issue:44387", + "issue:44423", "issue:44479", - "issue:44492", - "issue:44509", - "issue:44534", "issue:44610", "issue:44671", "issue:44743", + "issue:44756", "issue:44877", "issue:44898", "issue:44913", "issue:44918", "issue:44991", "issue:45059", + "issue:45072", "issue:45161", + "issue:45357", "issue:45406", "issue:45464", "issue:45701", "issue:45704" ], "soft_pairs": [ - "issue:43749|issue:44534", - "issue:43526|issue:43881", - "issue:44877|issue:44991", + "issue:43404|issue:45357", + "issue:40990|issue:45701", + "issue:43873|issue:44756", + "issue:44186|issue:45072", + "issue:44423|issue:45406", "issue:42617|issue:42915", + "issue:43577|issue:43708", + "issue:44387|issue:45704", + "issue:44877|issue:44991", + "issue:43526|issue:43881", + "issue:36331|issue:42617", + "issue:36331|issue:42915", "issue:43450|issue:45701", "issue:43550|issue:43720", - "issue:36331|issue:42915", "issue:44610|issue:44898", - "issue:44186|issue:44610", - "issue:44387|issue:45704", "issue:44479|issue:44918", - "issue:36331|issue:42617", + "issue:44186|issue:44610", "issue:43316|issue:44913", "issue:42915|issue:43317", - "issue:40990|issue:45701", - "issue:41720|issue:43322", - "issue:43295|issue:45701", "issue:43550|issue:43606", - "issue:41720|issue:45161", + "issue:43295|issue:45701", + "issue:41720|issue:42617", "issue:44671|issue:44743", + "issue:41720|issue:45161", "issue:43531|issue:43572", "issue:44292|issue:44479", - "issue:41720|issue:42617", "issue:36010|issue:42617", "issue:44186|issue:44361", "issue:43540|issue:43782", - "issue:45406|issue:45464", - "issue:41720|issue:44292", - "issue:43421|issue:45701", - "issue:30064|issue:34634", - "issue:44492|issue:44509", "issue:43388|issue:45059", + "issue:41720|issue:43873", "issue:42915|issue:43322", - "issue:43606|issue:43720" + "issue:45406|issue:45464", + "issue:36331|issue:41720" ], - "prepared_review_unit_hash": "8604d3d40b1a2b07cac2e63c172aeb3200af909eea5af0ad884b7905620ed748", + "prepared_review_unit_hash": "05eb5f197e9da3f9d27c98613ec833a41ef35085b41c73ab110e2b1d9457b087", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 36948, - "estimated_input_tokens": 9237, - "estimated_eval_tokens": 18730 + "serialized_chars": 37800, + "estimated_input_tokens": 9450, + "estimated_eval_tokens": 19156 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The soft pairs are only superficially similar; they span unrelated models, APIs, and failure modes. I would not merge any of these pairs as duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No clear canonical issue: this set is heterogeneous, with distinct bugs across loading, tokenization, multimodal processing, quantization, and serving paths.", + "summary": "The set is heterogeneous: most items are unrelated model-, tokenizer-, training-, or device-mapping bugs. None of the proposed soft edges look like the same underlying defect, so I would not merge any pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43873 is the broadest and most general issue in the set (offloading/quantization with multiple references), so it is the best representative if one must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue candidate; none of the items subsumes the others or represents the shared underlying defect.", + "best_issue_reason": "issue:43873 has the widest scope and strongest signal among otherwise unrelated issues, making it the best global issue representative.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43749", - "right": "issue:44534", + "left": "issue:43404", + "right": "issue:45357", "accept": false, - "reason": "FSDP CPU/RAM-efficient loading vs non-persistent buffer initialization are different code paths and symptoms." + "reason": "Both involve multimodal/model key handling, but they are different models and different bugs: lm_head tying vs incorrect saved visual encoder keys." }, { - "left": "issue:43526", - "right": "issue:43881", + "left": "issue:40990", + "right": "issue:45701", "accept": false, - "reason": "A BEiT image-processor label bug and a GLM-4V loading failure are unrelated model-specific issues." + "reason": "A perplexity regression on gpt-oss is not the same as a version-dependent tokenization change." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Both mention memory/offload, but one is quantization offloading behavior and the other is a Strix Halo mmap OOM workaround." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crashes on NER/batched padding are unrelated to dtype mismatches in bfloat16 inference." + }, + { + "left": "issue:44423", + "right": "issue:45406", "accept": false, - "reason": "Both involve loading regressions, but one is strict config handling and the other is tokenizer loading; not the same defect." + "reason": "Both are serve/multimodal crashes, but the failing code paths differ: continuous batching with string tensors vs Gemma4Processor missing _tokenizer." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "Different failures: a 3d_parallel.py runtime problem versus a Qwen3MoE FP8 config issue." + "reason": "3d_parallel.py startup failure and Qwen3Moe FP8 failure are unrelated model/runtime issues." }, { - "left": "issue:43450", - "right": "issue:45701", + "left": "issue:43577", + "right": "issue:43708", "accept": false, - "reason": "A batched video-shape bug is unrelated to a general tokenization-version regression." + "reason": "BLIP2 dtype loading behavior and Trainer checkpoint resume step accounting are unrelated." }, { - "left": "issue:43550", - "right": "issue:43720", + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Both are memory-related, but one is int4 reserved-memory growth in Transformers 5.x and the other is a T5 apex RMSNorm leak." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config loading for granite_speech and tokenizer loading for est-roberta are different failure modes in different subsystems." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "A BeitImageProcessorFast label-reduction bug is unrelated to a glm-4v-9b model loading failure." + }, + { + "left": "issue:36331", + "right": "issue:42617", "accept": false, - "reason": "These are different model/runtime bugs: torch.compile+SDPA for Bamba versus packed-weight loading for BitNet." + "reason": "CustomTrainer.compute_loss signature breakage is unrelated to 3d_parallel.py failing to run." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "Trainer API keyword mismatch and Qwen3MoE FP8 loading are distinct problems." + "reason": "A trainer API mismatch is not the same bug as Qwen3Moe failing with FineGrainedFP8Config." }, { - "left": "issue:44610", - "right": "issue:44898", + "left": "issue:43450", + "right": "issue:45701", "accept": false, - "reason": "Both mention image sizes, but they affect different models and different resize/interpolation logic." + "reason": "Video processor batched-shape regression and tokenization-version changes are unrelated." }, { - "left": "issue:44186", - "right": "issue:44610", + "left": "issue:43550", + "right": "issue:43720", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and OmDet-Turbo image-size mismatch are unrelated." + "reason": "torch.compile + SDPA failure in Bamba and packed-weight unpacking in BitNet are different code paths." }, { - "left": "issue:44387", - "right": "issue:45704", + "left": "issue:44610", + "right": "issue:44898", "accept": false, - "reason": "CUDA reserved-memory growth under int4 quantization is not the same as an apex RMSNorm memory leak." + "reason": "Both involve image size mismatches, but one is an OmDet processor/model expectation mismatch and the other is Perceiver interpolation logic." }, { "left": "issue:44479", "right": "issue:44918", "accept": false, - "reason": "One is a video-input regression in Qwen VL models; the other is embedding unpacking with TRL SFT." + "reason": "A video-input regression in Qwen VL models is unrelated to TRL SFT embedding unpacking." }, { - "left": "issue:36331", - "right": "issue:42617", + "left": "issue:44186", + "right": "issue:44610", "accept": false, - "reason": "Trainer compute_loss signature breakage is unrelated to the 3d_parallel.py failure." + "reason": "LayoutLMv2 tokenizer NER/padding crashes and OmDet input-size mismatch are different bugs." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Config API discrepancy in Gemma3TextConfig is not the same as GPTNeoX rotary_pct reload behavior." + "reason": "Gemma3TextConfig API discrepancy and GPTNeoXConfig rotary_pct reload behavior are unrelated config bugs." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "Both are loading-related, but one is FP8 Qwen3MoE and the other is dequantized model offload with device_map=auto." - }, - { - "left": "issue:40990", - "right": "issue:45701", - "accept": false, - "reason": "High perplexity on a model/dataset pair is not the same as a tokenizer behavior change." + "reason": "Qwen3Moe FP8 failure and device_map auto dequantized-model offload failure are different model-loading problems." }, { - "left": "issue:41720", - "right": "issue:43322", + "left": "issue:43550", + "right": "issue:43606", "accept": false, - "reason": "Qwen3 auto device-map assertions and Llava Next loading segfaults are different failures." + "reason": "Bamba torch.compile/SDPA failure and Bark CPU-offload device mismatch do not share the same underlying bug." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Processor/tokenizer API regression and general tokenization-version changes are not the same code path." - }, - { - "left": "issue:43550", - "right": "issue:43606", - "accept": false, - "reason": "Different models and different offload/compile issues; only the broad 'device mismatch' theme overlaps." + "reason": "A regression around processor.tokenizer/images in v4.57.5 is unrelated to general tokenization changes across versions." }, { "left": "issue:41720", - "right": "issue:45161", + "right": "issue:42617", "accept": false, - "reason": "Auto device mapping on Qwen3 is not the same as tensor-parallel GPT-OSS MoE behavior." + "reason": "Qwen3 auto device mapping cuda asserts and a 3d_parallel script failure are not the same issue." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT MLM prediction error and Qwen3.5 recurrent-state reset are unrelated." + "reason": "CamemBERT masked-LM prediction errors and Qwen3.5 recurrent-state reset in modular code are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both are model-parallelism related, but one is an A800 cudaErrorAssert during auto device mapping and the other is GPT-OSS TP 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 different bugs." + "reason": "A Qwen3-MoE sliding-window bug and a StableLmConfig pad_token_idx regression are unrelated." }, { "left": "issue:44292", "right": "issue:44479", "accept": false, - "reason": "NVFP4 run failure and Qwen VL video regression are unrelated model-specific issues." - }, - { - "left": "issue:41720", - "right": "issue:42617", - "accept": false, - "reason": "Auto device mapping assert and 3d_parallel.py runtime failure are distinct problems." + "reason": "Qwen-3-8B-NVFP4 runtime errors are unrelated to the video-input regression in Qwen VL models." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "GenerationMixin import failure is unrelated to the 3d_parallel.py issue." + "reason": "GenerationMixin import failure and 3d_parallel.py runtime failure are completely different issues." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Two different tokenizer bugs affecting different models and tasks." + "reason": "These are different tokenizer bugs affecting different tokenizers and tasks." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError and Qwen3VL weight_only load error are different code paths." + "reason": "Qwen3OmniMoe video processing and Qwen3VL weight_only loading are different failure paths." }, { - "left": "issue:45406", - "right": "issue:45464", + "left": "issue:43388", + "right": "issue:45059", "accept": false, - "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming inference failure are separate serving bugs." + "reason": "Metric gathering dropping tuple labels is unrelated to SAM3 text+box behavior." }, { "left": "issue:41720", - "right": "issue:44292", - "accept": false, - "reason": "Auto device mapping on Qwen3 and Qwen-3-8B-NVFP4 runtime failure are not the same defect." - }, - { - "left": "issue:43421", - "right": "issue:45701", - "accept": false, - "reason": "Runtime special-token post-processor updates and general tokenization-version changes overlap only loosely; not a duplicate." - }, - { - "left": "issue:30064", - "right": "issue:34634", - "accept": false, - "reason": "Void segmentation-map processing and Bark voice_preset issues are unrelated." - }, - { - "left": "issue:44492", - "right": "issue:44509", - "accept": false, - "reason": "A cache-strategy typo and stale docs for removed pipeline tasks are different documentation issues." - }, - { - "left": "issue:43388", - "right": "issue:45059", + "right": "issue:43873", "accept": false, - "reason": "Metric label truncation and SAM3 text/box behavior are unrelated." + "reason": "Both involve device/offload themes, but one is a Qwen3 cuda assert and the other is a broader quantization/offloading complaint." }, { - "left": "issue:42915", - "right": "issue:43322", + "left": "issue:45406", + "right": "issue:45464", "accept": false, - "reason": "Qwen3MoE FP8 loading and Llava Next segfaults are different model-loading failures." + "reason": "Gemma4Processor _tokenizer crash and Qwen3.5 streaming inference failure are different models and different bugs." }, { - "left": "issue:43606", - "right": "issue:43720", + "left": "issue:36331", + "right": "issue:41720", "accept": false, - "reason": "CPU offload device mismatch for Bark and packed-weight unpacking for BitNet are distinct bugs." + "reason": "Trainer compute_loss API breakage is unrelated to Qwen3 device-mapping crashes." } ] }, @@ -3960,7 +3905,8 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:33666", + "issue:30064", + "issue:33357", "issue:34567", "issue:34634", "issue:35707", @@ -3969,29 +3915,28 @@ "issue:42886", "issue:43381", "issue:43421", - "issue:43525", "issue:43531", "issue:43550", - "issue:43596", "issue:43606", "issue:43653", + "issue:43720", "issue:43746", "issue:43782", "issue:43844", "issue:43873", "issue:43927", - "issue:43994", "issue:44062", + "issue:44077", "issue:44186", - "issue:44220", "issue:44368", "issue:44451", + "issue:44492", + "issue:44509", "issue:44514", "issue:44545", "issue:44589", "issue:44610", "issue:44625", - "issue:44671", "issue:44843", "issue:44898", "issue:44918", @@ -4000,268 +3945,260 @@ "issue:44991", "issue:45042", "issue:45072", - "issue:45081", - "issue:45356", "issue:45357", "issue:45381", "issue:45479", "issue:45588", + "issue:45701", "issue:45704", "issue:45759" ], "soft_pairs": [ + "issue:43421|issue:45701", + "issue:30064|issue:34634", + "issue:33357|issue:35707", + "issue:44492|issue:44509", + "issue:41720|issue:43746", + "issue:43606|issue:43720", + "issue:43873|issue:45704", "issue:44514|issue:44545", "issue:43550|issue:43653", + "issue:44625|issue:45479", "issue:43531|issue:45759", "issue:43606|issue:43653", "issue:43421|issue:43927", + "issue:34634|issue:41720", "issue:34567|issue:35707", "issue:44368|issue:45357", - "issue:44898|issue:45072", "issue:44451|issue:44991", - "issue:43873|issue:45704", - "issue:41720|issue:43746", + "issue:44898|issue:45072", + "issue:43782|issue:45381", + "issue:44077|issue:44625", "issue:44062|issue:44451", "issue:43421|issue:43653", - "issue:42886|issue:44843", "issue:44987|issue:45042", + "issue:42886|issue:44843", "issue:45588|issue:45759", - "issue:33666|issue:35707", "issue:43531|issue:44589", "issue:43421|issue:44186", - "issue:43381|issue:43844", - "issue:39692|issue:44610", "issue:44918|issue:44936", - "issue:45081|issue:45356", - "issue:43525|issue:43531", - "issue:44625|issue:45479", - "issue:43782|issue:45381", - "issue:43994|issue:44671", - "issue:34634|issue:41720", - "issue:43381|issue:45704", - "issue:43596|issue:44220", - "issue:39692|issue:43873" + "issue:43381|issue:43844", + "issue:39692|issue:44610" ], - "prepared_review_unit_hash": "fa6969c618e3bee768fcbf33b464f80887d58bb27f95a72ed22e7c1411fa5b99", + "prepared_review_unit_hash": "9b18ae9013403b2979de2471d0adc1b13e4b3467a0d5c8372c5df6096f4c1c36", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37274, - "estimated_input_tokens": 9319, - "estimated_eval_tokens": 18894 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36272, + "estimated_input_tokens": 9068, + "estimated_eval_tokens": 18392 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous; only the two Qwen2_5_VLProcessor `apply_chat_template` reports are the same underlying bug. All other suggested pairs appear to be unrelated issues in different models/code paths.", - "confidence": 0.94, - "canonical_issue_reason": "issue:44514 is the earlier, more specific report of the Qwen2_5_VLProcessor batched `apply_chat_template` crash; issue:44545 looks like the later duplicate of the same failure.", + "summary": "Most pairs are unrelated; the only clear duplicate is the Qwen2_5_VLProcessor batched padding=False crash. No PRs are present.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44514 is the earliest report of the Qwen2_5_VLProcessor.apply_chat_template batched-input crash, and issue:44545 appears to be the same bug reported later.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44514 is the clearest representative of the only true duplicate pair in this set.", + "best_issue_reason": "issue:44514 is the best canonical issue candidate because it is the original report for the only true duplicate pair and describes the same concrete failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44514", - "right": "issue:44545", - "accept": true, - "reason": "Same component, same batched `apply_chat_template` crash when `padding=False`; titles describe the same bug." + "left": "issue:43421", + "right": "issue:45701", + "accept": false, + "reason": "Both concern tokenization-related behavior, but one is about runtime special-token updates and the other about tokenizer output changing across versions; not the same bug." }, { - "left": "issue:43550", - "right": "issue:43653", + "left": "issue:30064", + "right": "issue:34634", "accept": false, - "reason": "Different subsystems: torch.compile/SDPA model bug vs tokenizer special-token registration." + "reason": "Different subsystems and failures: image processor segmentation maps vs BarkProcessor voice_preset handling." }, { - "left": "issue:43531", - "right": "issue:45759", + "left": "issue:33357", + "right": "issue:35707", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to composite Qwen multimodal config unwrapping." + "reason": "Unrelated errors: MacOS bus error with a CLIP model vs progressive generation with inputs_embeds/past_key_values." }, { - "left": "issue:43606", - "right": "issue:43653", + "left": "issue:44492", + "right": "issue:44509", "accept": false, - "reason": "CPU offload device mismatch for Bark is unrelated to BigBirdTokenizer special-token decoding." + "reason": "Both are documentation/style-related, but they target different doc issues and are not the same change." }, { - "left": "issue:43421", - "right": "issue:43927", + "left": "issue:41720", + "right": "issue:43746", "accept": false, - "reason": "Both touch tokenizer state, but one is runtime post-processor refresh and the other is config persistence/load; not the same bug." + "reason": "Different problems: Qwen3 auto device mapping CUDA assert vs PEFT adapter loading from local checkpoints." }, { - "left": "issue:34567", - "right": "issue:35707", + "left": "issue:43606", + "right": "issue:43720", "accept": false, - "reason": "Trainer token counting and progressive generation with embeddings/past key values are unrelated failures." + "reason": "Both are CI/model-loading failures, but one is CPU offload device mismatch and the other is unpacking packed BitNet weights during accelerate loading." }, { - "left": "issue:44368", - "right": "issue:45357", + "left": "issue:43873", + "right": "issue:45704", "accept": false, - "reason": "A warning about `tie_word_embeddings` is not the same as incorrect visual-encoder keys being saved." + "reason": "Quantization/offloading vs T5 apex RMSNorm memory leak; different code paths and fixes." }, { - "left": "issue:44898", - "right": "issue:45072", - "accept": false, - "reason": "Perceiver resolution handling and bfloat16 dtype mismatch in different models are unrelated." + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same title and same concrete failure: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False." }, { - "left": "issue:44451", - "right": "issue:44991", + "left": "issue:43550", + "right": "issue:43653", "accept": false, - "reason": "Both are tokenizer-loading regressions, but they affect different models and likely different code paths." + "reason": "Different model-specific bugs: torch.compile/SDPA for Bamba vs BigBirdTokenizer special-token registration." }, { - "left": "issue:43873", - "right": "issue:45704", + "left": "issue:44625", + "right": "issue:45479", "accept": false, - "reason": "Quantization/offloading behavior is unrelated to a T5 apex RMSNorm memory leak." + "reason": "One is Qwen3.5 config propagation for num_labels; the other is a sequence-classification loss bug across models." }, { - "left": "issue:41720", - "right": "issue:43746", + "left": "issue:43531", + "right": "issue:45759", "accept": false, - "reason": "Qwen3 auto device-map CUDA assert is unrelated to GraniteSpeech PEFT checkpoint loading." + "reason": "Both involve Qwen3.x configs, but one is a sliding_window issue in MoE models and the other is from_config unwrapping text_config; not the same bug." }, { - "left": "issue:44062", - "right": "issue:44451", + "left": "issue:43606", + "right": "issue:43653", "accept": false, - "reason": "AddedToken construction error is unrelated to loading ScandiBERT." + "reason": "Distinct issues: CPU offload device mismatch vs tokenizer mask token registration/empty decode output." }, { "left": "issue:43421", - "right": "issue:43653", + "right": "issue:43927", "accept": false, - "reason": "Special-token post-processing updates are different from BigBirdTokenizer mask-token registration/decoding." + "reason": "Runtime special-token/post-processor behavior is different from DiaConfig losing custom token IDs on save/load." }, { - "left": "issue:42886", - "right": "issue:44843", + "left": "issue:34634", + "right": "issue:41720", "accept": false, - "reason": "Both involve offline mode, but one is generic cache loading while the other is a specific `_patch_mistral_regex` hub call." + "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping are unrelated." }, { - "left": "issue:44987", - "right": "issue:45042", + "left": "issue:34567", + "right": "issue:35707", "accept": false, - "reason": "Loading a specific model in 5.1.0 is unrelated to PIL image processors incorrectly requiring torchvision." + "reason": "TrainerState token counting and progressive generation with cache are unrelated bugs." }, { - "left": "issue:45588", - "right": "issue:45759", + "left": "issue:44368", + "right": "issue:45357", "accept": false, - "reason": "Flash-attention sinkless model crash is unrelated to Qwen multimodal config unwrapping." + "reason": "Different Qwen3.5 problems: a training warning about tie_word_embeddings vs incorrect visual encoder keys on save_pretrained." }, { - "left": "issue:33666", - "right": "issue:35707", + "left": "issue:44451", + "right": "issue:44991", "accept": false, - "reason": "Qwen2-VL multi-GPU training and progressive generation are different problems." + "reason": "Both are tokenizer-loading complaints, but they concern different models and likely different root causes." }, { - "left": "issue:43531", - "right": "issue:44589", + "left": "issue:44898", + "right": "issue:45072", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to the Float8 storage lookup error." + "reason": "Perceiver image resolution handling and dtype mismatch in SwitchTransformers/TimmWrapperModel are different failures." }, { - "left": "issue:43421", - "right": "issue:44186", + "left": "issue:43782", + "right": "issue:45381", "accept": false, - "reason": "Runtime special-token updates are not the same as LayoutLMv2Tokenizer NER/padding crashes." + "reason": "Qwen3VL weight_only loading and qwen2.5-vl video vision_position_ids are unrelated multimodal bugs." }, { - "left": "issue:43381", - "right": "issue:43844", + "left": "issue:44077", + "right": "issue:44625", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction is unrelated to abnormal gradient growth under ZeRO-3." + "reason": "PatchTSMixer post_init policy is unrelated to Qwen3.5 num_labels propagation." }, { - "left": "issue:39692", - "right": "issue:44610", + "left": "issue:44062", + "right": "issue:44451", "accept": false, - "reason": "Both mention processor/model mismatches, but they involve different models and distinct preprocessing bugs." + "reason": "AddedToken constructor argument conflict is different from a model-loading failure for ScandiBERT." }, { - "left": "issue:44918", - "right": "issue:44936", + "left": "issue:43421", + "right": "issue:43653", "accept": false, - "reason": "TRL SFT embedding unpacking and trainer evaluate-after-train failures are different training issues." + "reason": "Special-token runtime updates and BigBirdTokenizer mask-token registration are related to tokenizers in general, but not the same underlying defect." }, { - "left": "issue:45081", - "right": "issue:45356", + "left": "issue:44987", + "right": "issue:45042", "accept": false, - "reason": "Both reference mistral-regex-related tokenization, but one is an AttributeError crash and the other is a codec-handling regression." + "reason": "Different issues: loading physical-intelligence/fast on newer transformers vs PIL image processors requiring torchvision." }, { - "left": "issue:43525", - "right": "issue:43531", + "left": "issue:42886", + "right": "issue:44843", "accept": false, - "reason": "Missing `pad_token_id` on Llama4Config is unrelated to Qwen3-MoE sliding-window behavior." + "reason": "Both involve offline mode, but one is tokenizer cache loading and the other is an unconditional model_info() call in mistral regex patching." }, { - "left": "issue:44625", - "right": "issue:45479", + "left": "issue:45588", + "right": "issue:45759", "accept": false, - "reason": "Qwen3.5 config propagation and degenerate single-label classification loss are separate model-classification issues." + "reason": "Flash attention sink-less model crash and Qwen multimodal from_config handling are different code paths." }, { - "left": "issue:43782", - "right": "issue:45381", + "left": "issue:43531", + "right": "issue:44589", "accept": false, - "reason": "Qwen3VL weight-only loading failure is unrelated to Qwen2.5-VL video vision-position IDs." + "reason": "Sliding_window behavior for Qwen3-MoE is unrelated to missing Float8 storage lookup." }, { - "left": "issue:43994", - "right": "issue:44671", + "left": "issue:43421", + "right": "issue:44186", "accept": false, - "reason": "Incorrect SigLIP2 outputs and CamemBERT MLM regression are different model-specific failures." + "reason": "Special-token runtime updates and LayoutLMv2Tokenizer NER/batched padding crashes are different tokenizer bugs." }, { - "left": "issue:34634", - "right": "issue:41720", + "left": "issue:44918", + "right": "issue:44936", "accept": false, - "reason": "BarkProcessor voice preset handling is unrelated to Qwen3 auto device-mapping CUDA asserts." + "reason": "TRL SFT input-embedding unpacking and trainer.evaluate() after train() are different training workflow issues." }, { "left": "issue:43381", - "right": "issue:45704", - "accept": false, - "reason": "Gradient checkpointing in eval mode is unrelated to T5\u2019s apex RMSNorm leak." - }, - { - "left": "issue:43596", - "right": "issue:44220", + "right": "issue:43844", "accept": false, - "reason": "DeepSpeed ZeRO-3 init with BertModel and `_torch_extract_fbank_features()` are unrelated errors." + "reason": "Gradient checkpointing in eval mode and DeepSpeed ZeRO-3 gradient growth are unrelated training bugs." }, { "left": "issue:39692", - "right": "issue:43873", + "right": "issue:44610", "accept": false, - "reason": "SigLIP2 doc-example errors are unrelated to quantization/offloading behavior." + "reason": "SigLIP2 docs example errors and OmDet-Turbo processor/model resolution mismatch are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the 44514/44545 pair is the only clearly duplicated issue based on the near-identical titles, and the other pair rejections stay conservative rather than overstating sameness." + "feedback": "The analysis is grounded in the packet. The duplicate call on issue:44514/44545 is well supported by the matching titles and chronology, and the rest of the soft-edge rejections stay conservative rather than overclaiming equivalence. The note that no PRs are present is also consistent with the packet." }, "evaluator_used": true, "retried": false, @@ -4284,14 +4221,19 @@ "nodes": [ "issue:30064", "issue:36010", + "issue:39692", + "issue:40444", + "issue:41720", "issue:42915", + "issue:43322", "issue:43335", + "issue:43381", "issue:43421", + "issue:43525", "issue:43531", "issue:43540", "issue:43550", - "issue:43575", - "issue:43577", + "issue:43596", "issue:43618", "issue:43643", "issue:43653", @@ -4300,19 +4242,14 @@ "issue:43720", "issue:43746", "issue:43756", - "issue:43782", - "issue:43818", "issue:43819", "issue:43873", + "issue:43994", "issue:44008", "issue:44062", - "issue:44077", "issue:44186", "issue:44220", - "issue:44448", "issue:44488", - "issue:44561", - "issue:44625", "issue:44661", "issue:44671", "issue:44756", @@ -4320,256 +4257,252 @@ "issue:44843", "issue:44877", "issue:44898", - "issue:44977", - "issue:45020", "issue:45072", "issue:45081", "issue:45183", + "issue:45356", "issue:45381", "issue:45479", + "issue:45685", "issue:45704", "issue:45759" ], "soft_pairs": [ + "issue:43994|issue:44671", + "issue:43525|issue:43531", + "issue:45081|issue:45356", + "issue:43596|issue:44220", + "issue:41720|issue:45704", + "issue:43381|issue:45704", + "issue:39692|issue:43873", + "issue:43746|issue:43873", + "issue:40444|issue:43746", "issue:42915|issue:45759", + "issue:43746|issue:45381", "issue:44220|issue:44661", "issue:45183|issue:45479", "issue:43618|issue:43697", "issue:43540|issue:44008", "issue:43720|issue:43819", - "issue:44077|issue:44625", "issue:30064|issue:36010", - "issue:43575|issue:45704", "issue:44756|issue:45759", "issue:43716|issue:44008", "issue:43335|issue:43756", "issue:43653|issue:43720", - "issue:42915|issue:43746", - "issue:43746|issue:45381", - "issue:43746|issue:43873", - "issue:44843|issue:45081", + "issue:44062|issue:44488", + "issue:43322|issue:45685", "issue:43550|issue:45072", "issue:44186|issue:44898", "issue:43550|issue:43819", + "issue:44843|issue:45081", "issue:43421|issue:44811", - "issue:43577|issue:43818", - "issue:43643|issue:44877", - "issue:44062|issue:44488", - "issue:43531|issue:43643", - "issue:43540|issue:45381", - "issue:44448|issue:44671", - "issue:43653|issue:43819", - "issue:43782|issue:44977", - "issue:44561|issue:45020" + "issue:43643|issue:44877" ], - "prepared_review_unit_hash": "d816910637bf2f00fcef59a0745aa91171c2dedc220b02ad3503082186406b1e", + "prepared_review_unit_hash": "7efa06bd9315f4cb57142f25ab1f85e4ac08b1efb5f49745a22a1a716210983f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36372, - "estimated_input_tokens": 9093, - "estimated_eval_tokens": 18442 + "soft_pair_count": 29, + "serialized_chars": 36117, + "estimated_input_tokens": 9030, + "estimated_eval_tokens": 18316 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is a mix of unrelated model-, tokenizer-, config-, and runtime-specific bugs. None of the soft-pair candidates look like true duplicates or the same fixable code-path problem, and there are no PRs in the input.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43873 is the broadest, most active open report here, with multiple inbound references and discussion around a runtime behavior regression.", + "summary": "Mostly heterogeneous cluster. issue:45081 is the best anchor for the Mistral tokenizer regression family, but the Mistral-related pair with issue:45356 should stay rejected: they share a theme and keyword, not enough evidence that they are the same concrete bug.", + "confidence": 0.72, + "canonical_issue_reason": "issue:45081 is the best canonical issue because it is open, directly names the failing `_patch_mistral_regex` path, and is the clearest actionable report in the Mistral tokenizer area.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the strongest triage anchor because it is active, broadly scoped, and already attracting follow-up references.", + "best_issue_reason": "issue:45081 is the strongest representative issue: it is open, specific, and best captures the shared Mistral regex/tokenizer problem without over-merging distinct regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42915", - "right": "issue:45759", + "left": "issue:43994", + "right": "issue:44671", "accept": false, - "reason": "Different problems: Qwen3MoE FP8 config failure vs. AutoModelForCausalLM config unwrapping for composite Qwen multimodal configs." + "reason": "Different models and failures: SigLIP2 output/pipeline corruption is not the same bug as CamemBERT masked-LM regression." }, { - "left": "issue:44220", - "right": "issue:44661", + "left": "issue:43525", + "right": "issue:43531", "accept": false, - "reason": "Different code paths: audio feature extraction failure vs. tokenizer-model mapping logic in add-new-model-like." + "reason": "Separate Qwen3 issues with different symptoms: missing pad_token_id vs sliding_window behavior." }, { - "left": "issue:45183", - "right": "issue:45479", + "left": "issue:45081", + "right": "issue:45356", "accept": false, - "reason": "One is input validation/error messaging, the other is a classification loss bug with num_labels=1." + "reason": "Related Mistral tokenizer theme, but only title-level similarity; no shared files, targets, or concrete evidence of the same underlying bug." }, { - "left": "issue:43618", - "right": "issue:43697", + "left": "issue:43596", + "right": "issue:44220", "accept": false, - "reason": "Unrelated regressions in different models: CLIP output attentions vs. RTDetrV2 output differences in v5." + "reason": "Unrelated subsystems: DeepSpeed ZeRO-3/BERT init failure vs audio fbank extraction." }, { - "left": "issue:43540", - "right": "issue:44008", + "left": "issue:41720", + "right": "issue:45704", "accept": false, - "reason": "Different multimodal failures in different models; no shared concrete bug or fix path." + "reason": "Different code paths and failure modes: Qwen3 device-map CUDA assert vs T5 apex RMSNorm memory leak." }, { - "left": "issue:43720", - "right": "issue:43819", + "left": "issue:43381", + "right": "issue:45704", "accept": false, - "reason": "Different mechanisms: accelerate loading/unpacking packed weights vs. DAC latent conversion/STE mismatch." + "reason": "Gradient checkpointing eval-mode behavior is unrelated to the T5 RMSNorm memory leak." }, { - "left": "issue:44077", - "right": "issue:44625", + "left": "issue:39692", + "right": "issue:43873", "accept": false, - "reason": "One is a model post_init restriction, the other is config propagation for Qwen3.5 labels." + "reason": "Doc-example/model mismatch and quantization failure are not the same as generic offloading under quantization." }, { - "left": "issue:30064", - "right": "issue:36010", + "left": "issue:43746", + "right": "issue:43873", "accept": false, - "reason": "Completely unrelated: image processor segmentation-map handling vs. an import error for GenerationMixin." + "reason": "Both involve loading/quantization, but the concrete bugs are different." }, { - "left": "issue:43575", - "right": "issue:45704", + "left": "issue:40444", + "right": "issue:43746", "accept": false, - "reason": "Different issues: tensor-parallel OOM on Qwen2-57B vs. T5 using a leaky apex RMSNorm." + "reason": "Different multimodal loading failures: iterable dataset with multiple images vs GraniteSpeech checkpoint loading." }, { - "left": "issue:44756", + "left": "issue:42915", "right": "issue:45759", "accept": false, - "reason": "Different scope: mmap/OOM on Strix Halo vs. config unwrapping for Qwen multimodal models." - }, - { - "left": "issue:43716", - "right": "issue:44008", - "accept": false, - "reason": "Different models and failures: Mistral-3 dtype mismatch vs. Gemma 3n variable name collision." + "reason": "Qwen3Moe FP8 failure and composite config unwrapping are distinct bugs." }, { - "left": "issue:43335", - "right": "issue:43756", + "left": "issue:43746", + "right": "issue:45381", "accept": false, - "reason": "Unrelated model architecture bugs: sparse layer creation vs. RoPE layer retention." + "reason": "Different models and failures: GraniteSpeech checkpoint loading vs Qwen2.5-VL video position IDs." }, { - "left": "issue:43653", - "right": "issue:43720", + "left": "issue:44220", + "right": "issue:44661", "accept": false, - "reason": "Tokenizer special-token registration bug vs. BitNet packed-weight loading bug." + "reason": "Audio feature extraction and tokenizer mapping-name registration are unrelated." }, { - "left": "issue:42915", - "right": "issue:43746", + "left": "issue:45183", + "right": "issue:45479", "accept": false, - "reason": "Different models and failure modes: Qwen3MoE FP8 config vs. GraniteSpeech PEFT checkpoint loading." + "reason": "Generic input-validation messaging is distinct from degenerate zero-loss classification behavior." }, { - "left": "issue:43746", - "right": "issue:45381", + "left": "issue:43618", + "right": "issue:43697", "accept": false, - "reason": "Checkpoint loading with PEFT adapters is unrelated to Qwen2.5-VL video vision_position_ids." + "reason": "CLIP attentions assignment and RTDetrV2 output drift are separate regressions." }, { - "left": "issue:43746", - "right": "issue:43873", + "left": "issue:43540", + "right": "issue:44008", "accept": false, - "reason": "Unrelated: GraniteSpeech checkpoint loading vs. quantization/offloading behavior." + "reason": "Unrelated failures: Qwen3OmniMoe video processing vs Gemma3n variable collision." }, { - "left": "issue:44843", - "right": "issue:45081", + "left": "issue:43720", + "right": "issue:43819", "accept": false, - "reason": "Both touch _patch_mistral_regex, but they are different bugs: offline model_info call vs. backend_tokenizer AttributeError during regex fixing." + "reason": "Packed-weight loading in BitNet and DAC forward/from_latents mismatch are different concrete bugs." }, { - "left": "issue:43550", - "right": "issue:45072", + "left": "issue:30064", + "right": "issue:36010", "accept": false, - "reason": "Different subsystems and symptoms: torch.compile+SDPA failure vs. dtype mismatches in bfloat16 inference." + "reason": "Void segmentation-map processing has nothing to do with a GenerationMixin import error." }, { - "left": "issue:44186", - "right": "issue:44898", + "left": "issue:44756", + "right": "issue:45759", "accept": false, - "reason": "Distinct preprocessing/tokenization bug vs. image classification failure at non-default resolution." + "reason": "Strix Halo mmap/OOM and Qwen multimodal config unwrapping are unrelated." }, { - "left": "issue:43550", - "right": "issue:43819", + "left": "issue:43716", + "right": "issue:44008", "accept": false, - "reason": "Different model/runtime issues: compile/SDPA failure vs. DAC latent/forward mismatch." + "reason": "Mistral image-preprocessor dtype mismatch and Gemma3n audio-mask collision are different failures." }, { - "left": "issue:43421", - "right": "issue:44811", + "left": "issue:43335", + "right": "issue:43756", "accept": false, - "reason": "Post-processor updating for special tokens is not the same as Whisper batch_decode skip_special_tokens handling." + "reason": "Sparse-layer creation in SwitchTransformers and Smollm3 RoPE-layer dropping are unrelated." }, { - "left": "issue:43577", - "right": "issue:43818", + "left": "issue:43653", + "right": "issue:43720", "accept": false, - "reason": "Different models and bugs: BLIP2 dtype preservation vs. Video-LLaVA missing temporal attention/weight sharing." + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight unpacking are separate issues." }, { - "left": "issue:43643", - "right": "issue:44877", + "left": "issue:44062", + "right": "issue:44488", "accept": false, - "reason": "Both involve config loading, but one is missing fields from trust_remote_code AutoConfig and the other is strict granite_speech config rejection." + "reason": "AddedToken constructor duplication and model-loading failure for cjvt/sleng-bert are not the same bug." }, { - "left": "issue:44062", - "right": "issue:44488", + "left": "issue:43322", + "right": "issue:45685", "accept": false, - "reason": "Different tokenizer/loading failures: AddedToken keyword collision vs. a model load regression for cjvt/sleng-bert." + "reason": "Llava Next segfault and MPS histogram-on-Int error are unrelated." }, { - "left": "issue:43531", - "right": "issue:43643", + "left": "issue:43550", + "right": "issue:45072", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to AutoConfig trust_remote_code field loss." + "reason": "torch.compile SDPA failure in Bamba is not the same as dtype mismatches in SwitchTransformers/TimmWrapperModel." }, { - "left": "issue:44448", - "right": "issue:44671", + "left": "issue:44186", + "right": "issue:44898", "accept": false, - "reason": "Both are v5 regressions, but they affect different models and concrete outputs; not the same underlying bug." + "reason": "Tokenizer NER/padding crash and Perceiver non-default-resolution failure are different subsystems." }, { - "left": "issue:43653", + "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Tokenizer special-token registration and DAC STE mismatch are unrelated." + "reason": "Different model/runtime failures with no shared concrete code path." }, { - "left": "issue:43782", - "right": "issue:44977", + "left": "issue:44843", + "right": "issue:45081", "accept": false, - "reason": "Different Qwen variants and failure modes: from_pretrained weight_only error vs. flash-attention generation bug." + "reason": "Both touch `_patch_mistral_regex`, but one is HF offline handling and the other is a backend-attribute crash; not the same bug." }, { - "left": "issue:44561", - "right": "issue:45020", + "left": "issue:43421", + "right": "issue:44811", "accept": false, - "reason": "44561 is specifically the removal of is_torch_fx_available breaking remote_code models; 45020 is a broader report about recent versions breaking remote_code models, not the same concrete bug." + "reason": "Runtime special-token post-processor update bug is distinct from Whisper batch_decode skip_special_tokens handling." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, but it overstates one key soft-edge verdict. The packet\u2019s only Mistral-related pair, issue:45081 vs issue:45356, is marked deterministic_accept=false and is supported only by soft title similarity; there are no shared filenames, shared targets, or direct evidence that they are the same concrete bug. Treating that pair as accepted is not conservative enough. Everything else looks appropriately cautious, but this one verdict should remain rejected or at most tentative." + }, + "evaluator_used": true, + "retried": true, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -4591,23 +4524,21 @@ "issue:34634", "issue:36010", "issue:36331", + "issue:41720", "issue:42175", - "issue:43262", - "issue:43322", - "issue:43329", - "issue:43404", - "issue:43452", - "issue:43526", + "issue:43531", + "issue:43540", "issue:43577", "issue:43582", "issue:43606", + "issue:43643", "issue:43644", "issue:43650", "issue:43653", - "issue:43697", "issue:43720", "issue:43749", "issue:43782", + "issue:43818", "issue:43819", "issue:43825", "issue:43827", @@ -4615,246 +4546,255 @@ "issue:43994", "issue:44060", "issue:44186", + "issue:44291", "issue:44368", "issue:44442", + "issue:44448", "issue:44485", "issue:44509", + "issue:44561", "issue:44610", "issue:44671", "issue:44805", "issue:44821", - "issue:44857", - "issue:44912", "issue:44936", + "issue:44977", "issue:44991", + "issue:45020", "issue:45072", "issue:45081", "issue:45084", + "issue:45325", + "issue:45381", "issue:45406", - "issue:45538", - "issue:45600", - "issue:45685", "issue:45701" ], "soft_pairs": [ + "issue:43577|issue:43818", + "issue:43531|issue:43643", + "issue:43540|issue:45381", + "issue:44448|issue:44671", + "issue:43653|issue:43819", + "issue:43782|issue:44977", + "issue:44561|issue:45020", + "issue:43644|issue:43749", "issue:43825|issue:43994", - "issue:43322|issue:45685", + "issue:43577|issue:43582", "issue:44821|issue:45084", - "issue:43825|issue:43827", + "issue:41720|issue:44991", "issue:30064|issue:36331", - "issue:44485|issue:44509", - "issue:43577|issue:43582", "issue:44186|issue:44442", + "issue:43825|issue:43827", + "issue:44485|issue:44509", + "issue:44671|issue:45701", + "issue:44805|issue:44936", "issue:43606|issue:45072", - "issue:43452|issue:44991", "issue:34634|issue:36010", - "issue:44805|issue:44936", "issue:42175|issue:45081", "issue:44610|issue:45072", + "issue:43782|issue:45325", "issue:45081|issue:45406", "issue:44060|issue:44368", - "issue:43650|issue:43906", "issue:43720|issue:45072", - "issue:44671|issue:45701", - "issue:43404|issue:43782", - "issue:43582|issue:44857", - "issue:43262|issue:43819", - "issue:43526|issue:43697", - "issue:45084|issue:45538", - "issue:43329|issue:45600", - "issue:43644|issue:43749", - "issue:43653|issue:44912" + "issue:43650|issue:43906", + "issue:43749|issue:44291" ], - "prepared_review_unit_hash": "fbff41e56a679b16bda0d3690fce5af578a8dee00dba3fdef0be3728830f6735", + "prepared_review_unit_hash": "7fa13227828298125442195d6d93d96133d991c3daa9d7b41d8ed9aed9a65b1a", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 35989, - "estimated_input_tokens": 8998, - "estimated_eval_tokens": 18252 + "soft_pair_count": 28, + "serialized_chars": 36241, + "estimated_input_tokens": 9061, + "estimated_eval_tokens": 18378 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the items span unrelated bugs in processors, tokenizers, training, loading, and docs. I found no convincing duplicate issue/PR pairings.", - "confidence": 0.9, - "canonical_issue_reason": "No single issue is a good canonical representative because the cluster does not center on one underlying bug; the titles point to many distinct code paths and models.", + "summary": "This cluster is mostly a heterogeneous mix of Transformers bug reports that only share broad themes like v5 regressions, model loading, tokenization, or processor issues. None of the soft-similarity pairs look like true duplicates of the same underlying bug/change, so I rejected all of them. The broadest, most representative issue is the open remote_code regression tracker.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45020 is the broadest umbrella in the set: it captures recent-version regressions affecting remote_code models and is open with active discussion, making it the best representative issue for this noisy cluster.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong cluster anchor here. The most discussed items are still about different failures, so using any one as canonical would be misleading.", + "best_issue_reason": "issue:45020 is the most generally useful issue to anchor the cluster because it is broad, current, and close to the recurring theme across several reports about recent Transformers breakage.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43825", - "right": "issue:43994", + "left": "issue:43577", + "right": "issue:43818", "accept": false, - "reason": "Both mention pipeline behavior, but one is about a wrong v5 error message and the other about SigLIP2 producing bad outputs; different bugs." + "reason": "Different models and different failure modes: dtype loading for BLIP-2 vs missing temporal attention/shared weights in Video-LLaVA." }, { - "left": "issue:43322", - "right": "issue:45685", + "left": "issue:43531", + "right": "issue:43643", "accept": false, - "reason": "Different subsystems and failure modes: Llava Next loading segfault vs an MPS histogram/int dtype issue." + "reason": "Unrelated bugs: Qwen3-MoE sliding_window behavior vs AutoConfig missing fields with trust_remote_code." }, { - "left": "issue:44821", - "right": "issue:45084", + "left": "issue:43540", + "right": "issue:45381", "accept": false, - "reason": "Unrelated: AutoImageProcessor URL loading vs a template compilation TypeError." + "reason": "Both involve video inputs, but they target different model families and distinct bugs (Qwen3OmniMoe ValueError vs Qwen2.5-VL position-id scaling)." }, { - "left": "issue:43825", - "right": "issue:43827", + "left": "issue:44448", + "right": "issue:44671", "accept": false, - "reason": "Closely related theme (v5 pipeline removals), but one is runtime error messaging and the other is docs cleanup; not the same concrete bug." + "reason": "Both are v5 output regressions, but they affect different models and tasks, so they do not look like the same underlying bug." }, { - "left": "issue:30064", - "right": "issue:36331", + "left": "issue:43653", + "right": "issue:43819", "accept": false, - "reason": "Different areas: image segmentation-map processing vs CustomTrainer compute_loss signature mismatch." + "reason": "Completely different code paths: tokenizer special-token registration vs DAC latent decoding/STE mismatch." }, { - "left": "issue:44485", - "right": "issue:44509", + "left": "issue:43782", + "right": "issue:44977", "accept": false, - "reason": "GLM-5 RoPE implementation discussion is unrelated to pipeline docs references." + "reason": "Different Qwen variants and different stages of failure: loading with weight_only vs generation with flash-attention." }, { - "left": "issue:43577", - "right": "issue:43582", + "left": "issue:44561", + "right": "issue:45020", "accept": false, - "reason": "Blip2 dtype loading bug vs Apple Silicon allocator warmup TypeError are unrelated." + "reason": "Related theme, but not the same bug: one is a specific is_torch_fx_available removal breakage, the other is a broad remote_code regression report." }, { - "left": "issue:44186", - "right": "issue:44442", + "left": "issue:43644", + "right": "issue:43749", "accept": false, - "reason": "Both are tokenizer bugs, but they affect different tokenizers with different behaviors and fixes." + "reason": "Both are loading regressions, but one is non-persistent buffer initialization and the other is FSDP CPU RAM efficient loading; too different to merge." }, { - "left": "issue:43606", - "right": "issue:45072", + "left": "issue:43825", + "right": "issue:43994", "accept": false, - "reason": "Both involve device/dtype issues, but they are separate models and separate code paths." + "reason": "Different issues: an incorrect pipeline error message vs a model producing nonsensical outputs through AutoModel/pipeline." }, { - "left": "issue:43452", - "right": "issue:44991", + "left": "issue:43577", + "right": "issue:43582", "accept": false, - "reason": "Tokenizer/model loading regressions, but not the same concrete failure; one is gguf_file handling, the other is a specific tokenizer load breakage." + "reason": "No shared underlying bug: BLIP-2 dtype handling vs Apple Silicon TypeError in caching_allocator_warmup." }, { - "left": "issue:34634", - "right": "issue:36010", + "left": "issue:44821", + "right": "issue:45084", "accept": false, - "reason": "BarkProcessor voice_preset bug is unrelated to the GenerationMixin import error." + "reason": "Both are runtime errors, but one is AutoImageProcessor-from-URL loading and the other is template compilation; unrelated." }, { - "left": "issue:44805", - "right": "issue:44936", + "left": "issue:41720", + "right": "issue:44991", "accept": false, - "reason": "Mask shape IndexError and trainer.evaluate() after train() are different problems." + "reason": "Different symptoms and likely causes: Qwen3 auto device mapping CUDA assert vs tokenizer loading regression for est-roberta." }, { - "left": "issue:42175", - "right": "issue:45081", + "left": "issue:30064", + "right": "issue:36331", "accept": false, - "reason": "Package/backend install issue vs Mistral regex patch AttributeError; no shared underlying bug." + "reason": "Image processor segmentation-map failure and CustomTrainer compute_loss signature mismatch are unrelated." }, { - "left": "issue:44610", - "right": "issue:45072", + "left": "issue:44186", + "right": "issue:44442", "accept": false, - "reason": "OmDet-Turbo input-size mismatch is unrelated to dtype mismatch inference issues." + "reason": "Tokenizer crashes on NER/batched padding vs AutoTokenizer failing to load a specific tokenizer are not the same bug." }, { - "left": "issue:45081", - "right": "issue:45406", + "left": "issue:43825", + "right": "issue:43827", "accept": false, - "reason": "Both are AttributeErrors in tokenizer/processor code, but on different attributes and different components." + "reason": "Both mention removed pipeline tasks, but one is an error-message issue and the other is stale docs; not the same defect." }, { - "left": "issue:44060", - "right": "issue:44368", + "left": "issue:44485", + "right": "issue:44509", "accept": false, - "reason": "Related to Qwen tied-weights warnings, but the reported failures are not the same concrete bug or fix." + "reason": "GLM-5 RoPE implementation discussion is unrelated to docs still mentioning removed pipeline tasks." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:44671", + "right": "issue:45701", "accept": false, - "reason": "Placeholder issue title vs an isolated reproduction of a different issue; no substantive overlap." + "reason": "A specific CamemBERT masked-LM regression is not the same as a general tokenization-change report." }, { - "left": "issue:43720", + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: mask/index shape mismatch vs trainer.evaluate failing after training." + }, + { + "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Packed-weight loading bug vs bfloat16 dtype mismatch; unrelated." + "reason": "Different model/device issues: CPU offload device mismatch vs bfloat16 dtype mismatches." }, { - "left": "issue:44671", - "right": "issue:45701", + "left": "issue:34634", + "right": "issue:36010", "accept": false, - "reason": "CamemBERT MLM regression in v5 is not the same as the broad statement that version changes tokenization." + "reason": "BarkProcessor voice_preset behavior and a GenerationMixin import error are unrelated." }, { - "left": "issue:43404", - "right": "issue:43782", + "left": "issue:42175", + "right": "issue:45081", "accept": false, - "reason": "Different model families and different loading/tied-weight failures." + "reason": "Packaging/backend install issue vs Mistral tokenizer regex patch crash; no shared underlying bug." }, { - "left": "issue:43582", - "right": "issue:44857", + "left": "issue:44610", + "right": "issue:45072", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError is unrelated to a CUDA AMP loss crash." + "reason": "Processor image-size mismatch and bfloat16 dtype mismatch are unrelated." }, { - "left": "issue:43262", - "right": "issue:43819", + "left": "issue:43782", + "right": "issue:45325", "accept": false, - "reason": "Audio chat-template sampling-rate default bug vs DAC.from_latents mismatch; different functionality." + "reason": "Different Qwen VL bugs: weight_only load error vs incorrect still-image temporal position_ids." }, { - "left": "issue:43526", - "right": "issue:43697", + "left": "issue:45081", + "right": "issue:45406", "accept": false, - "reason": "Different models and different bugs: label reduction in BeitImageProcessorFast vs RTDetrV2 output drift." + "reason": "Both are AttributeErrors during loading/serve, but they involve different objects and causes, so they are not duplicates." }, { - "left": "issue:45084", - "right": "issue:45538", + "left": "issue:44060", + "right": "issue:44368", "accept": false, - "reason": "Template compilation error is unrelated to CLIPTokenizer model_max_length behavior." + "reason": "Both mention tied-weights warnings, but one is a model-internal incorrect warning and the other is a LoRA fine-tuning warning; not the same bug." }, { - "left": "issue:43329", - "right": "issue:45600", + "left": "issue:43720", + "right": "issue:45072", "accept": false, - "reason": "Multimodal token counting bug vs auto_mappings referencing removed configs; no shared underlying defect." + "reason": "BitNet packed-weight loading failure is unrelated to bfloat16 dtype mismatch issues." }, { - "left": "issue:43644", - "right": "issue:43749", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "Non-persistent buffer junk in v5 and broken FSDP CPU RAM efficient loading are different loading problems." + "reason": "A placeholder/data-request issue and an isolated reproduction report are not the same bug or change." }, { - "left": "issue:43653", - "right": "issue:44912", + "left": "issue:43749", + "right": "issue:44291", "accept": false, - "reason": "BigBird special-token decoding bug is unrelated to MXFP4 quantization fallback." + "reason": "Two distinct loading regressions: FSDP CPU RAM efficient loading vs init_empty_weights TypeError from _is_hf_initialized." } ] }, @@ -4878,29 +4818,26 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:34567", "issue:34634", "issue:36010", "issue:36331", "issue:38175", "issue:39692", - "issue:40990", - "issue:41720", "issue:43257", "issue:43262", "issue:43329", - "issue:43450", "issue:43452", "issue:43454", "issue:43526", "issue:43531", - "issue:43540", "issue:43550", "issue:43577", "issue:43582", "issue:43638", "issue:43643", "issue:43645", + "issue:43653", + "issue:43696", "issue:43697", "issue:43720", "issue:43746", @@ -4909,264 +4846,273 @@ "issue:43854", "issue:43927", "issue:43931", - "issue:44291", "issue:44479", "issue:44554", "issue:44568", "issue:44589", "issue:44610", + "issue:44756", "issue:44811", - "issue:44849", + "issue:44857", "issue:44912", "issue:44936", "issue:44977", + "issue:44991", "issue:45042", + "issue:45084", "issue:45356", - "issue:45357", "issue:45488", - "issue:45701", + "issue:45538", + "issue:45600", "issue:45759" ], "soft_pairs": [ + "issue:43582|issue:44857", + "issue:43526|issue:43697", + "issue:43262|issue:43819", + "issue:43526|issue:44554", + "issue:45084|issue:45538", + "issue:34634|issue:36331", + "issue:43329|issue:45600", + "issue:43653|issue:44912", + "issue:43452|issue:44991", "issue:43638|issue:44936", + "issue:43257|issue:43329", "issue:43746|issue:44811", - "issue:38175|issue:44610", "issue:44589|issue:45042", + "issue:38175|issue:44610", "issue:43454|issue:43577", "issue:43854|issue:44977", - "issue:43526|issue:44554", + "issue:43257|issue:43696", "issue:36010|issue:43645", "issue:44568|issue:45488", - "issue:34634|issue:36331", "issue:39692|issue:44912", - "issue:43257|issue:43329", "issue:43819|issue:43927", "issue:43720|issue:44610", - "issue:43749|issue:44291", "issue:43550|issue:44912", - "issue:43262|issue:44811", + "issue:43749|issue:44756", "issue:44568|issue:45356", + "issue:43262|issue:44811", "issue:43531|issue:45042", - "issue:43643|issue:45759", "issue:36010|issue:36331", - "issue:34567|issue:40990", - "issue:43329|issue:43450", + "issue:43643|issue:45759", "issue:43329|issue:44479", - "issue:44849|issue:45701", - "issue:43257|issue:43931", - "issue:43697|issue:45357", - "issue:43452|issue:43582", - "issue:41720|issue:43540", - "issue:44291|issue:45042" + "issue:43257|issue:43931" ], - "prepared_review_unit_hash": "dba4fec7fd2248aaac6fba3f39fc3f599bb0fd097a37cdcb185898679487d297", + "prepared_review_unit_hash": "758bc0f2b7333327127fd3cc7fa456c93205223bf6a69c2f1454a9124b927d7c", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 37051, - "estimated_input_tokens": 9263, - "estimated_eval_tokens": 18782 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of mostly unrelated Transformers regressions. The soft pairs share broad themes like loading, tokenization, video, or multimodal processing, but they do not look like the same concrete bug/change.", - "confidence": 0.88, - "canonical_issue_reason": "issue:43749 is the broadest, highest-impact loading/initialization regression in the set and has the strongest discussion/reference signal; the rest are mostly narrower model- or feature-specific bugs.", + "summary": "This cluster is highly heterogeneous: it spans unrelated bugs in tokenizers, audio/video processors, model loading, training APIs, quantization, and hardware-specific failures. None of the soft pairs look like the same underlying defect, so there are no duplicate merges here.", + "confidence": 0.98, + "canonical_issue_reason": "No true canonical issue: the items are not one bug cluster, but a mix of unrelated regressions across different subsystems and model families.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 best represents the cluster because it is a central infrastructure-loading failure rather than a model-specific edge case.", + "best_issue_reason": "If one representative must be chosen, issue:43749 is the broadest and most externally visible loading regression in the set, but it should not be treated as a duplicate canonical for the rest.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43638", - "right": "issue:44936", - "accept": false, - "reason": "Different trainer failures: Deepspeed ZeRO-3 index error vs trainer.evaluate() after train()." - }, - { - "left": "issue:43746", - "right": "issue:44811", + "left": "issue:43582", + "right": "issue:44857", "accept": false, - "reason": "Different code paths: PEFT/local checkpoint loading vs Whisper batch_decode skip_special_tokens handling." + "reason": "Different failures: Apple Silicon allocator warmup TypeError vs CUDA float16 AMP crash in LwDetrImageLoss." }, { - "left": "issue:38175", - "right": "issue:44610", + "left": "issue:43526", + "right": "issue:43697", "accept": false, - "reason": "Different models and bugs: SigLIP2 zero probabilities vs OmDet-Turbo processor size mismatch." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to RTDetrV2 output drift in v5." }, { - "left": "issue:44589", - "right": "issue:45042", + "left": "issue:43262", + "right": "issue:43819", "accept": false, - "reason": "Unrelated failures: Float8 storage lookup vs PIL image processor torchvision dependency regression." + "reason": "Audio chat template sampling-rate default bug is unrelated to DAC.from_latents / missing STE mismatch." }, { - "left": "issue:43454", - "right": "issue:43577", + "left": "issue:43526", + "right": "issue:44554", "accept": false, - "reason": "Both are model-loading issues, but one is lm_head tying in AyaVision and the other is Blip2 dtype propagation; not the same bug." + "reason": "Image processor label handling is unrelated to the MPS attention correctness issue." }, { - "left": "issue:43854", - "right": "issue:44977", + "left": "issue:45084", + "right": "issue:45538", "accept": false, - "reason": "Different issues: GLM-4.7-Flash test loading failure vs Qwen3.5 flash-attention generation behavior." + "reason": "Non-template node compilation error and CLIPTokenizer max_length bug are different code paths." }, { - "left": "issue:43526", - "right": "issue:44554", + "left": "issue:34634", + "right": "issue:36331", "accept": false, - "reason": "Different subsystems: BEiT label reduction bug vs MPS attention correctness issue." + "reason": "Bark voice_preset failure is unrelated to CustomTrainer.compute_loss signature changes." }, { - "left": "issue:36010", - "right": "issue:43645", + "left": "issue:43329", + "right": "issue:45600", "accept": false, - "reason": "Both are API regressions, but one is GenerationMixin import resolution and the other is custom model initialization in notebooks." + "reason": "Video multimodal token counting bug is unrelated to stale auto_mappings CI references." }, { - "left": "issue:44568", - "right": "issue:45488", + "left": "issue:43653", + "right": "issue:44912", "accept": false, - "reason": "Both concern tokenization, but they hit different tokenizer implementations and failure modes." + "reason": "BigBird special-token decode bug is unrelated to MXFP4 quantization loading fallback." }, { - "left": "issue:34634", - "right": "issue:36331", + "left": "issue:43452", + "right": "issue:44991", "accept": false, - "reason": "Different training/audio bugs: BarkProcessor voice_preset vs CustomTrainer.compute_loss signature change." + "reason": "GGUF from_pretrained breakage is not the same as the EMBEDDIA tokenizer load regression." }, { - "left": "issue:39692", - "right": "issue:44912", + "left": "issue:43638", + "right": "issue:44936", "accept": false, - "reason": "Unrelated: documentation/example errors for SigLIP2 vs MXFP4 quantization fallback for git-oss-20b." + "reason": "Zero3 non-pretrained BERT crash is a different problem from evaluate-after-train failure." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Different problems: Qwen3 MoE weight conversion/loading vs undefined helpers in multimodal video token counting." + "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed is unrelated to video token helper undefined names." }, { - "left": "issue:43819", - "right": "issue:43927", + "left": "issue:43746", + "right": "issue:44811", "accept": false, - "reason": "Different model bugs: DAC latent reconstruction mismatch vs DiaConfig losing custom token IDs on save/load." + "reason": "PEFT adapter local checkpoint loading and Whisper batch_decode skip_special_tokens are distinct bugs." }, { - "left": "issue:43720", - "right": "issue:44610", + "left": "issue:44589", + "right": "issue:45042", "accept": false, - "reason": "Different failures: BitNet packed-weight unpacking during accelerate loading vs OmDet-Turbo processor output shape/size mismatch." + "reason": "Float8 storage lookup TypeError is unrelated to PIL backend torchvision requirement regressions." }, { - "left": "issue:43749", - "right": "issue:44291", + "left": "issue:38175", + "right": "issue:44610", "accept": false, - "reason": "Both involve loading infrastructure, but one is FSDP CPU RAM efficient loading and the other is init_empty_weights/_is_hf_initialized signature breakage." + "reason": "SigLIP2 zero-probability outputs and OmDet 640px vs 224px processor mismatch are different issues." }, { - "left": "issue:43550", - "right": "issue:44912", + "left": "issue:43454", + "right": "issue:43577", "accept": false, - "reason": "Different model/runtime issues: torch.compile SDPA bug vs MXFP4 quantization fallback." + "reason": "AyaVision lm_head tying failure is unrelated to Blip2 dtype propagation." }, { - "left": "issue:43262", - "right": "issue:44811", + "left": "issue:43854", + "right": "issue:44977", "accept": false, - "reason": "Different audio processor bugs: chat-template sampling rate default vs Whisper decode special-token handling." + "reason": "GLM-4.7-Flash load test failure is not the same as Qwen3.5 flash-attention generation regression." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "MoE weight conversion/loading bug is unrelated to GPT-oss out-of-memory." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import error and Jupyter custom-model init regression are separate API breakages." }, { "left": "issue:44568", - "right": "issue:45356", + "right": "issue:45488", "accept": false, - "reason": "Different tokenizer regressions on different models and code paths; not the same defect." + "reason": "mdeberta add_special_tokens BOS/EOS regression is different from LlamaTokenizer ByteLevel-to-Metaspace override." }, { - "left": "issue:43531", - "right": "issue:45042", + "left": "issue:39692", + "right": "issue:44912", "accept": false, - "reason": "Unrelated: Qwen3-MoE sliding_window bug vs PIL backend torchvision requirement regression." + "reason": "SigLIP2 documentation example problems are unrelated to git-oss-20b MXFP4 quantization loading." }, { - "left": "issue:43643", - "right": "issue:45759", + "left": "issue:43819", + "right": "issue:43927", "accept": false, - "reason": "Both are config/loading-related, but one is trust_remote_code returning missing fields and the other is AutoModelForCausalLM.from_config not unwrapping text_config." + "reason": "DAC latent/forward mismatch is unrelated to DiaConfig losing custom token IDs on save/load." }, { - "left": "issue:36010", - "right": "issue:36331", + "left": "issue:43720", + "right": "issue:44610", "accept": false, - "reason": "Different APIs: import path for GenerationMixin vs Trainer.compute_loss keyword-argument compatibility." + "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to OmDet processor size mismatch." }, { - "left": "issue:34567", - "right": "issue:40990", + "left": "issue:43550", + "right": "issue:44912", "accept": false, - "reason": "Unrelated: TrainerState token counter not updating vs GPT-OSS perplexity regression." + "reason": "Bamba torch.compile + SDPA failure is unrelated to quantization fallback in git-oss-20b." }, { - "left": "issue:43329", - "right": "issue:43450", + "left": "issue:43749", + "right": "issue:44756", "accept": false, - "reason": "Both are video-related, but one is missing functions in token counting and the other is batched video processor shape handling." + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different loading/hardware issues." }, { - "left": "issue:43329", - "right": "issue:44479", + "left": "issue:44568", + "right": "issue:45356", "accept": false, - "reason": "Shared video theme only; one is undefined helpers in a video branch, the other is a broader v5.3.0 video input regression." + "reason": "Different tokenizer regressions: missing BOS/EOS addition vs codec handling / warning behavior in Kimi-K2.5." }, { - "left": "issue:44849", - "right": "issue:45701", + "left": "issue:43262", + "right": "issue:44811", "accept": false, - "reason": "Different regressions: Qwen3.5 hidden-state output bug vs version-dependent tokenization changes." + "reason": "apply_chat_template sampling-rate default bug is unrelated to Whisper batch_decode special-token handling." }, { - "left": "issue:43257", - "right": "issue:43931", + "left": "issue:43531", + "right": "issue:45042", "accept": false, - "reason": "Different Qwen-related loading bugs: MoE weight conversion vs VL weight shape mismatch." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to PIL backend torchvision dependency issues." }, { - "left": "issue:43697", - "right": "issue:45357", + "left": "issue:36010", + "right": "issue:36331", "accept": false, - "reason": "Different issues: RTDetrV2 output drift in v5 vs save_pretrained storing incorrect visual encoder keys." + "reason": "Importing GenerationMixin and compute_loss keyword-argument mismatch are unrelated failures." }, { - "left": "issue:43452", - "right": "issue:43582", + "left": "issue:43643", + "right": "issue:45759", "accept": false, - "reason": "Unrelated: gguf_file/from_pretrained loading breakage vs Apple Silicon caching_allocator_warmup TypeError." + "reason": "trust_remote_code missing fields in AutoConfig is not the same as unwrapping text_config in AutoModelForCausalLM.from_config." }, { - "left": "issue:41720", - "right": "issue:43540", + "left": "issue:43329", + "right": "issue:44479", "accept": false, - "reason": "Different Qwen3-family failures: auto device mapping cuda assert vs video input ValueError in Qwen3OmniMoe." + "reason": "Both involve video paths, but one is an internal undefined-name bug in token counting while the other is a model-specific video regression; not enough to treat as the same defect." }, { - "left": "issue:44291", - "right": "issue:45042", + "left": "issue:43257", + "right": "issue:43931", "accept": false, - "reason": "Different bugs in different layers: init_empty_weights argument handling vs image processor torchvision dependency." + "reason": "MoE loading/conversion under accelerate+deepspeed is unrelated to Qwen3-VL weight-shape mismatch." } ] }, @@ -5190,307 +5136,293 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:28282", "issue:33357", "issue:34567", - "issue:36010", - "issue:36331", + "issue:34689", "issue:38175", + "issue:40990", + "issue:41720", "issue:42175", "issue:42915", "issue:43257", "issue:43317", + "issue:43329", + "issue:43450", "issue:43452", - "issue:43493", "issue:43526", "issue:43540", "issue:43577", - "issue:43643", - "issue:43646", - "issue:43696", + "issue:43582", + "issue:43697", "issue:43742", - "issue:43749", "issue:43756", "issue:43824", "issue:43906", "issue:43927", - "issue:43994", "issue:44062", "issue:44291", "issue:44479", "issue:44488", "issue:44560", "issue:44737", - "issue:44756", - "issue:44792", "issue:44821", + "issue:44849", "issue:44898", "issue:44912", - "issue:44964", + "issue:44938", "issue:44987", "issue:44991", + "issue:45042", "issue:45072", - "issue:45081", "issue:45137", - "issue:45375", + "issue:45357", "issue:45381", "issue:45405", "issue:45479", "issue:45538", - "issue:45588" + "issue:45588", + "issue:45701" ], "soft_pairs": [ + "issue:34567|issue:40990", + "issue:43329|issue:43450", + "issue:44849|issue:45701", + "issue:43257|issue:43756", + "issue:43697|issue:45357", + "issue:44987|issue:45538", + "issue:43452|issue:43582", + "issue:41720|issue:43540", "issue:42915|issue:43742", + "issue:44291|issue:45042", + "issue:44479|issue:45381", "issue:33357|issue:34567", + "issue:43577|issue:45072", + "issue:42915|issue:44821", + "issue:42915|issue:45538", + "issue:42175|issue:43824", "issue:43317|issue:45538", - "issue:44987|issue:45538", - "issue:44479|issue:45381", "issue:43540|issue:44560", "issue:43927|issue:44912", - "issue:42915|issue:44821", - "issue:43257|issue:43756", - "issue:43577|issue:45072", "issue:43540|issue:44479", - "issue:42915|issue:45538", - "issue:43257|issue:43696", + "issue:44062|issue:44991", + "issue:42175|issue:44488", + "issue:42175|issue:44938", "issue:45137|issue:45405", + "issue:43526|issue:45479", "issue:38175|issue:44898", "issue:44737|issue:45588", - "issue:44062|issue:44991", - "issue:43742|issue:44291", "issue:43906|issue:45381", - "issue:43646|issue:43749", - "issue:43749|issue:44756", - "issue:44062|issue:45081", - "issue:36010|issue:42175", - "issue:34567|issue:36331", - "issue:43643|issue:45375", - "issue:43493|issue:43994", - "issue:42175|issue:43824", - "issue:43452|issue:44488", - "issue:43526|issue:45479", - "issue:44291|issue:44821", - "issue:44792|issue:44964" + "issue:28282|issue:34689" ], - "prepared_review_unit_hash": "10475a892ee88241aff13a2d3e0b91d5fea5da0d52da5e458f69c5482d49f8e9", + "prepared_review_unit_hash": "0c5a27ae8180f760e8178ac5d7c2ce828fe7346ce3f9e7d265c32461f0c76611", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36567, - "estimated_input_tokens": 9142, - "estimated_eval_tokens": 18540 + "soft_pair_count": 29, + "serialized_chars": 35926, + "estimated_input_tokens": 8982, + "estimated_eval_tokens": 18220 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: most items are distinct bug reports across unrelated models/subsystems. The only strong duplicate-like matches are the Qwen video-input regression pair and the SigLIP2 correctness pair.", - "confidence": 0.66, - "canonical_issue_reason": "issue:44479 is the broadest, most central issue in the set around the shared Qwen video-input regression, and it has the strongest umbrella framing for the accepted duplicate-like report.", + "summary": "This cluster is mostly heterogeneous transformer bug reports with only one plausible duplicate subcluster around Qwen video-input regressions. Most soft pairs are not the same underlying issue.", + "confidence": 0.33, + "canonical_issue_reason": "Issue 44479 is the best umbrella candidate because it describes the broad v5.3.0 video-input regression affecting several Qwen video models, and it can subsume the narrower Qwen2.5-VL report in 45381.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the best single representative of the cluster\u2019s overlapping video-regression theme; it is broader than the narrower follow-up report and is more suitable as the canonical issue.", + "best_issue_reason": "44479 is the most representative issue here: it is broad, regression-focused, and closest to the only plausible duplicate subcluster in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42915", - "right": "issue:43742", + "left": "issue:34567", + "right": "issue:40990", "accept": false, - "reason": "Different models and failure modes: Qwen3Moe FP8 config vs MobileLLM key error." + "reason": "TrainerState token counting and GPT-OSS perplexity are unrelated bugs in different code paths." }, { - "left": "issue:33357", - "right": "issue:34567", + "left": "issue:43329", + "right": "issue:43450", "accept": false, - "reason": "Unrelated areas: MacOS bus error with CLIP vs TrainerState token-count tracking." + "reason": "Undefined names in multimodal token counting vs batched video processor shape handling are different failures." }, { - "left": "issue:43317", - "right": "issue:45538", + "left": "issue:44849", + "right": "issue:45701", "accept": false, - "reason": "Different problems: offload/loading failure vs CLIPTokenizer max-length constant." + "reason": "Qwen3.5 hidden-states behavior and version-dependent tokenization changes are not the same bug." }, { - "left": "issue:44987", - "right": "issue:45538", + "left": "issue:43257", + "right": "issue:43756", "accept": false, - "reason": "Different loading bugs: model load regression vs tokenizer max-length issue." + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed and Smollm3 RoPE-layer dropping are unrelated model issues." }, { - "left": "issue:44479", - "right": "issue:45381", - "accept": true, - "reason": "Both describe the same Qwen video-input regression; 45381 looks like the narrower symptom/diagnosis of the broader v5.3.0 video-input bug." - }, - { - "left": "issue:43540", - "right": "issue:44560", + "left": "issue:43697", + "right": "issue:45357", "accept": false, - "reason": "Both involve video inputs, but they are different Qwen model variants with different errors; not enough to treat as the same bug." + "reason": "RTDetrV2 output differences and Qwen3.5 save_pretrained key persistence affect different models and code paths." }, { - "left": "issue:43927", - "right": "issue:44912", + "left": "issue:44987", + "right": "issue:45538", "accept": false, - "reason": "Different root causes: config token IDs being lost vs MXFP4 quantization fallback/loading behavior." + "reason": "Loading a specific model fails for one reason, while CLIPTokenizer max length is a separate tokenizer constant bug." }, { - "left": "issue:42915", - "right": "issue:44821", + "left": "issue:43452", + "right": "issue:43582", "accept": false, - "reason": "Different targets and failures: Qwen3Moe FP8 config vs AutoImageProcessor URL loading." + "reason": "gguf_file loading failures and Apple Silicon allocator TypeError are unrelated." }, { - "left": "issue:43257", - "right": "issue:43756", + "left": "issue:41720", + "right": "issue:43540", "accept": false, - "reason": "Distinct model-specific issues: MOE weight conversion under accelerate/deepspeed vs Smollm3 RoPE layer mismatch." + "reason": "Qwen3 auto device mapping CUDA assert and Qwen3OmniMoe video processing ValueError are different failures." }, { - "left": "issue:43577", - "right": "issue:45072", + "left": "issue:42915", + "right": "issue:43742", "accept": false, - "reason": "Both mention dtype, but they affect different models and code paths; too broad to merge." + "reason": "Qwen3Moe FP8 config failure and MobileLLM key error do not share the same underlying code path." }, { - "left": "issue:43540", - "right": "issue:44479", + "left": "issue:44291", + "right": "issue:45042", "accept": false, - "reason": "Same broad topic of video inputs, but the model families and symptoms differ; not clearly the same concrete bug." + "reason": "init_empty_weights argument handling and PIL backend torchvision dependency are separate regressions." }, { - "left": "issue:42915", - "right": "issue:45538", - "accept": false, - "reason": "Completely different scopes: FP8/Qwen3Moe loading vs CLIP tokenizer max length." + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both report the same class of v5.3.0 Qwen video-input regression; 45381 looks like a narrower manifestation of the broader 44479 bug." }, { - "left": "issue:43257", - "right": "issue:43696", + "left": "issue:33357", + "right": "issue:34567", "accept": false, - "reason": "Different underlying issue classes: MOE conversion/loading vs CUDA OOM." + "reason": "MacOS CLIP bus error and TrainerState token accounting are unrelated." }, { - "left": "issue:45137", - "right": "issue:45405", + "left": "issue:43577", + "right": "issue:45072", "accept": false, - "reason": "DeepSpeed ZeRO3 deque failure is unrelated to PEFT version bump." + "reason": "BLIP2 dtype persistence and SwitchTransformers/TimmWrapperModel bf16 mismatch are different bugs." }, { - "left": "issue:38175", - "right": "issue:44898", + "left": "issue:42915", + "right": "issue:44821", "accept": false, - "reason": "Both are image-model correctness bugs, but they concern different architectures and failure modes." + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor-from-URL are not the same issue." }, { - "left": "issue:44737", - "right": "issue:45588", + "left": "issue:42915", + "right": "issue:45538", "accept": false, - "reason": "XLNet CPU-device bug is unrelated to flash-attention sink-less model crash." + "reason": "Qwen3Moe FP8 configuration and CLIPTokenizer max_length are unrelated." }, { - "left": "issue:44062", - "right": "issue:44991", + "left": "issue:42175", + "right": "issue:43824", "accept": false, - "reason": "Tokenizer-related, but different models and insufficient evidence of the same special-token bug." + "reason": "TensorFlow backend packaging and missing Qwen2.5-VL import are different package/import problems." }, { - "left": "issue:43742", - "right": "issue:44291", + "left": "issue:43317", + "right": "issue:45538", "accept": false, - "reason": "Both are model-loading errors, but the specific causes are different (_is_hf_initialized vs MobileLLM key error)." + "reason": "device_map=auto offload failure and CLIPTokenizer model_max_length are unrelated." }, { - "left": "issue:43906", - "right": "issue:45381", + "left": "issue:43540", + "right": "issue:44560", "accept": false, - "reason": "Isolated reproduction of a different issue; not the same as the Qwen2.5-VL video regression." + "reason": "Both involve video, but they are different models and different error modes, so they are not clearly the same bug." }, { - "left": "issue:43646", - "right": "issue:43749", + "left": "issue:43927", + "right": "issue:44912", "accept": false, - "reason": "Custom model init breakage is unrelated to FSDP CPU RAM-efficient loading." + "reason": "DiaConfig save/load token IDs and MXFP4 loading fallback are different configuration bugs." }, { - "left": "issue:43749", - "right": "issue:44756", + "left": "issue:43540", + "right": "issue:44479", "accept": false, - "reason": "Different memory/loading failures with different triggers; not the same bug." + "reason": "Qwen3OmniMoe video ValueError and the broader Qwen video-input regression are too different in model family and symptom to merge confidently." }, { "left": "issue:44062", - "right": "issue:45081", + "right": "issue:44991", "accept": false, - "reason": "Tokenizer special-arg error vs Mistral regex/backend_tokenizer AttributeError are different failure points." + "reason": "AddedToken special-argument error and tokenizer loading failure for a specific model are distinct problems." }, { - "left": "issue:36010", - "right": "issue:42175", + "left": "issue:42175", + "right": "issue:44488", "accept": false, - "reason": "Import path break for GenerationMixin is unrelated to TensorFlow being absent from a torch extra install." + "reason": "TensorFlow backend packaging and model loading for cjvt/sleng-bert are unrelated." }, { - "left": "issue:34567", - "right": "issue:36331", + "left": "issue:42175", + "right": "issue:44938", "accept": false, - "reason": "TrainerState token accounting vs custom compute_loss signature mismatch are separate training issues." + "reason": "TensorFlow backend packaging and Python 3.14 loading failure are not the same bug." }, { - "left": "issue:43643", - "right": "issue:45375", + "left": "issue:45137", + "right": "issue:45405", "accept": false, - "reason": "Both concern config fields, but one is trust_remote_code config materialization and the other is a specific Qwen vision config field being dropped." + "reason": "DeepSpeed ZeRO3 deque error and PEFT version bump are unrelated." }, { - "left": "issue:43493", - "right": "issue:43994", - "accept": true, - "reason": "Both report the same SigLIP2 model discrepancy/correctness problem: HF outputs disagree with expected behavior and produce nonsensical results." - }, - { - "left": "issue:42175", - "right": "issue:43824", + "left": "issue:43526", + "right": "issue:45479", "accept": false, - "reason": "Package/dependency availability issue vs missing model class import; unrelated." + "reason": "BeitImageProcessorFast label reduction and zero-loss sequence classification are different bugs." }, { - "left": "issue:43452", - "right": "issue:44488", + "left": "issue:38175", + "right": "issue:44898", "accept": false, - "reason": "Different loading breakages for different models and APIs." + "reason": "Siglip2 zero probabilities and Perceiver non-default resolution failure are separate model-specific issues." }, { - "left": "issue:43526", - "right": "issue:45479", + "left": "issue:44737", + "right": "issue:45588", "accept": false, - "reason": "Different image/label handling issues: BeitImageProcessorFast reduction bug vs sequence-classification loss degeneracy." + "reason": "XLNet CPU tensor allocation and flash_attention sink-less AttributeError are unrelated." }, { - "left": "issue:44291", - "right": "issue:44821", + "left": "issue:43906", + "right": "issue:45381", "accept": false, - "reason": "General init_empty_weights loading bug vs AutoImageProcessor-from-URL failure are different code paths." + "reason": "A reproduction of another issue and a Qwen video-position-id regression are not the same underlying bug." }, { - "left": "issue:44792", - "right": "issue:44964", + "left": "issue:28282", + "right": "issue:34689", "accept": false, - "reason": "Janus generation test failure and Phi-4 multimodal loading failure are unrelated." + "reason": "Missing PyTorch import error and Llama 3.2 vision model loading regression are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The two accepted duplicate-like pairs are reasonably supported by the issue titles/descriptions, and the remaining soft-edge rejections are appropriately cautious. No overstatement stands out." + "feedback": "The cluster-level summary is grounded in the packet, and the only accepted soft pair (44479/45381) is a conservative, well-supported duplicate call based on the shared Qwen video-input regression theme. The rejected pairs are generally cautious and do not overclaim sameness across clearly different model/code paths. " }, "evaluator_used": true, "retried": false, @@ -5514,269 +5446,291 @@ "issue:24643", "issue:28282", "issue:29127", - "issue:34689", - "issue:38175", + "issue:34567", + "issue:36331", "issue:41762", - "issue:42175", - "issue:43295", "issue:43452", + "issue:43493", "issue:43531", "issue:43577", "issue:43582", "issue:43643", - "issue:43653", + "issue:43688", "issue:43697", "issue:43720", "issue:43742", + "issue:43749", "issue:43756", - "issue:43819", - "issue:44361", - "issue:44442", + "issue:43782", + "issue:43994", + "issue:44062", + "issue:44164", + "issue:44291", "issue:44466", "issue:44484", + "issue:44488", "issue:44514", "issue:44561", - "issue:44568", "issue:44589", "issue:44625", "issue:44704", - "issue:44811", + "issue:44792", "issue:44821", "issue:44843", "issue:44877", "issue:44898", "issue:44936", - "issue:44938", - "issue:44977", + "issue:44964", "issue:45005", "issue:45042", "issue:45081", "issue:45216", "issue:45290", "issue:45325", - "issue:45356", + "issue:45375", "issue:45381", "issue:45405", "issue:45600" ], "soft_pairs": [ - "issue:28282|issue:34689", + "issue:43749|issue:44164", + "issue:43742|issue:44291", + "issue:28282|issue:29127", + "issue:43756|issue:45325", + "issue:43582|issue:43782", + "issue:43493|issue:43994", + "issue:44062|issue:45081", + "issue:34567|issue:36331", + "issue:43643|issue:45375", + "issue:44291|issue:44821", + "issue:43452|issue:44488", + "issue:44792|issue:44964", "issue:45042|issue:45081", "issue:44484|issue:44704", "issue:45405|issue:45600", - "issue:43756|issue:45325", "issue:43742|issue:44589", + "issue:43531|issue:44821", + "issue:43688|issue:43749", "issue:43697|issue:45216", - "issue:28282|issue:29127", "issue:44514|issue:45381", - "issue:43531|issue:44821", "issue:43643|issue:44561", - "issue:43582|issue:44977", "issue:44625|issue:44877", "issue:43452|issue:43577", "issue:43720|issue:44898", "issue:44514|issue:45290", "issue:43452|issue:44843", - "issue:42175|issue:44938", - "issue:44466|issue:45005", - "issue:43819|issue:44811", - "issue:43295|issue:45356", "issue:41762|issue:44936", - "issue:43653|issue:44568", "issue:24643|issue:29127", - "issue:38175|issue:43295", - "issue:44361|issue:44442" + "issue:44466|issue:45005" ], - "prepared_review_unit_hash": "4bca7397a15a2ff095c2d0091c6669a6d94828502a7a1e11cf3547d8af6d16cf", + "prepared_review_unit_hash": "12cb4a235a328e662df6943e95021d5a14e16ecd09ebc1f171cad0268a3f6e5e", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 35227, - "estimated_input_tokens": 8807, - "estimated_eval_tokens": 17870 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36669, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and does not look like a single duplicate set. The only convincing duplicate pair is the tied-weights save_pretrained regression; the rest are unrelated tokenizer, config, model-loading, and vision bugs.", - "confidence": 0.88, - "canonical_issue_reason": "Issue 44466 is the most concrete anchor for the tied-weights regression: it describes the symptom, version boundary, and device-dependent serialization behavior more specifically than 45005.", + "summary": "This cluster is mostly heterogeneous. Nearly all soft pairs are adjacent but not true duplicates; only the two SigLIP2 reports look like the same underlying model-implementation discrepancy.", + "confidence": 0.89, + "canonical_issue_reason": "issue:43493 is the clearest, most specific report of the underlying SigLIP2 correctness problem and is the better anchor for the only plausible duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44466 is the strongest standalone issue in the cluster because it has the clearest repro target and the most actionable description of the underlying save_pretrained/tied-weights problem.", + "best_issue_reason": "issue:43493 has the strongest reproducible bug description and best matches the likely shared root cause for the SigLIP2 pair.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43749", + "right": "issue:44164", + "accept": false, + "reason": "Both involve loading/saving paths, but one is FSDP CPU RAM-efficient loading and the other is extra_state serialization; different bugs and code paths." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Both fail during model loading, but MobileLLM key errors and init_empty_weights/_is_hf_initialized errors are distinct causes." + }, { "left": "issue:28282", - "right": "issue:34689", + "right": "issue:29127", + "accept": false, + "reason": "Completely different problems: missing PyTorch import vs. a LayoutLMv3 box-validation error-message issue." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both mention RoPE/position handling, but they affect different models and different logic; not the same concrete bug." + }, + { + "left": "issue:43582", + "right": "issue:43782", "accept": false, - "reason": "Different failures: missing PyTorch import vs. model loading regression for a specific Llama vision model." + "reason": "Different failures in unrelated areas: Apple Silicon allocator warmup vs. Qwen3VL weight-only loading." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both describe SigLIP2 producing incorrect outputs relative to the expected implementation; one is the JAX-vs-HF discrepancy and the other is the same symptom in model/pipeline use." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer-related, but one is an AddedToken constructor conflict and the other is a Mistral regex patch crash; different root causes." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Both are trainer-related, but num_input_tokens_seen not updating and compute_loss signature mismatch are separate bugs." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both concern config handling, but one is generic trust_remote_code field loss and the other is a specific config field dropped by strict schema; not the same bug." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "Different loading paths and failures: init_empty_weights with an unexpected argument vs. AutoImageProcessor loading from URL." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading complaints, but gguf_file support and a specific sleng-bert load failure are not the same underlying issue." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Different multimodal model failures; a failing janus test case is not the same as loading Phi-4 multimodal instruct." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "Unrelated subsystems: image processor torchvision dependency vs. Mistral tokenizer regex crash." + "reason": "Unrelated bugs: PIL image processors wrongly requiring torchvision vs. a tokenizer regex crash." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "One is a save_pretrained shard-size question; the other is a cached_file kwargs propagation bug." + "reason": "Save-sharding defaults and AutoProcessor kwargs forwarding are unrelated." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "Release/versioning problem vs. broken auto_mappings references; not the same bug." - }, - { - "left": "issue:43756", - "right": "issue:45325", - "accept": false, - "reason": "Both involve positional/RoPE logic, but they affect different models and different code paths." + "reason": "Unreleased PEFT version pinning and removed auto_mappings config references are separate release/CI issues." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "Different loading/runtime failures: missing key for MobileLLM vs. Float8 storage lookup error." + "reason": "Different load-time errors with different causes and affected code paths." }, { - "left": "issue:43697", - "right": "issue:45216", + "left": "issue:43531", + "right": "issue:44821", "accept": false, - "reason": "Different regression surfaces: output drift in RTDetrV2 vs. incorrect checkpoint serialization for Qwen3.5." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to AutoImageProcessor URL loading." }, { - "left": "issue:28282", - "right": "issue:29127", + "left": "issue:43688", + "right": "issue:43749", "accept": false, - "reason": "ImportError for PyTorch is unrelated to the LayoutLMv3 box-validation error message issue." + "reason": "Both touch training/loading infrastructure, but auxiliary-loss normalization and FSDP loading are unrelated bugs." }, { - "left": "issue:44514", - "right": "issue:45381", + "left": "issue:43697", + "right": "issue:45216", "accept": false, - "reason": "Both are Qwen2.5-VL chat/vision bugs, but one is batched padding handling and the other is wrong video position ids." + "reason": "Different regressions in different models: RTDetrV2 output changes vs. Qwen3.5 checkpoint save_pretrained behavior." }, { - "left": "issue:43531", - "right": "issue:44821", + "left": "issue:44514", + "right": "issue:45381", "accept": false, - "reason": "Sliding-window behavior for Qwen3-MoE is unrelated to loading an AutoImageProcessor from a URL." + "reason": "Both are Qwen2.5-VL issues, but one is apply_chat_template batching and the other is video position IDs; not the same bug." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Both involve trust_remote_code, but one is missing fields in AutoConfig and the other is a removed helper breaking imports." - }, - { - "left": "issue:43582", - "right": "issue:44977", - "accept": false, - "reason": "AppleSilicon allocator TypeError is unrelated to Qwen3.5 flash-attention generation issues." + "reason": "Both affect trust_remote_code workflows, but missing config fields and removal of is_torch_fx_available are different failure modes." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Config propagation of num_labels is a different problem from strict config blocking granite_speech loading." + "reason": "Both involve config propagation/strictness, but they affect different models and different mechanisms." }, { "left": "issue:43452", "right": "issue:43577", "accept": false, - "reason": "gguf_file loading breakage and Blip2 dtype propagation are different model-loading bugs." + "reason": "gguf_file loading failures and Blip2 dtype propagation are unrelated." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "Packed-weight unpacking during accelerate loading is unrelated to Perceiver interpolation failure." + "reason": "BitNet packed-weight loading and Perceiver interpolation behavior are unrelated code paths." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Both hit apply_chat_template, but the triggers and failure modes differ enough that they are not the same bug." + "reason": "Same public API, but the crashes are on different inputs and edge cases; too broad to treat as one bug." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "Different failure points: gguf_file loading vs. unconditional model_info() call breaking offline mode." - }, - { - "left": "issue:42175", - "right": "issue:44938", - "accept": false, - "reason": "Packaging/backend dependency issue is unrelated to a Python 3.14 load failure." - }, - { - "left": "issue:44466", - "right": "issue:45005", - "accept": true, - "reason": "Both describe the same tied-weights/save_pretrained regression in v5, with 44466 giving the clearest device-dependent serialization symptom." - }, - { - "left": "issue:43819", - "right": "issue:44811", - "accept": false, - "reason": "DAC latent conversion mismatch and Whisper batch_decode skip_special_tokens behavior are unrelated." - }, - { - "left": "issue:43295", - "right": "issue:45356", - "accept": false, - "reason": "Different tokenizer regressions with different symptoms and likely different code paths." + "reason": "gguf_file loading issues and offline mistral-regex patch behavior are unrelated." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "DeepSpeed ZeRO-3 loading IndexError is unrelated to trainer.evaluate() failing after trainer.train()." - }, - { - "left": "issue:43653", - "right": "issue:44568", - "accept": false, - "reason": "Different tokenizer/special-token bugs affecting different models and behaviors." + "reason": "Both are training/loading related, but Gemma3 DeepSpeed ZeRO-3 loading and trainer.evaluate() after train() are distinct problems." }, { "left": "issue:24643", "right": "issue:29127", "accept": false, - "reason": "DeepSpeed weight-dimension RuntimeError is unrelated to LayoutLMv3 box-information validation." + "reason": "Completely different issues: DeepSpeed weight-shape runtime error vs. LayoutLMv3 box-validation messaging." }, { - "left": "issue:38175", - "right": "issue:43295", - "accept": false, - "reason": "Zero-probabilities output issue is unrelated to processor/tokenizer regression." - }, - { - "left": "issue:44361", - "right": "issue:44442", + "left": "issue:44466", + "right": "issue:45005", "accept": false, - "reason": "Different tokenizer load failures for different model families." + "reason": "Both concern tied weights, but one is inconsistent serialization by device and the other is translation-model tied-weight issues; not enough evidence for same bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is conservative and grounded in the packet: the tied-weights save_pretrained pair (44466/45005) is the only clearly mergeable duplicate, and the other pairwise rejections stay within the evidence suggested by the titles. No soft-edge verdict appears to overstate sameness." + "feedback": "The summary is grounded in the packet, and the soft-edge calls stay conservative. The only accepted pair, issue:43493 with issue:43994, is a reasonable same-bug match based on the shared SigLIP2 output-discrepancy symptom. The rejected pairs are mostly justified by distinct titles/bug classes without overstating evidence." }, "evaluator_used": true, "retried": false, @@ -5800,313 +5754,341 @@ "issue:28282", "issue:30064", "issue:32090", + "issue:33357", "issue:33666", "issue:34567", "issue:35141", - "issue:36010", + "issue:38175", "issue:41720", "issue:43122", - "issue:43262", + "issue:43295", "issue:43329", "issue:43452", - "issue:43493", "issue:43504", - "issue:43525", "issue:43531", "issue:43550", - "issue:43572", "issue:43577", + "issue:43646", "issue:43653", "issue:43688", "issue:43697", "issue:43720", "issue:43742", "issue:43749", - "issue:43756", - "issue:43825", + "issue:43819", "issue:43906", "issue:43927", "issue:44361", + "issue:44442", "issue:44451", "issue:44464", "issue:44479", + "issue:44554", + "issue:44568", "issue:44610", + "issue:44625", "issue:44743", - "issue:44821", + "issue:44811", "issue:44843", - "issue:44849", - "issue:44913", "issue:44936", "issue:44977", "issue:44991", - "issue:45084", "issue:45325", "issue:45356", - "issue:45357", "issue:45440", - "issue:45588" + "issue:45588", + "issue:45701", + "issue:45704" ], "soft_pairs": [ - "issue:30064|issue:33666", - "issue:43653|issue:44361", + "issue:43295|issue:45356", + "issue:43819|issue:44811", + "issue:43653|issue:44568", "issue:44464|issue:44743", - "issue:43577|issue:43697", - "issue:44843|issue:44991", + "issue:43688|issue:45440", + "issue:43688|issue:45704", "issue:41720|issue:43906", + "issue:38175|issue:43295", + "issue:44361|issue:44442", + "issue:43653|issue:44361", + "issue:30064|issue:33666", + "issue:28282|issue:32090", + "issue:43646|issue:43749", + "issue:43577|issue:43697", + "issue:43688|issue:45701", "issue:44479|issue:44936", + "issue:28282|issue:33357", + "issue:44843|issue:44991", + "issue:28282|issue:35141", "issue:44977|issue:45356", - "issue:43688|issue:45440", "issue:43577|issue:45588", "issue:43531|issue:43742", "issue:43653|issue:43927", + "issue:43329|issue:45325", + "issue:43122|issue:43906", "issue:44464|issue:44977", + "issue:43577|issue:44554", "issue:43550|issue:43927", + "issue:44479|issue:44625", "issue:43550|issue:44361", - "issue:43122|issue:43906", "issue:43504|issue:43720", "issue:32090|issue:34567", "issue:43504|issue:43550", + "issue:41720|issue:44464", "issue:32090|issue:33666", - "issue:28282|issue:35141", "issue:43504|issue:44610", - "issue:43452|issue:44451", - "issue:43756|issue:44913", - "issue:43329|issue:45325", - "issue:43742|issue:44821", - "issue:43688|issue:43749", - "issue:43525|issue:43572", - "issue:41720|issue:44464", - "issue:33666|issue:36010", - "issue:43493|issue:43825", - "issue:43262|issue:45084", - "issue:44849|issue:45357" + "issue:43452|issue:44451" ], - "prepared_review_unit_hash": "40735fe0efd11e59b73ef79c443cee48b614b9b4bb563b0e373d2f06ab3d48d7", + "prepared_review_unit_hash": "8293159639519b8ead9728c9615676121b2713b88a74606f95eefcfc08ff2841", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37710, - "estimated_input_tokens": 9428, - "estimated_eval_tokens": 19112 + "soft_pair_count": 37, + "serialized_chars": 38916, + "estimated_input_tokens": 9729, + "estimated_eval_tokens": 19714 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, trainer, vision, config, and generation regressions. None of the soft pairs look like the same underlying bug, so no dedup merges are justified.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue stands out; the items span multiple unrelated subsystems and model families, and the candidate pairs are not close enough to be duplicates.", + "summary": "This set is not a single duplicate cluster; it mixes many unrelated regressions across tokenization, loading, training, vision, and compilation. If one issue must represent the cluster, the broad tokenizer-regression report is the best umbrella, but the soft pairs do not look like true duplicates.", + "confidence": 0.31, + "canonical_issue_reason": "No strong duplicate anchor exists. Issue 45701 is the broadest representative of the recurring version-dependent tokenization complaints and has the most cross-references, so it is the best umbrella issue if one must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44843 is the clearest self-contained regression report, but it is still too specific to serve as a true canonical for this mixed cluster.", + "best_issue_reason": "45701 is the most general and central issue in this mixed set, so it is the best single triage representative even though it is not a confirmed duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:30064", - "right": "issue:33666", + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different behaviors and code paths: processor/tokenizer access vs codec/regex handling." + }, + { + "left": "issue:43819", + "right": "issue:44811", "accept": false, - "reason": "Image processor segmentation-map failure vs Qwen2-VL multi-GPU training; different models and failure modes." + "reason": "DAC latent decoding and Whisper batch decoding are unrelated features with different failure modes." }, { "left": "issue:43653", - "right": "issue:44361", + "right": "issue:44568", "accept": false, - "reason": "BigBirdTokenizer special-token decode bug vs MLukeTokenizer task AttributeError; not the same code path." + "reason": "Both mention special-token behavior, but one is BigBird decode registration and the other is mDeBERTa BOS/EOS insertion." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Both affect generation, but one is compiled-forward inconsistency and the other is recurrent-state reset with cache; distinct bugs." + "reason": "Chunked generation with compiled forward is a different bug from recurrent-state reset in modular_qwen3_5." }, { - "left": "issue:43577", - "right": "issue:43697", + "left": "issue:43688", + "right": "issue:45440", "accept": false, - "reason": "Blip2 dtype-loading issue vs RTDetrV2 output regression; unrelated subsystems and symptoms." + "reason": "Aux-loss normalization and DeepseekV3MoE divergence are different model/math bugs." }, { - "left": "issue:44843", - "right": "issue:44991", + "left": "issue:43688", + "right": "issue:45704", "accept": false, - "reason": "Both are tokenizer-load regressions, but one is an offline model_info call in Mistral regex patching and the other is a specific tokenizer load failure for a different model." + "reason": "Aux-loss normalization is unrelated to the T5 apex RMSNorm memory leak." }, { "left": "issue:41720", "right": "issue:43906", "accept": false, - "reason": "Insufficient evidence of the same defect; one is an A800 device-map cuda assert, the other is only a reproduction of another issue." + "reason": "Qwen3 auto device mapping cuda asserts are not the same as the isolated reproduction issue linked to another bug." }, { - "left": "issue:44479", - "right": "issue:44936", + "left": "issue:38175", + "right": "issue:43295", "accept": false, - "reason": "Video-input regression for several Qwen VL models vs trainer evaluate/train lifecycle failure; unrelated." + "reason": "Siglip2 zero probabilities and processor.tokenizer regression are different problems." }, { - "left": "issue:44977", - "right": "issue:45356", + "left": "issue:44361", + "right": "issue:44442", "accept": false, - "reason": "Flash-attention generation bug vs tokenizer codec/regex regression; different layers of the stack." + "reason": "MLukeTokenizer task errors and AutoTokenizer failing on FastSpeech2Conformer are distinct loading/attribute issues." }, { - "left": "issue:43688", - "right": "issue:45440", + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBird special-token decode and MLukeTokenizer AttributeError are unrelated tokenizer bugs." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Void segmentation maps in an image processor and Qwen2-VL multi-GPU training are different subsystems." + }, + { + "left": "issue:28282", + "right": "issue:32090", "accept": false, - "reason": "Both involve MoE models, but one is auxiliary-loss normalization and the other is native-vs-remote implementation divergence; not mergeable into one fix." + "reason": "Missing PyTorch import error and Trainer _gpu_broadcast_one NoneType error are unrelated runtime failures." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Custom model initialization breakage and FSDP CPU RAM efficient loading are different initialization/loading regressions." }, { "left": "issue:43577", - "right": "issue:45588", + "right": "issue:43697", "accept": false, - "reason": "Dtype propagation bug vs flash_attention crash on s_aux=None; different failure mechanisms." + "reason": "Blip2 dtype propagation and RTDetrV2 output drift are separate model-specific bugs." }, { - "left": "issue:43531", - "right": "issue:43742", + "left": "issue:43688", + "right": "issue:45701", "accept": false, - "reason": "Qwen3-MoE sliding_window issue vs MobileLLM load KeyError; unrelated." + "reason": "Only a broad regression theme overlaps; aux-loss normalization is not the same as tokenization changing across versions." }, { - "left": "issue:43653", - "right": "issue:43927", + "left": "issue:44479", + "right": "issue:44936", "accept": false, - "reason": "Both are token-related, but one is mask-token registration and the other is saved custom token IDs causing generation IndexError; distinct bugs." + "reason": "Video-input regression and trainer.evaluate after trainer.train affect different execution paths." }, { - "left": "issue:44464", - "right": "issue:44977", + "left": "issue:28282", + "right": "issue:33357", "accept": false, - "reason": "Compiled-forward inconsistency vs flash-attention generation failure; different concrete problems." + "reason": "PyTorch missing import and MacOS bus error are different environment/runtime issues." }, { - "left": "issue:43550", - "right": "issue:43927", + "left": "issue:44843", + "right": "issue:44991", "accept": false, - "reason": "torch.compile/SDPA failure vs config save/load token-ID loss; unrelated." + "reason": "HF_HUB_OFFLINE patch behavior and tokenizer loading for a specific model are different tokenizer-loading bugs." }, { - "left": "issue:43550", - "right": "issue:44361", + "left": "issue:28282", + "right": "issue:35141", "accept": false, - "reason": "Bamba compile+SDPA bug vs MLukeTokenizer AttributeError; no shared underlying issue." + "reason": "ImportError for PyTorch and post_init reinitializing embeddings are unrelated." }, { - "left": "issue:43122", - "right": "issue:43906", + "left": "issue:44977", + "right": "issue:45356", "accept": false, - "reason": "Different tokenization regression vs isolated reproduction of another issue; no evidence they are the same bug." + "reason": "Flash-attention generation issues and Kimi tokenizer codec handling are different regressions." }, { - "left": "issue:43504", - "right": "issue:43720", + "left": "issue:43577", + "right": "issue:45588", "accept": false, - "reason": "BEiT legacy-field load bug vs BitNet packed-weight unpacking bug; different loading paths." + "reason": "Blip2 dtype mismatch and flash_attention.py crashing on s_aux=None are unrelated." }, { - "left": "issue:32090", - "right": "issue:34567", + "left": "issue:43531", + "right": "issue:43742", "accept": false, - "reason": "Trainer broadcast NoneType error vs token-count bookkeeping not updating; both trainer-related but not the same defect." + "reason": "Qwen3-MoE sliding_window behavior and a MobileLLM key error are different bugs." }, { - "left": "issue:43504", - "right": "issue:43550", + "left": "issue:43653", + "right": "issue:43927", "accept": false, - "reason": "Legacy-field pretrained-model load bug vs Bamba torch.compile+SDPA failure; unrelated." + "reason": "BigBird special-token decode and DiaConfig losing custom token IDs are both token-related, but the root causes differ." }, { - "left": "issue:32090", - "right": "issue:33666", + "left": "issue:43329", + "right": "issue:45325", "accept": false, - "reason": "Trainer _gpu_broadcast_one TypeError vs Qwen2-VL multi-GPU training; unrelated." + "reason": "Undefined video helper names and rope index temporal scaling are separate video-VLM bugs." }, { - "left": "issue:28282", - "right": "issue:35141", + "left": "issue:43122", + "right": "issue:43906", "accept": false, - "reason": "Missing PyTorch import vs embedding resize reinitialization after post_init; different concerns entirely." + "reason": "Tokenization differences across versions and an isolated reproduction of another issue are not the same bug." }, { - "left": "issue:43504", - "right": "issue:44610", + "left": "issue:44464", + "right": "issue:44977", "accept": false, - "reason": "BEiT preset legacy-field bug vs OmDet-Turbo processor size mismatch; different models and code paths." + "reason": "Compiled-forward chunked generation and flash-attention generation are different failures." }, { - "left": "issue:43452", - "right": "issue:44451", + "left": "issue:43577", + "right": "issue:44554", "accept": false, - "reason": "gguf_file loading breakage vs a specific model load failure; both are load issues but not the same one." + "reason": "Blip2 dtype handling and an MPS attention correctness issue are unrelated." }, { - "left": "issue:43756", - "right": "issue:44913", + "left": "issue:43550", + "right": "issue:43927", "accept": false, - "reason": "RoPE layer-dropping bug vs rotary_pct reload bug; both config-related but clearly different regressions." + "reason": "Bamba torch.compile SDPA failure and DiaConfig token ID loss are different code paths." }, { - "left": "issue:43329", - "right": "issue:45325", + "left": "issue:44479", + "right": "issue:44625", "accept": false, - "reason": "Undefined video-branch helpers in multimodal token counting vs still-image temporal position scaling in Qwen2.5-VL; distinct code paths." + "reason": "Video-input regression and Qwen3.5 num_labels propagation are unrelated." }, { - "left": "issue:43742", - "right": "issue:44821", + "left": "issue:43550", + "right": "issue:44361", "accept": false, - "reason": "Model KeyError on load vs AutoImageProcessor URL-loading failure; unrelated." + "reason": "Bamba compile/S D P A issues and MLukeTokenizer AttributeError are unrelated." }, { - "left": "issue:43688", - "right": "issue:43749", + "left": "issue:43504", + "right": "issue:43720", "accept": false, - "reason": "Aux-loss normalization bug vs FSDP CPU RAM efficient loading breakage; different distributed-training issues." + "reason": "Beit legacy-field loading and BitNet packed-weight unpacking are different loading bugs." }, { - "left": "issue:43525", - "right": "issue:43572", + "left": "issue:32090", + "right": "issue:34567", "accept": false, - "reason": "Missing pad_token_id in Llama4Config vs missing pad_token_idx in StableLmConfig; similar theme, but different model/config regressions." + "reason": "Trainer broadcast NoneType and TrainerState token counter not updating are different trainer issues." }, { - "left": "issue:41720", - "right": "issue:44464", + "left": "issue:43504", + "right": "issue:43550", "accept": false, - "reason": "A800 auto device-map cuda assert vs chunked generation inconsistency with compiled forward; unrelated." + "reason": "Beit pretrained loading and Bamba compile/S D P A are unrelated." }, { - "left": "issue:33666", - "right": "issue:36010", + "left": "issue:41720", + "right": "issue:44464", "accept": false, - "reason": "Qwen2-VL multi-GPU training issue vs GenerationMixin import error; unrelated." + "reason": "Qwen3 auto device mapping asserts and chunked generation with compiled forward are different Qwen3 bugs." }, { - "left": "issue:43493", - "right": "issue:43825", + "left": "issue:32090", + "right": "issue:33666", "accept": false, - "reason": "SigLIP2 implementation discrepancy vs pipeline translation-task error message; different bugs." + "reason": "Trainer NoneType broadcast and Qwen2-VL multi-GPU training are different domains." }, { - "left": "issue:43262", - "right": "issue:45084", + "left": "issue:43504", + "right": "issue:44610", "accept": false, - "reason": "Audio processor sampling-rate default bug vs template compilation TypeError; unrelated." + "reason": "Legacy-field loading in Beit and OmDet processor size mismatch are unrelated." }, { - "left": "issue:44849", - "right": "issue:45357", + "left": "issue:43452", + "right": "issue:44451", "accept": false, - "reason": "Qwen3.5 output_hidden_states regression vs save_pretrained writing wrong visual-encoder keys; different parts of the model lifecycle." + "reason": "gguf_file breaking tokenizer/model loading and inability to load ScandiBERT are both loading failures, but not the same mechanism or model-specific bug." } ] }, @@ -6130,323 +6112,309 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:28282", - "issue:32090", "issue:33357", "issue:33666", + "issue:36010", "issue:36331", "issue:39692", "issue:43257", + "issue:43262", "issue:43278", "issue:43295", "issue:43299", "issue:43334", "issue:43493", "issue:43504", + "issue:43525", "issue:43540", "issue:43572", - "issue:43575", - "issue:43577", "issue:43606", "issue:43646", - "issue:43653", "issue:43688", "issue:43716", "issue:43720", + "issue:43742", "issue:43756", "issue:43824", + "issue:43825", "issue:43844", "issue:43927", - "issue:44291", "issue:44361", "issue:44423", "issue:44466", - "issue:44479", "issue:44488", - "issue:44554", "issue:44589", "issue:44625", "issue:44734", + "issue:44821", "issue:44843", + "issue:44849", + "issue:44913", "issue:44964", "issue:44977", "issue:45083", "issue:45200", "issue:45229", + "issue:45357", "issue:45405", "issue:45520", - "issue:45701", - "issue:45704", + "issue:45584", "issue:45759" ], "soft_pairs": [ + "issue:43278|issue:43688", + "issue:43742|issue:44821", + "issue:43756|issue:44913", + "issue:43525|issue:43572", + "issue:33666|issue:36010", + "issue:43493|issue:43825", + "issue:44849|issue:45357", + "issue:43262|issue:43295", "issue:43540|issue:43716", "issue:43720|issue:43927", "issue:43606|issue:43927", "issue:43334|issue:43572", - "issue:28282|issue:32090", - "issue:28282|issue:33357", - "issue:43577|issue:44554", - "issue:43688|issue:45704", - "issue:39692|issue:43493", + "issue:43278|issue:43844", "issue:33666|issue:36331", + "issue:43493|issue:43844", + "issue:39692|issue:43493", "issue:43334|issue:45759", "issue:43720|issue:44361", + "issue:43257|issue:43299", "issue:44964|issue:45405", - "issue:44479|issue:44625", "issue:43540|issue:44977", + "issue:44625|issue:45584", "issue:44488|issue:44843", - "issue:43493|issue:43844", "issue:43756|issue:45083", "issue:45200|issue:45229", - "issue:43257|issue:43299", - "issue:43278|issue:43844", "issue:43824|issue:44589", "issue:43646|issue:44466", - "issue:39692|issue:43295", "issue:44589|issue:45520", - "issue:43688|issue:45701", + "issue:39692|issue:43295", "issue:33357|issue:33666", "issue:43504|issue:43606", - "issue:44423|issue:44734", - "issue:43572|issue:44291", - "issue:43927|issue:44361", - "issue:43504|issue:43653", - "issue:44625|issue:45759", - "issue:43575|issue:45405" + "issue:44423|issue:44734" ], - "prepared_review_unit_hash": "0cf7a325194358c5398ae78978d12c74257fd0d106a55629ae6611e78624f5bc", + "prepared_review_unit_hash": "c4c876c1389763b07ff3e169931712d8c220928ef53513b6bf12360af585ce64", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 38205, - "estimated_input_tokens": 9552, - "estimated_eval_tokens": 19360 + "soft_pair_count": 32, + "serialized_chars": 37450, + "estimated_input_tokens": 9363, + "estimated_eval_tokens": 18982 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous mix of unrelated bugs (model loading, tokenizer/config regressions, dtype/device issues, training/serving failures). None of the soft pairs looks like the same underlying defect under a strict duplicate standard.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue: the cluster is too heterogeneous for duplicate consolidation, with only a few loose Qwen/config-loading themes.", + "summary": "This cluster is heterogeneous: it mixes many unrelated Transformers bugs across different models, configs, serving paths, and docs. None of the soft-edge pairs look like true duplicates, so I would not merge any of them.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out: the cluster spans unrelated regressions and model-specific bugs, so none is a safe representative.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue; at most, issue 45759 is the closest representative of a small Qwen multimodal config-loading subgroup, but it does not cover the rest of the cluster.", + "best_issue_reason": "No clear best issue for deduplication; several Qwen-related reports recur, but they target different code paths and failure modes.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Different bugs: dtype drift in evaluation vs auxiliary-loss normalization." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "Unrelated failures: MobileLLM load key error vs loading AutoImageProcessor from URL." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "Both mention RoPE, but one is a model design bug and the other is config reload serialization." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Same broad theme of missing token IDs, but different models and fields; not the same bug." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training issue and GenerationMixin import error are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy vs pipeline translation error messaging; different issues." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Hidden-states bug and save_pretrained visual-key bug are separate regressions." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio chat-template sampling-rate default and processor/tokenizer regression are different code paths." + }, { "left": "issue:43540", "right": "issue:43716", "accept": false, - "reason": "Different models and failures: Qwen3OmniMoe video-processing ValueError vs Mistral-3 image-preprocessor dtype mismatch." + "reason": "Video-input processing failure vs image preprocessor/model dtype mismatch; not duplicates." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "Unrelated code paths: packed-weight loading in BitNet vs DiaConfig token-id persistence/generation IndexError." + "reason": "Packed-weight unpacking during accelerate loading is unrelated to DiaConfig token-ID loss." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch is unrelated to DiaConfig save/load token ID loss." + "reason": "CPU offload device mismatch and config token-ID loss are different problems." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Both mention missing pad token fields, but they affect different models/configs and likely different fixes." - }, - { - "left": "issue:28282", - "right": "issue:32090", - "accept": false, - "reason": "ImportError about missing PyTorch is unrelated to Trainer broadcast TypeError." + "reason": "Different models and different missing config fields; not the same defect." }, { - "left": "issue:28282", - "right": "issue:33357", + "left": "issue:43278", + "right": "issue:43844", "accept": false, - "reason": "Different failures: missing PyTorch ImportError vs MacOS bus error with a CLIP model." + "reason": "Embedding dtype change and ZeRO-3 gradient growth are unrelated." }, { - "left": "issue:43577", - "right": "issue:44554", + "left": "issue:33666", + "right": "issue:36331", "accept": false, - "reason": "Different bug classes: BLIP2 dtype propagation vs MPS attention correctness when head dims differ." + "reason": "Training setup issue vs trainer signature incompatibility; distinct bugs." }, { - "left": "issue:43688", - "right": "issue:45704", + "left": "issue:43493", + "right": "issue:43844", "accept": false, - "reason": "Auxiliary-loss normalization in MoE models is unrelated to T5/apex RMSNorm memory leak." + "reason": "SigLIP2 discrepancy vs DeepSpeed gradient abnormality are unrelated." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "Doc example errors are not the same as HF-vs-JAX implementation discrepancy for SigLIP2." - }, - { - "left": "issue:33666", - "right": "issue:36331", - "accept": false, - "reason": "Multi-GPU training request is not the same as CustomTrainer compute_loss signature regression." + "reason": "Docs/example errors are not the same as a core HF-vs-JAX implementation discrepancy." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Both are Qwen multimodal config issues, but one is missing pad_token_id and the other is from_config text_config unwrapping; not the same bug." + "reason": "Both concern Qwen multimodal configs, but one is missing pad_token_id on load and the other is from_config unwrapping text_config." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "BitNet accelerate-loading issue is unrelated to MLukeTokenizer AttributeError." + "reason": "BitNet packed-weight loading and MLukeTokenizer AttributeError are unrelated." }, { - "left": "issue:44964", - "right": "issue:45405", + "left": "issue:43257", + "right": "issue:43299", "accept": false, - "reason": "Model loading regression for Phi-4 multimodal is unrelated to an unreleased PEFT minimum-version bump." + "reason": "Both are Qwen3 MoE-related, but one is accelerate/deepspeed weight conversion and the other is dev-version model loading." }, { - "left": "issue:44479", - "right": "issue:44625", + "left": "issue:44964", + "right": "issue:45405", "accept": false, - "reason": "Video-input regression across Qwen variants is distinct from num_labels propagation in Qwen3.5 config." + "reason": "Model load failure and a dependency version pin issue are unrelated." }, { "left": "issue:43540", "right": "issue:44977", "accept": false, - "reason": "Different Qwen3.5/Qwen3OmniMoe video-processing failure vs flash-attention generation bug." + "reason": "Video-input processing bug vs flash-attention generation bug; different failures." }, { - "left": "issue:44488", - "right": "issue:44843", + "left": "issue:44625", + "right": "issue:45584", "accept": false, - "reason": "Loading a specific model is unrelated to offline-mode breakage in AutoTokenizer._patch_mistral_regex." + "reason": "num_labels propagation and Whisper empty-transcription generation are unrelated." }, { - "left": "issue:43493", - "right": "issue:43844", + "left": "issue:44488", + "right": "issue:44843", "accept": false, - "reason": "SigLIP2 implementation mismatch is unrelated to ZeRO-3 gradient inflation." + "reason": "Model loading failure and offline-mode hub lookup bug are separate issues." }, { "left": "issue:43756", "right": "issue:45083", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch is unrelated to a helper-function behavior issue in qwen3_omni_moe." + "reason": "RoPE-layer bug and helper length computation bug are different code paths." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Text-only fine-tuning token-id defaulting is unrelated to Gemma4 multi-GPU CUDA OOM." - }, - { - "left": "issue:43257", - "right": "issue:43299", - "accept": false, - "reason": "Both are Qwen MOE-related, but one is weight conversion under accelerate/deepspeed and the other is composite model loading; not the same defect." - }, - { - "left": "issue:43278", - "right": "issue:43844", - "accept": false, - "reason": "Embedding dtype drift during eval is unrelated to gradient growth under HfDeepSpeedConfig + ZeRO-3." + "reason": "Text-only token-type default and multi-GPU OOM are not the same defect." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Missing Qwen2.5-VL export is unrelated to Float8 storage lookup failure." + "reason": "Import error for Qwen2.5-VL and float8 storage lookup error are unrelated." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model init breakage in v5 is unrelated to lm_head.weight serialization inconsistency." - }, - { - "left": "issue:39692", - "right": "issue:43295", - "accept": false, - "reason": "A broken doc example is not the same as a regression in custom model code using processor.tokenizer and image passing." + "reason": "Custom model initialization regression and lm_head serialization inconsistency are different bugs." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Different import/runtime problems: missing Float8 storage class vs flash_attn KeyError on Python 3.13." + "reason": "Float8 storage lookup and flash_attn KeyError on Python 3.13 are unrelated." }, { - "left": "issue:43688", - "right": "issue:45701", + "left": "issue:39692", + "right": "issue:43295", "accept": false, - "reason": "MoE auxiliary-loss normalization is unrelated to tokenization differences across transformer versions." + "reason": "Documentation example problems do not match the processor/tokenizer regression." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "MacOS bus error in CLIP loading is unrelated to Qwen2-VL multi-GPU training." + "reason": "MacOS CLIP bus error and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "BEiT legacy-field loading failure is unrelated to suno/bark-small CPU offload device mismatch." + "reason": "BEiT legacy-field load failure and bark CPU-offload device mismatch are distinct bugs." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both hit serve, but one is continuous-batching multimodal crash and the other is KV-cache continuation tensor indexing; different bugs." - }, - { - "left": "issue:43572", - "right": "issue:44291", - "accept": false, - "reason": "Missing pad_token_idx after 5.0 update is unrelated to init_empty_weights rejecting _is_hf_initialized." - }, - { - "left": "issue:43927", - "right": "issue:44361", - "accept": false, - "reason": "DiaConfig token ID persistence bug is unrelated to MLukeTokenizer task AttributeError." - }, - { - "left": "issue:43504", - "right": "issue:43653", - "accept": false, - "reason": "BEiT legacy-field loading failure is unrelated to BigBirdTokenizer special-token registration." - }, - { - "left": "issue:44625", - "right": "issue:45759", - "accept": false, - "reason": "Both concern Qwen3.5 config handling, but one is num_labels propagation and the other is from_config text_config unwrapping; not the same underlying issue." - }, - { - "left": "issue:43575", - "right": "issue:45405", - "accept": false, - "reason": "Tensor-parallel OOM for a large model is unrelated to a PEFT version requirement bump." + "reason": "Serve multimodal crash from string tensor handling and KV-cache continuation indexing bug are different failures." } ] }, @@ -6479,21 +6447,24 @@ "issue:43232", "issue:43278", "issue:43295", - "issue:43334", "issue:43493", - "issue:43525", + "issue:43504", "issue:43540", + "issue:43572", + "issue:43575", + "issue:43582", "issue:43606", - "issue:43688", + "issue:43653", "issue:43701", "issue:43844", - "issue:43873", + "issue:43927", "issue:43957", "issue:43994", "issue:44112", + "issue:44291", "issue:44315", + "issue:44361", "issue:44464", - "issue:44466", "issue:44493", "issue:44521", "issue:44554", @@ -6501,229 +6472,228 @@ "issue:44610", "issue:44792", "issue:44849", - "issue:44898", "issue:44964", + "issue:44977", "issue:45072", "issue:45084", - "issue:45127", "issue:45198", "issue:45200", "issue:45290", "issue:45325", "issue:45356", "issue:45405", - "issue:45412", "issue:45446", "issue:45520", "issue:45584", "issue:45701" ], "soft_pairs": [ + "issue:43572|issue:44291", + "issue:43927|issue:44361", + "issue:43504|issue:43653", + "issue:43575|issue:45405", + "issue:44464|issue:45584", + "issue:43994|issue:44493", + "issue:43278|issue:43701", "issue:44792|issue:45405", "issue:43606|issue:44112", "issue:44521|issue:45290", - "issue:43994|issue:44493", "issue:29942|issue:30333", - "issue:44464|issue:45584", "issue:43844|issue:44964", "issue:45084|issue:45520", "issue:44554|issue:45446", "issue:44568|issue:45701", - "issue:43688|issue:43873", - "issue:43540|issue:45200", "issue:15354|issue:38175", + "issue:43540|issue:45200", + "issue:43582|issue:44977", "issue:45072|issue:45198", "issue:43232|issue:44849", + "issue:37428|issue:43957", "issue:39401|issue:44568", - "issue:43278|issue:43701", "issue:44315|issue:44849", "issue:44554|issue:45356", - "issue:43295|issue:44610", - "issue:45200|issue:45325", - "issue:43334|issue:43525", - "issue:44898|issue:45412", - "issue:44466|issue:45127", "issue:43493|issue:43701", - "issue:37428|issue:43957" + "issue:43295|issue:44610", + "issue:45200|issue:45325" ], - "prepared_review_unit_hash": "1e34de14281a8fa704959989494164d873a1d9df73535df32823468d9d310a7e", + "prepared_review_unit_hash": "4be067b8df5b74014bed75d52f77247e9f4c83cc1cf62c91e30507cbb7b2b96a", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 35092, - "estimated_input_tokens": 8773, - "estimated_eval_tokens": 17802 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35863, + "estimated_input_tokens": 8966, + "estimated_eval_tokens": 18188 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is a loose grab bag of unrelated Transformers issues: tokenization regressions, generation bugs, multimodal processor failures, device/offload problems, and CI/test noise. The proposed soft links are mostly superficial theme matches rather than the same underlying bug.", - "confidence": 0.91, + "summary": "This cluster is highly heterogeneous: tokenizer regressions, generation bugs, loading/config issues, device/dtype mismatches, and unrelated CI failures are all mixed together. I do not see a single underlying bug/change unifying the items, so the soft edges should all be rejected.", + "confidence": 0.14, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If one issue must represent the set, issue:45701 is the broadest/meta report ('transformers version changes the tokenization') and best captures the recurring version-regression theme, though it is not a true duplicate of the others.", + "best_issue_reason": "No true canonical issue stands out because the cluster is noisy; if a placeholder is needed, issue:44521 is the most discussion-heavy and broadly representative of a multimodal chat-template regression, but it is not a real duplicate hub for the rest of the items.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44792", - "right": "issue:45405", + "left": "issue:43572", + "right": "issue:44291", "accept": false, - "reason": "Janus image-generation test failure vs an unreleased PEFT-version constraint; unrelated bugs." + "reason": "Both are transformer 5.x loading regressions, but one is a StableLmConfig padding-field issue and the other is an init_empty_weights/_is_hf_initialized constructor mismatch; different concrete code paths." }, { - "left": "issue:43606", - "right": "issue:44112", + "left": "issue:43927", + "right": "issue:44361", "accept": false, - "reason": "Both are CI/device-related, but one is a CPU offload mismatch on Bark and the other is a stale device override test in GraniteSpeech." + "reason": "One is a DiaConfig save/load regression causing generation IndexError, the other is an MLukeTokenizer AttributeError; both are tokenizer/config problems but not the same bug." }, { - "left": "issue:44521", - "right": "issue:45290", + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "BEiT semantic segmentation loading with a legacy field and BigBirdTokenizer special-token registration/empty decode are unrelated failures." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "Qwen2 tensor-parallel OOM and a PEFT minimum-version bump are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:45584", "accept": false, - "reason": "Both involve apply_chat_template, but one is multimodal assistant masks and the other is a crash on tool-call assistant messages; different concrete failures." + "reason": "Compiled-forward chunked generation inconsistency and Whisper empty-transcription failure after align_special_tokens are different generation bugs." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 nonsensical outputs vs a broad unexpected position-id key issue; different code paths and symptoms." + "reason": "SigLIP2 nonsense outputs and a generic position-id key warning are not the same underlying issue." }, { - "left": "issue:29942", - "right": "issue:30333", + "left": "issue:43278", + "right": "issue:43701", "accept": false, - "reason": "FlashAttention 2 test failures and MLflow job-reporting issues are unrelated." + "reason": "BF16-to-FP32 embedding dtype drift in eval is unrelated to resume_from_checkpoint key mismatches." }, { - "left": "issue:44464", - "right": "issue:45584", + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44112", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency vs Whisper empty-transcription generation failure; not the same bug." + "reason": "CPU-offload device mismatch in bark-small and a stale GraniteSpeech CI test both mention device behavior, but they are different code paths and failure modes." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal inputs and the other is a crash on tool-call assistant messages with no content; not the same concrete bug." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures and MLflow job-reporting after failed training are unrelated." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "ZeRO-3 gradient growth and Phi-4 multimodal load failures are separate training/loading problems." + "reason": "ZeRO-3 gradient growth and Phi-4-multimodal loading failure are different model/runtime issues." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Non-template node compilation error vs flash_attn import KeyError on Python 3.13; unrelated failures." + "reason": "Non-template node compile errors and a Python 3.13 flash_attn import KeyError are unrelated." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "MPS attention correctness issue and an AuxRequest version-check bug in flex_attention are different problems." + "reason": "An MPS attention correctness bug and a PyTorch version check for flex_attention imports are distinct issues." }, { "left": "issue:44568", "right": "issue:45701", "accept": false, - "reason": "Both are tokenizer regressions, but 44568 is a specific mdeberta BOS/EOS bug while 45701 is a broad version-change report; not one concrete duplicate." + "reason": "Specific BOS/EOS token omission for mdeberta-v3-base and a broad tokenization-change report are too general/different to be duplicates." }, { - "left": "issue:43688", - "right": "issue:43873", + "left": "issue:15354", + "right": "issue:38175", "accept": false, - "reason": "Aux-loss normalization and quantization offloading are different model-training/runtime issues." + "reason": "TorchScript GeneratorExp export failure and SigLIP2 zero probabilities are unrelated." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Qwen3OmniMoe video-processing error vs Gemma 4 mm_token_type_ids defaulting; different multimodal bugs." + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma 4 mm_token_type_ids defaults are different multimodal model bugs." }, { - "left": "issue:15354", - "right": "issue:38175", + "left": "issue:43582", + "right": "issue:44977", "accept": false, - "reason": "torch.jit.script GeneratorExp limitation vs SigLIP2 zero-probability outputs; unrelated." + "reason": "Apple Silicon caching_allocator_warmup TypeError and Qwen3.5 flash-attention generation issues are unrelated." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "BFloat16 dtype mismatches in inference and Wav2Vec2 save/tokenization failures do not share a concrete code-path." + "reason": "BF16 dtype mismatches in inference and Wav2Vec2 save/tokenization failures do not share the same root cause." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs sync issue and Qwen3.5 hidden-states bug are separate generation-path regressions." + "reason": "sync_gpus/_update_model_kwargs_for_generation behavior and Qwen3.5 output_hidden_states regression are different generation-path bugs." }, { - "left": "issue:39401", - "right": "issue:44568", + "left": "issue:37428", + "right": "issue:43957", "accept": false, - "reason": "Both are tokenizer bugs, but offset_mapping handling in Qwen3 and BOS/EOS insertion in mdeberta are distinct failures." + "reason": "Flash attention import failure and meta-device loading breakage are unrelated." }, { - "left": "issue:43278", - "right": "issue:43701", + "left": "issue:39401", + "right": "issue:44568", "accept": false, - "reason": "BF16-to-FP32 embedding dtype drift vs resume_from_checkpoint key mismatch; unrelated." + "reason": "Wrong offset_mapping in Qwen3 tokenizer and add_special_tokens behavior for mdeberta-v3-base are different tokenizer bugs." }, { "left": "issue:44315", "right": "issue:44849", "accept": false, - "reason": "Liger Kernel application with model_init and Qwen3.5 hidden-states output are not the same bug." - }, - { - "left": "issue:44554", - "right": "issue:45356", - "accept": false, - "reason": "MPS attention correctness vs Kimi-K2.5 tokenizer codec handling regression; different subsystems." + "reason": "Liger Kernel application with model_init and output_hidden_states regression are unrelated." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Processor/tokenizer API regression and OmDet-Turbo image-size mismatch are both processor-facing but not the same underlying issue." + "reason": "Processor/tokenizer regression in v4.57.5 and OmDet-Turbo input-size mismatch are different multimodal issues." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Gemma 4 token-type defaults and Qwen2.5-VL temporal position-id scaling are unrelated multimodal issues." - }, - { - "left": "issue:43334", - "right": "issue:43525", - "accept": false, - "reason": "Both mention missing pad_token_id, but they affect different model families and likely different config paths; too broad to merge." - }, - { - "left": "issue:44898", - "right": "issue:45412", - "accept": false, - "reason": "Perceiver resolution/interpolation failure and RT-DETR memory leak are unrelated." - }, - { - "left": "issue:44466", - "right": "issue:45127", - "accept": false, - "reason": "Serialization/tied-weights inconsistency and LoRA-merge collapse with extended vocab are different weight-handling bugs." - }, - { - "left": "issue:37428", - "right": "issue:43957", - "accept": false, - "reason": "Flash-attention import error and meta-device model-loading breakage are separate loading/runtime issues." + "reason": "Gemma 4 mm_token_type_ids defaults and Qwen2.5-VL rope-index scaling are unrelated multimodal bugs." } ] }, @@ -6753,16 +6723,18 @@ "issue:43262", "issue:43278", "issue:43295", + "issue:43334", + "issue:43388", "issue:43425", "issue:43450", "issue:43493", + "issue:43525", + "issue:43526", "issue:43572", "issue:43575", - "issue:43606", "issue:43653", "issue:43819", "issue:43824", - "issue:43825", "issue:43827", "issue:43844", "issue:43856", @@ -6776,285 +6748,284 @@ "issue:44360", "issue:44442", "issue:44464", + "issue:44466", "issue:44514", "issue:44521", "issue:44561", "issue:44610", "issue:44792", "issue:44857", + "issue:44898", "issue:44936", "issue:44987", "issue:45083", "issue:45127", "issue:45198", "issue:45245", - "issue:45290", - "issue:45381", "issue:45412", "issue:45563", - "issue:45584", - "issue:45593" + "issue:45584" ], "soft_pairs": [ + "issue:43334|issue:43525", + "issue:44898|issue:45412", + "issue:44466|issue:45127", + "issue:43873|issue:44464", + "issue:37428|issue:43824", "issue:43572|issue:44987", + "issue:44792|issue:45563", + "issue:43844|issue:44936", "issue:43572|issue:43957", - "issue:43873|issue:44464", + "issue:43388|issue:43526", "issue:43824|issue:43975", "issue:43295|issue:43450", - "issue:43844|issue:44936", "issue:43295|issue:43873", "issue:38175|issue:45412", "issue:44514|issue:44521", - "issue:44442|issue:45198", + "issue:42915|issue:43824", "issue:43278|issue:43856", + "issue:44442|issue:45198", "issue:43295|issue:43994", - "issue:37428|issue:43824", - "issue:42915|issue:43824", "issue:38175|issue:43493", "issue:43827|issue:43901", "issue:43653|issue:44112", + "issue:44206|issue:45584", "issue:43450|issue:43493", "issue:44360|issue:44561", "issue:43450|issue:43575", - "issue:44792|issue:45563", "issue:45127|issue:45245", "issue:43819|issue:45083", - "issue:44206|issue:45584", "issue:43262|issue:43450", "issue:43819|issue:44112", "issue:37428|issue:43425", - "issue:44610|issue:44857", - "issue:43278|issue:43825", - "issue:45290|issue:45381", - "issue:44857|issue:45593", - "issue:43606|issue:45083" + "issue:44610|issue:44857" ], - "prepared_review_unit_hash": "6c84a3c27a42b637b5bcee019b73325cd45a8bc6aaef65e34dbecf2b264f69b2", + "prepared_review_unit_hash": "f3ab976e8bee4b5fe23ff457308298160668759d7f6717760d8585acc14b6e9d", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 36641, - "estimated_input_tokens": 9161, - "estimated_eval_tokens": 18578 + "serialized_chars": 37401, + "estimated_input_tokens": 9351, + "estimated_eval_tokens": 18958 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated transformer bug reports that only share broad subsystem vocabulary. I only accept the SigLIP2 pair as likely the same underlying model discrepancy; the rest look like distinct issues.", - "confidence": 0.78, - "canonical_issue_reason": "issue:43493 is the best canonical issue because it describes the broad SigLIP2 implementation discrepancy at the root of the symptom-specific output failures, making it the most general and reusable duplicate target in this set.", + "summary": "Most pairs are clearly unrelated and should stay separate. The only plausible duplicate-like pair is the SigLIP2 implementation/output discrepancy; everything else differs by model, subsystem, or failure mode. No PRs are present.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43493 is the broadest and most diagnostic SigLIP2 report; it reads like the underlying implementation mismatch that could also explain the zero-probabilities symptom in issue:38175.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43493 is the strongest representative of the only plausible duplicate chain here: it frames the underlying SigLIP2 HF-vs-JAX mismatch, whereas nearby reports are narrower symptoms.", + "best_issue_reason": "issue:43493 best represents the cluster\u2019s only believable duplicate theme because it describes the core SigLIP2 correctness discrepancy rather than a single downstream symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Same general error shape ('missing pad_token_id'), but different model families and config classes; not the same bug." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver interpolation failure vs RT-DETR memory release issue; unrelated subsystems and symptoms." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "One is serialization instability by device, the other is LoRA merge collapse with tied embeddings; different failure paths." + }, + { + "left": "issue:43873", + "right": "issue:44464", + "accept": false, + "reason": "Quantization/offloading behavior and compiled chunked generation are separate code paths." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different missing symbols in different modules; not the same underlying breakage." + }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "Both are version/load failures, but one is a missing config field and the other is a model load OOM/regression; different code paths." + "reason": "Both involve model loading after v5 changes, but the reported failures are different and not clearly one bug." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "Janus test failure vs stale warning in paged generation; no shared code-path bug." + }, + { + "left": "issue:43844", + "right": "issue:44936", + "accept": false, + "reason": "Gradient explosion under ZeRO-3 is unrelated to trainer.evaluate() failing after train()." }, { "left": "issue:43572", "right": "issue:43957", "accept": false, - "reason": "Different failure modes: missing StableLm config field vs meta-device model loading breakage." + "reason": "Missing pad_token_idx in one config is not the same as meta-device loading breakage in others." }, { - "left": "issue:43873", - "right": "issue:44464", + "left": "issue:43388", + "right": "issue:43526", "accept": false, - "reason": "Quantization/offloading behavior and compiled chunked generation are unrelated bugs." + "reason": "Both touch label handling, but one is gather_for_metrics batching and the other is BeitImageProcessorFast reduction logic." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "ImportError for a missing Qwen2.5-VL symbol vs DeepSeek detokenization issue; not the same bug." + "reason": "Import failure for a VL class vs DeepSeek detokenization regression; unrelated problems." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor.tokenizer regression and batched video shape bug are different processor issues." - }, - { - "left": "issue:43844", - "right": "issue:44936", - "accept": false, - "reason": "DeepSpeed/ZeRO gradient inflation and trainer.evaluate() after train() are distinct training-loop problems." + "reason": "Processor.tokenizer regression and batched video shape bug affect different processor behaviors." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Custom processor regression and quantization offloading bug are unrelated." + "reason": "Custom processor API regression is unrelated to quantization/offloading behavior." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 zero probabilities and RT-DETR memory cleanup are unrelated models and symptoms." + "reason": "SigLIP2 inference issue and RT-DETR memory leak are unrelated." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Both involve apply_chat_template, but one is a batched-padding crash and the other is incorrect assistant masks; not enough evidence of one underlying bug." + "reason": "Both use apply_chat_template, but one crashes on batched padding=False input while the other returns wrong assistant masks; different bugs." }, { - "left": "issue:44442", - "right": "issue:45198", + "left": "issue:42915", + "right": "issue:43824", "accept": false, - "reason": "Different tokenizer/model families with separate tokenizer/save_pretrained failures." + "reason": "Qwen3MoE FP8 config failure and a Qwen2.5-VL import error are unrelated." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Embedding dtype drift in eval vs Qwen3 MoE memory efficiency are unrelated." - }, - { - "left": "issue:43295", - "right": "issue:43994", - "accept": false, - "reason": "Processor.tokenizer regression is not the same as SigLIP2 nonsensical outputs." + "reason": "Training/eval dtype mismatch is not the same as Qwen3 MoE memory inefficiency." }, { - "left": "issue:37428", - "right": "issue:43824", + "left": "issue:44442", + "right": "issue:45198", "accept": false, - "reason": "Different missing imports from different modules and model families." + "reason": "Different tokenizers and different failure modes; no shared duplicate basis." }, { - "left": "issue:42915", - "right": "issue:43824", + "left": "issue:43295", + "right": "issue:43994", "accept": false, - "reason": "FineGrainedFP8Config failure and missing Qwen2.5-VL export are unrelated." + "reason": "Processor API regression is unrelated to SigLIP2 wrong outputs." }, { "left": "issue:38175", "right": "issue:43493", "accept": true, - "reason": "Both point to the same SigLIP2 implementation problem: HF outputs are wrong/non-sensical relative to expected behavior, with 38175 showing a symptom and 43493 describing the broader mismatch." + "reason": "Both are SigLIP2 correctness complaints on the same model; 38175\u2019s zero probabilities plausibly stem from the broader HF-vs-JAX discrepancy in 43493." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs-related, but they concern different pipeline APIs and different stale-deprecation issues." + "reason": "Both are docs issues, but they concern different pipeline APIs and different changes." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Tokenizer special-token registration and a GraniteSpeech CI device override test failure are unrelated." + "reason": "Tokenizer special-token bug and CI stale-device test are unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Feature extractor center-arg crash and Whisper empty-transcription generation bug are separate issues." }, { "left": "issue:43450", "right": "issue:43493", "accept": false, - "reason": "Video batching shape bug and SigLIP2 implementation discrepancy affect different models and code paths." + "reason": "Batched video processor shape bug is unrelated to SigLIP2 implementation discrepancy." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "DSA indexer design discussion and removal of is_torch_fx_available are unrelated." + "reason": "DSA indexer discussion and trust_remote_code breakage have no shared code path." }, { "left": "issue:43450", "right": "issue:43575", "accept": false, - "reason": "Batched video processor shape bug and tensor-parallel OOM on Qwen2-57B are different issues." - }, - { - "left": "issue:44792", - "right": "issue:45563", - "accept": false, - "reason": "Janus image generation test failure and paged generate warning are unrelated generation issues." + "reason": "Video batching shape bug and Qwen2-57B tp OOM are unrelated." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "LoRA/tied-embeddings collapse and category-count RuntimeError are different model-training failures." + "reason": "LoRA/tied-weights collapse and category-limit runtime error are different model behaviors." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "DAC latent forward mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." - }, - { - "left": "issue:44206", - "right": "issue:45584", - "accept": false, - "reason": "Feature extractor center-arg crash and Whisper empty-transcription failure are different audio bugs." + "reason": "DAC latent/STE mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate default and video batched-shape bug are different processor issues." + "reason": "Audio apply_chat_template sample-rate defaulting and video batch-shape bugs are unrelated." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "DAC STE mismatch and GraniteSpeech CI device override failure are unrelated." + "reason": "DAC latent math issue and GraniteSpeech CI test flake are unrelated." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "Flash-attention import error and Torch 2.10 incompatibility are different compatibility issues." + "reason": "Specific import error for a missing symbol and a general Torch 2.10 incompatibility report are not the same bug." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "Processor output-size mismatch and CUDA AMP loss crash are unrelated detection bugs." - }, - { - "left": "issue:43278", - "right": "issue:43825", - "accept": false, - "reason": "Training/eval dtype drift and pipeline docs/error-message wording are unrelated." - }, - { - "left": "issue:45290", - "right": "issue:45381", - "accept": false, - "reason": "Chat-template tool-call crash and Qwen2.5-VL vision_position_ids bug are different failure modes." - }, - { - "left": "issue:44857", - "right": "issue:45593", - "accept": false, - "reason": "Float16/CUDA loss crash and missing auxiliary losses in denoising-off mode are different detection training bugs." - }, - { - "left": "issue:43606", - "right": "issue:45083", - "accept": false, - "reason": "CPU offload device mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss AMP crash are different issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it treats most pairs as unrelated and only accepts the SigLIP2 pair, which is plausible from the titles. The canonical choice of issue:43493 is also reasonable as the broader SigLIP2 discrepancy report. No obvious overstatement beyond what the packet supports." + "feedback": "The judgments are generally grounded in the packet and stay conservative: nearly all soft pairs are clearly separated by model, subsystem, or failure mode, and the single accepted SigLIP2 pair is plausibly the same theme without overstating certainty. The summary also matches the packet: all items are issues and there are no PRs." }, "evaluator_used": true, "retried": false, @@ -7078,14 +7049,12 @@ "issue:29127", "issue:29942", "issue:32090", - "issue:34689", - "issue:35141", "issue:37428", "issue:38175", "issue:39692", "issue:40444", + "issue:43278", "issue:43335", - "issue:43388", "issue:43404", "issue:43504", "issue:43526", @@ -7095,6 +7064,7 @@ "issue:43653", "issue:43720", "issue:43819", + "issue:43825", "issue:43931", "issue:44112", "issue:44292", @@ -7113,305 +7083,291 @@ "issue:45083", "issue:45200", "issue:45229", + "issue:45290", "issue:45310", - "issue:45372", + "issue:45381", "issue:45459", "issue:45478", "issue:45491", "issue:45520", - "issue:45563", "issue:45588", "issue:45593", "issue:45712" ], "soft_pairs": [ + "issue:43278|issue:43825", + "issue:44857|issue:45593", + "issue:45290|issue:45381", + "issue:43606|issue:45083", "issue:44448|issue:44743", "issue:45200|issue:45588", "issue:43335|issue:43550", - "issue:44462|issue:44561", - "issue:44561|issue:44779", + "issue:29942|issue:32090", "issue:37428|issue:44351", + "issue:44561|issue:44779", + "issue:44462|issue:44561", "issue:43611|issue:43931", - "issue:43388|issue:43526", "issue:39692|issue:43526", "issue:43819|issue:44493", "issue:45310|issue:45478", + "issue:29127|issue:32090", "issue:43335|issue:43653", - "issue:29942|issue:32090", - "issue:43404|issue:45200", - "issue:44855|issue:45083", "issue:44292|issue:45520", + "issue:44855|issue:45083", + "issue:43404|issue:45200", "issue:39692|issue:44493", - "issue:43335|issue:45083", "issue:43550|issue:44112", + "issue:43335|issue:45083", "issue:43335|issue:43504", "issue:44448|issue:44779", "issue:38175|issue:44493", "issue:43335|issue:45072", "issue:44977|issue:45229", "issue:43720|issue:44112", + "issue:40444|issue:43526", "issue:43335|issue:45593", "issue:45459|issue:45712", - "issue:44568|issue:45491", "issue:44448|issue:44977", - "issue:29127|issue:32090", - "issue:43606|issue:44857", - "issue:40444|issue:43526", - "issue:45372|issue:45588", - "issue:44857|issue:45072", - "issue:34689|issue:35141", - "issue:44977|issue:45563" + "issue:44568|issue:45491" ], - "prepared_review_unit_hash": "d9666ce86a147524e89affa9712db5eaa65089a2006b28968307198ed3d2cf94", + "prepared_review_unit_hash": "e123cc3dafbfee7eee9dbf19dba3fbae467a5f517eb12c0cce6068f8527badde", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 36, - "serialized_chars": 38870, - "estimated_input_tokens": 9718, - "estimated_eval_tokens": 19692 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 34, + "serialized_chars": 37570, + "estimated_input_tokens": 9393, + "estimated_eval_tokens": 19042 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated Transformers bug reports. The only clear duplicate pair is the two identical Qwen3.5 MoE from_pretrained errors; the rest share broad themes like model loading, tokenizers, or flash-attention but point to different code paths and should stay separate.", - "confidence": 0.94, - "canonical_issue_reason": "issue:45310 is the earlier report of the same Qwen3.5 MoE from_pretrained failure and is the cleaner canonical target.", + "summary": "This set is mostly a mix of unrelated Transformers bug reports with broad thematic overlap around v5 regressions. Only one soft pair looks like a true duplicate: the two identical Qwen3.5 MoE from_pretrained error issues. Everything else should stay separate.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45310 is the earlier, clearer report of the exact Qwen3.5 MoE from_pretrained regression, and it matches the later duplicate wording in issue:45478.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the best issue candidate because it is the earliest and directly matches the duplicate report in issue:45478.", + "best_issue_reason": "issue:45310 is the strongest representative because it is the original report for the only clear duplicate pair and describes the concrete failure mode succinctly.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Different bugs: dtype drift in train/eval vs a pipeline error-message regression." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Different models and failures: AMP float16 crash vs missing auxiliary losses when denoising is off." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Both involve chat/vision behavior, but they are different code paths and symptoms." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "Unrelated failures: CPU offload device mismatch vs a helper length-calculation bug." + }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Different problems: Pegasus output regression vs Qwen3.5 recurrent-state reset in modular code." + "reason": "Both mention output differences in v5-era Qwen/PEGASUS-like models, but they describe different mechanisms and model paths." }, { "left": "issue:45200", "right": "issue:45588", "accept": false, - "reason": "Unrelated: Gemma 4 token-type IDs defaulting vs flash-attention crash on sink-less models." + "reason": "Different feature areas: Gemma4 text-only mm_token_type_ids handling vs flash attention aux-state attribute crash." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "Different models and failures: SwitchTransformers sparse-layer config vs Bamba torch.compile/SDPA." + "reason": "Different configuration bug and runtime compile bug; same broad subsystem is not enough." }, { - "left": "issue:44462", - "right": "issue:44561", + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Both are training/runtime test failures, but the underlying faults are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "Tokenizer.json loading bug is unrelated to removal of is_torch_fx_available breaking trust_remote_code models." + "reason": "Both are import errors, but they are for different missing symbols and different modules." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "Different tokenizer/runtime regressions with different symptoms and likely different fixes." + "reason": "Both are v5 regressions affecting trust_remote_code/tokenization, but they are distinct API removals and tokenizer behavior issues." }, { - "left": "issue:37428", - "right": "issue:44351", + "left": "issue:44462", + "right": "issue:44561", "accept": false, - "reason": "Both are import errors, but for different missing symbols and different affected features." + "reason": "Tokenizer repo-loading bug vs removal of a helper function for trust_remote_code models." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Model loading breaks for different reasons: base_model_prefix handling vs weight-shape mismatch." - }, - { - "left": "issue:43388", - "right": "issue:43526", - "accept": false, - "reason": "Both concern labels, but one is gather_for_metrics truncation and the other is BeitImageProcessorFast reduce_labels behavior." + "reason": "Both are model loading regressions, but one is base_model_prefix handling and the other is Qwen3-VL weight-shape mismatch." }, { "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "Documentation/example errors are unrelated to the Beit label reduction bug." + "reason": "Both concern BEiT/SigLIP2 processing details, but one is a docs example failure and the other is a fast processor label bug." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "DAC latent conversion mismatch is unrelated to unexpected position-id keys." + "reason": "Different internal bugs: DAC latent reconstruction mismatch vs unexpected position-id keys." }, { "left": "issue:45310", "right": "issue:45478", "accept": true, - "reason": "Same title and same concrete failure: transformers>=5.4.0 Qwen3.5 MoE from_pretrained error." + "reason": "Same exact bug report: identical Qwen3.5 MoE from_pretrained error on transformers>=5.4.0." }, { - "left": "issue:43335", - "right": "issue:43653", + "left": "issue:29127", + "right": "issue:32090", "accept": false, - "reason": "Different tokenizer/model areas: SwitchTransformers config vs BigBirdTokenizer special-token registration." + "reason": "Different models and failures: LayoutLMv3 box validation messaging vs Trainer GPU broadcast NoneType error." }, { - "left": "issue:29942", - "right": "issue:32090", + "left": "issue:43335", + "right": "issue:43653", "accept": false, - "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType errors." + "reason": "Both are config/token-related, but one is sparse-layer creation and the other is tokenizer special-token registration." }, { - "left": "issue:43404", - "right": "issue:45200", + "left": "issue:44292", + "right": "issue:45520", "accept": false, - "reason": "Different bugs: untied lm_head weights vs missing mm_token_type_ids default." + "reason": "Both mention runtime/import problems, but the Qwen NVFP4 error and Python 3.13 flash_attn KeyError are unrelated." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 JIT parsing issue is unrelated to qwen3_omni_moe feature-length computation." + "reason": "Different failures: Python parsing/import error in DebertaV2 vs feature-extract length behavior in qwen3_omni_moe." }, { - "left": "issue:44292", - "right": "issue:45520", + "left": "issue:43404", + "right": "issue:45200", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime issue is not the same as a flash_attn import-utils KeyError on Python 3.13." + "reason": "Different model families and bugs: tied weights in Mistral3 vs Gemma4 mm_token_type_ids defaults." }, { "left": "issue:39692", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 doc/example mistakes are unrelated to the position-id key regression." + "reason": "Different issues despite both being model-related; one is docs/quantization, the other is position-id key handling." }, { - "left": "issue:43335", - "right": "issue:45083", + "left": "issue:43550", + "right": "issue:44112", "accept": false, - "reason": "Different subsystems and failure modes; no shared underlying bug." + "reason": "Both are CI/runtime failures, but one is torch.compile+SDPA and the other is a stale device override test." }, { - "left": "issue:43550", - "right": "issue:44112", + "left": "issue:43335", + "right": "issue:45083", "accept": false, - "reason": "Bamba SDPA compile failure is unrelated to a GraniteSpeech CI device-override test." + "reason": "Different bugs in different models: sparse layer creation vs feature length helper behavior." }, { "left": "issue:43335", "right": "issue:43504", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug is unrelated to BEiT legacy-field loading failure." + "reason": "Config creation bug vs pretrained preset loading failure; not the same underlying defect." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Different model regressions: Pegasus output mismatch vs DeepSeek tokenizer behavior change." + "reason": "Both are v5 behavior regressions, but one is Pegasus output divergence and the other is Deepseek tokenizer corruption." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 zero probabilities and unexpected position-id keys are different bugs." + "reason": "Different model symptoms: zero probabilities in SigLIP2 vs unexpected position-id keys." }, { "left": "issue:43335", "right": "issue:45072", "accept": false, - "reason": "SwitchTransformers config bug and bfloat16 dtype mismatch in inference are unrelated." + "reason": "Different dtype/runtime issues: sparse-layer config bug vs bfloat16 inference mismatch in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug is unrelated to Gemma4 multi-GPU CUDA OOM." + "reason": "Different models and symptoms: Qwen3.5 flash-attention generation bug vs Gemma4 multi-GPU CUDA OOM." }, { "left": "issue:43720", - "right": "issue:44112", - "accept": false, - "reason": "Packed-weight loading failure in BitNet is unrelated to a stale GraniteSpeech CI test." - }, - { - "left": "issue:43335", - "right": "issue:45593", - "accept": false, - "reason": "SwitchTransformers sparse-layer creation and D-FINE auxiliary-loss behavior are different issues." - }, - { - "left": "issue:45459", - "right": "issue:45712", - "accept": false, - "reason": "Tokenizer error masking due to missing protobuf is unrelated to dummy-class cleanup/check_repo failures." - }, - { - "left": "issue:44568", - "right": "issue:45491", - "accept": false, - "reason": "Tokenizer special-token behavior is unrelated to Gemma3 NaN embeddings from mixed-length batching." - }, - { - "left": "issue:44448", - "right": "issue:44977", - "accept": false, - "reason": "Different model families and symptoms: Pegasus output drift vs Qwen3.5 flash-attention generation issues." - }, - { - "left": "issue:29127", - "right": "issue:32090", - "accept": false, - "reason": "LayoutLMV3 error-message clarity is unrelated to Trainer broadcast NoneType errors." - }, - { - "left": "issue:43606", - "right": "issue:44857", + "right": "issue:44112", "accept": false, - "reason": "CPU-offload device mismatch and float16 AMP CUDA crash are distinct bugs." + "reason": "Different CI failures: BitNet packed-weight loading vs stale device override test." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "Qwen2.5-VL iterable-dataset finetuning failure is unrelated to BeitImageProcessorFast label reduction." + "reason": "Different bug classes: Qwen2.5-VL iterable dataset finetuning failure vs BEiT fast processor label reduction." }, { - "left": "issue:45372", - "right": "issue:45588", + "left": "issue:43335", + "right": "issue:45593", "accept": false, - "reason": "Gemma 4 processor loading ImportError is unrelated to flash-attention sink-less-model crash." + "reason": "Different model/config bugs: sparse-layer creation vs auxiliary-loss omission." }, { - "left": "issue:44857", - "right": "issue:45072", + "left": "issue:45459", + "right": "issue:45712", "accept": false, - "reason": "Different precision-related failures in different models and code paths." + "reason": "Different areas: tokenizer error hiding on missing protobuf vs dummy class cleanup and repo checks." }, { - "left": "issue:34689", - "right": "issue:35141", + "left": "issue:44448", + "right": "issue:44977", "accept": false, - "reason": "Model-loading regression for Llama 3.2 Vision is unrelated to token-embedding reinit after resize." + "reason": "Both concern generation regressions, but they affect different models and code paths." }, { - "left": "issue:44977", - "right": "issue:45563", + "left": "issue:44568", + "right": "issue:45491", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug is unrelated to a stale paging warning for num_return_sequences." + "reason": "Different tokenizer/model issues: add_special_tokens behavior vs Gemma3 NaN embeddings from sliding-window attention." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The duplicate call on issue:45310 / issue:45478 is supported by the identical title and matching failure description, and the other soft-edge rejections are framed narrowly enough to avoid overstating similarity." + "feedback": "The canonical pair is well-supported: issue:45310 and issue:45478 have identical titles and clearly refer to the same Qwen3.5 MoE from_pretrained regression. The remaining soft-edge rejections are conservative and stay within the evidence in the packet. No PR-pair concerns apply here." }, "evaluator_used": true, "retried": false, @@ -7434,64 +7390,77 @@ "nodes": [ "issue:25251", "issue:29942", - "issue:30064", "issue:32090", + "issue:33357", + "issue:34689", "issue:35141", - "issue:36010", "issue:37428", - "issue:42915", - "issue:42994", + "issue:41720", "issue:43065", "issue:43066", "issue:43278", + "issue:43329", "issue:43381", "issue:43412", - "issue:43421", "issue:43550", + "issue:43606", "issue:43653", "issue:43701", + "issue:43723", "issue:43873", "issue:43927", "issue:43976", "issue:43994", "issue:44186", - "issue:44206", - "issue:44265", - "issue:44361", "issue:44448", "issue:44451", "issue:44462", "issue:44466", - "issue:44479", "issue:44493", + "issue:44554", "issue:44743", "issue:44857", - "issue:44871", "issue:44977", + "issue:44987", + "issue:45072", "issue:45083", "issue:45198", + "issue:45229", "issue:45245", "issue:45372", + "issue:45406", "issue:45412", "issue:45440", "issue:45459", - "issue:45491", "issue:45520", "issue:45563", "issue:45588", "issue:45698" ], "soft_pairs": [ + "issue:34689|issue:35141", + "issue:43723|issue:44554", + "issue:43606|issue:44857", + "issue:44857|issue:45072", + "issue:45372|issue:45588", + "issue:37428|issue:44987", "issue:43701|issue:45083", + "issue:43873|issue:44493", + "issue:44977|issue:45563", "issue:43927|issue:45698", "issue:43550|issue:45083", - "issue:43066|issue:44462", + "issue:29942|issue:35141", "issue:43412|issue:45588", + "issue:32090|issue:33357", + "issue:43278|issue:43329", + "issue:43066|issue:44462", + "issue:43873|issue:45229", "issue:45245|issue:45563", - "issue:43873|issue:44493", - "issue:29942|issue:35141", + "issue:43976|issue:45440", + "issue:45372|issue:45406", + "issue:41720|issue:45229", + "issue:43065|issue:43994", "issue:45372|issue:45459", - "issue:36010|issue:37428", "issue:44448|issue:44466", "issue:43065|issue:45412", "issue:25251|issue:32090", @@ -7500,207 +7469,228 @@ "issue:44451|issue:45520", "issue:43653|issue:44743", "issue:44186|issue:45198", - "issue:43976|issue:45440", - "issue:43994|issue:45412", - "issue:44265|issue:45198", - "issue:44361|issue:45198", - "issue:42915|issue:42994", - "issue:43421|issue:45491", - "issue:44871|issue:44977", - "issue:29942|issue:30064", - "issue:43065|issue:43994", - "issue:44206|issue:44479" + "issue:43994|issue:45412" ], - "prepared_review_unit_hash": "9aa90e8e356efdb5534fa79bd40b129e10b955e842351496bc07b85b6bb90992", + "prepared_review_unit_hash": "e9fbe6a0c2ced4c8cccfe16a696c4fb8f90beeabf029143c657be8571f5b3af3", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36052, - "estimated_input_tokens": 9013, - "estimated_eval_tokens": 18282 + "soft_pair_count": 32, + "serialized_chars": 37273, + "estimated_input_tokens": 9319, + "estimated_eval_tokens": 18894 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: most items are unrelated bugs/regressions across tokenization, export, quantization, flash-attention, and model-specific behavior. I found no true duplicate pair among the soft edges.", + "summary": "This cluster is largely a false-positive similarity set: the issues span unrelated models, tokenizers, loading paths, attention/backends, and memory/offload problems. I would not merge any of the soft pairs.", "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue stands out because the cluster does not cohere around one underlying bug; the items span many distinct code paths and symptoms.", - "canonical_pr_reason": "No PRs are present in this cluster.", - "best_issue_reason": "If one issue must be used as a representative, issue 44462 is the cleanest self-contained report with a concrete, actionable loading symptom around AutoTokenizer/tokenizer.json.", - "best_pr_reason": "No PRs are present, so there is no best PR candidate.", + "canonical_issue_reason": "No single issue is a strong canonical; the items are heterogeneous and do not describe one shared bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue here is a reliable global representative for duplicate triage, because the cluster breaks into many unrelated problem areas.", + "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43701", - "right": "issue:45083", + "left": "issue:34689", + "right": "issue:35141", "accept": false, - "reason": "Checkpoint-resume key mismatch and qwen3_omni_moe feature-length behavior are different subsystems and different bugs." + "reason": "Different failures: Llama 3.2 Vision model loading vs embedding reinitialization after resize." }, { - "left": "issue:43927", - "right": "issue:45698", + "left": "issue:43723", + "right": "issue:44554", "accept": false, - "reason": "Both involve save/load, but one is about custom token IDs in config and the other about loading the wrong custom module." + "reason": "Tokenizer loading regression in v5 vs an MPS attention correctness issue; different code paths and symptoms." }, { - "left": "issue:43550", - "right": "issue:45083", + "left": "issue:43606", + "right": "issue:44857", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba-9B-v2 is unrelated to the qwen3_omni_moe helper-function issue." + "reason": "CPU offload device mismatch in bark-small vs a float16 AMP crash in LwDetr loss." }, { - "left": "issue:43066", - "right": "issue:44462", + "left": "issue:44857", + "right": "issue:45072", "accept": false, - "reason": "Both touch tokenizers, but one is a decoder-type mismatch in v5 while the other is ignoring tokenizer.json from the repo." + "reason": "Both mention dtype/precision, but the affected models and concrete bugs are different." }, { - "left": "issue:43412", + "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Executorch export/erfinv/training-guard failure and flash_attention s_aux=None crash are different code paths." + "reason": "Gemma 4 processor import failure vs a flash_attention.py AttributeError on sink-less models." }, { - "left": "issue:45245", - "right": "issue:45563", + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "Both are loading-related, but one is a missing internal flash-attention symbol and the other is a model-specific load failure." + }, + { + "left": "issue:43701", + "right": "issue:45083", "accept": false, - "reason": "Category-count overflow and a stale num_return_sequences warning are unrelated." + "reason": "Checkpoint resume key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:43873", "right": "issue:44493", "accept": false, - "reason": "Quantization offloading behavior and unexpected position-id key warnings are not the same bug." + "reason": "Offloading/quantization behavior is unrelated to unexpected position-id keys." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Flash-attention generation failure vs an unrelated stale warning in paged generate()." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Saving/loading custom token IDs is different from loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba is unrelated to qwen3_omni_moe length computation." }, { "left": "issue:29942", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 test failures and embedding reinit in post_init are distinct regressions." + "reason": "Failing Flash Attention 2 tests do not match the embedding reinit bug." }, { - "left": "issue:45372", - "right": "issue:45459", + "left": "issue:43412", + "right": "issue:45588", "accept": false, - "reason": "ReasoningEffort import breakage and protobuf error masking affect different tokenizer/processor paths." + "reason": "ExecuTorch export guard/erfinv support issues are not the same as the flash_attention None crash." }, { - "left": "issue:36010", - "right": "issue:37428", + "left": "issue:32090", + "right": "issue:33357", "accept": false, - "reason": "These are both import errors, but for different symbols in different modules." + "reason": "Trainer broadcast NoneType error and MacOS bus error on CLIP are different failures." }, { - "left": "issue:44448", - "right": "issue:44466", + "left": "issue:43278", + "right": "issue:43329", "accept": false, - "reason": "Pegasus output differences and lm_head.weight serialization inconsistency are separate model-generation vs save/load issues." + "reason": "Embedding dtype drift in eval is unrelated to undefined video-branch variables in multimodal token counting." }, { - "left": "issue:43065", - "right": "issue:45412", + "left": "issue:43066", + "right": "issue:44462", "accept": false, - "reason": "Sam3PixelDecoder dummy Conv2d and RT-DETR memory release are unrelated." + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are both tokenizer-related, but not the same defect." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:43873", + "right": "issue:45229", "accept": false, - "reason": "Pipeline top_k nesting behavior and Trainer _gpu_broadcast_one NoneType error are unrelated failures." + "reason": "Quantization/offload mismatch is not the same as Gemma4 multi-GPU CUDA OOM." }, { - "left": "issue:43278", - "right": "issue:43381", + "left": "issue:45245", + "right": "issue:45563", "accept": false, - "reason": "Embedding dtype drift in eval and gradient checkpointing being disallowed in eval are different eval-mode problems." + "reason": "Category-cardinality RuntimeError is unrelated to the stale num_return_sequences warning." }, { - "left": "issue:44857", - "right": "issue:45588", + "left": "issue:43976", + "right": "issue:45440", "accept": false, - "reason": "LwDetrImageLoss AMP/CUDA crash and flash_attention s_aux=None crash do not share the same underlying defect." + "reason": "Python version compatibility and DeepseekV3MoE output divergence are unrelated issues." }, { - "left": "issue:44451", - "right": "issue:45520", + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "Model load failure and Python 3.13 flash_attn KeyError are different import/runtime problems." + "reason": "Both touch Gemma 4, but one is a dependency import error and the other is a missing _tokenizer attribute in serve." }, { - "left": "issue:43653", - "right": "issue:44743", + "left": "issue:41720", + "right": "issue:45229", "accept": false, - "reason": "BigBirdTokenizer special-token decoding and Qwen recurrent-state reset are unrelated." + "reason": "Qwen3 device-map cudaErrorAssert and Gemma4 OOM are different models and failure modes." }, { - "left": "issue:44186", - "right": "issue:45198", + "left": "issue:43065", + "right": "issue:43994", "accept": false, - "reason": "LayoutLMv2Tokenizer padding/truncation crash and Wav2Vec2 save_pretrained/tokenization failure are different tokenizer bugs." + "reason": "Sam3PixelDecoder dummy Conv2d is unrelated to SigLIP2 producing nonsensical outputs." }, { - "left": "issue:43976", - "right": "issue:45440", + "left": "issue:45372", + "right": "issue:45459", "accept": false, - "reason": "Python-version incompatibility and DeepseekV3MoE divergence are not the same issue." + "reason": "Gemma4 processor import dependency break is not the same as protobuf-related tokenizer error masking." }, { - "left": "issue:43994", + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Pegasus v4/v5 output differences are unrelated to lm_head.weight serialization inconsistency." + }, + { + "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "Siglip2 nonsensical outputs and RT-DETR memory leakage are unrelated." + "reason": "A dummy Conv2d in Sam3PixelDecoder does not match RT-DETR memory release problems." }, { - "left": "issue:44265", - "right": "issue:45198", + "left": "issue:25251", + "right": "issue:32090", "accept": false, - "reason": "torch.export.export with torch_compilable_check and Wav2Vec2 save/tokenization failure are different defects." + "reason": "Pipeline top_k nesting change is unrelated to _gpu_broadcast_one rejecting NoneType." }, { - "left": "issue:44361", - "right": "issue:45198", + "left": "issue:43278", + "right": "issue:43381", "accept": false, - "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure are unrelated tokenizer issues." + "reason": "Embedding dtype changes in train/eval are unrelated to gradient checkpointing being disallowed in eval mode." }, { - "left": "issue:42915", - "right": "issue:42994", + "left": "issue:44857", + "right": "issue:45588", "accept": false, - "reason": "Both mention quantization, but one is a FineGrainedFP8Config runtime failure and the other is quantized model saving failure." + "reason": "Different layers: LwDetr AMP loss crash vs flash_attention sink-less model attribute error." }, { - "left": "issue:43421", - "right": "issue:45491", + "left": "issue:44451", + "right": "issue:45520", "accept": false, - "reason": "Runtime special-token post-processor updates and Gemma3 NaN embeddings are different bugs." + "reason": "ScandiBERT loading failure is unrelated to a Python 3.13 flash_attn import-utils KeyError." }, { - "left": "issue:44871", - "right": "issue:44977", + "left": "issue:43653", + "right": "issue:44743", "accept": false, - "reason": "Gemma eos_token_id mismatch and Qwen3.5 flash-attention generation regression are unrelated." + "reason": "BigBirdTokenizer decode/special-token bug is unrelated to Qwen3 recurrent state reset behavior." }, { - "left": "issue:29942", - "right": "issue:30064", + "left": "issue:44186", + "right": "issue:45198", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation map processing failure are unrelated." + "reason": "LayoutLMv2Tokenizer NER padding/truncation crash and Wav2Vec2 save/tokenization failure are different tokenizer bugs." }, { - "left": "issue:44206", - "right": "issue:44479", + "left": "issue:43994", + "right": "issue:45412", "accept": false, - "reason": "Unsupported center argument in LasrFeatureExtractor and Qwen VL video regression are different issues." + "reason": "SigLIP2 inference quality issue is unrelated to RT-DETR memory being retained after deletion." } ] }, @@ -7726,286 +7716,293 @@ "nodes": [ "issue:29127", "issue:29942", + "issue:30064", "issue:32090", + "issue:33290", "issue:33357", "issue:34567", + "issue:35141", "issue:36010", "issue:39692", - "issue:41720", "issue:42915", + "issue:42994", + "issue:43232", + "issue:43257", "issue:43262", "issue:43278", - "issue:43329", "issue:43334", "issue:43381", - "issue:43441", + "issue:43388", + "issue:43421", "issue:43526", "issue:43531", - "issue:43575", "issue:43653", "issue:43701", - "issue:43723", "issue:43756", "issue:43824", - "issue:43873", "issue:43931", "issue:44186", "issue:44206", + "issue:44265", "issue:44279", + "issue:44361", "issue:44442", "issue:44464", - "issue:44554", + "issue:44479", "issue:44610", - "issue:44734", - "issue:44743", - "issue:44811", - "issue:44857", + "issue:44871", "issue:44977", "issue:45072", "issue:45103", + "issue:45198", "issue:45229", "issue:45245", "issue:45372", - "issue:45406", "issue:45412", "issue:45440", "issue:45491", - "issue:45588", "issue:45593" ], "soft_pairs": [ - "issue:44734|issue:45406", + "issue:44265|issue:45198", + "issue:44361|issue:45198", + "issue:44206|issue:44479", + "issue:42915|issue:42994", + "issue:43421|issue:45491", + "issue:44871|issue:44977", + "issue:43257|issue:43278", + "issue:43756|issue:45440", + "issue:29942|issue:30064", "issue:33357|issue:39692", - "issue:44610|issue:45412", "issue:29127|issue:34567", + "issue:32090|issue:33290", + "issue:44610|issue:45412", "issue:45103|issue:45372", - "issue:43873|issue:45229", - "issue:43756|issue:45440", - "issue:45372|issue:45406", - "issue:43723|issue:44554", - "issue:43278|issue:43575", - "issue:43278|issue:43329", + "issue:43824|issue:44279", + "issue:45440|issue:45593", "issue:43381|issue:43701", + "issue:45072|issue:45229", "issue:43653|issue:44442", - "issue:45440|issue:45593", - "issue:41720|issue:45229", + "issue:43526|issue:44186", "issue:29942|issue:36010", - "issue:43824|issue:44279", - "issue:32090|issue:33357", + "issue:29127|issue:33357", "issue:43334|issue:43531", - "issue:43526|issue:44186", + "issue:44464|issue:45491", "issue:44977|issue:45245", - "issue:43526|issue:44857", + "issue:29127|issue:35141", "issue:42915|issue:43931", - "issue:43441|issue:45588", "issue:43262|issue:44206", - "issue:44464|issue:45491", - "issue:43526|issue:44610", - "issue:44743|issue:44811", + "issue:43232|issue:43388", "issue:44442|issue:45072" ], - "prepared_review_unit_hash": "9100251c14a0e352eff00225799b65ae5670905cd88fdd9b5532dabc3cdaf994", + "prepared_review_unit_hash": "83d93b41f4ed85d2c1d5ddcdb5a9a82b014603e310ba6b8df73d5c78bed5a57f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36476, - "estimated_input_tokens": 9119, - "estimated_eval_tokens": 18494 + "soft_pair_count": 30, + "serialized_chars": 36784, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a broad false-positive mix of unrelated issues across tokenizers, model loading, training, attention, quantization, and docs. None of the soft pairs describe the same concrete bug or change, so no duplicates should be merged here.", + "summary": "The cluster is heterogeneous: the issues span unrelated bugs in tokenizers, generation, vision, training, quantization, and docs. None of the soft pairs look like true duplicates.", "confidence": 0.98, - "canonical_issue_reason": "No canonical issue: the items are heterogeneous and do not share one underlying bug, code path, or fix surface.", + "canonical_issue_reason": "No canonical issue stands out; the items cover unrelated root causes across multiple subsystems.", "canonical_pr_reason": null, - "best_issue_reason": "No single best issue for deduplication; none stands out as the representative of a real duplicate group.", + "best_issue_reason": "No single issue is a strong global representative because the cluster is too diverse and the pairwise candidates do not share the same concrete bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44734", - "right": "issue:45406", + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "Torch export failure on torch_compilable_check is unrelated to Wav2Vec2 save_pretrained/tokenization failures." + }, + { + "left": "issue:44361", + "right": "issue:45198", "accept": false, - "reason": "Different failures: KV-cache continuation tensor indexing vs Gemma4Processor missing _tokenizer." + "reason": "MLukeTokenizer AttributeError on tasks is a different tokenizer bug from Wav2Vec2 save/tokenization failures." }, { - "left": "issue:33357", - "right": "issue:39692", + "left": "issue:44206", + "right": "issue:44479", "accept": false, - "reason": "Unrelated: MacOS CLIP bus error vs SigLIP2 doc example/model-processor mismatch and quantization issue." + "reason": "Unsupported center arg in LasrFeatureExtractor is unrelated to the Qwen VL video-input regression." }, { - "left": "issue:44610", - "right": "issue:45412", + "left": "issue:42915", + "right": "issue:42994", "accept": false, - "reason": "Different subsystems: processor image-size mismatch vs RT-DETR memory not released on deletion." + "reason": "FineGrainedFP8Config failure in Qwen3-MoE and quantized model saving are both quantization-related, but they are different concrete bugs." }, { - "left": "issue:29127", - "right": "issue:34567", + "left": "issue:43421", + "right": "issue:45491", "accept": false, - "reason": "Different problems: LayoutLMv3 error clarity vs TrainerState token-count tracking not updating." + "reason": "Runtime post-processor updates for special tokens are unrelated to Gemma3 NaNs from sliding-window all-padding windows." }, { - "left": "issue:45103", - "right": "issue:45372", + "left": "issue:44871", + "right": "issue:44977", "accept": false, - "reason": "Unrelated: auto_docstring crash with future annotations vs mistral_common import breakage for Gemma 4 loading." + "reason": "Gemma eos_token_id inconsistency and Qwen3.5 flash-attention generation failure are different issues." }, { - "left": "issue:43873", - "right": "issue:45229", + "left": "issue:43257", + "right": "issue:43278", "accept": false, - "reason": "Different issues: quantization offloading behavior vs Gemma4 multi-GPU CUDA OOM." + "reason": "MOE weight conversion with accelerate+deepspeed is unrelated to BF16-to-FP32 embedding dtype drift in eval." }, { "left": "issue:43756", "right": "issue:45440", "accept": false, - "reason": "Different model-specific bugs: Smollm3 RoPE layer handling vs DeepseekV3MoE divergence from remote implementation." + "reason": "Smollm3 RoPE-layer mismatch and DeepseekV3MoE divergence concern different models and different implementation bugs." }, { - "left": "issue:45372", - "right": "issue:45406", + "left": "issue:29942", + "right": "issue:30064", "accept": false, - "reason": "ImportError in Gemma4 processor loading is unrelated to serve-time _tokenizer attribute crash." + "reason": "Flash Attention 2 test failures are unrelated to void segmentation map processing." }, { - "left": "issue:43723", - "right": "issue:44554", + "left": "issue:33357", + "right": "issue:39692", "accept": false, - "reason": "Different areas: AutoTokenizer loading regression vs MPS attention correctness bug." + "reason": "MacOS CLIP bus error is a different failure mode from the SigLIP2 docs/model-processor mismatch and quantization issue." }, { - "left": "issue:43278", - "right": "issue:43575", + "left": "issue:29127", + "right": "issue:34567", "accept": false, - "reason": "Embedding dtype mismatch in eval vs tensor-parallel OOM on model load are unrelated." + "reason": "LayoutLMv3 error-message clarity and TrainerState num_input_tokens_seen not updating are unrelated bugs." }, { - "left": "issue:43278", - "right": "issue:43329", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "No shared code path: dtype change in evaluate vs multimodal token counting bug." + "reason": "NoneType broadcast in Trainer and Adafactor+deepspeed OOM are different training problems." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:44610", + "right": "issue:45412", "accept": false, - "reason": "Gradient checkpointing eval-mode restriction is unrelated to checkpoint resume key mismatch." + "reason": "OmDet-Turbo processor size mismatch is unrelated to RT-DETR memory not being released." }, { - "left": "issue:43653", - "right": "issue:44442", + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotation handling and Gemma4 processor import failure are unrelated code paths." + }, + { + "left": "issue:43824", + "right": "issue:44279", "accept": false, - "reason": "Different tokenizer bugs: special-token registration vs AutoTokenizer failing to load a specific tokenizer class." + "reason": "A specific Qwen2.5-VL import error is not the same as a generic dependency issue." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "Unrelated model correctness issue vs D-FINE auxiliary-loss behavior." + "reason": "DeepseekV3MoE implementation divergence and D-FINE auxiliary-loss omission are unrelated model bugs." }, { - "left": "issue:41720", + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction and resume_from_checkpoint key mismatch are different trainer behaviors." + }, + { + "left": "issue:45072", "right": "issue:45229", "accept": false, - "reason": "Different GPU inference failures: Qwen3 auto device mapping assert vs Gemma4 multi-GPU OOM." + "reason": "Inference dtype mismatches are unrelated to a multi-GPU CUDA OOM." }, { - "left": "issue:29942", - "right": "issue:36010", + "left": "issue:43653", + "right": "issue:44442", "accept": false, - "reason": "Flash Attention test failures are unrelated to a GenerationMixin import error." + "reason": "BigBirdTokenizer special-token registration and AutoTokenizer loading FastSpeech2ConformerTokenizer are different tokenizer issues." }, { - "left": "issue:43824", - "right": "issue:44279", + "left": "issue:43526", + "right": "issue:44186", "accept": false, - "reason": "Specific missing symbol import vs generic dependency issue; not the same bug." + "reason": "BeitImageProcessorFast reduce_labels and LayoutLMv2Tokenizer padding/NER crashes are unrelated preprocessing bugs." }, { - "left": "issue:32090", + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "Flash Attention 2 tests and GenerationMixin import errors are unrelated." + }, + { + "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "Trainer broadcast NoneType error and MacOS CLIP bus error are unrelated." + "reason": "LayoutLMv3 error messaging and CLIP bus error on MacOS have no shared root cause." }, { "left": "issue:43334", "right": "issue:43531", "accept": false, - "reason": "Different Qwen3 variants and different failures: missing pad_token_id vs sliding_window behavior." + "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding_window behavior affect different model components." }, { - "left": "issue:43526", - "right": "issue:44186", + "left": "issue:44464", + "right": "issue:45491", "accept": false, - "reason": "BEiT label reduction bug vs LayoutLMv2 tokenizer crash on NER/padding are unrelated." + "reason": "Compiled-forward chunked generation inconsistency is a different bug from Gemma3 NaNs caused by all-padding windows." }, { "left": "issue:44977", "right": "issue:45245", "accept": false, - "reason": "Flash-attention generation issue is unrelated to the 2^24 category limit runtime error." + "reason": "Flash-attention generation regression and category-count runtime limit are unrelated failures." }, { - "left": "issue:43526", - "right": "issue:44857", + "left": "issue:29127", + "right": "issue:35141", "accept": false, - "reason": "Different modules and symptoms: image-label reduction vs LwDetrImageLoss AMP crash." + "reason": "LayoutLMv3 error clarity and token-embedding reinitialization in post_init are unrelated bugs." }, { "left": "issue:42915", "right": "issue:43931", "accept": false, - "reason": "Qwen3Moe FP8 config failure is not the same as Qwen3-VL weight-shape mismatch." - }, - { - "left": "issue:43441", - "right": "issue:45588", - "accept": false, - "reason": "Both mention flash attention, but they affect different code paths and concrete crashes." + "reason": "Qwen3-MoE FP8 config failure and Qwen3-VL weight-shape mismatch are different model-loading problems." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Audio chat-template sample-rate defaulting is unrelated to an unsupported feature-extractor argument." - }, - { - "left": "issue:44464", - "right": "issue:45491", - "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is unrelated to Gemma3 NaN embeddings from sliding-window attention." - }, - { - "left": "issue:43526", - "right": "issue:44610", - "accept": false, - "reason": "Image label reduction bug and OmDet-Turbo input-size mismatch are distinct processor issues." + "reason": "Audio chat-template sampling-rate defaulting and unsupported center arg in LasrFeatureExtractor are unrelated." }, { - "left": "issue:44743", - "right": "issue:44811", + "left": "issue:43232", + "right": "issue:43388", "accept": false, - "reason": "Recurrent-state cache reset bug is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "Generation kwargs sync_gpus behavior and gather_for_metrics label dropping are different trainer/data-path bugs." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "Tokenizer-loading failure and dtype mismatches in bfloat16 inference are unrelated." + "reason": "AutoTokenizer load failure is unrelated to dtype mismatch in bfloat16 inference." } ] }, @@ -8029,26 +8026,25 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:29127", "issue:33290", - "issue:33357", "issue:34567", "issue:34689", "issue:35141", - "issue:38175", "issue:39692", "issue:42907", - "issue:43295", "issue:43316", "issue:43381", + "issue:43441", "issue:43452", "issue:43493", "issue:43525", + "issue:43526", "issue:43550", "issue:43575", "issue:43611", "issue:43646", "issue:43701", + "issue:43723", "issue:43761", "issue:43827", "issue:43873", @@ -8061,14 +8057,13 @@ "issue:44387", "issue:44442", "issue:44479", - "issue:44568", "issue:44743", + "issue:44811", + "issue:44857", "issue:44898", - "issue:44964", "issue:44977", - "issue:45072", + "issue:45003", "issue:45216", - "issue:45229", "issue:45290", "issue:45440", "issue:45479", @@ -8076,220 +8071,249 @@ "issue:45584", "issue:45588", "issue:45600", - "issue:45698" + "issue:45698", + "issue:45759" ], "soft_pairs": [ + "issue:43526|issue:44857", + "issue:43526|issue:43994", + "issue:43441|issue:45588", + "issue:44743|issue:44811", + "issue:43761|issue:44079", "issue:43452|issue:45588", "issue:43575|issue:43701", - "issue:29127|issue:35141", + "issue:45003|issue:45440", "issue:43646|issue:44077", - "issue:45072|issue:45229", + "issue:43723|issue:45479", "issue:43575|issue:44387", "issue:44977|issue:45584", + "issue:43761|issue:44188", "issue:45588|issue:45600", - "issue:33357|issue:38175", "issue:43316|issue:43525", "issue:44479|issue:45290", - "issue:44964|issue:45440", - "issue:43761|issue:44079", "issue:43931|issue:45216", + "issue:43994|issue:44206", "issue:42907|issue:45698", "issue:33290|issue:34567", "issue:34689|issue:43611", - "issue:39692|issue:45491", "issue:44442|issue:44898", - "issue:43761|issue:44188", + "issue:39692|issue:45491", + "issue:33290|issue:35141", + "issue:33290|issue:34689", "issue:43493|issue:43827", "issue:43381|issue:43873", - "issue:44568|issue:45479", "issue:43550|issue:44743", - "issue:43295|issue:45584", - "issue:43994|issue:44206" + "issue:44077|issue:44479", + "issue:43316|issue:45759" ], - "prepared_review_unit_hash": "2aa24714c0edb08e816c0448714ca132825e9b25012224bbdd5cf84104deffcb", + "prepared_review_unit_hash": "2fd93cadc358197ab422656a6b5619c01ff599e1d11dd641896f769469e8e1bc", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 26, - "serialized_chars": 35627, - "estimated_input_tokens": 8907, - "estimated_eval_tokens": 18070 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36167, + "estimated_input_tokens": 9042, + "estimated_eval_tokens": 18340 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: mostly unrelated bugs/regressions spanning loading, quantization, attention, tokenization, docs, and model-specific failures. None of the soft pairs look like the same underlying bug, so all are rejected.", - "confidence": 0.27, - "canonical_issue_reason": "No single canonical duplicate stands out; the issues do not converge on one concrete bug and instead cover many distinct failure modes.", + "summary": "This is not a true duplicate cluster: the items are mostly unrelated bugs/regressions across different models, APIs, and subsystems, so no single canonical issue emerges. The proposed soft edges are all too weak to merge.", + "confidence": 0.96, + "canonical_issue_reason": "No clear canonical issue: the set is heterogeneous, and none of the issues cleanly subsumes the others as the same underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "If forced to anchor the cluster, issue 45698 is the broadest representative of the recurring save/load/custom-module theme, but it is still not a true duplicate of the rest.", + "best_issue_reason": "No global best issue for deduping; the cluster spans multiple distinct failure modes, so picking one as representative would be misleading.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "Different failures: BeitImageProcessorFast label reduction bug vs AMP/CUDA loss crash in LwDetrImageLoss." + }, + { + "left": "issue:43526", + "right": "issue:43994", + "accept": false, + "reason": "Unrelated subsystems and symptoms: BeitImageProcessorFast label handling vs SigLIP2 model output quality." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both mention FlashAttention, but one is a Ministral-3 v5 RC regression and the other is an AttributeError on sink-less models; not the same concrete bug." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different code paths: Qwen recurrent-state reset vs Whisper processor.decode token skipping." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression is distinct from generic ModelOutput key assignment behavior." + }, { "left": "issue:43452", "right": "issue:45588", "accept": false, - "reason": "Different code paths: tokenizer/gguf loading vs flash-attention sink handling." + "reason": "Tokenizer/model loading with gguf_file is unrelated to a flash_attention AttributeError." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "Both are regressions, but one is tensor-parallel OOM and the other is a checkpoint key mismatch." + "reason": "OOM during tensor-parallel model load vs resume_from_checkpoint key mismatch are different problems." }, { - "left": "issue:29127", - "right": "issue:35141", + "left": "issue:45003", + "right": "issue:45440", "accept": false, - "reason": "Unrelated bugs: LayoutLMv3 error messaging vs embedding reinitialization after resize." + "reason": "sys.modules access bug in modeling_utils is unrelated to DeepseekV3MoE implementation divergence." }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "Both mention initialization, but they describe different issues and model-specific constraints." + "reason": "Custom model initialization regression is too broad to equate with a specific patchtsmixer post_init restriction." }, { - "left": "issue:45072", - "right": "issue:45229", + "left": "issue:43723", + "right": "issue:45479", "accept": false, - "reason": "Bfloat16 dtype mismatch in CI vs multi-GPU CUDA OOM are different failure modes." + "reason": "Tokenizer loading regression and zero-loss sequence-classification bug are unrelated." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "Both involve memory pressure, but the causes are different: tp load OOM vs int4 reserved-memory growth." + "reason": "Both end in OOM, but one is tensor-parallel load OOM and the other is int4 quantization reserved-memory growth; different root causes." }, { "left": "issue:44977", "right": "issue:45584", "accept": false, - "reason": "Different subsystems and symptoms: flash-attention generation vs empty-transcription chat-template failure." + "reason": "Qwen3.5 flash-attention generation regression is unrelated to Whisper empty-transcription handling." }, { - "left": "issue:45588", - "right": "issue:45600", + "left": "issue:43761", + "right": "issue:44188", "accept": false, - "reason": "Flash-attention sink handling and auto_mappings stale config references are unrelated." + "reason": "Hidden_states regression in CLIPVisionModel is unrelated to attention-kernel branching under torch.compile." }, { - "left": "issue:33357", - "right": "issue:38175", + "left": "issue:45588", + "right": "issue:45600", "accept": false, - "reason": "Different models and behaviors: MacOS bus error vs zero-probability output." + "reason": "FlashAttention crash and stale auto_mappings references are unrelated issues." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Both are API discrepancies, but for different configs and different missing fields." + "reason": "API inconsistency in Gemma3TextConfig is not the same bug as Llama4Config missing pad_token_id." }, { "left": "issue:44479", "right": "issue:45290", "accept": false, - "reason": "Video-input regression in Qwen VL models is unrelated to assistant tool-call handling in chat templates." - }, - { - "left": "issue:44964", - "right": "issue:45440", - "accept": false, - "reason": "Model loading failure for Phi-4 multimodal vs DeepSeekV3 implementation divergence are distinct bugs." + "reason": "Video-input regression for Qwen variants is unrelated to chat template crashes on tool-call assistant messages." }, { - "left": "issue:43761", - "right": "issue:44079", + "left": "issue:43931", + "right": "issue:45216", "accept": false, - "reason": "Hidden-states omission in CLIPVisionModel is not the same as ModelOutput key assignment logic." + "reason": "Model weight-shape mismatch on load is different from a save_pretrained checkpoint regression." }, { - "left": "issue:43931", - "right": "issue:45216", + "left": "issue:43994", + "right": "issue:44206", "accept": false, - "reason": "Both are Qwen3-family load/save issues, but one is a shape-mismatch load error and the other is a bad saved checkpoint." + "reason": "SigLIP2 bad outputs and LasrFeatureExtractor center-arg crash are separate issues." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Save failure for dequantized models vs wrong custom-module loading after save_pretrained are different save/load bugs." + "reason": "Dequantized model save failure is different from loading the wrong custom module after save_pretrained." }, { "left": "issue:33290", "right": "issue:34567", "accept": false, - "reason": "OOM in deepspeed/adafactor is unrelated to TrainerState token counting." + "reason": "OOM with Adafactor/DeepSpeed has no relation to TrainerState.num_input_tokens_seen not updating." }, { "left": "issue:34689", "right": "issue:43611", "accept": false, - "reason": "Both are loading regressions, but they involve different root causes and model-loading paths." + "reason": "Both are model-loading regressions, but they affect different mechanisms and model classes." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "FastSpeech2ConformerTokenizer loading and Perceiver interpolation behavior are unrelated." }, { "left": "issue:39692", "right": "issue:45491", "accept": false, - "reason": "Documentation example errors are unrelated to Gemma3 NaN embeddings in mixed-length batching." + "reason": "Documentation example errors in SigLIP2 do not match the Gemma3 NaN embedding bug." }, { - "left": "issue:44442", - "right": "issue:44898", + "left": "issue:33290", + "right": "issue:35141", "accept": false, - "reason": "Tokenizer loading failure and Perceiver resolution failure are separate model-specific bugs." + "reason": "Adafactor/DeepSpeed OOM is unrelated to embedding reinitialization after resize_token_embeddings." }, { - "left": "issue:43761", - "right": "issue:44188", + "left": "issue:33290", + "right": "issue:34689", "accept": false, - "reason": "CLIP hidden_states regression is unrelated to attention-kernel branching under torch.compile." + "reason": "OOM in training is not the same underlying bug as a Llama model-loading regression." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation mismatch and stale docs referencing pipeline() are unrelated." + "reason": "SigLIP2 implementation mismatch and docs still using pipeline() are different issue types and code paths." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient checkpointing in eval mode and quantization offloading are different behaviors." - }, - { - "left": "issue:44568", - "right": "issue:45479", - "accept": false, - "reason": "Tokenizer special-token behavior is unrelated to the zero-loss classification bug." + "reason": "Gradient-checkpointing eval-mode limitation is unrelated to quantization offloading behavior." }, { "left": "issue:43550", "right": "issue:44743", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is not the same as Qwen recurrent-state reset." + "reason": "torch.compile+SDPA failure in Bamba is unrelated to Qwen recurrent state reset." }, { - "left": "issue:43295", - "right": "issue:45584", + "left": "issue:44077", + "right": "issue:44479", "accept": false, - "reason": "Processor/tokenizer regression in save/load is unrelated to Whisper empty-transcription generation failure." + "reason": "A patchtsmixer post_init policy issue does not match a Qwen video-input regression." }, { - "left": "issue:43994", - "right": "issue:44206", + "left": "issue:43316", + "right": "issue:45759", "accept": false, - "reason": "SigLIP2 inference quality issue and LasrFeatureExtractor crash are different defects." + "reason": "Gemma3TextConfig API discrepancy is unrelated to AutoModelForCausalLM.from_config unwrapping text_config for Qwen multimodal configs." } ] }, @@ -8313,18 +8337,18 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:16998", + "issue:22355", "issue:25251", "issue:28282", - "issue:29127", - "issue:32090", "issue:33290", - "issue:33357", "issue:34689", "issue:36010", + "issue:36331", "issue:41628", "issue:41950", "issue:42175", - "issue:43232", + "issue:43065", "issue:43295", "issue:43316", "issue:43381", @@ -8333,268 +8357,239 @@ "issue:43441", "issue:43475", "issue:43540", - "issue:43550", - "issue:43606", "issue:43643", - "issue:43701", + "issue:43673", + "issue:43688", "issue:43720", "issue:43742", + "issue:43749", "issue:43825", "issue:43856", "issue:43873", + "issue:43931", "issue:44077", "issue:44162", "issue:44190", "issue:44442", - "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:45372", - "issue:45584", - "issue:45656", - "issue:45759" + "issue:45584" ], "soft_pairs": [ - "issue:43316|issue:45759", "issue:41628|issue:45372", "issue:43643|issue:44913", "issue:43720|issue:44442", - "issue:32090|issue:33290", + "issue:43749|issue:43931", "issue:44190|issue:44496", "issue:43873|issue:44871", + "issue:33290|issue:36331", "issue:34689|issue:36010", "issue:43381|issue:43856", - "issue:44077|issue:44479", "issue:44855|issue:45137", - "issue:33290|issue:45656", + "issue:43673|issue:44743", "issue:43540|issue:44514", "issue:43316|issue:44877", - "issue:29127|issue:33357", "issue:43381|issue:43742", "issue:42175|issue:43316", "issue:43404|issue:43441", - "issue:33290|issue:34689", "issue:25251|issue:28282", "issue:44162|issue:44792", "issue:41950|issue:43475", - "issue:43232|issue:43388", - "issue:43295|issue:43825", - "issue:44871|issue:45584", - "issue:45127|issue:45200", - "issue:43550|issue:43701", - "issue:43606|issue:43701", - "issue:43295|issue:44871" + "issue:43388|issue:43688", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43065|issue:44617", + "issue:44077|issue:44534", + "issue:44871|issue:45584" ], - "prepared_review_unit_hash": "7c89439a7dd20679f763df775b7ab3684c00bc9edd50dbfeea8717b8992341d9", + "prepared_review_unit_hash": "81a29fb1bcd5c811c2bfa0e89e535db594b0ccfbceb1395c895561e07d3f6ff0", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36559, - "estimated_input_tokens": 9140, - "estimated_eval_tokens": 18536 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 25, + "serialized_chars": 34485, + "estimated_input_tokens": 8622, + "estimated_eval_tokens": 17500 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is heterogeneous: it spans unrelated bugs in config loading, generation, multimodal processors, deepspeed, offloading, and training scripts. I don\u2019t see evidence that any soft pair is the same underlying issue, so all candidate duplicate links should be rejected.", + "summary": "This cluster is not a duplicate set: the soft pairs are mostly broad-theme lookalikes (loading, configs, processors, cache, memory) but each refers to a distinct concrete bug or model-specific regression. None are safe to merge as the same underlying issue.", "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue stands out because the items are not one bug cluster; if a representative issue must be chosen, issue:41950 is the clearest open, concrete, and actionable report.", + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:41950 is the best representative issue only in the sense that it is open, specific, and has a clear failure mode; however, it is not a duplicate of the rest of the set.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43316", - "right": "issue:45759", - "accept": false, - "reason": "Both involve config-related multimodal/model behavior, but they describe different bugs: one is a Gemma3TextConfig API mismatch, the other is from_config not unwrapping text_config for Qwen composite configs." - }, { "left": "issue:41628", "right": "issue:45372", "accept": false, - "reason": "Both are import errors, but they stem from different missing symbols and different code paths (AutoImageProcessor vs ReasoningEffort/Gemma 4 processor loading)." + "reason": "Both are import/processor-loading failures, but they involve different missing symbols and different dependency chains, so they are not the same bug." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "Both concern config fields surviving load/reload, but the reported failures are different: missing fields with trust_remote_code vs rotary_pct resetting on GPTNeoXConfig reload." + "reason": "Both concern config values not surviving load/reload, but one is a trust_remote_code field-loss issue and the other is a specific rotary_pct round-trip regression." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "Unrelated failures: packed BitNet weights during accelerate loading vs AutoTokenizer failing to load a specific tokenizer." + "reason": "Unrelated failures: BitNet accelerate-loading unpacking vs AutoTokenizer loading a custom tokenizer class." }, { - "left": "issue:32090", - "right": "issue:33290", + "left": "issue:43749", + "right": "issue:43931", "accept": false, - "reason": "Both mention Trainer/DeepSpeed, but one is a NoneType broadcast error and the other is an optimizer OOM; these are not the same bug." + "reason": "Both happen during model loading, but one is broken FSDP CPU RAM efficient loading and the other is a shape-mismatch for a specific Qwen3-VL model." }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Different loading failures: local dataset handling in a no-trainer script vs an unrecognized model config missing model_type." + "reason": "These are different problems in different code paths: local dataset loading in a training script versus unrecognized model/config loading." }, { "left": "issue:43873", "right": "issue:44871", "accept": false, - "reason": "One is a quantization/offloading memory behavior issue; the other is an eos_token_id configuration mismatch in Gemma-3." + "reason": "Quantization/offloading behavior and eos_token_id configuration are separate issues with different root causes." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "One is an OOM in Adafactor+DeepSpeed; the other is a Trainer API signature mismatch." }, { "left": "issue:34689", "right": "issue:36010", "accept": false, - "reason": "Both are model-loading/import regressions, but they are distinct problems affecting different APIs and code paths." + "reason": "A model-loading regression for Llama 3.2 Vision is unrelated to the GenerationMixin import error." }, { "left": "issue:43381", "right": "issue:43856", "accept": false, - "reason": "Gradient checkpointing in eval mode and Qwen3 MoE memory usage are separate training behaviors, not the same defect." - }, - { - "left": "issue:44077", - "right": "issue:44479", - "accept": false, - "reason": "Different model/runtime failures: optional post_init handling for patchtsmixer vs video-input regression for Qwen VL models." + "reason": "Gradient checkpointing in eval mode is not the same as inefficient MoE training memory usage." }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 decorator/comment parsing failure in DebertaV2Model is unrelated to the DeepSpeed ZeRO3 deque IndexError." + "reason": "Python 3.13 torch.jit parsing failure and DeepSpeed ZeRO3 deque underflow are unrelated." }, { - "left": "issue:33290", - "right": "issue:45656", + "left": "issue:43673", + "right": "issue:44743", "accept": false, - "reason": "Both involve DeepSpeed optimizer behavior, but one reports OOM with Adafactor while the other reports a duplicated optimizer step call." + "reason": "Both mention cache, but one is missing GenerationMixin cache in chunked_prefill and the other is recurrent-state reset in a Qwen3 modular model." }, { "left": "issue:43540", "right": "issue:44514", "accept": false, - "reason": "Both are multimodal/video-related, but one is a Qwen3OmniMoe video processing ValueError and the other is a batched apply_chat_template crash in Qwen2_5_VLProcessor." + "reason": "Different multimodal processor bugs: video input handling in Qwen3OmniMoe versus batched chat template crashes in Qwen2.5-VL." }, { "left": "issue:43316", "right": "issue:44877", "accept": false, - "reason": "Both involve config objects, but they are different bugs: API inconsistency in Gemma3TextConfig vs strict config blocking granite_speech loading." - }, - { - "left": "issue:29127", - "right": "issue:33357", - "accept": false, - "reason": "Completely unrelated issues: LayoutLMv3 error messaging vs a MacOS bus error on a CLIP model." + "reason": "One is an API discrepancy in Gemma3TextConfig, the other is strict config validation blocking granite_speech loading." }, { "left": "issue:43381", "right": "issue:43742", "accept": false, - "reason": "No overlap beyond generic training/loading context; gradient checkpointing eval behavior is unrelated to MobileLLM key errors." + "reason": "Gradient checkpointing/eval-mode behavior does not match a key error while loading MobileLLM-125M." }, { "left": "issue:42175", "right": "issue:43316", "accept": false, - "reason": "Install/backend packaging for TensorFlow is unrelated to a Gemma3TextConfig API discrepancy." + "reason": "TensorFlow packaging omission and Gemma3TextConfig API mismatch are unrelated." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Both touch Mistral-family models, but one is tied weights in model conversion/loading and the other is a FlashAttention runtime failure; not the same code-path bug." - }, - { - "left": "issue:33290", - "right": "issue:34689", - "accept": false, - "reason": "Different failures in different subsystems: DeepSpeed Adafactor OOM vs Llama 3.2 Vision model loading regression." + "reason": "Both are Mistral-family issues, but one is tied lm_head weights and the other is a FlashAttention failure; different code paths." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline top_k nesting behavior is unrelated to an AutoModel PyTorch import error." + "reason": "Pipeline top_k output-shape change and missing PyTorch import are entirely different issues." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "ESM2 breakage and a janus test failure are different model-specific issues." + "reason": "ESM2 breakage and a janus image-generation test failure are unrelated." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Both are vision/video model reports, but one is a pipeline processor-selection bug and the other is a missing attribute in Sam3VisionEncoderOutput." + "reason": "Video-classification processor selection and Sam3VisionEncoderOutput missing an attribute are different bugs." }, { - "left": "issue:43232", - "right": "issue:43388", + "left": "issue:43388", + "right": "issue:43688", "accept": false, - "reason": "Generation kwargs after sync_gpus and metric gathering label truncation are unrelated problems." + "reason": "The former is a gather_for_metrics last-batch label truncation bug; the latter is auxiliary-loss normalization in MoE models." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Processor/tokenizer regression and pipeline error-message wording are different bugs with different symptoms and fixes." - }, - { - "left": "issue:44871", - "right": "issue:45584", - "accept": false, - "reason": "Both involve special tokens, but one is a config inconsistency in Gemma-3 and the other is Whisper failing on empty transcription after align_special_tokens." + "reason": "Both are v5 regressions, but one is about processor.tokenizer/image-tokenizer behavior and the other is a pipeline error message about translation support." }, { - "left": "issue:45127", - "right": "issue:45200", + "left": "issue:16998", + "right": "issue:22355", "accept": false, - "reason": "LoRA merge with extended vocab causing collapse is unrelated to Gemma 4 requiring mm_token_type_ids defaults." + "reason": "Model_max_length on DeBERTa-V3 and a missing transformers.onnx module are unrelated." }, { - "left": "issue:43550", - "right": "issue:43701", + "left": "issue:43065", + "right": "issue:44617", "accept": false, - "reason": "A torch.compile/SDPA model failure is unrelated to a resume_from_checkpoint key mismatch." + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video CUDA OOM." }, { - "left": "issue:43606", - "right": "issue:43701", + "left": "issue:44077", + "right": "issue:44534", "accept": false, - "reason": "CPU offload device mismatch for bark-small is unrelated to the checkpoint key mismatch issue." + "reason": "Optional post_init handling in patchtsmixer is unrelated to non-persistent buffers being filled with junk in v5." }, { - "left": "issue:43295", - "right": "issue:44871", + "left": "issue:44871", + "right": "issue:45584", "accept": false, - "reason": "These are separate regressions: processor.tokenizer access in custom code vs Gemma-3 eos_token_id inconsistency." + "reason": "Gemma-3 eos_token_id inconsistency and Whisper empty-transcription generation failure are different model-specific bugs." } ] }, @@ -8618,17 +8613,18 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:33290", + "issue:32090", "issue:35141", - "issue:36331", "issue:41720", "issue:42617", "issue:43065", - "issue:43335", + "issue:43295", "issue:43452", + "issue:43550", "issue:43575", + "issue:43606", "issue:43646", - "issue:43673", + "issue:43701", "issue:43742", "issue:43761", "issue:43827", @@ -8637,7 +8633,6 @@ "issue:43931", "issue:43994", "issue:44060", - "issue:44077", "issue:44190", "issue:44242", "issue:44464", @@ -8648,257 +8643,265 @@ "issue:44610", "issue:44617", "issue:44734", - "issue:44743", "issue:44792", - "issue:44855", "issue:44871", "issue:44913", + "issue:44933", "issue:44936", "issue:44964", "issue:44977", "issue:45127", + "issue:45200", "issue:45245", "issue:45290", - "issue:45357", + "issue:45362", "issue:45381", + "issue:45406", + "issue:45563", "issue:45584", - "issue:45588", "issue:45600", "issue:45698" ], "soft_pairs": [ - "issue:43673|issue:44743", "issue:43761|issue:44479", - "issue:33290|issue:35141", - "issue:33290|issue:36331", - "issue:43065|issue:44617", + "issue:45127|issue:45200", + "issue:43550|issue:43701", + "issue:43606|issue:43701", + "issue:43295|issue:44871", "issue:43452|issue:44496", + "issue:44734|issue:45406", "issue:43575|issue:43742", - "issue:44936|issue:45698", "issue:43646|issue:44913", + "issue:44936|issue:45698", "issue:42617|issue:43575", "issue:43761|issue:44242", - "issue:43575|issue:43873", - "issue:44077|issue:44534", "issue:43931|issue:44060", - "issue:44479|issue:44734", "issue:44617|issue:45584", + "issue:43575|issue:43873", + "issue:32090|issue:35141", + "issue:45563|issue:45584", + "issue:44521|issue:45381", + "issue:44479|issue:44734", "issue:44464|issue:44871", "issue:43065|issue:44610", - "issue:43827|issue:43994", "issue:43906|issue:45290", + "issue:43827|issue:43994", + "issue:44933|issue:45362", "issue:44792|issue:44977", "issue:44534|issue:44913", + "issue:44464|issue:45563", "issue:41720|issue:45245", "issue:44190|issue:44964", - "issue:44792|issue:45600", - "issue:44855|issue:45588", - "issue:44521|issue:45381", - "issue:45127|issue:45357", - "issue:43335|issue:44855" + "issue:44792|issue:45600" ], - "prepared_review_unit_hash": "8424c1f4cb75b643a62837b83d6e81d61b5e4e1177be761e74dafa34ad37bdb4", + "prepared_review_unit_hash": "a90eaf00195ad706af739756993f85d791ceeca7134784d223a358beed820ce9", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35843, - "estimated_input_tokens": 8961, - "estimated_eval_tokens": 18178 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36551, + "estimated_input_tokens": 9138, + "estimated_eval_tokens": 18532 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a heterogeneous mix of unrelated bugs/regressions (OOMs, config/load failures, generation/cache issues, multimodal model regressions, and docs/CI problems). I do not see a meaningful duplicate set or a single canonical issue for the group.", + "summary": "This is a heterogeneous bundle of unrelated issue reports across training, loading, multimodal, serving, and docs regressions. None of the soft-similarity pairs look like the same underlying bug/change, so all soft edges are rejected.", "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue: the items span distinct code paths and failure modes, so the cluster looks like a false-positive mix rather than one underlying bug.", + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No global best issue is suitable because there is no coherent duplicate family to represent; the issues are too diverse in subsystem and root cause.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43673", - "right": "issue:44743", + "left": "issue:43761", + "right": "issue:44479", "accept": false, - "reason": "Both involve cache/state behavior, but one is a v5 chunked-prefill cache missing bug and the other is Qwen3.5 recurrent-state reset with seq_len>1; different model paths and causes." + "reason": "Both are regression-style failures, but one is CLIPVision hidden_states handling and the other is video-input position ids in Qwen VL models; different code paths." }, { - "left": "issue:43761", - "right": "issue:44479", + "left": "issue:45127", + "right": "issue:45200", "accept": false, - "reason": "Both are multimodal regressions, but CLIPVision hidden_states handling and Qwen video input position handling are different bugs in different models." + "reason": "LoRA/extended-vocab collapse with tied embeddings is unrelated to Gemma 4's mm_token_type_ids defaulting issue." }, { - "left": "issue:33290", - "right": "issue:35141", + "left": "issue:43550", + "right": "issue:43701", "accept": false, - "reason": "OOM with Adafactor/DeepSpeed is unrelated to token embedding reinitialization after resize and post_init." + "reason": "torch.compile SDPA failure for Bamba is unrelated to a resume_from_checkpoint key mismatch." }, { - "left": "issue:33290", - "right": "issue:36331", + "left": "issue:43606", + "right": "issue:43701", "accept": false, - "reason": "Memory blowup in training infrastructure is not the same as a CustomTrainer compute_loss signature mismatch." + "reason": "CPU offload device mismatch and checkpoint key mismatch are different bugs in different subsystems." }, { - "left": "issue:43065", - "right": "issue:44617", + "left": "issue:43295", + "right": "issue:44871", "accept": false, - "reason": "Same model family, but one is a dummy Conv2d implementation issue and the other is a CUDA OOM during video inference; not the same fix." + "reason": "Processor.tokenizer regression is not the same as inconsistent eos_token_id config on Gemma-3." }, { "left": "issue:43452", "right": "issue:44496", "accept": false, - "reason": "Both are model-loading failures, but one is gguf/AutoTokenizer-AutoModel handling and the other is missing model_type in config.json." + "reason": "gguf_file loading breakage and unrecognized model/config missing model_type are distinct model-loading failures." }, { - "left": "issue:43575", - "right": "issue:43742", + "left": "issue:44734", + "right": "issue:45406", "accept": false, - "reason": "Both fail on loading, but one is TP-related OOM for Qwen2-57B and the other is a config/key error for MobileLLM." + "reason": "KV-cache continuation indexing in serve is unrelated to Gemma4Processor missing _tokenizer." }, { - "left": "issue:44936", - "right": "issue:45698", + "left": "issue:43575", + "right": "issue:43742", "accept": false, - "reason": "trainer.evaluate after train and wrong custom module reload after save_pretrained are separate lifecycle bugs." + "reason": "Tensor-parallel OOM on Qwen3-57B is not the same as a key error loading MobileLLM-125M." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Transformers 5 custom model init breakage is unrelated to GPTNeoX rotary_pct not persisting on reload." + "reason": "Custom model initialization breakage and GPTNeoXConfig rotary_pct reload behavior are different config/init bugs." + }, + { + "left": "issue:44936", + "right": "issue:45698", + "accept": false, + "reason": "trainer.evaluate() after train() failure is unrelated to save_pretrained loading the wrong custom module." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "3d_parallel.py execution failure is unrelated to Qwen2 TP OOM during model loading." + "reason": "3d_parallel.py runtime failure is not the same underlying issue as Qwen3 tensor-parallel OOM." }, { "left": "issue:43761", "right": "issue:44242", "accept": false, - "reason": "CLIPVision hidden_states regression and missing load-balancing loss are different model components and failure modes." + "reason": "CLIPVision hidden_states being None is unrelated to MoE load-balancing loss omission." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-Next tied-weights warning point to different model/config problems." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video CUDA OOM is unrelated to Whisper failing on empty transcription after align_special_tokens." }, { "left": "issue:43575", "right": "issue:43873", "accept": false, - "reason": "Both mention loading/memory, but TP OOM and quantization offloading not working are distinct problems." + "reason": "Tensor-parallel OOM and quantization offload not working are both memory-related but not the same code-path bug." }, { - "left": "issue:44077", - "right": "issue:44534", + "left": "issue:32090", + "right": "issue:35141", "accept": false, - "reason": "Optional post_init restriction in patchtsmixer is unrelated to junk-filled non-persistent buffers in v5." + "reason": "Trainer _gpu_broadcast_one NoneType error is unrelated to post_init reinitializing output embeddings after resizing." }, { - "left": "issue:43931", - "right": "issue:44060", + "left": "issue:45563", + "right": "issue:45584", "accept": false, - "reason": "Weight shape mismatch in Qwen3-VL and incorrect tied-weights warning in Qwen3-Next are different bugs." + "reason": "Paged-generate warning behavior is unrelated to Whisper's empty-transcription crash." }, { - "left": "issue:44479", - "right": "issue:44734", + "left": "issue:44521", + "right": "issue:45381", "accept": false, - "reason": "Video-input regression in Qwen multimodal models is unrelated to KV-cache continuation tensor indexing in transformers serve." + "reason": "All-zero assistant masks for multimodal inputs and wrong video vision_position_ids are separate multimodal preprocessing bugs." }, { - "left": "issue:44617", - "right": "issue:45584", + "left": "issue:44479", + "right": "issue:44734", "accept": false, - "reason": "Sam3Video CUDA OOM and Whisper empty-transcription failure after align_special_tokens are unrelated." + "reason": "Video input regression in Qwen VL models is unrelated to serve KV-cache continuation tensor indexing." }, { "left": "issue:44464", "right": "issue:44871", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is unrelated to Gemma eos_token_id config mismatch." + "reason": "Chunked generation inconsistency under compiled forward is not the same as Gemma eos_token_id config mismatch." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Sam3PixelDecoder dummy layer issue is unrelated to OmDet-Turbo processor/model input-size mismatch." + "reason": "Dummy Conv2d in Sam3PixelDecoder and wrong processor image size are different SAM3 problems." + }, + { + "left": "issue:43906", + "right": "issue:45290", + "accept": false, + "reason": "An isolated reproduction of issue 38071 does not match the specific tokenize=True crash with tool-call assistant messages." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs still referencing pipeline() and Siglip2 nonsensical outputs are different issues." + "reason": "Docs still referencing pipeline() and nonsensical SigLIP2 outputs are unrelated." }, { - "left": "issue:43906", - "right": "issue:45290", + "left": "issue:44933", + "right": "issue:45362", "accept": false, - "reason": "A reproduction of issue 38071 is not the same as apply_chat_template crashing on assistant tool-call messages." + "reason": "Missing image_utils import and Qwen3.5 chat crash are separate failures." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation bug are unrelated." + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation bug are not the same bug." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer junk in v5 and GPTNeoX rotary_pct reload default are distinct configuration/runtime bugs." + "reason": "Non-persistent buffers filled with junk is unrelated to rotary_pct not persisting on reload." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and stale num_return_sequences warning are different issues." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert is unrelated to the 2^24 category limit RuntimeError." + "reason": "Qwen3 auto device-mapping CUDA assert is unrelated to the 2^24 category limit error." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Local dataset loading in no-trainer image classification and Phi-4 multimodal model loading are different paths." + "reason": "Local dataset loading in no-trainer script is unrelated to Phi-4 multimodal model loading failure." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "Janus test failure is unrelated to stale auto_mappings references for removed Sam3LiteText configs." - }, - { - "left": "issue:44855", - "right": "issue:45588", - "accept": false, - "reason": "Python 3.13 IndentationError in DebertaV2 import is unrelated to flash_attention's s_aux=None crash." - }, - { - "left": "issue:44521", - "right": "issue:45381", - "accept": false, - "reason": "Both are multimodal/chat-template issues, but all-zero assistant masks and wrong vision_position_ids are different concrete bugs." - }, - { - "left": "issue:45127", - "right": "issue:45357", - "accept": false, - "reason": "LoRA merge collapse with extended vocab is unrelated to incorrect visual encoder keys saved by Qwen3.5." - }, - { - "left": "issue:43335", - "right": "issue:44855", - "accept": false, - "reason": "SwitchTransformers sparse-layer creation and a Python 3.13 DebertaV2 import parse failure are unrelated." + "reason": "Janus image-generation test failure and auto_mappings Sam3LiteText CI breakage are unrelated." } ] }, @@ -8926,267 +8929,297 @@ "issue:38175", "issue:38617", "issue:41628", + "issue:41950", + "issue:42175", "issue:43065", "issue:43116", "issue:43295", "issue:43335", + "issue:43388", "issue:43404", "issue:43479", "issue:43540", "issue:43550", - "issue:43643", "issue:43644", "issue:43650", "issue:43761", "issue:43976", + "issue:44016", "issue:44077", + "issue:44079", "issue:44112", "issue:44186", "issue:44190", "issue:44220", - "issue:44246", - "issue:44464", + "issue:44279", "issue:44466", - "issue:44496", "issue:44561", "issue:44683", "issue:44734", "issue:44743", "issue:44855", - "issue:44877", "issue:44913", - "issue:44933", "issue:44938", - "issue:45003", + "issue:45127", "issue:45216", "issue:45245", "issue:45276", - "issue:45362", + "issue:45357", "issue:45375", - "issue:45406", "issue:45446", "issue:45542", - "issue:45563", "issue:45584", + "issue:45588", "issue:45600", "issue:45632" ], "soft_pairs": [ + "issue:44855|issue:45588", + "issue:43388|issue:44079", + "issue:43761|issue:44743", + "issue:45127|issue:45357", + "issue:43335|issue:44855", "issue:44938|issue:45542", + "issue:44077|issue:44079", "issue:43976|issue:44855", - "issue:44077|issue:44877", "issue:43644|issue:44913", "issue:43550|issue:44190", - "issue:43761|issue:44743", "issue:43540|issue:44734", "issue:45276|issue:45584", - "issue:43650|issue:44220", "issue:44112|issue:45600", + "issue:43650|issue:44220", "issue:43116|issue:44190", "issue:43295|issue:45276", "issue:43335|issue:44190", - "issue:45245|issue:45584", + "issue:41950|issue:44016", "issue:36246|issue:41628", + "issue:45245|issue:45584", "issue:43404|issue:45216", "issue:43479|issue:45375", - "issue:44933|issue:45362", + "issue:36246|issue:38617", "issue:44734|issue:45632", - "issue:44464|issue:45563", + "issue:42175|issue:44279", "issue:44186|issue:44734", "issue:38175|issue:43065", + "issue:44016|issue:44220", "issue:44561|issue:45632", "issue:44683|issue:45446", - "issue:44466|issue:44913", - "issue:44246|issue:45003", - "issue:36246|issue:38617", - "issue:45406|issue:45632", - "issue:43643|issue:44496", - "issue:45563|issue:45584" + "issue:44466|issue:44913" ], - "prepared_review_unit_hash": "5b0a488241905aeea3224192b954b039470a9e9718dea880e65c69d4e9af6f38", + "prepared_review_unit_hash": "d75f4dbc967ee3c049ce5ee3c1cfeab9090add6b12b00b3ee84149020caa87e3", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36765, - "estimated_input_tokens": 9192, - "estimated_eval_tokens": 18640 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36652, + "estimated_input_tokens": 9163, + "estimated_eval_tokens": 18582 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is highly heterogeneous: the soft pairs mostly share only broad themes like imports, configs, or regressions, but they do not describe the same concrete bug/change. I would not merge any of the proposed pairs.", - "confidence": 0.89, - "canonical_issue_reason": null, + "summary": "This cluster is mostly a grab-bag of unrelated issues rather than a duplicate set. None of the soft pairs look like the same underlying bug, so I would not merge any of them; if forced to pick an anchor, I\u2019d use the processor/tokenizer regression issue as the broadest representative.", + "confidence": 0.28, + "canonical_issue_reason": "issue:43295 is the most representative high-signal regression in the set: it is a concrete, user-facing break in core processor/tokenizer behavior, and it is broader than the many narrow model-specific or environment-specific bugs here.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong duplicate hub for this set. If a single representative is needed, issue 44246 is the broadest user-facing runtime/reporting bug with active discussion, but it is not a true stand-in for the others.", + "best_issue_reason": "As a global representative, issue:43295 is the best anchor because it is a clear upstream API regression with practical impact and a cleaner triage target than the surrounding one-off model, config, or tooling issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44938", - "right": "issue:45542", + "left": "issue:44855", + "right": "issue:45588", "accept": false, - "reason": "Different failures: Python 3.14 import/load problem vs missing TensorFlow backend when only tensorboard is installed." + "reason": "Different failure modes: Python import parsing on 3.13 vs flash_attention crashing on s_aux=None." }, { - "left": "issue:43976", + "left": "issue:43388", + "right": "issue:44079", + "accept": false, + "reason": "Different bugs: metric gathering drops tuple labels, while ModelOutput keys are assigned incorrectly when a field was previously None." + }, + { + "left": "issue:43761", + "right": "issue:44743", + "accept": false, + "reason": "Different models and code paths: CLIPVisionModel hidden_states regression vs Qwen3_5 recurrent-state reset with cache." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Both involve saving/weights, but one is LoRA merge with extended vocab collapse and the other is incorrect visual encoder keys in save_pretrained." + }, + { + "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "Both are Python-version-related import errors, but they involve different models, symptoms, and root causes." + "reason": "Unrelated: SwitchTransformers sparse-layer construction vs an import-time IndentationError in DebertaV2Model." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "One is a Python 3.14 load failure; the other is a TensorFlow-backend dependency/install issue." }, { "left": "issue:44077", - "right": "issue:44877", + "right": "issue:44079", + "accept": false, + "reason": "Different subsystems: patchtsmixer post_init handling vs ModelOutput key assignment logic." + }, + { + "left": "issue:43976", + "right": "issue:44855", "accept": false, - "reason": "Both touch config validation, but one is about optional post_init on patchtsmixer and the other about loading granite_speech config." + "reason": "Python version compatibility break vs an unrelated Deberta import parsing error." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "One is a non-persistent buffer initialization regression; the other is a config field not persisting on reload." + "reason": "Non-persistent buffer junk values and GPTNeoX rotary_pct reload are separate state/serialization bugs." }, { "left": "issue:43550", "right": "issue:44190", "accept": false, - "reason": "Completely different areas: SDPA/torch.compile on a model vs example script dataset loading." - }, - { - "left": "issue:43761", - "right": "issue:44743", - "accept": false, - "reason": "Different models and different bugs: CLIP hidden_states omission vs recurrent state reset with cache." + "reason": "Bamba torch.compile SDPA failure is unrelated to the local-dataset loading issue in the image classification script." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Video input validation in Qwen3OmniMoe is unrelated to serve KV-cache continuation indexing." + "reason": "Qwen3OmniMoe video-input handling and serve KV-cache continuation indexing are distinct code paths." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Token embedding resize wiring vs Whisper empty-transcription generation failure are distinct issues." + "reason": "Gemma4 embedding-resize wiring and Whisper empty-transcription generation are unrelated bugs." }, { - "left": "issue:43650", - "right": "issue:44220", + "left": "issue:44112", + "right": "issue:45600", "accept": false, - "reason": "The titles are unrelated; one is not a real technical bug description and the other is a feature-specific audio feature extraction issue." + "reason": "CI test flakiness in GraniteSpeech and stale auto_mappings references to removed configs are different problems." }, { - "left": "issue:44112", - "right": "issue:45600", + "left": "issue:43650", + "right": "issue:44220", "accept": false, - "reason": "Both mention CI, but they are different failures in different code paths and different models." + "reason": "Placeholder issue versus an actual _torch_extract_fbank_features() bug; no shared underlying defect." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Example-script multi-label classification output bug vs local dataset loading failure are not the same problem." + "reason": "Different example-script failures: empty multi-label classification results vs local dataset loading." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Processor/tokenizer regression is unrelated to token embedding resizing behavior." + "reason": "Both touch multimodal/embedding behavior, but one is a processor.tokenizer regression and the other is resize_token_embeddings not wiring outputs correctly." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "SwitchTransformers config layer construction bug is unrelated to the image-classification dataset loader issue." + "reason": "SwitchTransformers config initialization and a training-script dataset-loading failure are unrelated." }, { - "left": "issue:45245", - "right": "issue:45584", + "left": "issue:41950", + "right": "issue:44016", "accept": false, - "reason": "Generic category-count runtime error is different from Whisper post-processing/empty transcription handling." + "reason": "Video-classification pipeline bug and notebook syntax error are clearly different issues." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they expose different missing symbols from different modules and likely different regressions." + "reason": "Both are import errors, but they involve different missing symbols and different API surfaces, so they are not the same bug." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Runtime category-limit error and Whisper empty-transcription failure are unrelated." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Weight tying in Mistral3 is a different bug from incorrect Qwen3.5 checkpoint saving." + "reason": "lm_head weight tying in Mistral3ForConditionalGeneration is not the same as Qwen3.5 save_pretrained writing wrong visual encoder keys." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Both involve multimodal config fields, but they affect different models and different missing/defaulted fields." + "reason": "One is default config initialization from None, the other is a missing field being dropped by strict serialization." }, { - "left": "issue:44933", - "right": "issue:45362", + "left": "issue:36246", + "right": "issue:38617", "accept": false, - "reason": "A missing image_utils import is unrelated to Qwen3.5 chat crashes." + "reason": "Different import failures: Qwen2_5_VLImageProcessor missing vs layer_type_validation missing from configuration_utils." }, { "left": "issue:44734", "right": "issue:45632", "accept": false, - "reason": "Serve KV-cache indexing crash and trust_remote_code cache-path collision are unrelated." + "reason": "Serve KV-cache indexing crash and local-model cache-path collision are unrelated." }, { - "left": "issue:44464", - "right": "issue:45563", + "left": "issue:42175", + "right": "issue:44279", "accept": false, - "reason": "Compiled forward generation inconsistency is not the same as a stale warning about num_return_sequences." + "reason": "Both mention dependencies, but they are too vague and do not share a concrete package/version or identical failure mode." }, { "left": "issue:44186", "right": "issue:44734", "accept": false, - "reason": "Tokenizer padding/NER crash is unrelated to serve-side KV-cache tensor indexing." + "reason": "Tokenizer padding/truncation crash in NER is unrelated to serve KV-cache continuation indexing." }, { "left": "issue:38175", "right": "issue:43065", "accept": false, - "reason": "Zero probabilities in SigLIP2 and dummy Conv2d in Sam3PixelDecoder are entirely different model bugs." + "reason": "Unexpected zero probabilities in SigLIP2 and a dummy Conv2d in Sam3PixelDecoder are different model bugs." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error is unrelated to an fbank feature-extraction issue." }, { "left": "issue:44561", "right": "issue:45632", "accept": false, - "reason": "Removal of is_torch_fx_available affects remote-code imports broadly; cache-path collisions are a separate trust_remote_code bug." + "reason": "Removal of is_torch_fx_available and cache-path collisions for local models are different regressions." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Torch >= 2.9 flex_attention failure and incorrect version check for AuxRequest import are different flex_attention issues." + "reason": "Both involve flex_attention, but one is compiled flex_attention on torch>=2.9 and the other is a wrong version check for AuxRequest import; not the same concrete bug." }, { "left": "issue:44466", "right": "issue:44913", "accept": false, - "reason": "Serialization of tied lm_head weights is unrelated to GPTNeoX rotary_pct reload behavior." - }, - { - "left": "issue:44246", - "right": "issue:45003", - "accept": false, - "reason": "Import slowdowns and unsafe sys.modules access may be related in spirit, but they are not the same concrete bug." + "reason": "Tied-weight serialization inconsistency and GPTNeoX rotary_pct reset on reload are separate config/serialization problems." } ] }, @@ -9211,22 +9244,19 @@ "cluster_id": "cluster-15354-384", "nodes": [ "issue:41950", - "issue:42175", "issue:42907", "issue:43295", "issue:43299", "issue:43381", - "issue:43388", "issue:43441", "issue:43479", "issue:43525", "issue:43526", "issue:43531", "issue:43577", + "issue:43643", "issue:43673", - "issue:43688", "issue:43824", - "issue:43827", "issue:43854", "issue:43901", "issue:43976", @@ -9234,270 +9264,280 @@ "issue:44062", "issue:44075", "issue:44112", - "issue:44220", "issue:44242", + "issue:44246", "issue:44279", "issue:44336", "issue:44351", "issue:44368", "issue:44496", - "issue:44683", + "issue:44521", "issue:44933", "issue:44964", + "issue:44987", + "issue:45003", "issue:45020", "issue:45042", "issue:45200", "issue:45216", "issue:45278", + "issue:45325", "issue:45335", "issue:45357", "issue:45405", - "issue:45542", + "issue:45406", "issue:45584", "issue:45593", "issue:45600", + "issue:45632", "issue:45759" ], "soft_pairs": [ - "issue:41950|issue:44016", + "issue:44246|issue:45003", + "issue:45406|issue:45632", + "issue:43643|issue:44496", "issue:44964|issue:45600", - "issue:42907|issue:43441", - "issue:43388|issue:43688", + "issue:44016|issue:44062", "issue:43295|issue:45335", + "issue:42907|issue:43441", "issue:44112|issue:44964", "issue:44496|issue:45042", "issue:43854|issue:44964", - "issue:44016|issue:44220", "issue:43673|issue:44336", + "issue:43976|issue:44987", "issue:43299|issue:45216", - "issue:42175|issue:44279", "issue:43531|issue:45600", + "issue:43824|issue:44933", "issue:43526|issue:43901", "issue:45335|issue:45584", + "issue:43976|issue:44279", "issue:44075|issue:44368", "issue:43381|issue:45593", "issue:43299|issue:45357", "issue:44496|issue:45759", + "issue:44521|issue:45325", "issue:41950|issue:45020", "issue:43901|issue:44496", "issue:44242|issue:45200", - "issue:44016|issue:44062", "issue:43577|issue:45216", "issue:43976|issue:45405", "issue:44351|issue:45278", - "issue:43824|issue:44933", - "issue:43388|issue:43827", - "issue:43479|issue:43525", - "issue:44683|issue:45542" + "issue:43479|issue:43525" ], - "prepared_review_unit_hash": "cd2e966ad2fb706b9e7214bbaa23986451317590debb6c1e0c4cf1ed98d2c8b5", + "prepared_review_unit_hash": "94a3ef83fddb90a27760172557c17e33859e121aff98138d20d0237c8bf31512", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36062, - "estimated_input_tokens": 9016, - "estimated_eval_tokens": 18288 + "serialized_chars": 36669, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a mix of unrelated Transformers regressions, import errors, and docs/test issues. None of the soft-pair candidates look like true duplicates.", - "confidence": 0.84, - "canonical_issue_reason": "issue:45020 is the broadest representative report here: a general regression affecting remote_code model loading across recent Transformers versions.", + "summary": "These issues cluster around Transformers regressions, but the specific failures are too different to treat any pair as duplicates; I would keep the broad open regression report as the cluster anchor.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 45020 is the broadest open umbrella report in the set, covering recent-version breakage around remote-code/model loading.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the best umbrella issue in this set because it captures the widest upstream regression surface.", + "best_issue_reason": "Issue 45020 is the best representative because it is open, high-level, and can plausibly subsume several nearby regression reports without overfitting to one model.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:41950", - "right": "issue:44016", + "left": "issue:44246", + "right": "issue:45003", "accept": false, - "reason": "Different problems: video pipeline processor lookup vs notebook syntax error." + "reason": "Both touch import-time behavior, but one is intermittent slowness and the other is an unsafe sys.modules access bug; not the same defect." }, { - "left": "issue:44964", - "right": "issue:45600", + "left": "issue:45406", + "right": "issue:45632", "accept": false, - "reason": "Different model-loading bug vs stale auto_mappings CI references." + "reason": "Gemma4Processor _tokenizer crash and trust_remote_code cache-path collision are unrelated failure modes." }, { - "left": "issue:42907", - "right": "issue:43441", + "left": "issue:43643", + "right": "issue:44496", "accept": false, - "reason": "Different code paths: save/dequantization vs FlashAttention model loading." + "reason": "Both are config/loading problems, but one is missing fields with trust_remote_code and the other is an unrecognized model due to missing model_type; different root causes." }, { - "left": "issue:43388", - "right": "issue:43688", + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Phi-4 multimodal loading and stale auto_mappings references to Sam3LiteText are unrelated issues." + }, + { + "left": "issue:44016", + "right": "issue:44062", "accept": false, - "reason": "Metrics batching bug vs MoE auxiliary-loss normalization; unrelated." + "reason": "Notebook syntax error and AddedToken keyword collision are clearly different bugs." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "Processor/tokenizer regression vs t5gemma embedding resize bug." + "reason": "They affect different code paths: processor/tokenizer regression versus resize_token_embeddings not propagating to decoder embeddings." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Saving dequantized models and FlashAttention loading for Ministral-3 are different problems." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "CI device override test issue vs Phi-4 multimodal loading failure." + "reason": "A stale CI test override in GraniteSpeech is unrelated to Phi-4 multimodal model loading." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized model/config issue vs PIL image processor torchvision dependency bug." + "reason": "Model config loading and PIL image processor torchvision dependency are separate issues." }, { "left": "issue:43854", "right": "issue:44964", "accept": false, - "reason": "Both are load failures, but for different models and different underlying causes." + "reason": "GLM-4.7-Flash unit-test loading failure is not the same as Phi-4 multimodal loading." }, { - "left": "issue:44016", - "right": "issue:44220", + "left": "issue:43673", + "right": "issue:44336", "accept": false, - "reason": "Notebook syntax error is unrelated to audio feature extraction failure." + "reason": "Generation cache handling and ANSI output formatting are unrelated." }, { - "left": "issue:43673", - "right": "issue:44336", + "left": "issue:43976", + "right": "issue:44987", "accept": false, - "reason": "Cache/memory issue vs ANSI terminal reporting bug; unrelated." + "reason": "Python version compatibility and loading physical-intelligence/fast are different reports." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Qwen loading regression vs Qwen save_pretrained regression; different concrete bugs." + "reason": "Both involve Qwen3.5-era regressions, but one is a loading failure and the other is a save_pretrained checkpoint serialization bug." }, { - "left": "issue:42175", - "right": "issue:44279", + "left": "issue:43531", + "right": "issue:45600", "accept": false, - "reason": "Backend dependency packaging issue vs generic dependency problem; too broad to merge." + "reason": "Qwen3-MoE sliding_window behavior and removed Sam3LiteText config references are unrelated." }, { - "left": "issue:43531", - "right": "issue:45600", + "left": "issue:43824", + "right": "issue:44933", "accept": false, - "reason": "Qwen3-MoE sliding window bug vs removed Sam3LiteText config references." + "reason": "Both are import errors, but they refer to different missing symbols and different causes." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "ImageProcessorFast label reduction bug vs documentation mismatch." + "reason": "reduce_labels behavior in BeitImageProcessorFast is unrelated to a docs mismatch in TextClassificationPipeline." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "Embedding resize bug vs Whisper empty-transcription generation failure." + "reason": "t5gemma embedding resizing and Whisper empty-transcription generation are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "A Python support regression and a generic dependency issue are not the same underlying change." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Optimizer argument handling vs LoRA warning about tie_word_embeddings." + "reason": "Unused SGD optimizer args and a tie_word_embeddings warning in Qwen3.5 are unrelated." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing eval-mode bug vs D-FINE auxiliary-loss behavior." + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary-loss handling are different code paths." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Loading failure vs save_pretrained checkpoint-key regression; not same bug." + "reason": "Both are Qwen3.5-related, but one is a loading regression and the other is incorrect visual-encoder key saving; not the same bug." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Unrecognized model registration vs from_config text_config unwrapping; related area but different defect." + "reason": "An unrecognized-model error and a composite-config unwrap issue are distinct config-handling bugs." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "Multimodal assistant masks and Qwen2.5-VL rope-index scaling are different multimodal behaviors." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "General version-regression theme only; different subsystems and concrete failures." + "reason": "Video-classification processor lookup and remote_code model breakage are not the same issue." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Docs behavior note vs model-loading config error; unrelated." + "reason": "A documentation discrepancy is not the same as a model-loading failure." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "MoE load-balancing loss bug vs Gemma 4 mm_token_type_ids defaulting issue." - }, - { - "left": "issue:44016", - "right": "issue:44062", - "accept": false, - "reason": "Notebook syntax error vs AddedToken keyword-argument conflict." + "reason": "Load-balancing loss omission and mm_token_type_ids defaulting are separate model-training bugs." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "BLIP2 dtype propagation vs Qwen save_pretrained checkpoint regression." - }, - { - "left": "issue:43976", - "right": "issue:45405", - "accept": false, - "reason": "Python support breakage vs unreleased PEFT version bump." + "reason": "BLIP2 dtype persistence and Qwen3.5 save_pretrained corruption are unrelated." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "Specific HybridCache import failure vs broad many-import-errors report; not the same bug." + "reason": "45278 is an umbrella report about many import errors, while 44351 is one specific missing export; the root cause is not established as the same." }, { - "left": "issue:43824", - "right": "issue:44933", + "left": "issue:43976", + "right": "issue:45405", "accept": false, - "reason": "Different missing imports from different modules and causes." + "reason": "Python compatibility and the unreleased PEFT minimum-version bump are unrelated." }, { - "left": "issue:43388", - "right": "issue:43827", + "left": "issue:44351", + "right": "issue:45278", "accept": false, - "reason": "Training metric batching bug vs docs still referencing removed pipeline API." + "reason": "Same as above: one specific import failure versus a broad update-breakage report is too loose to merge." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Different config initialization issue vs missing pad_token_id attribute." - }, - { - "left": "issue:44683", - "right": "issue:45542", - "accept": false, - "reason": "Compiled flex_attention torch-version bug vs TensorBoard/TensorFlow packaging issue." + "reason": "Phi4MultimodalConfig default initialization and Llama4Config missing pad_token_id are different config bugs." } ] }, @@ -9521,14 +9561,13 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:32090", - "issue:35141", + "issue:33290", "issue:38175", "issue:39692", "issue:41950", "issue:42994", "issue:43065", - "issue:43122", + "issue:43262", "issue:43295", "issue:43299", "issue:43317", @@ -9536,18 +9575,19 @@ "issue:43450", "issue:43650", "issue:43653", + "issue:43673", "issue:43723", + "issue:43761", "issue:43784", "issue:43854", "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", @@ -9562,261 +9602,247 @@ "issue:45081", "issue:45276", "issue:45290", - "issue:45325", "issue:45397", "issue:45405", "issue:45542", "issue:45593", "issue:45600", + "issue:45656", "issue:45712" ], "soft_pairs": [ + "issue:44683|issue:45542", "issue:44462|issue:45081", "issue:44279|issue:45405", "issue:43299|issue:43854", "issue:39692|issue:43901", - "issue:32090|issue:35141", - "issue:41950|issue:43650", "issue:43295|issue:44779", "issue:44297|issue:44462", "issue:38175|issue:44617", - "issue:43976|issue:44279", + "issue:41950|issue:43650", + "issue:33290|issue:45656", "issue:44462|issue:44779", - "issue:44521|issue:45325", - "issue:43976|issue:44987", "issue:45600|issue:45712", + "issue:44016|issue:44291", "issue:43450|issue:44279", - "issue:43065|issue:44898", "issue:42994|issue:43317", + "issue:43065|issue:44898", "issue:44466|issue:45276", "issue:44261|issue:44263", "issue:43441|issue:44683", "issue:43650|issue:45290", "issue:44623|issue:45397", "issue:43723|issue:44162", - "issue:43784|issue:45542", "issue:44623|issue:44749", + "issue:43784|issue:45542", + "issue:43262|issue:44521", "issue:44671|issue:44779", + "issue:43673|issue:43761", "issue:43653|issue:45593", - "issue:44987|issue:45542", - "issue:43901|issue:43994", - "issue:43122|issue:43650", - "issue:44361|issue:45593" + "issue:44987|issue:45542" ], - "prepared_review_unit_hash": "50c5b93e896db6ce1eb9065d8b60b2ebe2ca3b4b652291773692f750d61fa29d", + "prepared_review_unit_hash": "3039bd251b5b247ff5466506d98b3119c3ab3ec5341b6323b91868fb24ea6172", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36962, - "estimated_input_tokens": 9241, - "estimated_eval_tokens": 18738 + "soft_pair_count": 29, + "serialized_chars": 36367, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This looks like a loose collection of Transformers regressions, with several tokenizer/model-loading and multimodal-processing complaints, but the proposed soft edges mostly share only broad subsystem overlap rather than the same concrete bug. No pair here is a safe duplicate merge.", - "confidence": 0.63, - "canonical_issue_reason": "issue:43723 is the broadest and most central tokenizer-loading regression in the set; it can anchor the nearby tokenizer/processor complaints better than the model-specific or unrelated issues.", - "canonical_pr_reason": null, - "best_issue_reason": "issue:43723 is the clearest representative issue: generic, regression-oriented, and closer in scope to several tokenizer-related reports than the others.", - "best_pr_reason": null, + "summary": "The cluster is mostly a set of unrelated Transformers issues that only share broad v5/regression vocabulary. The soft-similarity pairs are all false positives: they point to different subsystems, different failure modes, and usually different code paths.", + "confidence": 0.95, + "canonical_issue_reason": "No canonical issue: the items do not share one underlying bug, so picking any single issue would misrepresent the cluster.", + "canonical_pr_reason": "No pull requests are present in this cluster.", + "best_issue_reason": "No global best issue: the set is heterogeneous (tokenizers, processors, model loading, serialization, docs, deepspeed, attention, etc.), so there is no good representative duplicate target.", + "best_pr_reason": "No pull requests are present in this cluster.", "soft_edge_verdicts": [ + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Different bugs: compiled flex_attention failure on torch>=2.9 vs tensorboard installed without TensorFlow causing an undefined tf backend error." + }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "Both concern tokenizer loading, but one is AutoTokenizer ignoring repository tokenizer.json while the other is a Mistral regex patch crash; different failure points and fixes." + "reason": "Different tokenizer paths: AutoTokenizer ignoring repository tokenizer.json vs _patch_mistral_regex crashing because Tokenizer lacks backend_tokenizer." }, { "left": "issue:44279", "right": "issue:45405", "accept": false, - "reason": "Unrelated topics: a dependency problem vs a bumped PEFT minimum version. No shared bug." + "reason": "Unrelated: a generic transformers dependency issue vs MIN_PEFT_VERSION being bumped to an unreleased PEFT version." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Both are model-loading failures, but they target different models and likely different load-time issues." + "reason": "Different failures: Qwen3VL MoE model loading breakage vs NameError importing sentence-transformers with latest transformers." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "One is a SigLIP2 example/model mismatch and quantization issue; the other is a docs mismatch about return_all_scores. Not the same bug." - }, - { - "left": "issue:32090", - "right": "issue:35141", - "accept": false, - "reason": "Trainer gpu broadcast TypeError vs output embedding reinit after resizing embeddings. Different code paths and symptoms." - }, - { - "left": "issue:41950", - "right": "issue:43650", - "accept": false, - "reason": "Video-classification pipeline processor lookup vs a placeholder issue title; no meaningful overlap." + "reason": "Both involve docs/examples, but not the same bug: SigLIP2 example/model-processor mismatch and quantization failure vs stale TextClassificationPipeline docs mentioning return_all_scores." }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Both mention tokenizer behavior, but one is a processor/tokenizer API regression and the other is Deepseek tokenization correctness; too model-specific to merge." + "reason": "Different regressions: custom code accessing processor.tokenizer/passing images to tokenizer vs Deepseek tokenizer producing incorrect outputs in v5." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Related to tokenizer save/load metadata, but one is tokenizer_class mismatch on save and the other is AutoTokenizer ignoring tokenizer.json on load." + "reason": "Related to tokenizer serialization, but distinct: tokenizer.save_pretrained writes a mismatched tokenizer_class, while AutoTokenizer ignores tokenizer.json from the repo." }, { "left": "issue:38175", "right": "issue:44617", "accept": false, - "reason": "SigLIP2 zero probabilities and Sam3Video CUDA OOM are completely different issues." + "reason": "Different models and symptoms: unexpected zero probabilities in SigLIP2 vs Sam3Video CUDA out of memory." }, { - "left": "issue:43976", - "right": "issue:44279", + "left": "issue:41950", + "right": "issue:43650", "accept": false, - "reason": "Python-version support breakage vs a generic dependency issue; not the same underlying bug." + "reason": "No shared bug: video-classification pipeline using image processors vs an unrelated 'ADD THE DATA' issue." }, { - "left": "issue:44462", - "right": "issue:44779", + "left": "issue:33290", + "right": "issue:45656", "accept": false, - "reason": "Both are tokenizer-related, but one is repository tokenizer.json being ignored and the other is incorrect Deepseek tokenization results; different regressions." + "reason": "Both mention deepspeed/optimizer, but the concrete problems differ: AdaFactor OOM vs optimizer.step being called twice." }, { - "left": "issue:44521", - "right": "issue:45325", + "left": "issue:44462", + "right": "issue:44779", "accept": false, - "reason": "Both touch multimodal chat/VL behavior, but assistant mask generation and temporal rope position IDs are separate problems." + "reason": "Different tokenizer issues: ignoring tokenizer.json in AutoTokenizer vs Deepseek tokenizer producing incorrect results." }, { - "left": "issue:43976", - "right": "issue:44987", + "left": "issue:45600", + "right": "issue:45712", "accept": false, - "reason": "Environment/version incompatibility vs a specific model load failure; not duplicates." + "reason": "Different maintenance problems: stale auto_mappings.py entries vs leftover dummy classes leaking into dummy_pt_objects.py and dir(transformers)." }, { - "left": "issue:45600", - "right": "issue:45712", + "left": "issue:44016", + "right": "issue:44291", "accept": false, - "reason": "Both are CI/repo hygiene problems, but one is stale auto_mappings references and the other is leftover dummy classes. Different fixes." + "reason": "Unrelated: notebook syntax error in the intro notebook vs init_empty_weights breaking on an unexpected _is_hf_initialized argument." }, { "left": "issue:43450", "right": "issue:44279", "accept": false, - "reason": "Video processor shape bug vs dependency issue; unrelated." + "reason": "Different scopes: batched video processor shape bug vs a generic dependency issue with transformers." }, { - "left": "issue:43065", - "right": "issue:44898", + "left": "issue:42994", + "right": "issue:43317", "accept": false, - "reason": "Sam3PixelDecoder dummy layer vs Perceiver non-default-resolution failure. Different models and failures." + "reason": "Both are quantization-related, but not the same bug: saving a quantized model fails vs dequantized model loading with device_map=auto and cpu/gpu offload." }, { - "left": "issue:42994", - "right": "issue:43317", + "left": "issue:43065", + "right": "issue:44898", "accept": false, - "reason": "Both are quantization/load related, but saving a quantized model and loading a dequantized model with device_map=auto are different paths." + "reason": "Different model components: a dummy Conv2d in Sam3PixelDecoder vs Perceiver image classification failing for non-default resolution despite interpolate_pos_encoding=True." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Both involve embeddings/tied weights, but one is device-dependent serialization and the other is resize_token_embeddings not updating Gemma4 embeddings/output heads." + "reason": "Both touch weights/embeddings, but one is inconsistent lm_head serialization by device while the other is resize_token_embeddings not updating Gemma4 per-layer embeddings/output embeddings." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "MLA rms_norm_eps precision concern vs torch.split return values in GlmMoeDsaIndexer. Different components." + "reason": "Different code paths and symptoms: MLA q_a_layernorm eps precision issue vs torch.split return handling in GlmMoeDsaIndexer." }, { "left": "issue:43441", "right": "issue:44683", "accept": false, - "reason": "Attention backend failures, but one is Ministral-3 with FlashAttention in v5 RC and the other is compiled flex_attention on torch >= 2.9. Not the same bug." + "reason": "Both mention attention, but one is Ministral FlashAttention failure and the other is compiled flex_attention breaking on torch>=2.9; not the same fix." }, { "left": "issue:43650", "right": "issue:45290", "accept": false, - "reason": "Placeholder title vs a concrete chat-template crash with tool calls. No duplicate signal." + "reason": "Completely unrelated: a placeholder 'ADD THE DATA' issue vs apply_chat_template crashing on assistant tool-call messages with no content." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Processor save_pretrained missing files vs Gemma-4 zero3 from_pretrained loading. Related area, different issue." + "reason": "Different loading/saving failures: processor.save_pretrained missing files vs gemma-4 zero3 from_pretrained failing." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "Tokenizer-loading regression vs ESM2 model breakage; different model families and failure modes." - }, - { - "left": "issue:43784", - "right": "issue:45542", - "accept": false, - "reason": "Import-time NameError in sentence-transformers integration vs backend selection error when only tensorboard is installed. Different cause." + "reason": "Different models and failure modes: AutoTokenizer.from_pretrained loading issue vs ESM2 being broken for users." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained files missing vs a tokenizer performance regression after upgrade. Not duplicates." + "reason": "Different problems: missing files from processor.save_pretrained vs a performance regression when filtering data after upgrading Transformers." }, { - "left": "issue:44671", - "right": "issue:44779", + "left": "issue:43784", + "right": "issue:45542", "accept": false, - "reason": "Both are v5 correctness regressions in NLP models, but CamemBERT masked LM output and Deepseek tokenization are separate concrete bugs." + "reason": "Unrelated runtime errors: importing sentence-transformers hits NameError: nn is not defined vs tensorboard-only installation causing an undefined tf backend error." }, { - "left": "issue:43653", - "right": "issue:45593", + "left": "issue:43262", + "right": "issue:44521", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug vs D-FINE auxiliary-loss behavior. Unrelated." + "reason": "Both involve chat templates, but the bugs differ: audio apply_chat_template defaults to 16kHz instead of config value vs multimodal assistant_masks becoming all zero." }, { - "left": "issue:44987", - "right": "issue:45542", + "left": "issue:44671", + "right": "issue:44779", "accept": false, - "reason": "Model loading failure vs TensorFlow backend/installation issue. No overlap." + "reason": "Different tokenizer/model behavior: CamemBERT masked LM predictions wrong in v5 vs Deepseek tokenizer outputs incorrect results." }, { - "left": "issue:43901", - "right": "issue:43994", + "left": "issue:43673", + "right": "issue:43761", "accept": false, - "reason": "Documentation/API mismatch vs SigLIP2 nonsensical outputs. Different scope." + "reason": "Different regressions: GenerationMixin cache missing during chunked_prefill vs CLIPVisionModel.forward returning hidden_states=None." }, { - "left": "issue:43122", - "right": "issue:43650", + "left": "issue:43653", + "right": "issue:45593", "accept": false, - "reason": "Tokenizer behavior changed across versions vs placeholder issue title. No duplicate." + "reason": "No common bug: BigBirdTokenizer special-token/decode issue vs D-FINE not applying auxiliary losses when denoising is off." }, { - "left": "issue:44361", - "right": "issue:45593", + "left": "issue:44987", + "right": "issue:45542", "accept": false, - "reason": "MLukeTokenizer AttributeError vs D-FINE denoising/aux-loss behavior. Completely different bugs." + "reason": "Different failures: loading physical-intelligence/fast on transformers>=5.1.0 vs tensorboard without TensorFlow causing backend errors." } ] }, @@ -9840,295 +9866,294 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:16998", - "issue:22355", "issue:36246", "issue:41628", "issue:41950", + "issue:42175", "issue:42491", "issue:42757", - "issue:43262", + "issue:43122", "issue:43295", "issue:43299", "issue:43352", "issue:43475", "issue:43525", - "issue:43673", - "issue:43761", + "issue:43526", + "issue:43531", + "issue:43650", "issue:43824", "issue:43881", "issue:43901", "issue:43976", + "issue:43994", "issue:44016", "issue:44188", "issue:44220", "issue:44242", "issue:44246", - "issue:44291", "issue:44295", "issue:44297", "issue:44336", "issue:44351", + "issue:44361", "issue:44373", "issue:44462", "issue:44485", "issue:44496", - "issue:44521", - "issue:44561", "issue:44589", "issue:44617", "issue:44661", - "issue:44857", - "issue:44871", - "issue:44964", + "issue:45003", "issue:45250", "issue:45356", "issue:45362", "issue:45397", "issue:45542", + "issue:45561", + "issue:45563", "issue:45593", "issue:45759" ], "soft_pairs": [ - "issue:44016|issue:44291", - "issue:43262|issue:44521", + "issue:43122|issue:43650", + "issue:43901|issue:43994", + "issue:42175|issue:43976", + "issue:44361|issue:45593", "issue:43475|issue:44016", "issue:44297|issue:45397", - "issue:43673|issue:43761", + "issue:43824|issue:43976", "issue:44188|issue:44336", "issue:43525|issue:45759", "issue:43525|issue:43881", "issue:44589|issue:45362", "issue:43295|issue:44617", + "issue:45561|issue:45563", "issue:42491|issue:43299", "issue:41628|issue:43976", "issue:43901|issue:44246", "issue:44336|issue:44373", "issue:43352|issue:45250", - "issue:16998|issue:22355", "issue:44242|issue:45593", + "issue:41950|issue:44220", "issue:44295|issue:45397", "issue:44661|issue:45542", - "issue:41950|issue:44220", "issue:44246|issue:44496", + "issue:43531|issue:45362", "issue:44462|issue:45356", + "issue:43526|issue:45003", "issue:36246|issue:42757", "issue:44336|issue:44485", "issue:36246|issue:44351", - "issue:43673|issue:44561", - "issue:43824|issue:43976", - "issue:43976|issue:44964", - "issue:44246|issue:44871", - "issue:44242|issue:44857" + "issue:45003|issue:45561" ], - "prepared_review_unit_hash": "ce782dec63463b2cbd8429aa5adcf7090407f73c3f25f89d868382405a98a69e", + "prepared_review_unit_hash": "71d952df8f26c6ae5b10b65f9b92daae6c7398c5c8b2929b8b749bc3c5837801", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36525, - "estimated_input_tokens": 9132, - "estimated_eval_tokens": 18520 + "serialized_chars": 35748, + "estimated_input_tokens": 8937, + "estimated_eval_tokens": 18130 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These soft-similarity pairs are mostly superficial; they share broad themes like imports, loading, tokenizers, or multimodal processing, but they do not appear to be the same underlying bug. I would reject all soft edges.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue: the set spans unrelated bugs (imports, model loading, tokenizer/pipeline regressions, docs), so no item cleanly represents the whole cluster.", + "summary": "This is a loose similarity cluster of mostly unrelated Transformers issues. The only common thread is broad import/load/regression language, but none of the soft pairs appear to describe the same concrete bug or change.", + "confidence": 0.83, + "canonical_issue_reason": "issue:43824 is the most representative of the repeated import/export failure pattern in this set, with concrete runtime impact and relatively strong discussion/reference activity.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44521 is the strongest standalone representative because it is a current, well-documented multimodal regression with inbound references, but it still does not unify the rest of the set.", + "best_issue_reason": "Among the mixed issues here, issue:43824 is the clearest, most central example of the recurring model-import/load regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44016", - "right": "issue:44291", + "left": "issue:43122", + "right": "issue:43650", "accept": false, - "reason": "Notebook syntax error vs. init_empty_weights TypeError; unrelated failures." + "reason": "Tokenizer behavior regression and a vague data request are unrelated; no shared code path." }, { - "left": "issue:43262", - "right": "issue:44521", + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Docs mismatch for a pipeline option is not the same bug as bad SigLIP2 outputs." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Missing TensorFlow from a torch install and a Python-version compatibility break are different issues." + }, + { + "left": "issue:44361", + "right": "issue:45593", "accept": false, - "reason": "Both touch apply_chat_template, but one is audio sampling-rate defaulting and the other is multimodal assistant masks; different bugs." + "reason": "MLukeTokenizer task-time AttributeError and D-FINE auxiliary-loss gating are unrelated model bugs." }, { "left": "issue:43475", "right": "issue:44016", "accept": false, - "reason": "SAM 3 vision encoder attribute error is unrelated to a notebook syntax error." + "reason": "SAM3 video encoder attribute failure and a notebook syntax error are not the same problem." }, { "left": "issue:44297", "right": "issue:45397", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to gemma-4 zero3 loading." + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma-4 ZeRO-3 from_pretrained loading." }, { - "left": "issue:43673", - "right": "issue:43761", + "left": "issue:43824", + "right": "issue:43976", "accept": false, - "reason": "Chunked prefill cache issue and CLIP hidden_states regression are different code paths." + "reason": "An import/export symbol error and a Python 3.9/3.10 support issue are different failure modes." }, { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "torch.compile attention-kernel divergence vs ANSI codes in loading_report; no overlap beyond general utilities." + "reason": "torch.compile attention-kernel branching and stray ANSI codes in a report helper are unrelated." }, { "left": "issue:43525", "right": "issue:45759", "accept": false, - "reason": "Both are model-loading/config issues, but Llama4 pad_token_id and Qwen multimodal text_config unwrapping are distinct failures." + "reason": "Both are config-loading regressions, but they affect different models and different missing attributes/unwrapping paths." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Different loading errors for different models; no clear shared concrete bug." + "reason": "Llama4Config pad_token_id and GLM-4V loading failure are distinct model-loading bugs." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Float8 storage deserialization and Qwen3.5 chat crash are unrelated." + "reason": "Missing Float8 storage support and a Qwen3.5 chat crash are not the same underlying issue." }, { "left": "issue:43295", "right": "issue:44617", "accept": false, - "reason": "processor.tokenizer regression vs CUDA OOM; not the same underlying problem." + "reason": "Processor/tokenizer regression and SAM3Video OOM are different code-path problems." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "A pytest-xdist file race and a stale generate() warning are unrelated testing/runtime issues." }, { "left": "issue:42491", "right": "issue:43299", "accept": false, - "reason": "Both involve Qwen/MoE loading regressions, but they affect different model families and failure modes." + "reason": "Qwen3-MoE LoRA compatibility on hf4.x and Qwen3VL MoE loading are different model-family regressions." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "ImportError for AutoImageProcessor is unrelated to Python version support incompatibility." + "reason": "AutoImageProcessor import failure and Python version incompatibility are not the same bug." }, { "left": "issue:43901", "right": "issue:44246", "accept": false, - "reason": "Docs mismatch for return_all_scores vs intermittent import slowness; different concerns." + "reason": "A docs/behavior mismatch and intermittent slow import are unrelated." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "ANSI color handling in loading_report is unrelated to a docstring typo." + "reason": "ANSI output in a helper and a wrong docstring are separate documentation/UX issues." }, { "left": "issue:43352", "right": "issue:45250", "accept": false, - "reason": "Specific model lacks Flash Attention 2 support; the other is just a generic Flash Attention topic." + "reason": "Model-specific FA2 unsupported error and a generic Flash Attention 2.0 topic are not the same concrete fix." }, { - "left": "issue:16998", - "right": "issue:22355", + "left": "issue:44242", + "right": "issue:45593", "accept": false, - "reason": "model_max_length question and missing transformers.onnx module are unrelated." + "reason": "Missing router load-balancing loss and missing D-FINE auxiliary losses happen in different models and flags." }, { - "left": "issue:44242", - "right": "issue:45593", + "left": "issue:41950", + "right": "issue:44220", "accept": false, - "reason": "Both are loss-related when an option is disabled, but they affect different models and mechanisms." + "reason": "Video-classification pipeline image-processor lookup and audio fbank extraction are unrelated pipeline bugs." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "position_ids buffer access error vs zero3 from_pretrained issue; no shared code path." + "reason": "Reading a buffered position_ids tensor and Gemma-4 zero3 loading are different errors." }, { "left": "issue:44661", "right": "issue:45542", "accept": false, - "reason": "add-new-model-like tokenizer mapping bug is unrelated to the TensorFlow backend import error." + "reason": "TOKENIZER_MAPPING_NAMES model-generation failure and missing TensorFlow backend when only tensorboard is installed are unrelated." }, { - "left": "issue:41950", - "right": "issue:44220", + "left": "issue:44246", + "right": "issue:44496", "accept": false, - "reason": "Video-classification processor lookup and _torch_extract_fbank_features() are different pipeline components." + "reason": "Intermittent import slowness and unrecognized model/config loading are different problems." }, { - "left": "issue:44246", - "right": "issue:44496", + "left": "issue:43531", + "right": "issue:45362", "accept": false, - "reason": "Slow import is unrelated to unrecognized model config loading." + "reason": "Qwen3-MoE sliding_window behavior and a Qwen3.5 chat crash are not the same bug." }, { "left": "issue:44462", "right": "issue:45356", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and Kimi-K2.5 codec regression are different tokenizer bugs." + "reason": "AutoTokenizer ignoring tokenizer.json and a Kimi-K2.5 tokenizer regression are separate tokenizer issues." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and unsafe sys.modules access are unrelated." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Different missing imports from different packages; same symptom class, not the same bug." + "reason": "Different import-name failures for different symbols and modules; same symptom class only." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI-output bug and GLM-5 RoPE discussion are unrelated." + "reason": "ANSI-code leakage and GLM-5 RoPE implementation are unrelated." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different missing symbols and modules." - }, - { - "left": "issue:43673", - "right": "issue:44561", - "accept": false, - "reason": "cache-related chunked prefill regression and removed helper breaking trust_remote_code are unrelated." - }, - { - "left": "issue:43824", - "right": "issue:43976", - "accept": false, - "reason": "Missing Qwen2_5_VL export and Python version support issue are unrelated." - }, - { - "left": "issue:43976", - "right": "issue:44964", - "accept": false, - "reason": "Python compatibility regression and Phi-4 multimodal loading failure are different issues." - }, - { - "left": "issue:44246", - "right": "issue:44871", - "accept": false, - "reason": "Import timing problem vs Gemma-3 eos_token_id mismatch; no shared bug." + "reason": "Both are import errors, but for different missing exports and different underlying fixes." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:45003", + "right": "issue:45561", "accept": false, - "reason": "Different loss/AMP failures in different models; too broad to merge." + "reason": "Unsafe sys.modules access and a captured_info.txt race are unrelated implementation bugs." } ] }, @@ -10163,24 +10188,27 @@ "issue:43408", "issue:43441", "issue:43475", - "issue:43526", "issue:43540", "issue:43596", "issue:43653", "issue:43673", "issue:43901", + "issue:44060", + "issue:44077", "issue:44162", + "issue:44242", "issue:44246", "issue:44295", "issue:44336", "issue:44368", "issue:44466", "issue:44496", - "issue:44734", + "issue:44561", "issue:44749", "issue:44779", + "issue:44857", + "issue:44871", "issue:44877", - "issue:44977", "issue:44987", "issue:44995", "issue:45003", @@ -10189,265 +10217,247 @@ "issue:45229", "issue:45292", "issue:45335", - "issue:45362", "issue:45375", - "issue:45397", "issue:45459", - "issue:45561", "issue:45563", "issue:45584", "issue:45588", - "issue:45656", "issue:45701", "issue:45759" ], "soft_pairs": [ - "issue:45362|issue:45375", + "issue:43673|issue:44561", + "issue:44246|issue:44871", + "issue:44242|issue:44857", + "issue:36246|issue:43596", "issue:44162|issue:45200", "issue:44295|issue:44749", "issue:44779|issue:45292", "issue:43441|issue:43540", - "issue:45292|issue:45584", "issue:44877|issue:45375", - "issue:36246|issue:43596", + "issue:42886|issue:45229", + "issue:45292|issue:45584", "issue:44987|issue:45459", - "issue:45292|issue:45701", "issue:41628|issue:41950", + "issue:45292|issue:45701", "issue:44779|issue:45701", - "issue:45561|issue:45563", "issue:43901|issue:45003", "issue:45375|issue:45759", - "issue:43065|issue:43408", "issue:43232|issue:44336", + "issue:43065|issue:43408", "issue:43257|issue:45137", - "issue:43526|issue:45003", "issue:44995|issue:45200", "issue:44496|issue:45003", + "issue:44060|issue:44077", "issue:43262|issue:44246", - "issue:42886|issue:45229", "issue:43475|issue:43540", "issue:44466|issue:45335", "issue:44995|issue:45588", "issue:43653|issue:44162", - "issue:44368|issue:45563", - "issue:45397|issue:45656", - "issue:44368|issue:44977", - "issue:43673|issue:44734" + "issue:44368|issue:45563" ], - "prepared_review_unit_hash": "d035c946b33a38e7aefba61d37ac04b970a6d8e17c2a11c151e9e80669c52969", + "prepared_review_unit_hash": "7526e9249fe665baab816697c58e74c20dbcf7fed155c901ad8de904b8f3e91f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37083, - "estimated_input_tokens": 9271, - "estimated_eval_tokens": 18798 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36020, + "estimated_input_tokens": 9005, + "estimated_eval_tokens": 18266 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are distinct Transformers regressions across tokenization, config loading, generation, DeepSpeed, and SAM3/Qwen/Gemma code paths. None of the soft pairs look like the same concrete bug/change, so I would not merge any of them as duplicates.", - "confidence": 0.68, - "canonical_issue_reason": "issue:45292 is the best representative of the small embedding/weight-resizing regression subset: it is concrete, broadly applicable, and closer to other nearby resize/tied-weight reports than the more model-specific tickets.", + "summary": "This is a noisy, heterogeneous cluster of Transformers regressions. The items share broad library-loading/modeling themes, but none of the soft pairs appear to be the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "issue:41628 is the cleanest top-level user-facing import/export failure in the set and the best weak anchor for the cluster, though the cluster is still quite mixed.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45292 is the clearest standalone bug in the set and the most reusable triage anchor for this cluster; it has a direct API symptom and broad model impact.", + "best_issue_reason": "issue:41628 is the most representative broad issue among these, but only loosely; it does not truly cover the many unrelated model/config/generation bugs in the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45362", - "right": "issue:45375", + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Different regressions: chunked_prefill cache handling vs removal of is_torch_fx_available breaking trust_remote_code models." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowdown is unrelated to Gemma eos_token_id/config mismatch." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "MoE load-balancing loss bug is unrelated to an AMP/CUDA crash in LwDetrImageLoss." + }, + { + "left": "issue:36246", + "right": "issue:43596", "accept": false, - "reason": "Both mention Qwen 3.5, but one is a chat/runtime crash and the other is a strict-config field drop; different code paths." + "reason": "Qwen2.5-VL import export failure and a DeepSpeed ZeRO3 deque underflow are different bugs." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "ESM2 loading is unrelated to Gemma 4 mm_token_type_ids defaulting; different models and failures." + "reason": "ESM2 breakage is unrelated to Gemma4 mm_token_type_ids defaults." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "One is a position_ids buffer bug, the other is a post-upgrade tokenization slowdown; not the same issue." + "reason": "Buffer/position_ids access bug is unrelated to a tokenization performance regression." }, { "left": "issue:44779", "right": "issue:45292", "accept": false, - "reason": "DeepSeek tokenizer output regression and token embedding resize propagation are different bugs." + "reason": "Tokenizer correctness regression is different from resize_token_embeddings not propagating to output_embeddings." }, { "left": "issue:43441", "right": "issue:43540", "accept": false, - "reason": "FlashAttention crash in Ministral-3 and Qwen3OmniMoe video-input validation are unrelated." - }, - { - "left": "issue:45292", - "right": "issue:45584", - "accept": false, - "reason": "Resize-token-embeddings affects embedding weights, while Whisper empty-transcription failure is about special-token/chat-template handling." + "reason": "FlashAttention failure in Ministral and Qwen3OmniMoe video-input processing are separate problems." }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Both involve config handling, but one is strict loading of granite_speech and the other is a missing Qwen3_5MoeVisionConfig field." + "reason": "Both involve strict config handling, but they affect different models, fields, and failure modes." }, { - "left": "issue:36246", - "right": "issue:43596", + "left": "issue:42886", + "right": "issue:45229", "accept": false, - "reason": "Importing a missing Qwen2.5 VL processor and a ZeRO3/BertModel index error are clearly different bugs." + "reason": "Offline tokenizer cache loading is unrelated to multi-GPU CUDA OOM." }, { - "left": "issue:44987", - "right": "issue:45459", + "left": "issue:45292", + "right": "issue:45584", "accept": false, - "reason": "Model loading failure on transformers>=5.1.0 and protobuf-related tokenizer error masking are different failure points." + "reason": "Embedding resize propagation and Whisper empty-transcription handling are different bugs." }, { - "left": "issue:45292", - "right": "issue:45701", + "left": "issue:44987", + "right": "issue:45459", "accept": false, - "reason": "Resize-token-embeddings behavior and version-dependent tokenization changes are not the same concrete regression." + "reason": "Loading a specific repo under v5 is not the same as protobuf error masking in tokenizer decoding." }, { "left": "issue:41628", "right": "issue:41950", "accept": false, - "reason": "AutoImageProcessor import failure and video-classification pipeline choosing image processors are related only by naming, not the same bug." + "reason": "A missing top-level AutoImageProcessor import is not the same as the video-classification pipeline choosing image processors." }, { - "left": "issue:44779", + "left": "issue:45292", "right": "issue:45701", "accept": false, - "reason": "Both concern tokenization, but one is DeepSeek-specific incorrect output and the other is a broader version-dependent tokenization change." + "reason": "resize_token_embeddings propagation is unrelated to generic version-dependent tokenization changes." }, { - "left": "issue:45561", - "right": "issue:45563", + "left": "issue:44779", + "right": "issue:45701", "accept": false, - "reason": "A pytest-xdist file race and a stale generation warning are unrelated test vs generation issues." + "reason": "DeepSeek tokenizer regression and broad tokenization changes are not the same concrete bug." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "Docs drift for return_all_scores is unrelated to the sys.modules access bug in modeling_utils." + "reason": "Documentation mismatch is unrelated to unsafe sys.modules access in modeling_utils." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "One is a missing config field being dropped, the other is AutoModelForCausalLM.from_config not unwrapping text_config." - }, - { - "left": "issue:43065", - "right": "issue:43408", - "accept": false, - "reason": "SAM3 pixel-decoder dummy conv and sam3_video/sam3_tracker warning are different SAM3 issues with different symptoms." + "reason": "Missing deepstack_visual_indexes and unwrapping text_config are different Qwen multimodal config bugs." }, { "left": "issue:43232", "right": "issue:44336", "accept": false, - "reason": "Generation kwargs after sync_gpus and ANSI codes in loading_report are unrelated." + "reason": "Generation kwargs after sync_gpus is unrelated to ANSI codes in loading_report." }, { - "left": "issue:43257", - "right": "issue:45137", + "left": "issue:43065", + "right": "issue:43408", "accept": false, - "reason": "Both mention DeepSpeed, but Qwen3 MOE weight conversion and ZeRO3 deque underflow are different code paths." + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as a sam3_video vs sam3_tracker config mismatch warning." }, { - "left": "issue:43526", - "right": "issue:45003", + "left": "issue:43257", + "right": "issue:45137", "accept": false, - "reason": "BeitImageProcessorFast label reduction and unsafe sys.modules access are unrelated." + "reason": "Qwen3 MoE weight conversion and ZeRO3 deque underflow are distinct DeepSpeed-related failures." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "GlmMoeDsa stale indexer cache and Gemma 4 token-type defaults are unrelated." + "reason": "GlmMoeDsa stale cache crash is unrelated to Gemma4 token-type id requirements." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Unrecognized model config and sys.modules access are separate core-loading failures." + "reason": "Unrecognized model/config loading and sys.modules access are different loading-path bugs." }, { - "left": "issue:43262", - "right": "issue:44246", + "left": "issue:44060", + "right": "issue:44077", "accept": false, - "reason": "Audio processor sampling-rate defaults and occasional slow imports are unrelated." + "reason": "Incorrect tied-weights warning in Qwen3-Next is unrelated to patchtsmixer post_init policy." }, { - "left": "issue:42886", - "right": "issue:45229", + "left": "issue:43262", + "right": "issue:44246", "accept": false, - "reason": "Offline tokenizer cache loading and Gemma 4 multi-GPU OOM are different problems." + "reason": "Audio chat-template sampling rate defaults have nothing to do with import latency." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "SAM3 vision encoder output attribute error and Qwen3OmniMoe video input ValueError are distinct model bugs." + "reason": "SAM 3 Video missing output attribute and Qwen3OmniMoe video-input validation are unrelated." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Tied-weight serialization inconsistency is not the same as resize_token_embeddings failing to update decoder.embed_tokens." + "reason": "Serialization of tied weights by device is different from resize_token_embeddings not reaching decoder.embed_tokens." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "A stale cache crash in GlmMoeDsa and flash_attention.py s_aux=None handling are different issues." + "reason": "Stale forward-pass cache in GlmMoeDsa is unrelated to flash-attention s_aux handling." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token registration and ESM2 broken loading are unrelated." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are different issues." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "Both are warnings, but one is tie_word_embeddings config noise and the other is num_return_sequences warning drift." - }, - { - "left": "issue:45397", - "right": "issue:45656", - "accept": false, - "reason": "Gemma-4 ZeRO3 loading and duplicate optimizer.step() under DeepSpeed are different failures." - }, - { - "left": "issue:44368", - "right": "issue:44977", - "accept": false, - "reason": "Tie-word-embeddings warning noise is unrelated to Qwen3.5 flash-attention generation failure." - }, - { - "left": "issue:43673", - "right": "issue:44734", - "accept": false, - "reason": "GenerationMixin cache missing in chunked_prefill and KV-cache continuation indexing are both cache-related but not the same concrete bug." + "reason": "Tie_word_embeddings warning noise is unrelated to a stale num_return_sequences warning in paged generate()." } ] }, @@ -10479,7 +10489,6 @@ "issue:42673", "issue:42994", "issue:43232", - "issue:43452", "issue:43475", "issue:43482", "issue:43525", @@ -10487,27 +10496,29 @@ "issue:43646", "issue:43673", "issue:43704", - "issue:43827", + "issue:43873", "issue:43931", "issue:44038", - "issue:44060", - "issue:44077", "issue:44162", + "issue:44188", "issue:44297", "issue:44361", "issue:44368", "issue:44623", "issue:44661", + "issue:44734", "issue:44749", "issue:44869", "issue:44933", - "issue:45003", + "issue:44945", + "issue:44977", "issue:45081", "issue:45092", "issue:45103", "issue:45137", "issue:45200", "issue:45230", + "issue:45278", "issue:45310", "issue:45356", "issue:45375", @@ -10515,25 +10526,24 @@ "issue:45397", "issue:45459", "issue:45478", - "issue:45561", "issue:45600", "issue:45656" ], "soft_pairs": [ - "issue:44060|issue:44077", - "issue:45003|issue:45561", + "issue:45397|issue:45656", + "issue:44368|issue:44977", + "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:43673|issue:45278", "issue:44297|issue:44623", "issue:43646|issue:44162", "issue:44361|issue:44749", - "issue:43232|issue:43673", "issue:44623|issue:45656", "issue:28282|issue:30990", - "issue:43673|issue:43827", - "issue:43452|issue:45003", "issue:42994|issue:43482", "issue:44749|issue:45310", "issue:43475|issue:43643", @@ -10549,222 +10559,236 @@ "issue:42673|issue:43704", "issue:42994|issue:43525", "issue:45310|issue:45375", + "issue:43232|issue:44188", + "issue:43873|issue:44945", "issue:42222|issue:45200" ], - "prepared_review_unit_hash": "8a45655a08e6d8a93d8fc4e2698ffe6d9edd405b96aa877f87921227b76fef7e", + "prepared_review_unit_hash": "0c70e05ee33c862d58f307374d658269893435cd2597fad35f63e84947036817", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36049, - "estimated_input_tokens": 9013, - "estimated_eval_tokens": 18282 + "soft_pair_count": 32, + "serialized_chars": 36557, + "estimated_input_tokens": 9140, + "estimated_eval_tokens": 18536 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a heterogeneous grab bag of Transformers/model-loading regressions, not one tight duplicate cluster. The only clear mini-cluster is the Qwen3.5-MoE loading/config issue, where one report names the likely root cause and the others look like symptom duplicates.", - "confidence": 0.37, - "canonical_issue_reason": "issue:45375 is the best anchor because it identifies the likely root cause (missing `deepstack_visual_indexes` being dropped by `@strict`) behind the Qwen3.5-MoE `from_pretrained` failures; the other Qwen3.5-MoE reports are downstream symptom duplicates.", + "summary": "This cluster is mostly a grab bag of distinct Transformers issues, with a broad concentration of v5 regressions, model-loading failures, tokenizer/config breakages, and unrelated performance/runtime bugs. Only one soft pair looks like a true duplicate; the rest are thematically similar but not the same underlying bug.", + "confidence": 0.84, + "canonical_issue_reason": "issue:45278 is the broadest representative of the cluster\u2019s main theme: user-facing breakages after upgrading to Transformers v5, especially import/loading failures. It is a good umbrella issue, though not an exact duplicate of most others.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45375 is the most actionable and root-cause-oriented report in the only clear duplicate subcluster, and it best explains the related loading errors in 45310/45478.", + "best_issue_reason": "issue:45278 is the best anchor for the cluster because it is the most general and representative report of the recurring v5 regression/import problem class seen across multiple items.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44060", - "right": "issue:44077", + "left": "issue:45397", + "right": "issue:45656", "accept": false, - "reason": "Different model families and different warnings; no shared concrete bug path." + "reason": "Different problems: one is a Gemma-4 Zero3 from_pretrained failure, the other is a DeepSpeed optimizer-step duplication bug." }, { - "left": "issue:45003", - "right": "issue:45561", + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen3.5, but one is a warning about tie_word_embeddings and the other is a flash-attention generation failure." + }, + { + "left": "issue:43673", + "right": "issue:44734", "accept": false, - "reason": "`sys.modules` access bug vs xdist race on `captured_info.txt`; unrelated failures." + "reason": "Both involve generation/cache paths, but chunked_prefill cache handling and /v1/responses KV-cache continuation are different code paths and failures." }, { "left": "issue:45137", "right": "issue:45600", "accept": false, - "reason": "DeepSpeed ZeRO3 deque error vs removed Sam3LiteText config references; different code paths." + "reason": "Completely different: DeepSpeed ZeRO3 deque pop versus auto_mappings referencing removed Sam3LiteText configs." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Both mention Qwen, but one is a weight-shape load error and the other is a tied-weights warning; not the same bug." + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a Qwen3.5 embedding-tie warning." }, { "left": "issue:44933", "right": "issue:45459", "accept": false, - "reason": "Import missing from `image_utils` vs protobuf exception masking tokenizer errors; different root causes." + "reason": "Both are import-related, but image_utils import breakage is unrelated to tokenizer protobuf decode-error masking." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper timestamp decode crash vs Kimi tokenizer codec/regression warning; unrelated." + "reason": "Whisper timestamp decoding crash and Kimi-K2.5 tokenizer regression are different subsystems and symptoms." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both touch generation internals, but sync_gpus kwargs updating and chunked_prefill cache missing are not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:45278", + "accept": false, + "reason": "Generation cache handling is unrelated to broad import errors after upgrade; different subsystems and failure modes." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Tokenizer save metadata mismatch vs processor save missing files; same area, but not the same concrete bug." + "reason": "Both concern save_pretrained behavior, but tokenizer_config mismatch and missing processor files are distinct save-path bugs." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Broad v5 custom-model init breakage vs an ESM2-specific breakage; too different to merge." + "reason": "Both are v5 breakages, but custom model initialization and ESM2 breakage are not the same code-path problem." }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Tokenizer AttributeError vs a data-filtering slowdown regression; not the same failure." - }, - { - "left": "issue:43232", - "right": "issue:43673", - "accept": false, - "reason": "Different generation internals (`sync_gpus` kwargs update vs chunked prefill cache missing)." + "reason": "Tokenizer AttributeError on tasks is unrelated to a performance slowdown after upgrading Transformers." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Processor serialization bug vs DeepSpeed optimizer stepping twice; unrelated." + "reason": "Processor save_pretrained file loss and DeepSpeed optimizer-step duplication are unrelated." }, { "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "ImportError for PyTorch vs a loading hang in Sentence Transformers; different symptoms and likely causes." - }, - { - "left": "issue:43673", - "right": "issue:43827", - "accept": false, - "reason": "Generation cache regression vs docs still referencing removed `pipeline()` APIs; not a duplicate bug." - }, - { - "left": "issue:43452", - "right": "issue:45003", - "accept": false, - "reason": "GGUF loading breakage vs unsafe `sys.modules` access; different code paths." + "reason": "PyTorch-missing ImportError and Sentence Transformers loading hang are different issues." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized saving failure vs Qwen2.5-GGUF loading failure; opposite sides of the workflow and different bugs." + "reason": "Quantized model saving failure and GGUF loading failure are both model-format related but not the same concrete bug." }, { "left": "issue:44749", "right": "issue:45310", "accept": false, - "reason": "Performance regression on upgraded Transformers vs Qwen3.5-MoE loading error; unrelated." + "reason": "A slowdown regression after upgrade is not the same as a Qwen3.5 Moe from_pretrained error." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM3 vision encoder attribute error vs missing fields returned from `AutoConfig`; different models and failures." + "reason": "Missing Sam3 vision output attribute and missing fields in trust_remote_code AutoConfig are different object initialization bugs." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance regression and DeepSpeed optimizer-step duplication are different symptoms and code paths." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "Generic bug report vs Gemma-4 DeepSpeed ZeRO3 loading issue; no evidence they match." + "reason": "Generic bug report and Gemma-4 Zero3 from_pretrained failure do not point to the same underlying bug." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe bug vs processor save missing files; not the same underlying problem." + "reason": "Qwen3-VL-Moe transformers v5 issue is unrelated to processor.save_pretrained missing files." }, { "left": "issue:42371", "right": "issue:45092", "accept": false, - "reason": "TF32 settings docs issue vs InternVL2 meta-init incompatibility; documentation vs runtime bug." + "reason": "TF32 API-settings warning and InternVL2 meta-initialization incompatibility are not the same problem." }, { "left": "issue:43475", "right": "issue:45381", "accept": false, - "reason": "Different video-model failures (`Sam3VisionEncoderOutput` attribute vs Qwen2.5-VL position IDs)." + "reason": "SAM3 missing attribute and Qwen2.5-VL video position-id bug are separate model-specific regressions." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Tokenizer regex crash vs docstring/annotations crash; unrelated subsystems." + "reason": "Tokenizer regex patch crash and auto-docstring annotation crash are unrelated failures." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Different import failures in different modules; not the same missing symbol bug." + "reason": "Both are import errors, but they affect different symbols and likely different export paths; not the same bug." }, { "left": "issue:45375", "right": "issue:45478", - "accept": true, - "reason": "Same Qwen3.5-MoE loading regression; 45375 explains the likely root cause that would surface as 45478's `from_pretrained` error." + "accept": false, + "reason": "Both involve Qwen3.5 Moe, but one is a missing config field and the other is a general from_pretrained error; too broad to treat as duplicates." }, { "left": "issue:43475", "right": "issue:44661", "accept": false, - "reason": "SAM3 output attribute issue vs tokenizer mapping failure when adding a model; unrelated." + "reason": "SAM3 vision output attribute error is unrelated to add-new-model-like failing on TOKENIZER_MAPPING_NAMES." }, { "left": "issue:42673", "right": "issue:43704", - "accept": false, - "reason": "Same title, but they are separate reports with no evidence of a shared code-path bug from the provided text." + "accept": true, + "reason": "Same exact title and same described bug: Qwen3ForCausalLM leaking VRAM across multiple dataloader threads." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized save failure vs missing `pad_token_id` on `Llama4Config`; different bugs." + "reason": "Quantized model saving failure and Llama4Config missing pad_token_id are unrelated." }, { "left": "issue:45310", "right": "issue:45375", - "accept": true, - "reason": "Same Qwen3.5-MoE failure family; 45375 pinpoints the missing config field that plausibly causes 45310's `from_pretrained` error." + "accept": false, + "reason": "Both are Qwen3.5-related, but one is a from_pretrained error and the other is a specific missing-field/strictness issue; not the same concrete bug." }, { - "left": "issue:42222", - "right": "issue:45200", + "left": "issue:43232", + "right": "issue:44188", "accept": false, - "reason": "VitPose model breakage vs Gemma-4 token-type-id defaulting; different model-specific issues." + "reason": "Generation kwargs syncing and torch.compile attention-kernel divergence are unrelated." }, { - "left": "issue:44623", - "right": "issue:45656", + "left": "issue:43873", + "right": "issue:44945", "accept": false, - "reason": "Processor save regression vs DeepSpeed optimizer-step duplication; not the same bug." + "reason": "Quantization offloading and pipeline-parallel output corruption are different runtime behaviors and fixes." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "VitPose model breakage and Gemma-4 mm_token_type_ids requirements are unrelated model-specific issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded overall and conservative in the soft-edge calls. The summary appropriately treats the packet as mostly heterogeneous and limits the duplicate claim to the Qwen3.5-MoE loading/config trio, which is the only clearly supported mini-cluster from the titles provided." + "feedback": "The summary is grounded in the packet and stays conservative: it identifies 45278 as a broad umbrella for the v5 import/regression cluster, and the soft-pair calls are appropriately cautious. The only accepted pair (42673/43704) is well supported by identical titles and matching bug descriptions; the rest are reasonably rejected as related-but-distinct issues." }, "evaluator_used": true, "retried": false, @@ -10799,11 +10823,13 @@ "issue:43278", "issue:43377", "issue:43479", + "issue:43504", + "issue:43526", + "issue:43673", "issue:43723", "issue:43784", "issue:43824", - "issue:43874", - "issue:43901", + "issue:43976", "issue:44038", "issue:44060", "issue:44079", @@ -10811,289 +10837,274 @@ "issue:44188", "issue:44222", "issue:44242", - "issue:44295", "issue:44351", "issue:44373", - "issue:44448", "issue:44496", "issue:44568", "issue:44623", "issue:44734", "issue:44779", "issue:44792", + "issue:44869", + "issue:44933", + "issue:44945", "issue:44964", - "issue:45092", "issue:45103", "issue:45125", "issue:45137", "issue:45310", "issue:45356", - "issue:45405", "issue:45561", - "issue:45600", - "issue:45656", - "issue:45712" + "issue:45712", + "issue:45753" ], "soft_pairs": [ "issue:42831|issue:44222", "issue:41628|issue:44351", + "issue:44945|issue:45753", "issue:44222|issue:45310", "issue:42994|issue:44164", "issue:42175|issue:45712", + "issue:43673|issue:44188", "issue:43377|issue:44242", "issue:36010|issue:41628", "issue:44038|issue:45310", + "issue:43526|issue:45561", + "issue:43232|issue:44792", + "issue:43504|issue:45561", "issue:44496|issue:45561", - "issue:45561|issue:45600", "issue:44964|issue:45137", "issue:42898|issue:44568", + "issue:43479|issue:44079", "issue:30990|issue:35141", - "issue:43901|issue:45561", - "issue:43232|issue:44792", - "issue:43232|issue:44188", - "issue:43784|issue:43824", + "issue:43479|issue:44869", + "issue:42175|issue:44933", + "issue:43976|issue:44933", "issue:44188|issue:44373", "issue:43278|issue:45137", + "issue:43784|issue:43824", "issue:44623|issue:45310", - "issue:44734|issue:45103", "issue:30333|issue:33290", + "issue:44734|issue:45103", "issue:42898|issue:45356", "issue:43723|issue:44568", - "issue:44060|issue:45125", "issue:42898|issue:44779", - "issue:45405|issue:45712", - "issue:43874|issue:45092", - "issue:42898|issue:44448", - "issue:43479|issue:44079", - "issue:44188|issue:44964", - "issue:44295|issue:45656" + "issue:44060|issue:45125" ], - "prepared_review_unit_hash": "45df2ddf32ced97ef49075c2cb6c421075e9ed25609c76dd9efc1b94bbe9cb4e", + "prepared_review_unit_hash": "4d1d3a59fc177c1151b0933dcfe3caf48321b0436391f152eb37351bf8c23e94", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37267, - "estimated_input_tokens": 9317, - "estimated_eval_tokens": 18890 + "soft_pair_count": 31, + "serialized_chars": 36841, + "estimated_input_tokens": 9211, + "estimated_eval_tokens": 18678 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: mostly unrelated bug reports spanning tokenizer regressions, import errors, save/load issues, multimodal model failures, and DeepSpeed/compile edge cases. None of the soft-edge pairs look like the same underlying defect, so all are rejected.", - "confidence": 0.92, - "canonical_issue_reason": "issue:42898 is the broadest, most central regression in the set: a v5 behavior change affecting tokenization semantics rather than a model-specific one-off. It best represents the general \u2018transformers v5 breaking behavior\u2019 theme visible across several nearby issues.", + "summary": "This set is not a true duplicate cluster: it mixes unrelated issues across imports, tokenizers, generation, FP8, save/load, multimodal, and testing. I rejected every soft edge.", + "confidence": 0.81, + "canonical_issue_reason": "issue:45561 is the strongest standalone report by activity and repro detail, but it is not a real representative of the rest of this heterogeneous set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the strongest representative issue because it is concrete, broadly relevant, and not tied to a single model family or integration path.", + "best_issue_reason": "issue:45561 has the highest external signal (inbound refs) and a concrete race-condition description, so it is the best single issue to keep if one must be chosen.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "Both mention FP8, but one is an accuracy issue and the other is a save_pretrained failure for MoE; different code paths." + "reason": "Both mention FP8, but one is an accuracy regression and the other is a save_pretrained bug for MoE; different failure modes." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different modules." + "reason": "Both are import errors, but they concern different missing symbols (`AutoImageProcessor` vs `HybridCache`) and different code paths." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption and an image-triggered infinite loop in Qwen3_5 are unrelated bugs." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "One is a FP8 save_pretrained bug and the other is a Qwen3.5 from_pretrained error; not the same defect." + "reason": "Both involve Qwen3.5/MoE serialization/loading, but one is FP8 save_pretrained and the other is a from_pretrained regression on newer transformers." }, { "left": "issue:42994", "right": "issue:44164", "accept": false, - "reason": "Both involve serialization, but one is quantized model saving and the other is extra_state handling in save/from_pretrained." + "reason": "Both are save/load issues, but `quantized model saving` and `extra_state` handling are distinct serialization failures." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Packaging/backend availability versus dummy class leakage into torch-less environments are different problems." + "reason": "TensorFlow backend packaging and leftover dummy class cleanup are different maintenance problems." + }, + { + "left": "issue:43673", + "right": "issue:44188", + "accept": false, + "reason": "Generation cache missing in chunked prefill and attention-kernel divergence under torch.compile are different internals and symptoms." }, { "left": "issue:43377", "right": "issue:44242", "accept": false, - "reason": "Missing padding-mask support in an encoder is unrelated to load-balancing loss gating on router logits." + "reason": "Missing padding-mask support in MIMI encoder vs missing load-balancing loss when router logits are disabled are not the same bug." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Different missing imports (`GenerationMixin` vs `AutoImageProcessor`) and different failure surfaces." + "reason": "Both are import failures, but they target different exported classes and different regression sources." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Both touch Qwen/MoE and v5, but the reported failures are distinct and no shared code path is shown." + "reason": "Both mention Qwen3.5/VL MoE compatibility, but one is a broad transformers 5.0 model issue and the other is a specific newer-version from_pretrained error." }, { - "left": "issue:44496", + "left": "issue:43526", "right": "issue:45561", "accept": false, - "reason": "A model config loading error and a pytest-xdist file race are unrelated." - }, - { - "left": "issue:45561", - "right": "issue:45600", - "accept": false, - "reason": "Concurrency on captured_info.txt is unrelated to stale auto_mappings references." - }, - { - "left": "issue:44964", - "right": "issue:45137", - "accept": false, - "reason": "Multimodal model loading and a DeepSpeed ZeRO3 deque error are separate bugs." + "reason": "ImageProcessor label reduction and a pytest-xdist captured_info race are completely unrelated." }, { - "left": "issue:42898", - "right": "issue:44568", - "accept": false, - "reason": "Both are tokenization regressions in v5, but they affect different tokenizer behaviors and models." - }, - { - "left": "issue:30990", - "right": "issue:35141", + "left": "issue:43232", + "right": "issue:44792", "accept": false, - "reason": "A loading hang and an embedding reinitialization bug are not the same underlying issue." + "reason": "One is a generation-kwarg update bug after sync_gpus; the other is a janus image-generation test failure." }, { - "left": "issue:43901", + "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "A documentation mismatch is unrelated to a test utility race condition." + "reason": "BEiT pretrained loading with a legacy field is unrelated to a testing-utils file race." }, { - "left": "issue:43232", - "right": "issue:44792", + "left": "issue:44496", + "right": "issue:45561", "accept": false, - "reason": "Generation kwargs handling and a Janus image-generation test failure do not share the same concrete bug." + "reason": "Model config loading error and xdist capture-file race are different layers of the stack." }, { - "left": "issue:43232", - "right": "issue:44188", + "left": "issue:44964", + "right": "issue:45137", "accept": false, - "reason": "Generation bookkeeping after sync_gpus is unrelated to torch.compile-dependent attention kernel divergence." + "reason": "Phi-4 multimodal loading and a DeepSpeed ZeRO3 deque underflow do not share a code path." }, { - "left": "issue:43784", - "right": "issue:43824", + "left": "issue:42898", + "right": "issue:44568", "accept": false, - "reason": "Both are import failures, but for different names in different packages/models." + "reason": "Both are tokenizer behavior regressions in v5, but one is whitespace cleanup and the other is add_special_tokens/BOS-EOS handling." }, { - "left": "issue:44188", - "right": "issue:44373", + "left": "issue:43479", + "right": "issue:44079", "accept": false, - "reason": "A runtime kernel divergence bug is unrelated to a wrong docstring." + "reason": "Multimodal config defaulting and ModelOutput key assignment are unrelated object-initialization bugs." }, { - "left": "issue:43278", - "right": "issue:45137", + "left": "issue:30990", + "right": "issue:35141", "accept": false, - "reason": "BF16/FP32 dtype drift during eval is unrelated to a DeepSpeed deque underflow." + "reason": "Sentence-Transformers loading hang vs embedding resize reinitialization are unrelated model-loading vs model-shape issues." }, { - "left": "issue:44623", - "right": "issue:45310", + "left": "issue:43479", + "right": "issue:44869", "accept": false, - "reason": "Missing files in processor.save_pretrained and a Qwen3.5 from_pretrained error are different operations and bugs." + "reason": "Phi-4 multimodal config init and Whisper timestamp decoding crash are unrelated." }, { - "left": "issue:44734", - "right": "issue:45103", + "left": "issue:42175", + "right": "issue:44933", "accept": false, - "reason": "A serving KV-cache indexing crash and an auto-docstring annotation crash are unrelated." + "reason": "Packaging dependency selection and a missing image_utils import are different issues." }, { - "left": "issue:44779", - "right": "issue:44568", + "left": "issue:43976", + "right": "issue:44933", "accept": false, - "reason": "Both are tokenizer regressions, but they involve different models and different tokenization behaviors." + "reason": "Python version compatibility and a nonexistent image_utils import are not the same defect." }, { - "left": "issue:44792", - "right": "issue:44964", + "left": "issue:44188", + "right": "issue:44373", "accept": false, - "reason": "Different multimodal models and different failure modes; not mergeable into one fix." + "reason": "A torch.compile-dependent kernel branch and a wrong docstring are not duplicates." }, { - "left": "issue:45092", + "left": "issue:43278", "right": "issue:45137", "accept": false, - "reason": "Remote-code/meta-init incompatibility is unrelated to a ZeRO3 deque error." + "reason": "Embedding dtype drift during evaluation and a ZeRO3 deque pop error are unrelated." }, { - "left": "issue:45103", - "right": "issue:44734", + "left": "issue:43784", + "right": "issue:43824", "accept": false, - "reason": "Reversed pair of the same two unrelated issues; still not a duplicate." + "reason": "Both are import errors, but they involve different packages and different missing names (`nn` vs `Qwen2_5_VLForConditionalGeneration`)." }, { - "left": "issue:45125", + "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen3.5 MoE, but one is tensor-parallel metadata and the other is a from_pretrained failure; too different to treat as one bug." - }, - { - "left": "issue:45356", - "right": "issue:42898", - "accept": false, - "reason": "Two separate tokenizer regressions affecting different models and behaviors." + "reason": "Processor save_pretrained missing files and Qwen3.5 MoE from_pretrained are distinct save/load problems." }, { - "left": "issue:45405", - "right": "issue:45712", + "left": "issue:30333", + "right": "issue:33290", "accept": false, - "reason": "A dependency version bump issue is unrelated to dummy class cleanup." + "reason": "MLFlow reporting on failed training jobs and deepspeed OOM with Adafactor are unrelated." }, { - "left": "issue:43874", - "right": "issue:45092", + "left": "issue:44734", + "right": "issue:45103", "accept": false, - "reason": "Different multimodal model families and different missing-method/meta-init problems." + "reason": "KV-cache indexing in `transformers serve` and an auto-docstring future-annotations crash are different subsystems." }, { "left": "issue:42898", - "right": "issue:44448", + "right": "issue:45356", "accept": false, - "reason": "A tokenizer behavior change and a model output mismatch across v4/v5 are not the same defect." + "reason": "Both are tokenizer regressions, but they affect different models and different underlying codec/whitespace behavior." }, { - "left": "issue:43479", - "right": "issue:44079", + "left": "issue:43723", + "right": "issue:44568", "accept": false, - "reason": "Default config initialization and ModelOutput key assignment are unrelated internals." + "reason": "Tokenizer loading in v5 and add_special_tokens behavior for mdeberta are separate tokenizer regressions." }, { - "left": "issue:44188", - "right": "issue:44964", + "left": "issue:42898", + "right": "issue:44779", "accept": false, - "reason": "Compile-dependent attention divergence does not match a model loading failure." + "reason": "Both are v5 tokenizer regressions, but one is cleanup_spaces behavior and the other is Deepseek codec/output correctness." }, { - "left": "issue:44295", - "right": "issue:45656", + "left": "issue:44060", + "right": "issue:45125", "accept": false, - "reason": "A position_ids buffer access error and double optimizer stepping under DeepSpeed are different bugs." + "reason": "Tied-weights warning misassignment and missing `_tp_plan` for tensor parallelism are different Qwen3-related model bugs." } ] }, @@ -11117,6 +11128,7 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:22355", "issue:28282", "issue:30333", "issue:30990", @@ -11125,19 +11137,15 @@ "issue:36331", "issue:38175", "issue:41628", - "issue:42175", "issue:42757", "issue:42831", - "issue:42886", "issue:42898", - "issue:42994", "issue:43295", "issue:43479", "issue:43644", "issue:43723", "issue:43824", - "issue:43873", - "issue:43976", + "issue:43874", "issue:44206", "issue:44222", "issue:44261", @@ -11146,38 +11154,43 @@ "issue:44297", "issue:44360", "issue:44373", + "issue:44442", + "issue:44448", "issue:44483", "issue:44484", "issue:44485", "issue:44514", "issue:44623", - "issue:44869", "issue:44871", "issue:44913", "issue:44933", - "issue:44945", "issue:44977", + "issue:44991", "issue:45092", "issue:45103", "issue:45310", "issue:45357", + "issue:45405", "issue:45406", "issue:45459", "issue:45656", "issue:45712" ], "soft_pairs": [ + "issue:45405|issue:45712", + "issue:43874|issue:45092", + "issue:42898|issue:44448", + "issue:44295|issue:45656", "issue:42898|issue:43295", "issue:44933|issue:45712", - "issue:42175|issue:44933", "issue:41628|issue:45459", - "issue:43976|issue:44933", "issue:44222|issue:44297", "issue:44297|issue:45656", - "issue:43479|issue:44869", + "issue:43479|issue:43644", "issue:44206|issue:45103", "issue:42831|issue:44623", "issue:44222|issue:44623", + "issue:22355|issue:28282", "issue:30333|issue:36331", "issue:41628|issue:42757", "issue:30333|issue:34689", @@ -11188,246 +11201,257 @@ "issue:44222|issue:44484", "issue:44483|issue:45406", "issue:44263|issue:44360", - "issue:43873|issue:44945", "issue:44263|issue:45092", "issue:44514|issue:45103", - "issue:43479|issue:43644", "issue:44373|issue:44485", "issue:44295|issue:44623", "issue:43723|issue:44871", + "issue:43723|issue:44991", "issue:45092|issue:45357", - "issue:44261|issue:45092", "issue:43479|issue:44913", + "issue:44261|issue:45092", + "issue:43723|issue:44442", "issue:30990|issue:38175", - "issue:41628|issue:43824", - "issue:42886|issue:42994" + "issue:41628|issue:43824" ], - "prepared_review_unit_hash": "14f28dffa56d704361441560d585f26a54f614daa74bb28b78392701f258f765", + "prepared_review_unit_hash": "8f99af6788e864c0cac33c73f001b1683e88d66762a9fd98f70a12f3f576a3d9", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37709, - "estimated_input_tokens": 9428, - "estimated_eval_tokens": 19112 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 36, + "serialized_chars": 37826, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are unrelated Transformers regressions spanning imports, tokenizers, multimodal processors, training, quantization, and save/load behavior. The soft-similarity pairs share broad themes only, not the same concrete bug or change, so none should be merged.", - "confidence": 0.82, - "canonical_issue_reason": "issue:41628 is the most representative user-facing import regression in the set and sits near several other import/load-related reports, making it the best single issue to anchor the cluster.", + "summary": "This cluster is highly heterogeneous: it mixes unrelated import errors, tokenizer/model-loading regressions, multimodal processor issues, config serialization bugs, and training/runtime problems. I found no true duplicate pairs.", + "confidence": 0.98, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:41628 is the broadest and clearest representative of the import/load subset; it is actionable and more general than the model-specific or training-specific items around it.", + "best_issue_reason": "No single issue is a strong cluster anchor; the closest broad theme is tokenizer/model-loading regressions, but the items are still too diverse to treat as one duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Unrelated: PEFT version pin mismatch vs leftover dummy classes/import surface cleanup." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Different failures: missing image-processor method vs remote-code/meta-init incompatibility." + }, { "left": "issue:42898", - "right": "issue:43295", + "right": "issue:44448", "accept": false, - "reason": "Both are v5 regressions, but one is tokenization cleanup behavior and the other is processor/tokenizer attribute access for custom model code; different code paths." + "reason": "Both are v5 behavior changes, but one is tokenization cleanup and the other is Pegasus output drift." }, { - "left": "issue:44933", - "right": "issue:45712", + "left": "issue:44295", + "right": "issue:45656", "accept": false, - "reason": "Both involve image_utils-related symbols, but one is a bad import target while the other is dummy PT classes leaking into dir() without torch; not the same bug." + "reason": "Different code paths: position_ids buffer access vs DeepSpeed double optimizer step." }, { - "left": "issue:42175", - "right": "issue:44933", + "left": "issue:42898", + "right": "issue:43295", "accept": false, - "reason": "TensorFlow backend packaging and a missing image_utils import are different dependency/import problems." + "reason": "Tokenizer cleanup change vs processor/tokenizer regression on batched image input." }, { - "left": "issue:41628", - "right": "issue:45459", + "left": "issue:44933", + "right": "issue:45712", "accept": false, - "reason": "Different missing symbols in different modules: transformers import exposure vs tokenizer protobuf error handling." + "reason": "Different cleanup areas: nonexistent image_utils import vs dummy_pt_objects leakage." }, { - "left": "issue:43976", - "right": "issue:44933", + "left": "issue:41628", + "right": "issue:45459", "accept": false, - "reason": "Python version compatibility regression is unrelated to the image_utils import issue." + "reason": "Import errors in different modules with different causes: AutoImageProcessor vs protobuf decode handling." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "FP8 save_pretrained failure and tokenizer save metadata mismatch affect different save paths." + "reason": "Different features: FP8 save_pretrained for MoE vs tokenizer_class metadata mismatch." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer config serialization and deepspeed optimizer-step behavior are unrelated." + "reason": "Tokenizer save metadata bug vs optimizer stepping twice under DeepSpeed." }, { "left": "issue:43479", - "right": "issue:44869", + "right": "issue:43644", "accept": false, - "reason": "Multimodal config default initialization and Whisper timestamp decoding are different model-specific bugs." + "reason": "Config default initialization bug vs non-persistent buffer corruption." }, { "left": "issue:44206", "right": "issue:45103", "accept": false, - "reason": "Feature extractor argument regression and auto_docstring annotation crash are separate code paths." + "reason": "Feature-specific image extractor argument bug vs docstring/introspection crash." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FP8 accuracy issues and processor.save_pretrained file omissions are not the same underlying defect." + "reason": "Accuracy regression in FP8 vs processor.save_pretrained missing files." }, { "left": "issue:44222", "right": "issue:44623", "accept": false, - "reason": "Both affect saving, but FP8/moe serialization and processor artifact saving are distinct failures." + "reason": "Both involve saving, but FP8 MoE save bug is not the same as processor artifact omission." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Different missing-import problems: transformers.onnx vs PyTorch availability for AutoModel." }, { "left": "issue:30333", "right": "issue:36331", "accept": false, - "reason": "MLflow reporting on failed jobs and a Trainer compute_loss signature break are unrelated." + "reason": "MLflow reporting on failed training vs Trainer compute_loss signature change." }, { "left": "issue:41628", "right": "issue:42757", "accept": false, - "reason": "Missing AutoImageProcessor import and missing is_offline_mode from huggingface_hub are different import regressions." + "reason": "Both are import errors, but in different packages and unrelated symbols." }, { "left": "issue:30333", "right": "issue:34689", "accept": false, - "reason": "MLflow status reporting and Llama 3.2 model loading are unrelated bugs." + "reason": "Training failure reporting vs a model-loading regression." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Tokenizer loading from cache and flash-attention generation errors are different failure modes." + "reason": "Tokenizer loading failure vs flash-attention generation issue; different subsystems." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer save metadata mismatch and Qwen3.5 MoE from_pretrained regression are separate issues." + "reason": "Tokenizer config mismatch vs Qwen3.5 MoE from_pretrained regression." }, { "left": "issue:28282", "right": "issue:30333", "accept": false, - "reason": "PyTorch missing ImportError and MLflow job reporting are not the same problem." + "reason": "PyTorch missing import error vs MLflow status reporting bug." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "Failed-job reporting and embedding resize/post_init reinitialization are unrelated." + "reason": "Training logging issue vs embedding resize/post_init reinitialization bug." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 save failure and a max_shard_size behavior question are not duplicates." + "reason": "FP8 save_pretrained bug is unrelated to max_shard_size behavior." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Chat completions request handling and Gemma4Processor _tokenizer access crash are different API paths." + "reason": "Chat-completions request handling vs Gemma4Processor missing _tokenizer attribute." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Torch.split indexing behavior and a missing ReLU in the DSA indexer are different bugs in different logic." - }, - { - "left": "issue:43873", - "right": "issue:44945", - "accept": false, - "reason": "Quantization offloading and pipeline parallelism output corruption are separate distributed-inference issues." + "reason": "Different GLM/DSA indexer problems: split return handling vs missing ReLU." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "torch.split return handling and remote-code meta initialization incompatibility do not share the same concrete defect." + "reason": "Torch split return bug vs remote-code/meta-init incompatibility." }, { "left": "issue:44514", "right": "issue:45103", "accept": false, - "reason": "Batched chat template padding crash and auto_docstring annotation crash are unrelated processor/docstring issues." - }, - { - "left": "issue:43479", - "right": "issue:43644", - "accept": false, - "reason": "Default multimodal config initialization and non-persistent buffer junk are different initialization bugs." + "reason": "Batched chat-template crash vs auto-docstring annotation crash." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Wrong docstring for position_ids and GLM-5 RoPE implementation are not the same bug." + "reason": "Docstring wording issue vs RoPE implementation discussion." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Buffer readback after registration and processor.save_pretrained missing files are different serialization/storage issues." + "reason": "Position_ids buffer read error vs processor.save_pretrained missing files." }, { "left": "issue:43723", "right": "issue:44871", "accept": false, - "reason": "Tokenizer loading failure and eos_token_id config inconsistency are distinct tokenizer-related bugs." + "reason": "Both concern tokenizers, but one is a general v5 load issue and the other is eos_token_id mismatch." + }, + { + "left": "issue:43723", + "right": "issue:44991", + "accept": false, + "reason": "Same broad area, but the concrete failures/models differ; not a clear duplicate." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Transformers v5 meta-init incompatibility and incorrect visual encoder keys on save_pretrained are separate model lifecycle problems." + "reason": "Different multimodal regressions: meta initialization incompatibility vs incorrect visual encoder keys on save." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Different config persistence bugs: multimodal defaults vs rotary_pct reload reset." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA layernorm epsilon precision error and remote-code meta initialization incompatibility are unrelated." + "reason": "RMS norm epsilon precision/config issue vs remote-code meta-init incompatibility." }, { - "left": "issue:30990", - "right": "issue:38175", + "left": "issue:44514", + "right": "issue:45103", "accept": false, - "reason": "Sentence Transformers loading hang and zero probabilities in siglip2-base-patch16-224 are different model behaviors." + "reason": "Batched processor crash vs future-annotations docstring crash; unrelated." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they concern different symbols and model families; not the same missing code path." - }, - { - "left": "issue:42886", - "right": "issue:42994", - "accept": false, - "reason": "Offline tokenizer cache loading and quantized model saving failures are unrelated." + "reason": "Different missing-class imports: AutoImageProcessor vs Qwen2_5_VLForConditionalGeneration." } ] }, @@ -11451,16 +11475,17 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:22355", - "issue:28282", + "issue:29127", "issue:30990", "issue:33357", + "issue:39401", "issue:41628", "issue:42371", "issue:42757", "issue:42831", "issue:42886", "issue:42907", + "issue:42994", "issue:43066", "issue:43116", "issue:43122", @@ -11468,10 +11493,10 @@ "issue:43408", "issue:43475", "issue:43576", - "issue:43723", "issue:43756", "issue:43784", "issue:43937", + "issue:43976", "issue:44038", "issue:44060", "issue:44117", @@ -11481,14 +11506,13 @@ "issue:44297", "issue:44360", "issue:44373", - "issue:44442", "issue:44485", + "issue:44488", "issue:44514", "issue:44521", "issue:44625", "issue:44704", "issue:44743", - "issue:44779", "issue:44829", "issue:44945", "issue:45200", @@ -11501,252 +11525,247 @@ "issue:45753" ], "soft_pairs": [ - "issue:22355|issue:28282", + "issue:42886|issue:42994", "issue:42831|issue:45310", "issue:44263|issue:44373", - "issue:44945|issue:45753", "issue:42831|issue:44038", + "issue:43232|issue:45593", + "issue:43756|issue:44060", "issue:42757|issue:45278", + "issue:42831|issue:45753", "issue:43576|issue:43784", "issue:42886|issue:43066", - "issue:43232|issue:45593", + "issue:43122|issue:44521", "issue:44117|issue:44704", + "issue:43232|issue:44743", "issue:30990|issue:33357", "issue:42371|issue:44263", "issue:44704|issue:45290", - "issue:43756|issue:44060", "issue:42907|issue:44945", "issue:43122|issue:44514", + "issue:43976|issue:44945", "issue:44295|issue:44297", "issue:43232|issue:45245", - "issue:43723|issue:44779", + "issue:39401|issue:44488", "issue:43408|issue:43475", - "issue:43723|issue:44442", - "issue:43122|issue:44521", "issue:43784|issue:45600", "issue:42907|issue:43937", "issue:41628|issue:45278", "issue:44261|issue:44373", "issue:44360|issue:44485", - "issue:43232|issue:44743", + "issue:39401|issue:44521", "issue:43116|issue:45200", "issue:44625|issue:44829", - "issue:43122|issue:44779" + "issue:29127|issue:30990" ], - "prepared_review_unit_hash": "5c489127a4c42cad715d8a6e163ae4dda5e9d789bbe6f427a4da87f76567a1d2", + "prepared_review_unit_hash": "62ac4993119e57a9d823013d127a85b6b1741ce920c15a6a1392508029e0f64f", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36587, - "estimated_input_tokens": 9147, - "estimated_eval_tokens": 18550 + "soft_pair_count": 32, + "serialized_chars": 36856, + "estimated_input_tokens": 9214, + "estimated_eval_tokens": 18684 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This looks like a loose bundle of mostly unrelated Transformers issues, not a tight duplicate cluster. No soft edge is convincing enough to merge; the closest theme is the broad import-regression report, which is the best anchor if one must be chosen.", - "confidence": 0.63, - "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report in the set (many import errors after upgrading), and it is the closest match to the other import-related regressions.", + "summary": "This cluster is highly heterogeneous: it mixes unrelated Transformers bugs around imports, tokenization, multimodal processors, generation, quantization, and model-specific regressions. None of the soft pairs appear to be the same underlying defect, so there is no strong duplicate group here.", + "confidence": 0.38, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report in the set ('Many import errors after update from 4.57.0 to 5.5.0') and is the closest thing to a cluster representative, though the cluster is still too mixed to treat it as a true canonical duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best global representative because it is the most general and likely to subsume other import-related failures in this bundle.", + "best_issue_reason": "issue:45278 is the best representative only because it captures a general v5 regression theme; it is not a strong canonical for the whole cluster since most other issues are model- or API-specific and unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:22355", - "right": "issue:28282", + "left": "issue:42886", + "right": "issue:42994", "accept": false, - "reason": "Both are import/load failures, but one is about transformers.onnx and the other about missing PyTorch; different root causes." + "reason": "Offline tokenizer cache loading and quantized model saving failures are different problems with different code paths." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy drift and a Qwen3.5 MOE from_pretrained error are unrelated problems." + "reason": "FineGrainedFP8 accuracy drift and Qwen3.5 from_pretrained failure are unrelated bugs." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "A torch.split return-value issue and a docstring bug are not the same underlying defect." + "reason": "A torch.split return-value issue is not the same as a wrong docstring." }, { - "left": "issue:44945", - "right": "issue:45753", + "left": "issue:42831", + "right": "issue:44038", "accept": false, - "reason": "Pipeline-parallelism output corruption and a specific Qwen3_5 image infinite loop are distinct model/runtime bugs." + "reason": "FP8 accuracy regression and Qwen3-VL-Moe bug are distinct model/runtime issues." }, { - "left": "issue:42831", - "right": "issue:44038", + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation kwargs handling and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:44060", "accept": false, - "reason": "Different symptoms and code paths: FP8 accuracy vs Qwen3-VL-Moe loading/usage." + "reason": "Smollm3 RoPE layer count and Qwen3-Next tied-weights warning do not describe the same defect." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "45278 is a broad import-regression report, but 42757 is a specific huggingface_hub symbol import error; not enough to call them the same bug." + "reason": "Both are import-related, but one is a specific missing symbol and the other is a broad upgrade-wide import regression." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy issues and an infinite loop on a specific image are not the same bug." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "An env command regression and a sentence-transformers import NameError are different failures." + "reason": "Broken env command and a NameError when importing sentence-transformers are separate failures." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline cache loading and wrong tokenizer decoder type are different tokenizer issues." + "reason": "Offline cache loading and decoder-type mismatch are unrelated tokenizer issues." }, { - "left": "issue:43232", - "right": "issue:45593", + "left": "issue:43122", + "right": "issue:44521", "accept": false, - "reason": "Generation state handling after sync_gpus is unrelated to D-FINE auxiliary-loss behavior." + "reason": "Tokenization differences across versions and assistant mask generation for multimodal inputs are different problems." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES/ from_pretrained logic differs from AutoProcessor kwargs forwarding." + "reason": "A mapping-name None assumption bug and AutoProcessor kwargs forwarding are not the same code-path." + }, + { + "left": "issue:43232", + "right": "issue:44743", + "accept": false, + "reason": "General generation kwarg updates and Qwen3.5 recurrent-state reset are distinct generation bugs." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Sentence Transformers loading hang and a MacOS bus error on CLIP are separate issues." + "reason": "Sentence-Transformers loading stall and a MacOS bus error on CLIP are different failures." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 settings guidance and torch.split behavior have no direct overlap." + "reason": "TF32 settings guidance and torch.split return values are unrelated." }, { "left": "issue:44704", "right": "issue:45290", "accept": false, - "reason": "Cached-file kwargs forwarding is unrelated to apply_chat_template crashing on tool-call assistant messages." - }, - { - "left": "issue:43756", - "right": "issue:44060", - "accept": false, - "reason": "Smollm3 RoPE layer handling and Qwen3-Next tied-weights warnings are different model-specific bugs." + "reason": "AutoProcessor caching kwargs and chat-template crashes with tool calls are separate APIs and failure modes." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Saving dequantized Ministrals and incorrect pipeline-parallel output are unrelated." + "reason": "Saving dequantized models and incorrect pipeline-parallel outputs are not the same issue." }, { "left": "issue:43122", "right": "issue:44514", "accept": false, - "reason": "Both touch tokenization/chat templating, but one is a version-to-version tokenization difference and the other is a batched multimodal crash." + "reason": "Version-to-version tokenization drift and chat-template batching crashes are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version support regression and pipeline-parallel output corruption are unrelated." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "Reading a registered position_ids buffer and saving tokenizer_class metadata are unrelated." + "reason": "Reading a buffer-backed position_ids tensor and tokenizer_class mismatch on save are different tokenizer/model serialization issues." }, { "left": "issue:43232", "right": "issue:45245", "accept": false, - "reason": "Generation cache behavior and the 2^24 category limit error are not the same bug." + "reason": "Generation kwargs handling and a category-count runtime limit are unrelated." }, { - "left": "issue:43723", - "right": "issue:44779", + "left": "issue:39401", + "right": "issue:44488", "accept": false, - "reason": "A generic tokenizer load issue and incorrect DeepSeek tokenization are different failure modes." + "reason": "Qwen3 tokenizer offset mapping and failing to load a specific checkpoint are different issues." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Model-type mismatch warning and missing fpn_position_embeddings are related to the same family, but not the same concrete defect." - }, - { - "left": "issue:43723", - "right": "issue:44442", - "accept": false, - "reason": "Generic AutoTokenizer load regression and FastSpeech2ConformerTokenizer load failure are distinct." - }, - { - "left": "issue:43122", - "right": "issue:44521", - "accept": false, - "reason": "Different tokenization across versions and zero assistant masks for multimodal inputs are not the same issue." + "reason": "A model-type warning and a missing attribute on Sam3VisionEncoderOutput are not the same defect." }, { "left": "issue:43784", "right": "issue:45600", "accept": false, - "reason": "Sentence-transformers import NameError and broken auto_mappings CI are unrelated." + "reason": "Missing nn import in sentence-transformers and stale auto_mappings references are separate import/maintenance problems." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Dequantized save behavior and invalid GenerationConfig for GLM-5 are separate problems." + "reason": "Quantized model saving failure and invalid GenerationConfig are unrelated." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "Both are import-related, but 41628 is a single missing AutoImageProcessor import while 45278 is a broad upgrade regression; not enough evidence of one root cause." + "reason": "Missing AutoImageProcessor import and a broad upgrade-wide import-error report are related only at a very high level, not the same bug." }, { "left": "issue:44261", "right": "issue:44373", "accept": false, - "reason": "A missing rms_norm_eps config value and a wrong docstring do not share the same bug." + "reason": "A missing rms_norm_eps configuration and a wrong docstring are not the same bug." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "DSA indexer activation choice and GLM-5 RoPE implementation are unrelated." - }, - { - "left": "issue:43232", - "right": "issue:44743", - "accept": false, - "reason": "Both involve generation/cache semantics, but the sync_gpus bug and recurrent-state reset bug are different code paths." - }, - { - "left": "issue:43116", - "right": "issue:45200", - "accept": false, - "reason": "Example-script multi-label output is unrelated to mm_token_type_ids defaults for Gemma 4 fine-tuning." + "reason": "A missing ReLU in the DSA indexer and GLM-5 RoPE implementation are unrelated." }, { - "left": "issue:44625", - "right": "issue:44829", + "left": "issue:39401", + "right": "issue:44521", "accept": false, - "reason": "num_labels propagation and flash_attention_3 degenerate training are different sequence-classification issues." + "reason": "Tokenizer offset mapping and assistant mask generation are different tokenizer/processor behaviors." }, { - "left": "issue:43122", - "right": "issue:44779", + "left": "issue:29127", + "right": "issue:30990", "accept": false, - "reason": "Generic version-to-version tokenization changes and DeepSeek-specific incorrect tokenization are not clearly the same defect." + "reason": "LayoutLMv3 error messaging and Sentence Transformers loading are completely different issues." } ] }, @@ -11770,21 +11789,22 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:22355", "issue:29127", - "issue:30990", "issue:33453", "issue:36296", "issue:38617", "issue:39401", "issue:42222", "issue:42371", + "issue:42673", "issue:42757", - "issue:42831", "issue:42907", "issue:43064", "issue:43066", "issue:43116", "issue:43122", + "issue:43278", "issue:43441", "issue:43452", "issue:43493", @@ -11803,13 +11823,13 @@ "issue:44297", "issue:44351", "issue:44485", - "issue:44488", - "issue:44521", "issue:44704", "issue:44779", "issue:44936", + "issue:44945", "issue:44964", "issue:44977", + "issue:45072", "issue:45083", "issue:45216", "issue:45278", @@ -11819,273 +11839,273 @@ "issue:45753" ], "soft_pairs": [ - "issue:29127|issue:30990", + "issue:43122|issue:44779", "issue:42907|issue:44964", "issue:44038|issue:44297", "issue:42907|issue:43452", "issue:44038|issue:44295", - "issue:39401|issue:44488", "issue:36296|issue:43116", "issue:43824|issue:44351", "issue:44038|issue:44222", "issue:43066|issue:44779", - "issue:39401|issue:44521", "issue:45216|issue:45478", "issue:42757|issue:44351", + "issue:43278|issue:44945", "issue:43122|issue:44977", "issue:43992|issue:44704", + "issue:43493|issue:45753", "issue:42907|issue:44222", "issue:43784|issue:45446", + "issue:42673|issue:45072", + "issue:38617|issue:45278", + "issue:22355|issue:29127", "issue:43937|issue:44485", "issue:42222|issue:43441", + "issue:43064|issue:43673", + "issue:43116|issue:43493", "issue:42371|issue:43874", "issue:43976|issue:45446", - "issue:43493|issue:45753", - "issue:42831|issue:45753", "issue:44075|issue:44936", "issue:43116|issue:43825", "issue:44222|issue:45478", "issue:44038|issue:45478", - "issue:43116|issue:43493", - "issue:43064|issue:43673", "issue:45083|issue:45446", "issue:43937|issue:45563", "issue:33453|issue:39401", - "issue:39401|issue:44779", - "issue:38617|issue:45278" + "issue:39401|issue:44779" ], - "prepared_review_unit_hash": "47783cbf09764d6fe4b3847189916d7374639fb0d3e5c6d7deffa2ff6333faac", + "prepared_review_unit_hash": "d079340d6fd74993fc1b68ba38daeea6ab934b515166292703aa96b3808123e7", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 36776, - "estimated_input_tokens": 9194, - "estimated_eval_tokens": 18644 + "serialized_chars": 37369, + "estimated_input_tokens": 9343, + "estimated_eval_tokens": 18942 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: import errors, tokenizer regressions, model-specific loading/saving issues, training bugs, and unrelated warnings. None of the soft edges look like the same concrete bug or change, so all candidate duplicate links should be rejected.", - "confidence": 0.87, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella for the import-error subset, but the overall cluster is too mixed to justify a true canonical duplicate.", + "summary": "This cluster is heterogeneous: mostly unrelated Transformers regression/import/tokenizer issues with no true duplicate pairs among the soft candidates. The broadest representative is the open import-regression report, but it is not a true canonical duplicate for the rest.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella for the repeated import-regression reports, though the cluster overall is too mixed to have a single true canonical duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "45278 is the best representative only for the import-regression subset; it is broader than the others and most likely to absorb similar reports.", + "best_issue_reason": "Issue 45278 is the best overall representative because it spans the widest common theme here: version-related import failures after upgrading Transformers.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:29127", - "right": "issue:30990", + "left": "issue:43122", + "right": "issue:44779", "accept": false, - "reason": "Different subsystems and symptoms: layoutlmv3 box validation vs sentence-transformers loading hang." + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; no evidence of the same concrete bug." }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Both are model-loading/saving related, but they are different model families and failure modes." + "reason": "One is about saving dequantized models, the other about loading a multimodal model; different code paths." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe runtime bug vs tokenizer save metadata mismatch; not the same code path." + "reason": "Qwen3-VL-Moe loading failure and tokenizer_config metadata mismatch are distinct issues." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Dequantized save failure vs gguf/AutoTokenizer loading breakage are different bugs." + "reason": "Dequantized save issues and gguf AutoTokenizer/AutoModel loading failures are unrelated." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Model-specific generation/load issue vs position_ids buffer read error; unrelated." - }, - { - "left": "issue:39401", - "right": "issue:44488", - "accept": false, - "reason": "Tokenizer offset mapping bug vs inability to load a specific model checkpoint." + "reason": "A model-specific loading bug and a position_ids buffer access bug are not the same defect." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor-parallel training state bug vs example-script multi-label inference issue." + "reason": "Tensor-parallel training failure and empty multi-label classification outputs are unrelated." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different missing symbols from different modules." + "reason": "Both are import errors, but they reference different missing symbols from different modules." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Qwen3-VL-Moe bug vs FP8 save_pretrained bug; same broad area, different concrete failure." + "reason": "Both mention MoE, but one is a model load bug and the other is an FP8 save_pretrained bug." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Different tokenizer regressions affecting different models and code paths." - }, - { - "left": "issue:39401", - "right": "issue:44521", - "accept": false, - "reason": "Tokenizer offset mapping bug vs multimodal chat-template assistant mask bug." + "reason": "Tokenizer decoder-type mismatch and Deepseek tokenization regression are different concrete failures." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Both mention Qwen3.5 and v5.4+, but one is save_pretrained corruption and the other is from_pretrained failure." + "reason": "save_pretrained checkpoint corruption and from_pretrained loading errors are separate directions with different symptoms." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Import errors from different packages: huggingface_hub vs transformers." + "reason": "Different missing imports from different packages; only the generic class of error matches." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Embedding dtype drift in evaluate and wrong outputs under pipeline parallelism are not the same bug." }, { "left": "issue:43122", "right": "issue:44977", "accept": false, - "reason": "Tokenization regression vs flash-attention generation failure; not the same bug." + "reason": "Tokenizer regression and flash-attention generation failure are unrelated." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "Encoder weight loading issue vs AutoProcessor kwargs propagation issue." + "reason": "Missing embed_tokens.weight in one encoder and kwargs not forwarded in AutoProcessor are different code-path issues." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Implementation discrepancy for SigLIP2 and infinite-loop generation on one image are not duplicates." }, { "left": "issue:42907", "right": "issue:44222", "accept": false, - "reason": "Different save/pretrained bugs in different quantization formats and model families." + "reason": "Both involve save_pretrained, but they target different models and formats with different failures." }, { "left": "issue:43784", "right": "issue:45446", "accept": false, - "reason": "Sentence-transformers import NameError vs flex_attention version check bug are unrelated." + "reason": "Sentence-transformers import NameError and a flex_attention version-check bug are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leakage in dataloader threads and dtype mismatches in bfloat16 inference are different problems." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "A specific missing import and a broad import-error roundup overlap only loosely; not the same underlying bug." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "ONNX module import failure and a LayoutLMv3 error-message clarity issue are unrelated." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "GenerationConfig validation problem vs RoPE implementation discussion; different concrete issues." + "reason": "Invalid GenerationConfig and GLM-5 RoPE discussion are distinct issues." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "vitpose breakage vs Ministral FlashAttention failure; unrelated model paths." + "reason": "vitpose breakage and Ministral-3 FlashAttention failure are unrelated model-specific bugs." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Optimizer-state corruption in training and missing GenerationMixin cache in chunked prefill are different code paths." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Classification-script empty results and SigLIP2 implementation mismatch are unrelated." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 API deprecation notice vs missing image-patch helper; not the same bug." + "reason": "TF32 behavior settings and a missing image-patch-count method are not the same defect." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "Python version compatibility problem vs a specific PyTorch import/version check." - }, - { - "left": "issue:43493", - "right": "issue:45753", - "accept": false, - "reason": "SigLIP2 implementation discrepancy vs Qwen3_5 infinite-loop-on-image issue." - }, - { - "left": "issue:42831", - "right": "issue:45753", - "accept": false, - "reason": "FP8 accuracy issue vs multimodal generation loop are different failures." + "reason": "Python compatibility/version support and an incorrect PyTorch version check are different issues." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Optimizer argument handling vs trainer.evaluate after train; both trainer-related but not the same bug." + "reason": "Ignored SGD arguments and trainer.train()/evaluate() lifecycle failure are unrelated." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Empty multi-label results vs stale translation-task error message; unrelated example/pipeline issues." + "reason": "Empty multi-label outputs and a misleading pipeline translation-task error message are different bugs." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5/MoE, but one is FP8 save_pretrained and the other is from_pretrained loading." + "reason": "Both are MoE-related, but one is save_pretrained FP8 and the other is from_pretrained loading; not mergeable as one fix." }, { "left": "issue:44038", "right": "issue:45478", "accept": false, - "reason": "Different Qwen model variants and different failure modes; not mergeable as one fix." - }, - { - "left": "issue:43116", - "right": "issue:43493", - "accept": false, - "reason": "Classification example bug vs SigLIP2 model discrepancy are unrelated." - }, - { - "left": "issue:43064", - "right": "issue:43673", - "accept": false, - "reason": "Trainer/FSDP optimizer-state bug vs GenerationMixin cache issue; different code paths." + "reason": "Different Qwen model families and different API paths; not the same concrete bug." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "qwen3_omni_moe feature-length helper bug vs flex_attention import/version check bug." + "reason": "qwen3_omni_moe length-helper behavior and a flex_attention import/version-check issue are unrelated." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "GLM-5 generation-config validation vs paged-generate warning text are unrelated." + "reason": "Invalid GenerationConfig and a stale warning in paged generate are different behaviors." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Generic tokenizer loading regression vs Qwen3 offset_mapping bug; not the same defect." + "reason": "Both are tokenizer-related, but one is loading regression and the other is wrong offset mapping; different defects." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and outputs, so not one concrete bug." - }, - { - "left": "issue:38617", - "right": "issue:45278", - "accept": false, - "reason": "45278 is a broad import-error umbrella, but 38617 is a specific missing-symbol report and not clearly the same underlying bug." + "reason": "Different models and different tokenizer failure modes; no shared underlying bug is evident." } ] }, @@ -12109,25 +12129,27 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ - "issue:22355", - "issue:29127", "issue:38617", "issue:39401", "issue:41628", "issue:42175", "issue:42222", "issue:42371", + "issue:42617", "issue:42673", + "issue:42831", "issue:43012", "issue:43066", "issue:43122", - "issue:43381", "issue:43408", "issue:43421", "issue:43450", "issue:43577", + "issue:43582", + "issue:43673", "issue:43742", "issue:43827", + "issue:43906", "issue:43937", "issue:43986", "issue:44075", @@ -12140,257 +12162,262 @@ "issue:44484", "issue:44496", "issue:44829", - "issue:44945", - "issue:44964", + "issue:44933", "issue:44977", "issue:45071", - "issue:45072", "issue:45081", "issue:45216", "issue:45278", "issue:45310", + "issue:45341", "issue:45405", "issue:45446", - "issue:45459", "issue:45464", - "issue:45478", "issue:45542", + "issue:45561", "issue:45588", "issue:45656" ], "soft_pairs": [ "issue:44117|issue:45588", - "issue:22355|issue:29127", "issue:43827|issue:45446", - "issue:43381|issue:44945", "issue:43937|issue:44977", "issue:43450|issue:43986", + "issue:44291|issue:45071", + "issue:42673|issue:43906", "issue:45405|issue:45446", + "issue:45341|issue:45561", "issue:39401|issue:43421", "issue:44117|issue:45081", - "issue:42673|issue:45072", "issue:38617|issue:41628", "issue:44829|issue:45216", "issue:43408|issue:43742", "issue:42222|issue:45542", "issue:44410|issue:45310", - "issue:44291|issue:45071", "issue:42175|issue:42222", + "issue:42673|issue:43582", "issue:44484|issue:45310", + "issue:42371|issue:44261", + "issue:42673|issue:43577", "issue:43937|issue:44496", + "issue:38617|issue:44220", + "issue:42617|issue:45341", + "issue:44933|issue:45278", "issue:44483|issue:45464", "issue:44075|issue:45656", "issue:43066|issue:43122", - "issue:44964|issue:45278", + "issue:42371|issue:42831", "issue:43012|issue:43577", - "issue:38617|issue:44220", - "issue:42371|issue:44261", - "issue:42222|issue:45459", - "issue:44410|issue:45478", - "issue:44484|issue:45478" + "issue:43673|issue:45341" ], - "prepared_review_unit_hash": "1394739772ea01ba1b9c4f5cdbe55d98fc530467a74e2c5eea30d8e351a0e513", + "prepared_review_unit_hash": "d2b66a3f4451818dfceb613bc68e499156b4c336bf6ceec0bd93949d0aa8db42", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36336, - "estimated_input_tokens": 9084, - "estimated_eval_tokens": 18424 + "soft_pair_count": 30, + "serialized_chars": 36482, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: most items are unrelated model, tokenizer, optimizer, or import-regression bugs. None of the soft pairs look like true duplicates, so all soft edges are rejected. The closest thing to a canonical issue is the broad upgrade/import-regression report, but overall the cluster is not a clean duplicate set.", - "confidence": 0.33, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella report in the set, covering multiple import failures after upgrading to 5.5.0, but the cluster is too mixed for a strong canonical duplicate.", + "summary": "The candidates are mostly unrelated closed issues with only superficial lexical overlap; none of the soft pairs looks like the same concrete bug/change, so I reject all pairings. The closest umbrella topic is the open import-regression report, but it still does not duplicate the specific import-error tickets here.", + "confidence": 0.8, + "canonical_issue_reason": "issue:45278 is the broadest tracking-style issue in the set: it summarizes multiple import regressions after upgrading from 4.57.0 to 5.5.0 and is the most plausible umbrella representative.", "canonical_pr_reason": null, - "best_issue_reason": "45278 is the most general, user-facing issue and the best fit if one must anchor the cluster, though it still only matches a subset of the items.", + "best_issue_reason": "issue:45278 is the best representative issue because it is general enough to subsume several of the import-related reports, while the others are narrower point bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44117", "right": "issue:45588", "accept": false, - "reason": "Different failure modes: tokenizer mapping fallback vs flash-attention crash on sink-less models." - }, - { - "left": "issue:22355", - "right": "issue:29127", - "accept": false, - "reason": "Unrelated bugs: missing transformers.onnx module vs LayoutLMv3 error-message clarity." + "reason": "Different failure modes: tokenizer mapping None vs flash-attention s_aux=None crash in integration code." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs/pipeline removal mismatch is unrelated to the flex_attention PyTorch version check bug." - }, - { - "left": "issue:43381", - "right": "issue:44945", - "accept": false, - "reason": "Gradient checkpointing in eval mode and incorrect pipeline-parallel output are different code paths." + "reason": "Docs mentioning removed pipelines is unrelated to the flex_attention PyTorch version-check bug." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "Both affect generation, but one is invalid GenerationConfig and the other is a Qwen3.5 flash-attention issue." + "reason": "One is GenerationConfig validation, the other is Qwen3.5 flash-attention generation behavior; not the same bug." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Batched video processor shape bug is unrelated to AutoProcessor crashing without torchvision." + "reason": "Batched video-processor shape output vs AutoProcessor import crash without torchvision are distinct issues." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized TypeError and PretrainedConfig type-checking are separate compatibility problems." + }, + { + "left": "issue:42673", + "right": "issue:43906", + "accept": false, + "reason": "VRAM leak across dataloader threads is not the same as an isolated reproduction of a different issue." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "PEFT version pinning and a flex_attention import/version check are unrelated." + "reason": "PEFT version pinning and flex_attention import-version gating are unrelated." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "A testing_utils bug and an xdist race in captured_info.txt are different code paths and failure causes." }, { "left": "issue:39401", "right": "issue:43421", "accept": false, - "reason": "Tokenizer offset_mapping regression is different from post-processor not updating after runtime special-token changes." + "reason": "Tokenizer offset_mapping correctness is separate from runtime special-token/post-processor update behavior." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "Tokenizer mapping None and Mistral regex patch backend_tokenizer errors are different bugs." - }, - { - "left": "issue:42673", - "right": "issue:45072", - "accept": false, - "reason": "VRAM leak across dataloader threads is unrelated to dtype mismatches in inference." + "reason": "Tokenizer mapping returning None and the Mistral regex backend_tokenizer AttributeError are not the same defect." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different root causes." + "reason": "Both are import errors, but for different missing symbols from different modules; they do not share a concrete code-path fix." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "Flash-attention training degeneration is unrelated to a save_pretrained checkpoint regression." + "reason": "FlashAttention-3 training degeneration and Qwen3.5 save_pretrained regression are unrelated model-behavior bugs." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "Model-type warning for sam3_tracker is unrelated to a MobileLLM loading key error." + "reason": "A model-type warning for sam3_tracker/sam3_video is not the same as a key error when loading MobileLLM." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "Broken vitpose models and missing TensorFlow backend from tensorboard-only install are unrelated." + "reason": "Broken vitpose models and an undefined TensorFlow backend when only tensorboard is installed are different problems." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Missing Qwen3Next projections and Qwen3.5 MoE from_pretrained error are different model-specific loading bugs." + "reason": "Missing projections in qwen3next and a Qwen3.5 Moe from_pretrained error are distinct model-loading issues." }, { - "left": "issue:44291", - "right": "issue:45071", + "left": "issue:42175", + "right": "issue:42222", "accept": false, - "reason": "init_empty_weights argument handling and PreTrainedConfig type checking are separate regressions." + "reason": "Missing TensorFlow in the torch extra and broken vitpose models are unrelated." }, { - "left": "issue:42175", - "right": "issue:42222", + "left": "issue:42673", + "right": "issue:43582", "accept": false, - "reason": "TensorFlow backend packaging issue and vitpose model breakage are unrelated." + "reason": "A dataloader-thread VRAM leak is unrelated to the Apple Silicon caching_allocator_warmup TypeError." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "Chat/completions API failure is a different surface from Qwen3.5 MoE from_pretrained loading error." + "reason": "A question about max_shard_size defaults is not the same concrete bug as the Qwen3.5 Moe from_pretrained failure." }, { - "left": "issue:43937", - "right": "issue:44496", + "left": "issue:42371", + "right": "issue:44261", "accept": false, - "reason": "Invalid GenerationConfig and unrecognized model/config problems are different validation failures." + "reason": "TF32 API-setting guidance and MLA rms_norm_eps precision issues are unrelated." }, { - "left": "issue:44483", - "right": "issue:45464", + "left": "issue:42673", + "right": "issue:43577", "accept": false, - "reason": "These both touch chat/completions, but one is a request-acceptance bug and the other is streaming inference failure." + "reason": "VRAM leakage and Blip2 dtype remaining float32 are different code paths and symptoms." }, { - "left": "issue:44075", - "right": "issue:45656", + "left": "issue:43937", + "right": "issue:44496", "accept": false, - "reason": "SGD args ignored and deepspeed double-step are distinct optimizer bugs." + "reason": "GenerationConfig invalidation and unrecognized model/model_type loading are separate loader validation failures." }, { - "left": "issue:43066", - "right": "issue:43122", + "left": "issue:38617", + "right": "issue:44220", "accept": false, - "reason": "Both are tokenizer regressions, but one is wrong decoder type and the other is changed tokenization output." + "reason": "A missing import from configuration_utils is unrelated to _torch_extract_fbank_features()." }, { - "left": "issue:44964", + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "Failure to run 3d_parallel.py and a testing_utils bug are not the same underlying issue." + }, + { + "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "A single model-loading failure is not the same as the broad multi-import regression report." + "reason": "A specific missing import from image_utils is not the same concrete bug as a broad umbrella of import errors after upgrade." }, { - "left": "issue:43012", - "right": "issue:43577", + "left": "issue:44483", + "right": "issue:45464", "accept": false, - "reason": "Bfloat16 compilation warning and BLIP2 dtype remaining float32 are different dtype-related bugs." + "reason": "Generic chat/completions request rejection and a Qwen3.5-0.8B streaming inference failure are not the same defect." }, { - "left": "issue:38617", - "right": "issue:44220", + "left": "issue:44075", + "right": "issue:45656", "accept": false, - "reason": "Missing import symbol and _torch_extract_fbank_features failure are unrelated." + "reason": "SGD arguments being ignored and optimizer.step() being called twice under deepspeed are different optimizer bugs." }, { - "left": "issue:42371", - "right": "issue:44261", + "left": "issue:43066", + "right": "issue:43122", "accept": false, - "reason": "TF32 behavior warning and MLA rms_norm_eps precision mismatch are unrelated." + "reason": "Wrong decoder type in v5 and changed tokenization between versions are related only broadly; not a single concrete bug." }, { - "left": "issue:42222", - "right": "issue:45459", + "left": "issue:42371", + "right": "issue:42831", "accept": false, - "reason": "Vitpose model breakage and protobuf-hidden tokenizer errors do not share the same underlying bug." + "reason": "TF32 behavior warnings and FineGrainedFP8 accuracy problems are distinct precision-related issues." }, { - "left": "issue:44410", - "right": "issue:45478", + "left": "issue:43012", + "right": "issue:43577", "accept": false, - "reason": "Both mention Qwen models, but they are different loading failures with different model paths." + "reason": "A PyTorch warning during bfloat16 compilation and Blip2 dtype not propagating are different dtype issues." }, { - "left": "issue:44484", - "right": "issue:45478", + "left": "issue:43673", + "right": "issue:45341", "accept": false, - "reason": "Chat API request handling and Qwen3.5 MoE model loading are unrelated." + "reason": "GenerationMixin cache missing in chunked_prefill and a testing_utils bug do not share the same code path." } ] }, @@ -12416,19 +12443,19 @@ "nodes": [ "issue:38617", "issue:41628", - "issue:42371", - "issue:42673", + "issue:42222", "issue:42757", - "issue:42831", "issue:43012", "issue:43064", + "issue:43066", "issue:43097", "issue:43335", "issue:43408", "issue:43441", "issue:43452", "issue:43502", - "issue:43577", + "issue:43531", + "issue:43824", "issue:43873", "issue:43874", "issue:43906", @@ -12445,299 +12472,320 @@ "issue:44351", "issue:44368", "issue:44410", - "issue:44464", "issue:44484", "issue:44485", "issue:44493", "issue:44704", "issue:44829", - "issue:44945", - "issue:44964", - "issue:44977", "issue:44987", + "issue:45003", "issue:45071", "issue:45103", "issue:45245", "issue:45341", + "issue:45372", "issue:45375", + "issue:45459", + "issue:45478", "issue:45542", "issue:45588" ], "soft_pairs": [ + "issue:42222|issue:45459", + "issue:44410|issue:45478", "issue:43097|issue:44493", + "issue:44484|issue:45478", "issue:43874|issue:44261", "issue:44263|issue:44485", "issue:43012|issue:43408", "issue:43441|issue:44829", "issue:41628|issue:45071", "issue:45071|issue:45542", - "issue:42673|issue:44977", - "issue:42371|issue:42831", - "issue:42673|issue:43906", + "issue:43931|issue:44410", "issue:43986|issue:44484", + "issue:45003|issue:45341", "issue:44117|issue:44987", "issue:43097|issue:43452", + "issue:43986|issue:45341", "issue:43931|issue:44038", "issue:43335|issue:45071", "issue:43408|issue:44368", - "issue:43976|issue:44945", "issue:43874|issue:44263", + "issue:43986|issue:45372", + "issue:43531|issue:45341", + "issue:43976|issue:45071", "issue:44829|issue:45245", + "issue:44261|issue:44485", "issue:38617|issue:42757", "issue:43064|issue:44315", - "issue:44464|issue:45071", "issue:44410|issue:45375", "issue:44062|issue:44117", "issue:43097|issue:45588", - "issue:42673|issue:43577", "issue:38617|issue:44351", "issue:44704|issue:45103", "issue:44038|issue:44410", - "issue:43931|issue:44410", "issue:43986|issue:44117", - "issue:44261|issue:44485", - "issue:44964|issue:45341", "issue:43502|issue:43992", - "issue:43097|issue:43873" + "issue:43097|issue:43873", + "issue:43824|issue:43986", + "issue:43066|issue:43906" ], - "prepared_review_unit_hash": "05aff5baaaf86eb57dcafb34aa43f2b5ba3838cc3897b3275706e720027949a6", + "prepared_review_unit_hash": "306082c946f0dae028a391fd24e69da0b373af4e410c6715de88daa17f023fdc", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 38166, - "estimated_input_tokens": 9542, - "estimated_eval_tokens": 19340 + "soft_pair_count": 38, + "serialized_chars": 39082, + "estimated_input_tokens": 9771, + "estimated_eval_tokens": 19798 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Loose similarity cluster of many unrelated Transformers issues (import errors, model-specific load bugs, warnings, and training regressions). None of the soft pairs look like true duplicates.", - "confidence": 0.94, - "canonical_issue_reason": "No single canonical issue: the cluster spans multiple unrelated failure modes and model families, so one issue does not cleanly subsume the others.", + "summary": "This is a loose collection of Transformers issues around v5 regressions, model-loading failures, and unrelated warnings; I do not see any true duplicate pairs in the provided soft edges.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43097 is the clearest representative of the broad v5-breaking-change theme: it is concrete, user-facing, and has inbound references, while most other issues are narrower model-specific bugs or unrelated import/type warnings.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43097 is the closest representative anchor because it is a broad v5 regression and has inbound references, but it still does not explain the rest of the cluster.", + "best_issue_reason": "Issue 43097 is the strongest overall issue to keep as the cluster representative because it captures a clear core API regression with wider impact than the more specialized reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42222", + "right": "issue:45459", + "accept": false, + "reason": "Different failures: vitpose model breakage vs tokenizer/protobuf error." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen-related, but one is missing projections in qwen3next and the other is a Qwen3.5 MoE from_pretrained load failure." + }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Removed API warning vs unexpected position-id key warning; different regression and code path." + "reason": "Different bugs: removed config attribute vs unexpected position-id key warning." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to a Qwen3.5 MoE loading error." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "Different models and failures: missing image-patch method vs missing rms_norm_eps in config." + "reason": "Unrelated code paths: missing image patch method vs MLA RMS norm epsilon precision issue." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "GlmMoeDsaIndexer split behavior and GLM-5 RoPE implementation are distinct bugs." + "reason": "torch.split return-value issue in GLM MoE indexer vs RoPE implementation bug." }, { "left": "issue:43012", "right": "issue:43408", "accept": false, - "reason": "Float32 precision warning during compile is unrelated to the sam3 tracker/video model-type warning." + "reason": "Both mention warnings, but they are different model/compiler warning paths with no shared bug." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Both mention flash attention, but one is a load-time failure and the other is degenerate training; not the same bug." + "reason": "Different failures: Ministral FlashAttention crash vs degenerate training with flash_attention_3." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "Missing AutoImageProcessor import and PretrainedConfig type-checking breakage are unrelated issues." + "reason": "ImportError for AutoImageProcessor is not the same as a PretrainedConfig type-checking regression." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "Config type-checking regression vs undefined TF backend from tensorboard-only install; different causes." - }, - { - "left": "issue:42673", - "right": "issue:44977", - "accept": false, - "reason": "VRAM leak in multi-threaded dataloading vs generation failure with flash-attention; distinct symptoms and paths." + "reason": "Type-checking breakage and undefined tf backend error are unrelated." }, { - "left": "issue:42371", - "right": "issue:42831", + "left": "issue:43931", + "right": "issue:44410", "accept": false, - "reason": "TF32 API-setting guidance and FineGrainedFP8 accuracy loss are different concerns." + "reason": "Both are Qwen family loading issues, but they affect different models and different concrete failures." }, { - "left": "issue:42673", - "right": "issue:43906", + "left": "issue:43986", + "right": "issue:44484", "accept": false, - "reason": "VRAM leak report vs reproduction of another issue; no concrete shared defect." + "reason": "AutoProcessor/torchvision crash is unrelated to the save_pretrained shard-size discussion." }, { - "left": "issue:43986", - "right": "issue:44484", + "left": "issue:45003", + "right": "issue:45341", "accept": false, - "reason": "AutoProcessor/video-model crash without torchvision is unrelated to save_pretrained shard sizing." + "reason": "Unsafe sys.modules access is unrelated to the testing_utils bug." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES None handling and a model load failure for physical-intelligence/fast are unrelated." + "reason": "TOKENIZER_MAPPING_NAMES None handling is unrelated to the physical-intelligence/fast loading failure." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "Config removal warning and gguf_file loading breakage are different failures." + "reason": "Removed tie_embeddings behavior is not the same as gguf_file breaking tokenizer/model loading." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "AutoProcessor without torchvision and a testing_utils bug are unrelated." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Qwen3-VL weight mismatch and Qwen3-VL-Moe loading bug are different model-specific issues." + "reason": "Both are Qwen3-VL-related, but they are distinct model variants with different loading problems." }, { "left": "issue:43335", "right": "issue:45071", "accept": false, - "reason": "SwitchTransformers sparse-layer creation bug and config type-checking regression are unrelated." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to PretrainedConfig type checking." }, { "left": "issue:43408", "right": "issue:44368", "accept": false, - "reason": "Different warnings on different models: sam3 tracker/type mismatch vs tie_word_embeddings guidance." + "reason": "Different warnings on different models and settings; not the same underlying issue." }, { - "left": "issue:43976", - "right": "issue:44945", + "left": "issue:43874", + "right": "issue:44263", "accept": false, - "reason": "Python-version compatibility problem vs incorrect output under pipeline parallelism; not the same defect." + "reason": "Missing image-patch helper and torch.split return handling are different bugs." }, { - "left": "issue:43874", - "right": "issue:44263", + "left": "issue:43986", + "right": "issue:45372", "accept": false, - "reason": "Missing image-patch method and torch.split return-value issue are separate GLM bugs." + "reason": "Torchvision absence crash is unrelated to the mistral_common ReasoningEffort import error." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to the testing_utils issue." + }, + { + "left": "issue:43976", + "right": "issue:45071", + "accept": false, + "reason": "Python-version compatibility problem is not the same as a config type-check regression." }, { "left": "issue:44829", "right": "issue:45245", "accept": false, - "reason": "FlashAttention training degeneration and category-count runtime limit are unrelated." + "reason": "Degenerate training with flash_attention_3 and category-count overflow are unrelated." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "Missing rms_norm_eps in MLA q_a_layernorm is unrelated to GLM-5 RoPE implementation." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols from different packages." + "reason": "Both are import errors, but they are for different missing symbols and different modules." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "FSDP2/PEFT optimizer-state mismatch and Liger Kernel not applied with model_init are separate training-stack issues." - }, - { - "left": "issue:44464", - "right": "issue:45071", - "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is unrelated to PretrainedConfig type checking." + "reason": "FSDP2 optimizer-state corruption is unrelated to Liger Kernel not being applied with model_init." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Missing projections in qwen3next and missing deepstack_visual_indexes in config are distinct model/config bugs." + "reason": "Different Qwen vision config bugs: missing projections vs missing deepstack_visual_indexes field." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "AddedToken keyword duplication and TOKENIZER_MAPPING_NAMES None are unrelated tokenizer problems." + "reason": "Tokenizer AddedToken argument collision is unrelated to TOKENIZER_MAPPING_NAMES returning None." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "Embedding-tying warning and flash_attention.py s_aux=None crash are different code paths." - }, - { - "left": "issue:42673", - "right": "issue:43577", - "accept": false, - "reason": "VRAM leak under dataloader threads and Blip2 dtype staying float32 are unrelated." + "reason": "Removed tie_embeddings support is unrelated to the flash_attention s_aux=None crash." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "Different missing imports from different modules; same broad category only, not the same bug." - }, - { - "left": "issue:44704", - "right": "issue:45103", - "accept": false, - "reason": "Kwarg forwarding to cached_file and auto_docstring AttributeError are unrelated utility bugs." + "reason": "Both are import errors, but they concern different missing exports and different code paths." }, { - "left": "issue:44038", - "right": "issue:44410", + "left": "issue:44704", + "right": "issue:45103", "accept": false, - "reason": "Qwen3-VL-Moe loading issue and qwen3next missing projections are different model failures." + "reason": "Passing kwargs to cached_file is unrelated to the auto_docstring annotations crash." }, { - "left": "issue:43931", + "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Weight-shape mismatch for Qwen3-VL-30B and missing projections in qwen3next are not the same defect." + "reason": "Both involve Qwen variants, but the concrete bugs and affected components differ." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Video-model torchvision crash and tokenizer-mapping None handling are unrelated." + "reason": "AutoProcessor/torchvision failure is unrelated to tokenizer mapping None handling." }, { - "left": "issue:44261", - "right": "issue:44485", + "left": "issue:43502", + "right": "issue:43992", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm and GLM-5 RoPE implementation are different issues." + "reason": "local_files_only network leakage is unrelated to missing embed_tokens.weight in UMT5Encoder." }, { - "left": "issue:44964", - "right": "issue:45341", + "left": "issue:43097", + "right": "issue:43873", "accept": false, - "reason": "Phi-4 multimodal load failure and a testing_utils bug have no shared underlying cause." + "reason": "Config removal and offloading/quantization behavior are different issues." }, { - "left": "issue:43502", - "right": "issue:43992", + "left": "issue:43824", + "right": "issue:43986", "accept": false, - "reason": "Unexpected API requests with local_files_only and missing embed_tokens.weight on UMT5Encoder are unrelated." + "reason": "Missing Qwen2.5_VLForConditionalGeneration import is unrelated to the torchvision-less AutoProcessor crash." }, { - "left": "issue:43097", - "right": "issue:43873", + "left": "issue:43066", + "right": "issue:43906", "accept": false, - "reason": "Embedding-tying deprecation warning and quantization offloading behavior are different problems." + "reason": "Wrong tokenizer decoder type in v5 is unrelated to the isolated reproduction of issue 38071." } ] }, @@ -12761,9 +12809,9 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-15354-384", "nodes": [ + "issue:36010", "issue:36683", "issue:38617", - "issue:42617", "issue:42757", "issue:42886", "issue:42907", @@ -12779,321 +12827,331 @@ "issue:43475", "issue:43502", "issue:43526", - "issue:43531", "issue:43550", + "issue:43582", "issue:43618", + "issue:43756", "issue:43761", "issue:43824", "issue:43825", "issue:43867", "issue:43901", - "issue:43906", "issue:43937", - "issue:43986", + "issue:43976", "issue:43992", "issue:44038", + "issue:44077", "issue:44220", "issue:44230", "issue:44368", "issue:44380", "issue:44483", "issue:44484", + "issue:44485", "issue:44514", "issue:44655", "issue:44683", "issue:45216", "issue:45290", - "issue:45341", - "issue:45372", "issue:45381", - "issue:45561", - "issue:45588", + "issue:45479", "issue:45753" ], "soft_pairs": [ - "issue:43066|issue:43906", - "issue:43986|issue:45372", - "issue:43825|issue:44655", - "issue:43986|issue:45290", - "issue:45341|issue:45561", - "issue:43986|issue:45341", "issue:43097|issue:43901", - "issue:42617|issue:45341", + "issue:43825|issue:44655", + "issue:43352|issue:44380", "issue:43992|issue:44484", - "issue:45341|issue:45588", "issue:43012|issue:44368", "issue:42886|issue:43381", "issue:43012|issue:43550", "issue:43208|issue:43550", "issue:42907|issue:45753", + "issue:44380|issue:45479", "issue:42907|issue:44655", - "issue:43531|issue:45341", + "issue:43352|issue:43976", + "issue:43064|issue:44380", + "issue:36683|issue:43582", "issue:43208|issue:43526", "issue:44483|issue:44514", + "issue:43618|issue:43761", + "issue:43618|issue:44380", "issue:43012|issue:44683", "issue:42757|issue:43824", - "issue:43618|issue:43761", "issue:44483|issue:45290", - "issue:43064|issue:44380", "issue:44038|issue:44230", "issue:43299|issue:43867", "issue:38617|issue:43502", "issue:44483|issue:45381", + "issue:43756|issue:44485", "issue:43901|issue:44655", + "issue:43299|issue:45753", "issue:36683|issue:43475", - "issue:44220|issue:44655", "issue:43867|issue:45216", "issue:43352|issue:43937", "issue:42994|issue:44655", - "issue:43066|issue:45381" + "issue:44077|issue:44380", + "issue:43066|issue:45381", + "issue:36683|issue:44077", + "issue:36010|issue:42757", + "issue:44220|issue:44655" ], - "prepared_review_unit_hash": "702b219bd17b29be1bf6a7c13d7b0c1a056e0507f0ce983b37cffe09e6e2460c", + "prepared_review_unit_hash": "7bf07544d8ff79df8fbc396bde8ea9516a6119e6ede37d5716db139263476e6c", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 37893, - "estimated_input_tokens": 9474, - "estimated_eval_tokens": 19204 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 37, + "serialized_chars": 37951, + "estimated_input_tokens": 9488, + "estimated_eval_tokens": 19232 }, "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 bug reports. The strongest duplicate candidate is the pair of CLIP v5 regression issues, but the titles alone do not prove they are the same underlying bug, so it should remain unmatched unless the bodies show the same regression path.", - "confidence": 0.58, - "canonical_issue_reason": "issue:43761 is the cleaner representative of the CLIP regression because it states the concrete `CLIPVisionModel.forward` symptom and the v5 hidden-state failure more explicitly; issue:43618 reads like a narrower symptom report.", + "summary": "This cluster is heterogeneous: it mixes unrelated Transformers v5 regressions, import errors, model-specific load/save bugs, and docs/behavior mismatches. There is no real duplicate group here, so all soft links should be rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue is a true canonical duplicate for this mixed cluster; issue 44483 is the least-bad representative because it is a broad, user-facing v5 API regression.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43761 is the best issue to keep as the representative bug report because it is more specific and directly describes the observed regression path, while issue:43618 is less explicit.", + "best_issue_reason": "Issue 44483 is the best cluster representative because it affects a core public endpoint and has broader impact than the more model-specific or doc-only items.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43066", - "right": "issue:43906", - "accept": false, - "reason": "Both are model-loading/tokenization-related, but they point to different failures and different code paths." - }, - { - "left": "issue:43986", - "right": "issue:45372", + "left": "issue:43097", + "right": "issue:43901", "accept": false, - "reason": "Different root causes: missing `torchvision` versus a `ReasoningEffort` import failure." + "reason": "Removed embedding-tie API vs a docs/behavior mismatch for text classification; unrelated problems." }, { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "`pipeline()` error messaging and `save_pretrained()` for Pipeline objects are different functionalities." - }, - { - "left": "issue:43986", - "right": "issue:45290", - "accept": false, - "reason": "Processor loading without torchvision is unrelated to `apply_chat_template` crashing on tool-call messages." - }, - { - "left": "issue:45341", - "right": "issue:45561", - "accept": false, - "reason": "A vague testing-utils bug and an xdist race on `captured_info.txt` are not clearly the same bug." + "reason": "Pipeline translation-task messaging vs pipeline save_pretrained persistence; different code paths and failures." }, { - "left": "issue:43986", - "right": "issue:45341", - "accept": false, - "reason": "Only superficial overlap in testing/utils wording; the symptoms and code paths differ." - }, - { - "left": "issue:43097", - "right": "issue:43901", - "accept": false, - "reason": "A v5 API removal complaint and a docs mismatch for `return_all_scores` are not the same issue." - }, - { - "left": "issue:42617", - "right": "issue:45341", + "left": "issue:43352", + "right": "issue:44380", "accept": false, - "reason": "A 3D parallel script failure and a testing-utils race are unrelated." + "reason": "Nemotron flash-attn support gate vs GPT2 attention scaling under SDPA/FlashAttention; different models and bugs." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Missing checkpoint weights on load and the `max_shard_size` default in save_pretrained are different bugs." - }, - { - "left": "issue:45341", - "right": "issue:45588", - "accept": false, - "reason": "Testing-utils race and a flash-attention crash in sink-less models are different code paths." + "reason": "Missing encoder weight in load path vs default shard size choice in save_pretrained; not the same defect." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "A PyTorch precision warning during compilation is not the same as a `tie_word_embeddings` config warning." + "reason": "PyTorch precision warning during compile vs tie_word_embeddings warning from config handling; only superficially similar." }, { "left": "issue:42886", "right": "issue:43381", "accept": false, - "reason": "Offline cache loading and gradient-checkpointing eval-mode behavior are unrelated." + "reason": "Offline cache loading regression vs gradient checkpointing eval-mode restriction; unrelated behavior." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "One is a warning, the other is a `torch.compile` SDPA failure; different bugs." + "reason": "bfloat16 compile warning vs Bamba torch.compile SDPA failure; different symptom and root cause." }, { "left": "issue:43208", "right": "issue:43550", "accept": false, - "reason": "xLSTM training bugs and Bamba SDPA compile failures do not match." + "reason": "xLSTM training bugs vs Bamba compile/SDPA failure; distinct models and issues." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Saving dequantized models and an infinite-loop image bug are different problems." + "reason": "Saving dequantized models vs image-specific infinite-loop generation; no shared underlying bug." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention scaling ignored in attention backends vs zero-loss sequence classification bug; unrelated subsystems." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Quantized/dequantized model saving is unrelated to Pipeline serialization." + "reason": "Quantized/dequantized model save failure vs saving Pipeline objects; both involve saving but not the same bug." }, { - "left": "issue:43531", - "right": "issue:45341", + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model-specific flash-attn support vs Python 3.9/3.10 compatibility break; different failure domains." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2 optimizer-state mismatch vs GPT2 attention scaling; completely different training vs attention logic." + }, + { + "left": "issue:36683", + "right": "issue:43582", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and testing-utils bugs do not share an underlying failure." + "reason": "Gemma3Config missing vocab_size vs Apple Silicon allocator TypeError; unrelated errors." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training issues and BEiT label-reduction behavior are unrelated." + "reason": "xLSTM training bugs vs BeitImageProcessorFast label reduction bug; different components and symptoms." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Chat-completions API validation and batched `apply_chat_template` crashes are different failures." + "reason": "v1 chat/completions request rejection vs batched chat-template crash; both chat-related but not the same concrete bug." }, { - "left": "issue:43012", - "right": "issue:44683", + "left": "issue:43618", + "right": "issue:43761", "accept": false, - "reason": "A bfloat16 warning is unrelated to compiled flex-attention failures on newer torch versions." + "reason": "CLIP attentions assignment regression vs CLIPVisionModel hidden_states regression; related area but different outputs and code paths." }, { - "left": "issue:42757", - "right": "issue:43824", + "left": "issue:43618", + "right": "issue:44380", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and different packages." + "reason": "CLIP output regression vs GPT2 attention scaling; no meaningful overlap beyond model internals." }, { - "left": "issue:43618", - "right": "issue:43761", + "left": "issue:43012", + "right": "issue:44683", "accept": false, - "reason": "They are likely related CLIP v5 regressions, but the titles alone do not establish they are definitely the same bug." + "reason": "Compile precision warning vs flex_attention failure on newer torch; different failure modes." }, { - "left": "issue:44483", - "right": "issue:45290", + "left": "issue:42757", + "right": "issue:43824", "accept": false, - "reason": "Chat-completions request validation and `apply_chat_template` tool-call crashes are different issues." + "reason": "Missing hub helper import vs missing Transformers model import; both import errors but distinct symbols and fixes." }, { - "left": "issue:43064", - "right": "issue:44380", + "left": "issue:44483", + "right": "issue:45290", "accept": false, - "reason": "Trainer/FSDP optimizer-state corruption is unrelated to GPT2 attention scaling being ignored." + "reason": "Chat/completions endpoint rejection vs assistant tool-call chat-template crash; different APIs and failure points." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "A general Qwen3-VL-MoE breakage and an fp8 support request are not the same bug." + "reason": "Qwen3-VL-Moe loading bug vs fp8 support request; one is a regression, the other is capability support." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Qwen3VL-MoE loading breakage and sorted-state_dict load errors are different issues." + "reason": "Qwen3VL-Moe loading break vs sorted-state_dict load error; both loading-related but not the same issue." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "An import failure and accidental API requests despite `local_files_only=True` are unrelated." + "reason": "Missing configuration_utils import vs local_files_only still making network requests; unrelated import vs I/O behavior." }, { "left": "issue:44483", "right": "issue:45381", "accept": false, - "reason": "Different Qwen2.5-VL failures with different symptoms and code paths." + "reason": "Chat/completions request rejection vs Qwen2.5-VL video vision_position_ids bug; different endpoints and model logic." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch vs GLM-5 RoPE implementation discussion; related topic but not the same concrete bug." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Pipeline docs mismatch and Pipeline save_pretrained failure are different problems." + "reason": "Docs mention stale flag vs inability to save Pipeline objects; unrelated." }, { - "left": "issue:36683", - "right": "issue:43475", + "left": "issue:43299", + "right": "issue:45753", "accept": false, - "reason": "Gemma3 config missing `vocab_size` and SAM3 output missing `fpn_position_embeddings` are unrelated attribute errors." + "reason": "Qwen3VL-Moe loading regression vs infinite-loop generation on a specific image; distinct symptoms." }, { - "left": "issue:44220", - "right": "issue:44655", + "left": "issue:36683", + "right": "issue:43475", "accept": false, - "reason": "Feature-extraction helper behavior and Pipeline serialization do not share an underlying bug." + "reason": "Gemma3 config attribute missing vs SAM3 vision encoder output attribute missing; both missing attributes but different models and fields." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A load error tied to state_dict ordering is not the same as a Qwen3.5 save_pretrained checkpoint regression." + "reason": "General load-model error with sorted state_dict vs Qwen3.5 save_pretrained checkpoint corruption; different save/load sides." }, { "left": "issue:43352", "right": "issue:43937", "accept": false, - "reason": "Nemotron Flash Attention support and GLM-5 generation-config validation are unrelated." + "reason": "Flash-attention support error vs invalid GenerationConfig for GLM-5; unrelated model/config issues." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model saving and Pipeline saving are different serialization paths and bugs." + "reason": "Quantized model save failure vs Pipeline save_pretrained failure; both saving-related but not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "patchtsmixer post_init validation vs GPT2 attention scaling backend bug; unrelated." }, { "left": "issue:43066", "right": "issue:45381", "accept": false, - "reason": "Wrong tokenizer decoder type and Qwen2.5-VL video position-id issues are unrelated." + "reason": "Wrong decoder type in v5 tokenizer handling vs video vision_position_ids bug; different components." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Gemma3 config missing vocab_size vs patchtsmixer post_init policy; no shared underlying issue." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Missing GenerationMixin import vs missing is_offline_mode import; both import errors but different modules and fixes." + }, + { + "left": "issue:44220", + "right": "issue:44655", + "accept": false, + "reason": "Audio fbank extraction bug vs Pipeline save_pretrained bug; unrelated code paths." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "Most of the rejects are reasonably conservative, but the one accepted edge is a bit too strong given only the packet titles: issue:43618 (\u2018CLIPOutput attentions is no longer assigned\u2019) and issue:43761 (\u2018CLIPVisionModel.forward returns hidden_states=None even when output_hidden_states=True\u2019) may be related, but the output does not establish they are definitely the same bug. I\u2019d want that verdict softened to \u2018likely related\u2019 or left unmatched unless the bodies show the same regression path. The cluster summary also slightly overcommits by calling this the \u2018only credible duplicate edge\u2019 instead of \u2018the strongest possible duplicate candidate.\u2019" - }, - "evaluator_used": true, - "retried": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -13113,28 +13171,26 @@ "nodes": [ "issue:28282", "issue:33453", - "issue:36010", "issue:36296", "issue:36683", "issue:41669", - "issue:42757", "issue:42890", + "issue:43010", "issue:43064", - "issue:43097", - "issue:43299", "issue:43352", - "issue:43441", + "issue:43502", "issue:43525", "issue:43575", "issue:43618", - "issue:43756", + "issue:43723", "issue:43867", - "issue:43881", - "issue:43906", + "issue:43931", "issue:43976", "issue:43992", + "issue:44062", "issue:44077", - "issue:44117", + "issue:44112", + "issue:44188", "issue:44242", "issue:44246", "issue:44263", @@ -13142,306 +13198,297 @@ "issue:44360", "issue:44380", "issue:44484", - "issue:44485", "issue:44492", - "issue:44655", + "issue:44556", "issue:44855", "issue:44857", "issue:44938", - "issue:44977", + "issue:44964", + "issue:45071", "issue:45216", + "issue:45278", "issue:45310", "issue:45357", "issue:45375", + "issue:45381", "issue:45440", "issue:45478", - "issue:45479", "issue:45561", "issue:45588", - "issue:45753" + "issue:45593", + "issue:45600" ], "soft_pairs": [ - "issue:36010|issue:42757", "issue:44263|issue:45440", - "issue:43756|issue:44485", + "issue:43976|issue:44938", "issue:36683|issue:43525", + "issue:43352|issue:45593", + "issue:43064|issue:44492", + "issue:44188|issue:44380", "issue:43618|issue:45216", - "issue:43906|issue:45753", "issue:36683|issue:44242", - "issue:36683|issue:43881", - "issue:43299|issue:45753", - "issue:43352|issue:43976", + "issue:43352|issue:43931", + "issue:43010|issue:45278", "issue:41669|issue:44246", - "issue:43441|issue:44380", "issue:43992|issue:45310", - "issue:43352|issue:44380", - "issue:44380|issue:45479", "issue:43867|issue:45375", - "issue:43064|issue:44492", - "issue:43618|issue:44380", - "issue:43352|issue:45478", + "issue:36683|issue:43064", "issue:44355|issue:44855", - "issue:36683|issue:44117", "issue:42890|issue:45561", - "issue:43097|issue:43352", + "issue:44938|issue:45071", "issue:43867|issue:45357", - "issue:43976|issue:44938", "issue:43992|issue:45478", + "issue:43010|issue:43976", + "issue:44938|issue:45381", + "issue:43618|issue:44077", "issue:44360|issue:45440", - "issue:43064|issue:44655", "issue:36296|issue:43575", - "issue:44077|issue:44380", "issue:43618|issue:44857", - "issue:43867|issue:44484", "issue:43618|issue:45588", + "issue:43867|issue:44484", "issue:28282|issue:33453", - "issue:43618|issue:44977" + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43723|issue:44556" ], - "prepared_review_unit_hash": "6e9ce916a630d98b7cc23a92a8105b929410b3cc234f90d5f37625c08c679928", + "prepared_review_unit_hash": "4d35b1c31327a139888132bf4056ee5bb95888904a531de3476e081144c9919b", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 37796, - "estimated_input_tokens": 9449, - "estimated_eval_tokens": 19154 + "soft_pair_count": 33, + "serialized_chars": 37242, + "estimated_input_tokens": 9311, + "estimated_eval_tokens": 18878 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely noisy: several issues share broad keywords like configs, loading, flash-attention, or import/runtime failures, but they do not look like the same underlying bug or change. No pull requests are present.", - "confidence": 0.59, - "canonical_issue_reason": "issue:36683 is the best representative of the small config-attribute/load-path subtheme and is the most connected issue in the set, but the overall cluster is still heterogeneous.", + "summary": "The cluster is mostly a grab bag of unrelated model, config, training, and test failures. The only clear duplicate-like soft edge is import-time slowness around transformers imports; most other pairs are too model- or code-path-specific to merge.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45278 is the broadest umbrella in the set: it describes upgrade-related import/load regressions affecting multiple models, making it the best anchor for the import-regression subset of this cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:36683 is the strongest single issue to keep as the cluster representative because it is active and central to the few genuinely related config-attribute reports.", + "best_issue_reason": "issue:45278 is the best issue to keep as the cluster representative because it is the most general user-facing regression report and can subsume nearby import/loading failures better than the narrower model-specific bugs.", "best_pr_reason": null, - "soft_edge_verdicts": [ - { - "left": "issue:36010", - "right": "issue:42757", - "accept": false, - "reason": "Both are import errors, but they involve different missing symbols from different packages and different failure modes." - }, - { - "left": "issue:44263", - "right": "issue:45440", - "accept": false, - "reason": "These are unrelated model-path bugs: one is a torch.split/indexer issue, the other is a DeepSeekV3 implementation divergence." - }, - { - "left": "issue:43756", - "right": "issue:44485", - "accept": false, - "reason": "Both mention RoPE, but they concern different models and distinct implementation problems." - }, - { - "left": "issue:36683", - "right": "issue:43525", - "accept": false, - "reason": "Both are missing-config-attribute reports, but on different models and different fields, so they are not the same bug." - }, - { - "left": "issue:43618", - "right": "issue:45216", - "accept": false, - "reason": "One is about CLIPOutput attentions no longer being assigned; the other is a Qwen3.5 save_pretrained regression." - }, + "soft_edge_verdicts": [ { - "left": "issue:43906", - "right": "issue:45753", + "left": "issue:44263", + "right": "issue:45440", "accept": false, - "reason": "The first is an isolated reproduction of another issue; the second is a specific-image infinite loop in Qwen3_5, which is a different symptom." + "reason": "Different problems in different components: DSA indexer behavior vs DeepSeekV3 MoE divergence." }, { - "left": "issue:36683", - "right": "issue:44242", + "left": "issue:43976", + "right": "issue:44938", "accept": false, - "reason": "A missing Gemma3Config attribute is unrelated to load-balancing loss handling when output_router_logits=False." + "reason": "Both mention Python-version issues, but one is a v5.1.0 compatibility problem and the other is a Python 3.14 load failure; no same code path." }, { "left": "issue:36683", - "right": "issue:43881", + "right": "issue:43525", "accept": false, - "reason": "Different models and different loading failures; only the generic \"loading failed\" theme overlaps." + "reason": "Different missing config attributes on different models; same general symptom, not the same bug." }, { - "left": "issue:43299", - "right": "issue:45753", + "left": "issue:43352", + "right": "issue:45593", "accept": false, - "reason": "Both involve Qwen variants, but one is a v5.dev0 loading break and the other is an inference infinite-loop bug." + "reason": "Unrelated model-specific bugs: FlashAttention support gating vs D-FINE auxiliary loss behavior." }, { - "left": "issue:43352", - "right": "issue:43976", + "left": "issue:43064", + "right": "issue:44492", "accept": false, - "reason": "A FlashAttention support limitation is unrelated to the Python-version compatibility failure." + "reason": "Training/FSDP optimizer-state bug is unrelated to a cache-strategy typo." }, { - "left": "issue:41669", - "right": "issue:44246", + "left": "issue:44188", + "right": "issue:44380", "accept": false, - "reason": "Import-time performance and occasional import slowness are too broad here; these do not describe the same code path or bug." + "reason": "Both are attention-related, but they affect different mechanisms and code paths." }, { - "left": "issue:43441", - "right": "issue:44380", + "left": "issue:43618", + "right": "issue:45216", "accept": false, - "reason": "Both touch attention backends, but they are different model-specific bugs with different root causes." + "reason": "CLIP attentions assignment regression and Qwen3.5 save_pretrained checkpoint regression are different issues." }, { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:36683", + "right": "issue:44242", "accept": false, - "reason": "Missing embed_tokens.weight during UMT5Encoder loading is unrelated to the Qwen3.5 MoE from_pretrained regression." + "reason": "Different failures: missing Gemma3Config field vs MoE load-balancing loss omission." }, { "left": "issue:43352", - "right": "issue:44380", + "right": "issue:43931", "accept": false, - "reason": "FlashAttention support in Nemotron and GPT2 attention scaling being ignored are separate backend/model issues." + "reason": "Different models and failures: FA2 support check vs Qwen3-VL weight-shape mismatch." }, { - "left": "issue:44380", - "right": "issue:45479", + "left": "issue:43010", + "right": "issue:45278", "accept": false, - "reason": "Attention scaling behavior and a degenerate zero-loss classification bug are unrelated." + "reason": "No overlap in code path: no_grad cache update vs broad import errors after upgrading." }, { - "left": "issue:43867", - "right": "issue:45375", - "accept": false, - "reason": "A sorted-state_dict loading error is not the same as a missing deepstack_visual_indexes config field." + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Both describe import-time slowdown in transformers; 41669\u2019s import * regression plausibly explains the intermittent slow import reported in 44246." }, { - "left": "issue:43064", - "right": "issue:44492", + "left": "issue:43992", + "right": "issue:45310", "accept": false, - "reason": "Optimizer-state corruption during FSDP2/PEFT training is unrelated to a cache-strategy typo." + "reason": "Both are from_pretrained failures, but they affect different models and different loading paths." }, { - "left": "issue:43618", - "right": "issue:44380", + "left": "issue:43867", + "right": "issue:45375", "accept": false, - "reason": "These affect different models and different attention-related behaviors." + "reason": "Generic state_dict ordering error vs missing Qwen3.5 vision config field; not the same defect." }, { - "left": "issue:43352", - "right": "issue:45478", + "left": "issue:36683", + "right": "issue:43064", "accept": false, - "reason": "Both are Qwen/Nemotron-era model bugs, but the concrete failure modes are unrelated." + "reason": "Completely different areas: config attribute access vs distributed training optimizer state handling." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Compiled Python file errors and a Python 3.13 torch.jit.script parsing issue are distinct import/runtime problems." - }, - { - "left": "issue:36683", - "right": "issue:44117", - "accept": false, - "reason": "A missing model config attribute is not the same as TOKENIZER_MAPPING_NAMES returning None." + "reason": "Different import/compile failures with different root causes; one is compiled Python files, the other is torch.jit parsing a commented decorator block." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Both are test flakiness, but one is missing set_seed and the other is an xdist race on a captured file." + "reason": "Both hit tests, but one is seed setup flakiness and the other is xdist file races; different failures." }, { - "left": "issue:43097", - "right": "issue:43352", + "left": "issue:44938", + "right": "issue:45071", "accept": false, - "reason": "A removed config option and a FlashAttention support issue are unrelated." + "reason": "Python 3.14 load failure and PretrainedConfig type-checking breakage are unrelated regressions." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "These both concern saving/loading, but one is a state_dict ordering problem and the other is incorrect visual encoder keys." + "reason": "Both touch Qwen3.5 loading/saving, but one is a load error and the other is incorrect save keys; not the same concrete bug." }, { - "left": "issue:43976", - "right": "issue:44938", + "left": "issue:43992", + "right": "issue:45478", "accept": false, - "reason": "Both are Python-version compatibility reports, but they target different versions and likely different breakages." + "reason": "Different model-loading errors on different architectures and code paths." }, { - "left": "issue:43992", - "right": "issue:45478", + "left": "issue:43010", + "right": "issue:43976", "accept": false, - "reason": "A missing embedding weight during load is unrelated to a Qwen3.5 MoE from_pretrained error." + "reason": "No relation between cache update behavior and Python-version compatibility." }, { - "left": "issue:44360", - "right": "issue:45440", + "left": "issue:44938", + "right": "issue:45381", "accept": false, - "reason": "A missing ReLU in the DSA indexer is unrelated to DeepSeekV3 implementation divergence." + "reason": "Python 3.14 import failure vs Qwen2.5-VL video position IDs are unrelated." }, { - "left": "issue:43064", - "right": "issue:44655", + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different model issues: CLIP output fields vs patchtsmixer post_init validation." + }, + { + "left": "issue:44360", + "right": "issue:45440", "accept": false, - "reason": "Training optimizer-state bugs and saving Pipeline objects are different code paths." + "reason": "Unrelated algorithmic bugs in different model components." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Tensor-parallel training bugs and TP-related OOM during model load are not the same defect." + "reason": "Both mention tensor parallelism, but one is a training bug and the other is an OOM on loading a specific model; too different to merge." }, { - "left": "issue:44077", - "right": "issue:44380", + "left": "issue:43618", + "right": "issue:44857", "accept": false, - "reason": "An optional post_init restriction in patchtsmixer has nothing to do with GPT2 attention scaling." + "reason": "CLIP attentions regression and float16 AMP crash are unrelated." }, { "left": "issue:43618", - "right": "issue:44857", + "right": "issue:45588", "accept": false, - "reason": "Missing CLIPOutput attentions and a float16 AMP CUDA crash are unrelated bugs." + "reason": "Different subsystems: CLIP outputs vs flash_attention sink-less crash." }, { "left": "issue:43867", "right": "issue:44484", "accept": false, - "reason": "State_dict sorting failures and the max_shard_size default question are not the same issue." + "reason": "Generic load failure vs save_pretrained shard-size question; not the same bug." }, { - "left": "issue:43618", - "right": "issue:45588", + "left": "issue:28282", + "right": "issue:33453", "accept": false, - "reason": "The CLIPOutput attentions regression and the s_aux=None flash_attention crash affect different code paths." + "reason": "Both are loading-related, but one is a missing-PyTorch ImportError and the other is a tokenizer-loading regression." }, { - "left": "issue:28282", - "right": "issue:33453", + "left": "issue:42890", + "right": "issue:44964", "accept": false, - "reason": "AutoModel missing PyTorch and tokenizer-loading regression are completely different failures." + "reason": "Test seeding flakiness is unrelated to a multimodal model loading failure." }, { - "left": "issue:43618", - "right": "issue:44977", + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Different CI/test issues with different failure modes." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Unrelated test flake vs auto_mappings CI breakage." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Local-files-only API leakage and AddedToken duplicate-arg TypeError are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44556", "accept": false, - "reason": "These are unrelated: one concerns a missing attention field, the other is a Qwen3.5 flash-attention generation bug." + "reason": "Both are v5 loading regressions, but one is tokenizer loading and the other is checkpoint reload compatibility; not the same code path." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The representative choice and summary are grounded in the packet, and the lone accepted soft edge (41669 \u2194 44246) is a conservative, well-supported import-slowdown match. The other refusals stay appropriately cautious about differing models or code paths." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -13465,25 +13512,27 @@ "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:43723", + "issue:43673", "issue:43824", "issue:43827", "issue:43874", "issue:43901", - "issue:43976", - "issue:44062", "issue:44075", - "issue:44112", "issue:44222", "issue:44230", + "issue:44261", "issue:44315", "issue:44351", "issue:44355", @@ -13491,329 +13540,343 @@ "issue:44368", "issue:44387", "issue:44485", + "issue:44492", "issue:44556", "issue:44655", "issue:44749", - "issue:44938", + "issue:44908", "issue:44961", - "issue:44964", "issue:45071", "issue:45125", - "issue:45278", "issue:45341", "issue:45375", "issue:45468", - "issue:45478", - "issue:45600", - "issue:45656" + "issue:45478" ], "soft_pairs": [ - "issue:43901|issue:44938", - "issue:42890|issue:44964", - "issue:44655|issue:44938", - "issue:42890|issue:44112", - "issue:42890|issue:45600", - "issue:43502|issue:44062", "issue:42890|issue:45341", "issue:43010|issue:44075", + "issue:44387|issue:44556", "issue:42548|issue:44351", "issue:44222|issue:45125", + "issue:43010|issue:43065", + "issue:43010|issue:43232", "issue:43502|issue:44485", "issue:44355|issue:44749", "issue:45125|issue:45478", "issue:30990|issue:31795", "issue:45125|issue:45375", - "issue:43010|issue:45278", - "issue:44387|issue:44556", "issue:30333|issue:33453", - "issue:43723|issue:44556", "issue:43010|issue:44368", + "issue:43299|issue:44556", "issue:43502|issue:44222", "issue:44315|issue:44360", "issue:43502|issue:44961", - "issue:44355|issue:44655", "issue:42831|issue:43874", + "issue:44355|issue:44655", "issue:30333|issue:36296", - "issue:44355|issue:45071", "issue:44222|issue:44230", + "issue:44355|issue:45071", "issue:43827|issue:44961", - "issue:43010|issue:45656", "issue:33453|issue:34689", - "issue:43299|issue:44556", "issue:42757|issue:43576", "issue:33453|issue:36296", "issue:43576|issue:45071", "issue:43576|issue:43824", - "issue:43010|issue:43976", + "issue:43673|issue:44492", "issue:44961|issue:45468", - "issue:31795|issue:38617" + "issue:31795|issue:38617", + "issue:41628|issue:42548", + "issue:38617|issue:43576", + "issue:43576|issue:44351", + "issue:44908|issue:44961", + "issue:44485|issue:44961", + "issue:36296|issue:43901", + "issue:41628|issue:44355", + "issue:36296|issue:43526", + "issue:44261|issue:44360" ], - "prepared_review_unit_hash": "0dfd9c2b3bc8ef5417dc950ce3be516006b9d6124d60a8d03096a8fd7b799427", + "prepared_review_unit_hash": "17bce9ea23288d16d79d8dc5be12e9af4c4da3661ff57b5e3cb42852471cd966", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 39, - "serialized_chars": 38861, - "estimated_input_tokens": 9716, - "estimated_eval_tokens": 19688 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 41, + "serialized_chars": 38887, + "estimated_input_tokens": 9722, + "estimated_eval_tokens": 19700 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed bag of unrelated bug reports and documentation regressions around Transformers v5; only one soft edge looks like a plausible duplicate pair. I\u2019d use the broad v5 import-regression report as the representative issue, but there is no PR in this cluster.", - "confidence": 0.42, - "canonical_issue_reason": "issue:45278 is the broadest umbrella report here ('Many import errors after update from 4.57.0 to 5.5.0') and best represents the general v5 import/load regression theme, even though the cluster is not a set of true duplicates.", + "summary": "The set is mostly a loose bundle of unrelated Transformers v5 regressions, import errors, model-loading issues, and docs/test bugs. I do not see any soft pair that clearly refers to the same underlying bug or mergeable change.", + "confidence": 0.9, + "canonical_issue_reason": "issue:43576 is the broadest representative of the recurring v5 breakage theme and has the strongest discussion/inbound signal among the set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most general and representative issue among the mostly unrelated reports; it captures the shared version-upgrade breakage theme better than the narrower model-specific or doc-only items.", + "best_issue_reason": "issue:43576 is the best single representative issue here because it is broad and central, while the others are narrower, model-specific, or unrelated.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43901", - "right": "issue:44938", - "accept": false, - "reason": "Docs about a removed pipeline argument and a Python 3.14 import/load failure are different problems." - }, { "left": "issue:42890", - "right": "issue:44964", - "accept": false, - "reason": "Seed-setting flakiness in a SAM HQ test is unrelated to Phi-4 multimodal model loading." - }, - { - "left": "issue:44655", - "right": "issue:44938", + "right": "issue:45341", "accept": false, - "reason": "Saving Pipeline objects and Python 3.14 import failure are not the same code-path bug." + "reason": "Different scope: flaky SamHQ integration tests vs a generic testing_utils bug." }, { - "left": "issue:42890", - "right": "issue:44112", + "left": "issue:43010", + "right": "issue:44075", "accept": false, - "reason": "Test nondeterminism from missing set_seed() is separate from a GraniteSpeech CI device-override failure." + "reason": "Unrelated code paths: cache/layer no_grad behavior vs SGD optimizer args." }, { - "left": "issue:42890", - "right": "issue:45600", + "left": "issue:44387", + "right": "issue:44556", "accept": false, - "reason": "A flaky integration test is unrelated to stale auto_mappings references causing CI breakage." + "reason": "Both are v5 regressions, but they describe different failures: int4 memory growth vs checkpoint reload compatibility." }, { - "left": "issue:43502", - "right": "issue:44062", + "left": "issue:42548", + "right": "issue:44351", "accept": false, - "reason": "local_files_only leaking network requests and AddedToken keyword duplication are distinct loader/tokenizer issues." + "reason": "Both are import errors, but for different exported symbols and likely different fixes." }, { - "left": "issue:42890", - "right": "issue:45341", + "left": "issue:44222", + "right": "issue:45125", "accept": false, - "reason": "Both are test-related, but they describe different failures and different fixes." + "reason": "Different problems: FP8 save_pretrained behavior vs tensor-parallel plan missing on Qwen3.5 MoE." }, { "left": "issue:43010", - "right": "issue:44075", - "accept": false, - "reason": "Cache update/no_grad behavior is unrelated to SGD optimizer arguments being ignored." - }, - { - "left": "issue:42548", - "right": "issue:44351", + "right": "issue:43065", "accept": false, - "reason": "Both are import errors, but for different missing symbols and likely different fixes." + "reason": "Different models and bugs: cache update semantics vs Sam3PixelDecoder Conv2d stub." }, { - "left": "issue:44222", - "right": "issue:45125", + "left": "issue:43010", + "right": "issue:43232", "accept": false, - "reason": "FP8 save_pretrained on MoE models is not the same as missing _tp_plan for tensor parallelism." + "reason": "Different generation-related issues with separate code paths and symptoms." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "local_files_only network leakage and GLM-5 RoPE discussion are unrelated." + "reason": "Unrelated: offline/local-files-only networking behavior vs GLM-5 RoPE implementation." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled Python file errors are unrelated to a slowdown after upgrading Transformers." + "reason": "Unrelated: compiled Python file errors vs performance slowdown after upgrade." }, { "left": "issue:45125", "right": "issue:45478", - "accept": true, - "reason": "Both point to the same Qwen3.5 MoE loading failure path in v5, plausibly caused by missing tensor-parallel support." + "accept": false, + "reason": "Related model family, but one is missing _tp_plan and the other is a from_pretrained failure; not clearly the same bug." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "SentenceTransformer loading issues and model.forward documentation confusion are unrelated." + "reason": "Different concerns: loading hang vs documentation wording." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Missing _tp_plan and a missing deepstack_visual_indexes config field are different model-configuration bugs." + "reason": "Different Qwen3.5 MoE config/load issues involving separate missing fields." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "Different failures: MLFlow reporting on job failure vs tokenizer loading regression." }, { "left": "issue:43010", - "right": "issue:45278", + "right": "issue:44368", "accept": false, - "reason": "Cache no_grad/update behavior is unrelated to broad import errors after upgrading to v5.5.0." + "reason": "Different issues: no_grad cache updates vs tied-embeddings warning in LoRA fine-tuning." }, { - "left": "issue:44387", + "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Int4 quantization memory regression and checkpoint reload failure after upgrade are different bugs." + "reason": "Both are version-related loading regressions, but for different models and failure modes." }, { - "left": "issue:30333", - "right": "issue:33453", + "left": "issue:43502", + "right": "issue:44222", "accept": false, - "reason": "MLflow reporting on failed training jobs and tokenizer loading regression are unrelated." + "reason": "Different bugs: local_files_only network leakage vs FP8 moe save_pretrained." }, { - "left": "issue:43723", - "right": "issue:44556", + "left": "issue:44315", + "right": "issue:44360", "accept": false, - "reason": "Tokenizer loading in v5 and reloading old checkpoints after upgrade are related broadly, but not the same concrete bug." + "reason": "Completely different subsystems: Liger kernel application vs DSA indexer activation." }, { - "left": "issue:43010", - "right": "issue:44368", + "left": "issue:43502", + "right": "issue:44961", "accept": false, - "reason": "Cache update/no_grad behavior is unrelated to a tie_word_embeddings warning during Qwen LoRA fine-tuning." + "reason": "No overlap beyond being issues; networking bug vs unrelated raccoon title." }, { - "left": "issue:43502", - "right": "issue:44222", + "left": "issue:42831", + "right": "issue:43874", "accept": false, - "reason": "local_files_only network calls and FP8 MoE save_pretrained are different issues." + "reason": "Different model behaviors: FP8 accuracy vs multimodal token counting error." }, { - "left": "issue:44315", - "right": "issue:44360", + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Different areas: compiled-file errors vs pipeline save_pretrained support." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "Unrelated training concerns: MLFlow reporting vs tensor-parallel training bug." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Related to FP8, but one is save_pretrained and the other is inference support; not the same bug." + }, + { + "left": "issue:44355", + "right": "issue:45071", "accept": false, - "reason": "Liger Kernel not applying with model_init is unrelated to the DSA indexer ReLU discussion." + "reason": "Different failure types: compiled file errors vs PretrainedConfig type-checking regression." }, { - "left": "issue:43502", + "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "local_files_only network leakage and a one-word issue title do not describe the same bug." + "reason": "Docs/pipeline removal issue vs unrelated issue title." }, { - "left": "issue:44355", - "right": "issue:44655", + "left": "issue:33453", + "right": "issue:34689", "accept": false, - "reason": "Compiled Python file errors and inability to save Pipeline objects are different failure modes." + "reason": "Both are loading regressions, but they target different models and symptoms." }, { - "left": "issue:42831", - "right": "issue:43874", + "left": "issue:42757", + "right": "issue:43576", "accept": false, - "reason": "FP8 accuracy regression and missing image-processor method are unrelated." + "reason": "Different symptoms: missing hub import vs broken env command." }, { - "left": "issue:30333", + "left": "issue:33453", "right": "issue:36296", "accept": false, - "reason": "MLflow reporting failure and tensor-parallel training bug are different training-time issues." + "reason": "Different problems: tokenizer loading regression vs tensor-parallel training bug." }, { - "left": "issue:44355", + "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "Compiled file loading issues and PretrainedConfig type-checking breakage are unrelated." + "reason": "Different code paths: env CLI regression vs config type-checking." }, { - "left": "issue:44222", - "right": "issue:44230", + "left": "issue:43576", + "right": "issue:43824", "accept": false, - "reason": "Both involve FP8 MoE, but one is save_pretrained and the other is inference support; not the same concrete bug." + "reason": "Different bugs: env command broken vs missing Qwen2.5-VL export." }, { - "left": "issue:43827", - "right": "issue:44961", + "left": "issue:43673", + "right": "issue:44492", "accept": false, - "reason": "Pipeline docs references and a one-word title are unrelated." + "reason": "Cache missing in generation vs a typo in cache strategies." }, { - "left": "issue:43010", - "right": "issue:45656", + "left": "issue:44961", + "right": "issue:45468", "accept": false, - "reason": "Cache update decoration and optimizer double-step behavior under deepspeed are different code paths." + "reason": "No apparent relation; the titles point to unrelated issues." }, { - "left": "issue:33453", - "right": "issue:34689", + "left": "issue:31795", + "right": "issue:38617", "accept": false, - "reason": "Tokenizer loading regression and Llama 3.2 Vision model loading failure are separate model-loading bugs." + "reason": "Documentation clarification vs import error in configuration_utils." }, { - "left": "issue:43299", - "right": "issue:44556", + "left": "issue:41628", + "right": "issue:42548", "accept": false, - "reason": "Qwen3VL MoE loading failure and old-checkpoint reload failure are both upgrade-related, but not the same underlying bug." + "reason": "Both are import errors, but for different symbols and likely different export fixes." }, { - "left": "issue:42757", + "left": "issue:38617", "right": "issue:43576", "accept": false, - "reason": "Missing huggingface_hub export and a broken transformers env command are unrelated." + "reason": "Different layers: missing helper import vs env CLI breakage." }, { "left": "issue:43576", - "right": "issue:45071", + "right": "issue:44351", "accept": false, - "reason": "env command breakage and PretrainedConfig type checking are unrelated." + "reason": "Both mention import-related breakage, but the affected symbols and failure contexts differ." }, { - "left": "issue:43576", - "right": "issue:43824", + "left": "issue:44908", + "right": "issue:44961", "accept": false, - "reason": "Broken env command and missing Qwen2_5_VLForConditionalGeneration import are different regressions." + "reason": "Scheduler kwargs bug vs unrelated issue title." }, { - "left": "issue:43010", - "right": "issue:43976", + "left": "issue:44485", + "right": "issue:44961", "accept": false, - "reason": "Cache update behavior and Python 3.9 compatibility failure are unrelated." + "reason": "Unrelated model implementation discussion vs unrelated issue title." }, { - "left": "issue:44961", - "right": "issue:45468", + "left": "issue:36296", + "right": "issue:43901", "accept": false, - "reason": "Uninformative title and Gemma-4 audio positional encoding bug are unrelated." + "reason": "Tensor-parallel training bug vs documentation mismatch for text classification pipeline." }, { - "left": "issue:31795", - "right": "issue:38617", + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Import export regression vs compiled-file execution errors." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Different model/training bug vs BeitImageProcessorFast label handling bug." + }, + { + "left": "issue:44261", + "right": "issue:44360", "accept": false, - "reason": "Documentation confusion in model.forward and a missing import symbol are unrelated." + "reason": "Precision/config discussion for MLA vs a DSA indexer activation issue." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "Grounded and conservative overall. The representative issue choice is supported by the packet, the summary correctly notes this is a mixed cluster rather than true duplicates, and the one accepted soft-edge pair looks plausibly mergeable as the same Qwen3.5 MoE loading regression." - }, - "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 }, @@ -13832,63 +13895,54 @@ "nodes": [ "issue:31795", "issue:33453", - "issue:36296", - "issue:38617", - "issue:41628", + "issue:34689", + "issue:37428", "issue:41669", - "issue:42548", + "issue:42371", "issue:42890", - "issue:43010", - "issue:43232", "issue:43352", "issue:43408", - "issue:43526", + "issue:43502", + "issue:43519", "issue:43576", "issue:43673", + "issue:43704", "issue:43827", - "issue:43901", "issue:43976", "issue:44038", "issue:44077", "issue:44246", - "issue:44261", + "issue:44263", "issue:44273", "issue:44295", "issue:44297", "issue:44315", "issue:44351", - "issue:44355", "issue:44360", - "issue:44373", "issue:44393", + "issue:44485", "issue:44492", "issue:44556", "issue:44623", + "issue:44655", "issue:44704", "issue:44829", "issue:44861", "issue:44908", - "issue:44961", + "issue:44945", "issue:45003", "issue:45071", "issue:45092", "issue:45125", "issue:45230", + "issue:45310", "issue:45341", + "issue:45468", "issue:45478", "issue:45542", "issue:45753" ], "soft_pairs": [ - "issue:41628|issue:42548", - "issue:38617|issue:43576", - "issue:43576|issue:44351", - "issue:44908|issue:44961", - "issue:43010|issue:43232", - "issue:36296|issue:43901", - "issue:41628|issue:44355", - "issue:36296|issue:43526", - "issue:44261|issue:44360", "issue:44273|issue:45003", "issue:44623|issue:45230", "issue:44273|issue:45542", @@ -13898,7 +13952,7 @@ "issue:44393|issue:45071", "issue:44038|issue:45125", "issue:44556|issue:45092", - "issue:44373|issue:44961", + "issue:42371|issue:44492", "issue:44704|issue:44908", "issue:42890|issue:44908", "issue:44246|issue:44273", @@ -13907,290 +13961,359 @@ "issue:43352|issue:44393", "issue:43576|issue:44908", "issue:44297|issue:45230", - "issue:43673|issue:44492", "issue:44908|issue:45071", "issue:44295|issue:45230", "issue:41669|issue:44351", "issue:41669|issue:45341", + "issue:44492|issue:44945", "issue:41669|issue:44492", "issue:44393|issue:45478", "issue:44315|issue:44829", "issue:44077|issue:44861", "issue:43827|issue:44908", "issue:31795|issue:33453", - "issue:41669|issue:44908" + "issue:41669|issue:44908", + "issue:43576|issue:44273", + "issue:44655|issue:44861", + "issue:44360|issue:45468", + "issue:45230|issue:45310", + "issue:44273|issue:44908", + "issue:43519|issue:44485", + "issue:41669|issue:43704", + "issue:43519|issue:44908", + "issue:31795|issue:34689", + "issue:31795|issue:37428", + "issue:44315|issue:45092", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:42371|issue:43704", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:44263|issue:45468" ], - "prepared_review_unit_hash": "10a6187d61c96c91a4cd9b4ea586188c5acc338f5507daced519f4dd45865222", + "prepared_review_unit_hash": "a5e04012ed0db84461f5b5d5bbce1b389117dc2521d7f28e12ce30945b378553", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 39, - "serialized_chars": 38746, - "estimated_input_tokens": 9687, - "estimated_eval_tokens": 19630 + "soft_pair_count": 48, + "serialized_chars": 41424, + "estimated_input_tokens": 10356, + "estimated_eval_tokens": 20968 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of separate Transformers issues spanning imports, lazy loading, generation/cache behavior, docs, and model-specific regressions. No soft pair is strongly supported as the same underlying bug; the lazy-loading pair in particular lacks enough evidence to merge confidently.", - "confidence": 0.62, - "canonical_issue_reason": "issue:44273 is the broadest open report in the cluster and the safest representative for the lazy-loading subgroup, but the evidence is not strong enough to treat it as the same bug as issue:45003.", + "summary": "This is a loose Transformers issue cluster with repeated loading/import/model-initialization themes, but the items are mostly distinct regressions, docs, or model-specific bugs rather than true duplicates.", + "confidence": 0.52, + "canonical_issue_reason": "issue:44556 is the clearest umbrella-style loading regression: checkpoint reload failures after upgrading to v5.2/v5.3. It is broad enough to represent the cluster\u2019s general model-loading breakage theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44273 is the most central open issue and the best cluster representative by scope, even though the duplicate linkage to issue:45003 is unproven.", + "best_issue_reason": "issue:45092 is the most actionable and specific report: it describes a concrete incompatibility for old InternVL2 remote-code checkpoints during v5 meta initialization, which makes it the strongest standalone issue in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:41628", - "right": "issue:42548", + "left": "issue:44273", + "right": "issue:45003", "accept": false, - "reason": "Both are import failures, but they involve different missing symbols and different export paths; not the same concrete bug." + "reason": "Both touch core runtime behavior, but one is lazy loading and the other is unsafe sys.modules access; different bugs and fixes." }, { - "left": "issue:38617", - "right": "issue:43576", + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "44623 is a concrete processor.save_pretrained bug; 45230 is a generic bug report with no matching failure mode." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading regressions and a TensorFlow backend error are unrelated issues." + }, + { + "left": "issue:44393", + "right": "issue:45753", "accept": false, - "reason": "Missing internal import vs broken env command; different features and failure modes." + "reason": "Both mention Qwen3-VL/Qwen3.5 image behavior, but one is bbox output hallucination and the other is an infinite loop on a specific image." }, { "left": "issue:43576", - "right": "issue:44351", + "right": "issue:44556", "accept": false, - "reason": "An env command regression and a missing top-level cache export are unrelated." + "reason": "One is the v5 env command, the other is checkpoint reload compatibility; same era, different functionality and root cause." }, { - "left": "issue:44908", - "right": "issue:44961", + "left": "issue:41669", + "right": "issue:43673", "accept": false, - "reason": "Scheduler kwargs bug and an unrelated placeholder issue; no shared code-path." + "reason": "Import-star cleanup causing import slowdown is not the same bug as missing cache state during chunked prefill." }, { - "left": "issue:43010", - "right": "issue:43232", + "left": "issue:44393", + "right": "issue:45071", "accept": false, - "reason": "Both touch generation/cache code, but they describe different defects and different call sites." + "reason": "Qwen3-VL bbox output issues are unrelated to PretrainedConfig type-checking regressions." }, { - "left": "issue:36296", - "right": "issue:43901", + "left": "issue:44038", + "right": "issue:45125", "accept": false, - "reason": "Tensor-parallel training bug vs documentation mismatch; not the same issue." + "reason": "Both are Qwen3.5/Moe related, but one is a model support bug and the other is a missing tensor-parallel plan." }, { - "left": "issue:41628", - "right": "issue:44355", + "left": "issue:44556", + "right": "issue:45092", "accept": false, - "reason": "Top-level import failure for a symbol vs compiled-file execution error; different problems." + "reason": "Both concern v5 loading breakage, but they describe different incompatibilities and code paths; not the same concrete bug." }, { - "left": "issue:36296", - "right": "issue:43526", + "left": "issue:42371", + "right": "issue:44492", "accept": false, - "reason": "Tensor parallel training bug and a BeitImageProcessorFast label-return bug are unrelated." + "reason": "TF32 API guidance and a cache-strategy typo are unrelated." }, { - "left": "issue:44261", - "right": "issue:44360", + "left": "issue:44704", + "right": "issue:44908", "accept": false, - "reason": "Layer norm precision/config issue vs DSA indexer missing ReLU; different subsystems." + "reason": "AutoProcessor kwargs forwarding and inverse_sqrt scheduler kwargs are different APIs with different failures." }, { - "left": "issue:44273", - "right": "issue:45003", + "left": "issue:42890", + "right": "issue:44908", "accept": false, - "reason": "The titles are loosely related around lazy loading/import behavior, but there is not enough evidence here to say they are the same concrete bug." + "reason": "Missing set_seed in tests is a flaky-test issue, not a scheduler configuration bug." }, { - "left": "issue:44623", - "right": "issue:45230", + "left": "issue:44246", + "right": "issue:44273", "accept": false, - "reason": "Missing files on processor save vs a vague bug report; no evidence of the same defect." + "reason": "Import-time slowdown and lazy-loading correctness are adjacent but not the same bug." }, { - "left": "issue:44273", - "right": "issue:45542", + "left": "issue:43408", + "right": "issue:44315", "accept": false, - "reason": "Lazy-loading failure and TensorFlow backend detection error are different bugs." + "reason": "Model-type warning in sam3_tracker and Liger Kernel application with model_init are unrelated." }, { - "left": "issue:44393", - "right": "issue:45753", + "left": "issue:43976", + "right": "issue:44393", "accept": false, - "reason": "Same model family, but different symptoms and likely different root causes; too broad to merge." + "reason": "Python version compatibility and Qwen3-VL bounding-box output are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention support gating for Nemotron and Qwen3-VL bbox hallucination are different model issues." }, { "left": "issue:43576", - "right": "issue:44556", + "right": "issue:44908", "accept": false, - "reason": "Broken env command vs checkpoint reload incompatibility after v5 upgrade; different user flows." + "reason": "The env command regression and scheduler kwargs regression do not share a concrete fix." }, { - "left": "issue:41669", - "right": "issue:43673", + "left": "issue:44297", + "right": "issue:45230", "accept": false, - "reason": "Import-star performance regression vs missing generation cache in chunked prefill; not the same code-path." + "reason": "Tokenizer class metadata mismatch and a generic bug report are not duplicates." }, { - "left": "issue:44393", + "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Qwen3-VL output bug and PretrainedConfig type-checking regression are unrelated." + "reason": "Scheduler kwargs handling and PretrainedConfig type checking are unrelated." }, { - "left": "issue:44038", - "right": "issue:45125", + "left": "issue:44295", + "right": "issue:45230", "accept": false, - "reason": "Both are Qwen3-related, but one is a general bug and the other is a tensor-parallel missing-plan issue." + "reason": "Position_ids buffer access error and a generic bug report are not the same underlying issue." }, { - "left": "issue:44556", - "right": "issue:45092", + "left": "issue:41669", + "right": "issue:44351", "accept": false, - "reason": "Both concern v5 checkpoint incompatibilities, but they target different models and different failure causes." + "reason": "Import-star cleanup and HybridCache import failure are different import-path problems." }, { - "left": "issue:44373", - "right": "issue:44961", + "left": "issue:41669", + "right": "issue:45341", "accept": false, - "reason": "Docstring issue vs an unrelated placeholder issue; no overlap." + "reason": "Model import bloat and a testing_utils bug are unrelated." }, { - "left": "issue:44704", + "left": "issue:44492", + "right": "issue:44945", + "accept": false, + "reason": "Cache-strategy typo and incorrect pipeline-parallel output are separate concerns." + }, + { + "left": "issue:41669", + "right": "issue:44492", + "accept": false, + "reason": "The import-star cleanup issue is not the same as a cache-strategy typo." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Qwen3-VL bbox errors and Qwen3.5 Moe from_pretrained failures are different model regressions." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied in model_init and flash_attention_3 degenerate training are not the same code-path failure." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "A post_init policy change for patchtsmixer is unrelated to an AttributeError in tied-weight key handling." + }, + { + "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Processor kwargs forwarding bug and scheduler kwargs bug are different APIs." + "reason": "Docs still referencing pipeline() and scheduler kwargs are unrelated." }, { - "left": "issue:42890", + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion about forward args and a tokenizer loading regression are different issues." + }, + { + "left": "issue:41669", "right": "issue:44908", "accept": false, - "reason": "Flaky test due to missing seeds vs scheduler configuration bug; unrelated." + "reason": "Import-time bloat and inverse_sqrt scheduler kwargs are unrelated." }, { - "left": "issue:44246", + "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "Slow import sometimes and lazy loading not working properly are related in theme, but the reports do not describe the same concrete defect." + "reason": "Broken env command and lazy-loading malfunction are separate v5 issues." }, { - "left": "issue:43408", - "right": "issue:44315", + "left": "issue:44655", + "right": "issue:44861", "accept": false, - "reason": "Model-type warning vs model_init/Liger integration issue; different code paths." + "reason": "Saving pipeline objects and tied-weight key attribute handling do not share the same failure mode." }, { - "left": "issue:43976", - "right": "issue:44393", + "left": "issue:44360", + "right": "issue:45468", "accept": false, - "reason": "Python version compatibility and a Qwen3-VL inference bug are unrelated." + "reason": "DSA indexer activation logic and Gemma-4 audio positional encoding are unrelated model internals." }, { - "left": "issue:43352", - "right": "issue:44393", + "left": "issue:45230", + "right": "issue:45310", "accept": false, - "reason": "Flash-Attention-2 support error and Qwen3-VL bounding-box hallucination are different issues." + "reason": "Generic bug report and Qwen3.5 Moe from_pretrained error are not the same problem." }, { - "left": "issue:43576", + "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Broken env command and scheduler kwargs regression are unrelated." + "reason": "Lazy loading and scheduler kwargs handling are unrelated." }, { - "left": "issue:44297", - "right": "issue:45230", + "left": "issue:43519", + "right": "issue:44485", "accept": false, - "reason": "Tokenizer save metadata mismatch vs a vague bug report; no same underlying change." + "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are different model-specific bugs." }, { - "left": "issue:43673", - "right": "issue:44492", + "left": "issue:41669", + "right": "issue:43704", "accept": false, - "reason": "Generation cache bug and a typo in cache strategies are not the same defect." + "reason": "Import-star cleanup does not match a VRAM leak in multithreaded dataloaders." }, { - "left": "issue:44908", - "right": "issue:45071", + "left": "issue:43519", + "right": "issue:44908", "accept": false, - "reason": "Scheduler kwargs handling and config type checking are unrelated." + "reason": "Timestamp math in Qwen3VL and inverse_sqrt scheduler kwargs are unrelated." }, { - "left": "issue:44295", - "right": "issue:45230", + "left": "issue:31795", + "right": "issue:34689", "accept": false, - "reason": "position_ids buffer access error vs vague bug report; no evidence of duplication." + "reason": "Forward-argument docs and model-loading breakage for Llama 3.2 are not duplicates." }, { - "left": "issue:41669", - "right": "issue:44351", + "left": "issue:31795", + "right": "issue:37428", "accept": false, - "reason": "Import-star cleanup/performance issue vs missing HybridCache export; related to imports, but not the same bug." + "reason": "Documentation wording and an import error from flash attention utils are unrelated." }, { - "left": "issue:41669", - "right": "issue:45341", + "left": "issue:44315", + "right": "issue:45092", "accept": false, - "reason": "Import performance regression and a testing_utils bug are unrelated." + "reason": "Both concern model initialization/loading, but one is Liger Kernel application and the other is remote-code checkpoint meta-init incompatibility." }, { - "left": "issue:41669", - "right": "issue:44492", + "left": "issue:43704", + "right": "issue:44485", "accept": false, - "reason": "Import-star performance problem and a cache-strategy typo are different issues." + "reason": "VRAM leakage in dataloader threads and GLM-5 RoPE implementation are unrelated." }, { - "left": "issue:44393", - "right": "issue:45478", + "left": "issue:44485", + "right": "issue:45468", "accept": false, - "reason": "Different Qwen3.5/Moe bugs with different symptoms and no shared concrete failure mode." + "reason": "GLM-5 RoPE math and Gemma-4 audio positional encoding are different model implementations." }, { - "left": "issue:44315", - "right": "issue:44829", + "left": "issue:42371", + "right": "issue:43704", "accept": false, - "reason": "Liger-kernel application bug vs flash_attention_3 degenerate training; different paths." + "reason": "TF32 API guidance and a VRAM leak are unrelated." }, { - "left": "issue:44077", - "right": "issue:44861", + "left": "issue:43502", + "right": "issue:43519", "accept": false, - "reason": "Optional post_init validation issue vs tied-weight-key AttributeError; unrelated." + "reason": "local_files_only network calls and Qwen3VL timestamp calculation are different bugs." }, { - "left": "issue:43827", - "right": "issue:44908", + "left": "issue:43519", + "right": "issue:44623", "accept": false, - "reason": "Docs still referencing pipeline() vs scheduler kwargs bug; different kinds of changes." + "reason": "Timestamp calculation and processor.save_pretrained file omission are unrelated." }, { - "left": "issue:31795", - "right": "issue:33453", + "left": "issue:43519", + "right": "issue:43704", "accept": false, - "reason": "Documentation confusion in model.forward vs tokenizer loading regression; unrelated." + "reason": "Qwen3VL timestamp math and dataloader-thread VRAM leakage are different issues." }, { - "left": "issue:41669", - "right": "issue:44908", + "left": "issue:44263", + "right": "issue:45468", "accept": false, - "reason": "Import-star performance regression and scheduler kwargs bug are unrelated." + "reason": "torch.split return handling in GlmMoeDsaIndexer and Gemma-4 audio positional encoding do not share the same underlying bug." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The cluster summary is broadly reasonable, but it overstates the evidence for the one accepted soft pair. In the packet, issue:44273 and issue:45003 only have titles and a modest similarity score; there are no explicit links or body details proving they are the same bug. Describing 45003 as a concrete cause/root-cause for 44273 is too strong on this evidence. The rest of the soft-edge rejections look conservative, but this accepted duplicate verdict is not sufficiently grounded, so I can\u2019t accept the analyst output." - }, - "evaluator_used": true, - "retried": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -14212,77 +14335,43 @@ "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:43576", "issue:43606", + "issue:43618", "issue:43704", - "issue:43827", "issue:44230", "issue:44261", - "issue:44263", - "issue:44273", - "issue:44315", - "issue:44360", - "issue:44485", "issue:44492", + "issue:44556", "issue:44623", - "issue:44655", "issue:44861", "issue:44908", "issue:44998", - "issue:45092", - "issue:45230", "issue:45290", "issue:45310", "issue:45341", "issue:45468" ], "soft_pairs": [ - "issue:42371|issue:44492", - "issue:43576|issue:44273", - "issue:44655|issue:44861", - "issue:44360|issue:45468", - "issue:45230|issue:45310", - "issue:43827|issue:44492", - "issue:44273|issue:44908", - "issue:43519|issue:44485", - "issue:41669|issue:43704", - "issue:43519|issue:44908", - "issue:31795|issue:34689", - "issue:31795|issue:37428", - "issue:44315|issue:45092", - "issue:43704|issue:44485", - "issue:44485|issue:45468", - "issue:42371|issue:43704", - "issue:43502|issue:43519", - "issue:43519|issue:44623", - "issue:43519|issue:43704", - "issue:44263|issue:45468", "issue:43606|issue:45468", "issue:44261|issue:45468", "issue:43704|issue:45341", "issue:42913|issue:43066", "issue:44908|issue:44998", "issue:44861|issue:45290", - "issue:41669|issue:44998", + "issue:44492|issue:44998", + "issue:44998|issue:45341", + "issue:43618|issue:44998", + "issue:44556|issue:44998", "issue:43525|issue:44861", - "issue:43704|issue:44998", - "issue:43066|issue:44998", "issue:43572|issue:44861", "issue:31515|issue:38617", - "issue:44360|issue:44998", "issue:44230|issue:44623", "issue:30333|issue:31515", "issue:31515|issue:36296", @@ -14291,274 +14380,154 @@ "issue:31515|issue:33453", "issue:43298|issue:44230" ], - "prepared_review_unit_hash": "6d99e37ee71e9b1c63b7091e95ed0a0e99b3cb77020de6f8a3a9afb3445b46b5", + "prepared_review_unit_hash": "dde875dd58897301f324d18ceeed74a25e15e4f4b743ddf5e96aedc13fa37388", "component_budget": { "node_count": 384, "item_count": 384, "soft_pair_count": 1441, - "serialized_chars": 742684, - "estimated_input_tokens": 185671, - "estimated_eval_tokens": 371598 + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 }, "budget": { - "node_count": 40, - "item_count": 40, - "soft_pair_count": 40, - "serialized_chars": 34369, - "estimated_input_tokens": 8593, - "estimated_eval_tokens": 17442 + "node_count": 26, + "item_count": 26, + "soft_pair_count": 20, + "serialized_chars": 20575, + "estimated_input_tokens": 5144, + "estimated_eval_tokens": 10544 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: the soft pairs span unrelated bugs in tokenizers, loading, docs, schedulers, model internals, and utilities. None of the candidate pairs look like the same underlying issue/change.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "The cluster is mostly heterogeneous. A small subgroup concerns Transformers v5 migration/loading/tokenizer/config regressions, but several items are unrelated model-specific bugs and one is clearly off-topic, so there is no safe duplicate merge within the provided pairs.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 44556 is the broadest representative of the coherent v4\u2192v5 checkpoint/loading regression theme that overlaps with several other token/config issues; it is the best cluster anchor among the related reports.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 44556 is the most representative issue for the strongest shared theme in this set (version-upgrade loading failures and related tokenizer/config regressions).", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42371", - "right": "issue:44492", - "accept": false, - "reason": "TF32 behavior/settings API vs a cache-strategy typo; unrelated symptoms and code paths." - }, - { - "left": "issue:43576", - "right": "issue:44273", - "accept": false, - "reason": "Broken env command in v5 vs lazy-loading failure; different feature areas." - }, - { - "left": "issue:44655", - "right": "issue:44861", - "accept": false, - "reason": "Saving Pipeline objects vs tied-weights key handling; both involve save/load but not the same bug." - }, - { - "left": "issue:44360", - "right": "issue:45468", - "accept": false, - "reason": "DSA indexer ReLU issue vs Gemma4 audio positional encoding; different model internals." - }, - { - "left": "issue:45230", - "right": "issue:45310", - "accept": false, - "reason": "Generic bug report vs a specific Qwen3.5 MoE from_pretrained regression; not the same concrete issue." - }, - { - "left": "issue:43827", - "right": "issue:44492", - "accept": false, - "reason": "Docs still referencing pipeline() vs cache-strategy typo; unrelated." - }, - { - "left": "issue:44273", - "right": "issue:44908", - "accept": false, - "reason": "Lazy loading bug vs inverse_sqrt scheduler kwargs bug; different subsystems." - }, - { - "left": "issue:43519", - "right": "issue:44485", - "accept": false, - "reason": "Qwen3VL timestamp calculation vs GLM-5 RoPE implementation; both model-specific but distinct defects." - }, - { - "left": "issue:41669", - "right": "issue:43704", - "accept": false, - "reason": "Import-star cleanup/performance issue vs VRAM leak in Qwen3ForCausalLM; unrelated." - }, - { - "left": "issue:43519", - "right": "issue:44908", - "accept": false, - "reason": "Timestamp math in processor vs scheduler kwargs handling; no overlap in code path." - }, - { - "left": "issue:31795", - "right": "issue:34689", - "accept": false, - "reason": "Documentation confusion in model.forward vs a model-loading regression; unrelated." - }, - { - "left": "issue:31795", - "right": "issue:37428", - "accept": false, - "reason": "Docs around input_ids/past_key_values vs flash-attention import error; different topics." - }, - { - "left": "issue:44315", - "right": "issue:45092", - "accept": false, - "reason": "Liger kernel not applied with model_init vs meta-init incompatibility for old remote-code checkpoints; not the same bug." - }, - { - "left": "issue:43704", - "right": "issue:44485", - "accept": false, - "reason": "VRAM leak in dataloader-threaded causal LM use vs GLM-5 RoPE implementation; distinct issues." - }, - { - "left": "issue:44485", - "right": "issue:45468", - "accept": false, - "reason": "GLM-5 RoPE math vs Gemma4 audio positional encoding; different models and failures." - }, - { - "left": "issue:42371", - "right": "issue:43704", - "accept": false, - "reason": "TF32 configuration warning vs VRAM leak; unrelated." - }, - { - "left": "issue:43502", - "right": "issue:43519", - "accept": false, - "reason": "local_files_only still making API calls vs timestamp miscalculation; different code paths." - }, - { - "left": "issue:43519", - "right": "issue:44623", - "accept": false, - "reason": "Processor timestamp bug vs processor.save_pretrained missing files; both processor-related but not the same defect." - }, - { - "left": "issue:43519", - "right": "issue:43704", - "accept": false, - "reason": "Timestamp calculation vs VRAM leak; unrelated." - }, - { - "left": "issue:44263", - "right": "issue:45468", - "accept": false, - "reason": "torch.split return handling in GlmMoeDsaIndexer vs Gemma4 audio positional encoding; different components." - }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch for bark-small vs Gemma4 audio positional encoding; different models/problems." + "reason": "Different model-specific bugs: CPU offload device mismatch vs Gemma-4 audio positional encoding behavior." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "Missing rms_norm_eps in MLA q_a_layernorm vs Gemma4 audio positional encoding; unrelated." + "reason": "Both mention precision/config issues, but they target different architectures and different code paths." }, { "left": "issue:43704", "right": "issue:45341", "accept": false, - "reason": "VRAM leak in model use vs a testing_utils bug; no concrete overlap." + "reason": "VRAM leakage in Qwen3ForCausalLM and a testing_utils bug are unrelated." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "General tokenizer behavior difference across v4/v5 vs wrong decoder type in v5; related area but not clearly the same change." + "reason": "Both concern tokenizer behavior in v5, but one is a general behavior difference and the other is a specific decoder-type regression; not enough to treat as the same bug." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwarg bug vs unrelated issue titled 'Unemployment'; no relation." + "reason": "Scheduler keyword handling and an off-topic issue are unrelated." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weight-key crash during loading vs chat-template crash with tool calls; different APIs and failure modes." + "reason": "One is a tied-weights helper crash; the other is a chat-template/tool-call crash. Different subsystems and failure modes." }, { - "left": "issue:41669", + "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "Import-star cleanup/performance vs unrelated issue; no shared bug." + "reason": "A cache-strategy typo and an unrelated off-topic issue are not duplicates." }, { - "left": "issue:43525", - "right": "issue:44861", + "left": "issue:44998", + "right": "issue:45341", "accept": false, - "reason": "Missing pad_token_id on Llama4Config vs tied-weight-key AttributeError; unrelated." + "reason": "Off-topic issue versus a testing_utils bug; no shared underlying defect." }, { - "left": "issue:43704", + "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "VRAM leak vs unrelated issue; no overlap." + "reason": "CLIPOutput attentions regression is unrelated to the off-topic issue." }, { - "left": "issue:43066", + "left": "issue:44556", "right": "issue:44998", "accept": false, - "reason": "Tokenizer decoder type bug vs unrelated issue; no relation." + "reason": "Checkpoint reload regression and an unrelated off-topic issue are not duplicates." }, { - "left": "issue:43572", + "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "StableLmConfig pad_token_idx regression vs tied-weight-key crash; different config and code paths." + "reason": "Missing pad_token_id in Llama4Config is a config attribute regression, not the tied-weights list crash." }, { - "left": "issue:31515", - "right": "issue:38617", + "left": "issue:43572", + "right": "issue:44861", "accept": false, - "reason": "Slow from_pretrained checkpoint loading vs ImportError for layer_type_validation; unrelated loader problems." + "reason": "Missing pad_token_idx in StableLmConfig is a different config regression than the tied-weights helper error." }, { - "left": "issue:44360", - "right": "issue:44998", + "left": "issue:31515", + "right": "issue:38617", "accept": false, - "reason": "DSA indexer math issue vs unrelated issue; no common bug." + "reason": "Slow from_pretrained checkpoint loading and an ImportError in configuration_utils are distinct problems." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "FP8 inference support for Qwen3-VL/Qwen3.5 MoE vs processor.save_pretrained file omission; unrelated." + "reason": "FP8 model support issue and missing processor save files are unrelated." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow reporting on failed training jobs vs slow checkpoint loading; unrelated." + "reason": "MLflow job reporting failure and slow checkpoint loading are not the same bug." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Slow checkpoint loading vs tensor parallel training bug; different subsystems." + "reason": "Checkpoint loading performance and tensor-parallel training bug are different code paths." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Loading performance regression vs documentation confusion in model.forward; unrelated." + "reason": "Checkpoint loading slowness and documentation confusion are unrelated." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "FP8 support request for Qwen3-VL/Qwen3.5 MoE infer vs a from_pretrained error on transformers>=5.4.0; different concrete problems." + "reason": "Both involve Qwen3.5 MoE/fp8/versioning, but one is fp8 inference support and the other is a from_pretrained error; too different to merge as one bug." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Slow from_pretrained loading vs tokenizer-loading regression; not the same bug." + "reason": "Both mention loading, but one is slow checkpoint loading and the other is a tokenizer-loading regression; not the same concrete defect." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching not supporting audio models vs FP8 support for Qwen3-VL/Qwen3.5 MoE; different features and models." + "reason": "Continuous batching audio-model support and fp8 inference support are distinct feature/bug areas." } ] }, @@ -14588,12 +14557,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "b16df79519325d4d44019b581c01e8b7983539a7c1b09b039b5d16e5f1202548", + "prepared_review_unit_hash": "14d51eae4584217817909e37006e4234b1da0ffb1000c08a3b16207b10221ed3", "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 }, @@ -14601,7 +14570,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1590, + "serialized_chars": 1589, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -14610,18 +14579,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate feature requests for different model/runtime integration work: one asks for Flash Attention 2 support across more architectures, the other for GGUF support across more architectures. They share the same contribution framing but not the same underlying change.", - "confidence": 0.98, + "summary": "These are two separate community-contribution feature requests: one for Flash Attention 2 support across more architectures, and one for GGUF support across more architectures. They share a loose pattern but not the same underlying change, so they should not be merged as duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 26350 has the higher discussion activity and is a clearer standalone tracker, but it is not a duplicate of 33260; the requests target different features/code paths.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different features and code paths: Flash Attention 2 support vs GGUF support. Similar wording about architecture coverage does not make them duplicates." + "reason": "Both are broad 'support more architectures' requests, but Flash Attention 2 and GGUF are different features with different implementation paths and goals, so they are not the same underlying issue." } ] }, @@ -14658,12 +14627,12 @@ "issue:43834|issue:44365", "issue:44365|issue:44370" ], - "prepared_review_unit_hash": "536033348200ca16ef36127a4729cd4bce6a19ee705e4804a5339a55471f796e", + "prepared_review_unit_hash": "5875a0c81c38295cacd1e437c7a34f3fc542b44d509076bd7cb10ae564113cba", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4454, + "serialized_chars": 4455, "estimated_input_tokens": 1114, "estimated_eval_tokens": 2484 }, @@ -14671,7 +14640,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4454, + "serialized_chars": 4455, "estimated_input_tokens": 1114, "estimated_eval_tokens": 2484 }, @@ -14680,48 +14649,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster looks like repeated i18n-doc translation request templates, not true duplicates. The only concrete request is Turkish; the others are generic placeholder-language variants that likely refer to different languages, so they should stay separate.", - "confidence": 0.87, - "canonical_issue_reason": "issue:27088 is the most concrete and identifiable request: it names Turkish explicitly, while the others are generic template-style placeholders with no specific language shown.", + "summary": "These look like separate docs-translation requests, not duplicates of one bug/change. Issue 27088 is specifically for Turkish, while the others are generic i18n template issues that likely refer to different languages.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical duplicate stands out: the only concrete language is Turkish in 27088, while the others are generic placeholders and likely distinct language requests.", "canonical_pr_reason": null, - "best_issue_reason": "issue:27088 is the best representative to keep because it is specific, human-readable, and clearly scoped to one translation request; the others appear to be less concrete template instances.", + "best_issue_reason": "27088 is the best representative only because it is the most specific and earliest concrete request, but it should not absorb the others as duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Same translation-request template, but no evidence they target the same language; placeholder titles suggest different underlying requests." + "reason": "Both are translation-request templates, but there is no evidence they target the same language or same underlying work item." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Both are i18n-doc translation requests, but the specific language is different/unspecified, so this is template similarity rather than a duplicate." + "reason": "27088 is specifically Turkish; the other issue is a generic placeholder likely for a different language, so they are not the same request." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Turkish is a concrete language request; the other issue is a generic placeholder. They do not appear to be the same underlying issue." + "reason": "Different concrete localization targets are implied; same template text is not enough to treat them as duplicates." }, { "left": "issue:43834", "right": "issue:44370", "accept": false, - "reason": "Highly similar template wording, but nothing indicates the same language target or the same exact translation task." + "reason": "These appear to be separate instances of the same issue template, likely for different languages rather than the same underlying task." }, { "left": "issue:43834", "right": "issue:44365", "accept": false, - "reason": "These look like separate instances of the same template for different languages, not one duplicated issue." + "reason": "No evidence they refer to the same language or translation branch; similarity is only from the template." }, { "left": "issue:44365", "right": "issue:44370", "accept": false, - "reason": "Only template-level similarity is visible; without a shared concrete language or task, they should not be merged as duplicates." + "reason": "They share the same placeholder title, but that indicates a template family, not a duplicate issue." } ] }, @@ -14781,16 +14750,16 @@ "issue:43671|issue:45027", "issue:43671|issue:45177", "issue:43671|issue:44121", - "issue:43837|issue:45322", + "issue:43671|issue:43909", "issue:43837|issue:43909", + "issue:43837|issue:45322", + "issue:43837|issue:44121", "issue:43909|issue:44205", - "issue:43671|issue:43909", "issue:44121|issue:45322", - "issue:43837|issue:44121", "issue:43837|issue:45027", "issue:43837|issue:45177" ], - "prepared_review_unit_hash": "168ed3567329b21338d6486f8c4ce8b16cb462c8b4212c63d7c878503e11e042", + "prepared_review_unit_hash": "4d59586a97e1a311b15f907290a154371d56d810423ed2e934ab37bb97d952f1", "component_budget": { "node_count": 14, "item_count": 14, @@ -14812,180 +14781,180 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are all model-request issues in the same broad transformers ecosystem, but they describe different concrete models/features rather than duplicate bugs or the same change.", - "confidence": 0.28, - "canonical_issue_reason": "Issue 41115 is the most representative broad architecture-add request in the set, with stronger activity than most others and a generic title that fits the cluster\u2019s overall theme.", + "summary": "These items are all model-addition/support requests, but they target different architectures and modalities. The soft-similarity links are thematic, not duplicates of the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": "No true canonical duplicate: the cluster mixes distinct model requests (different models, modalities, and implementation paths), so there isn\u2019t one shared issue to collapse into.", "canonical_pr_reason": null, - "best_issue_reason": "41115 is the best standalone issue to anchor the cluster because it is concrete, broadly phrased, and closest to the shared pattern of requesting new model support.", + "best_issue_reason": "issue:37015 is the broadest/high-activity open model-addition request, so it works best as a representative anchor if one must be chosen, though it is not a duplicate sink.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model families (EuroBERT vs DEIMv2); same add-model pattern only." + "reason": "EuroBert config support vs DEIMv2 support are different model additions with separate code paths." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBERT config request and Qwen3-TTS support are different models and modalities." + "reason": "EuroBert config support and Qwen3-TTS support target unrelated model families/modalities." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are unrelated concrete additions." + "reason": "NeoBERT and Qwen3-ASR are separate model requests; not the same concrete change." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT support and weight-sparse transformer support are different code/model changes." + "reason": "NeoBERT vs weight-sparse transformer/circuit-sparsity are unrelated model architectures." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT vs LFM2.5 Audio are different model requests." + "reason": "NeoBERT and LFM2.5 Audio are different model support requests, not duplicates." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are distinct model additions." + "reason": "NeoBERT and ModernVBERT concern different model families and implementations." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are different model families and tasks." + "reason": "NeoBERT and Voxtral TTS are distinct model additions with no shared fix." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture work and Qwen3-ASR support are not the same underlying change." + "reason": "MiniCPM3 and Qwen3-ASR are different models and code paths." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 support and the sparse-transformer request target different implementations." + "reason": "MiniCPM3 support and circuit-sparsity support are unrelated requests." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are separate model-support requests." + "reason": "MiniCPM3 and ModernVBERT are separate model support items." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 architecture support and Voxtral-TTS support are unrelated changes." + "reason": "MiniCPM3 and Voxtral TTS are different model-modalities and cannot be merged as one fix." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are different model additions." + "reason": "DEIMv2 and Qwen3-TTS are unrelated model additions." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are distinct model requests." + "reason": "ModernVBERT and Qwen3-ASR are different model-specific requests." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are different concrete supports." + "reason": "ModernVBERT and Qwen3-TTS target different architectures/modalities." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "One is a specific model request, the other is a broad BERT/RoPE feature proposal." + "reason": "ModernVBERT support and BERT-like-with-RoPE support are not the same concrete change." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like models with RoPE and Voxtral-TTS are unrelated requests." + "reason": "BERT-like models with RoPE and Voxtral TTS are unrelated additions." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Both are Qwen3 audio requests, but TTS and ASR are separate concrete changes." + "reason": "Qwen3-TTS and Qwen3-ASR share a prefix but are different modalities and implementation paths." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and Voxtral-TTS are different model integrations." + "reason": "Qwen3-TTS and Voxtral TTS are distinct TTS model requests, not one duplicate." }, { "left": "issue:43671", - "right": "issue:43909", + "right": "issue:45177", "accept": false, - "reason": "TTS support for Qwen3 and LFM2.5 Audio 1.5B are different model additions." + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model support requests." }, { "left": "issue:43671", - "right": "issue:45177", + "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS support and DeepSeek-OCR2 are different modalities and code paths." + "reason": "Qwen3-TTS support and circuit-sparsity support are different model/code changes." }, { "left": "issue:43671", - "right": "issue:44121", + "right": "issue:43909", "accept": false, - "reason": "Audio model support and sparse-transformer support are unrelated." + "reason": "Qwen3-TTS and LFM2.5 Audio are separate audio model requests, not one fix." }, { "left": "issue:43837", - "right": "issue:45322", + "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and EUPE are different model requests." + "reason": "Qwen3-ASR and LFM2.5 Audio are different model additions despite both being audio-related." }, { "left": "issue:43837", - "right": "issue:43909", + "right": "issue:45322", "accept": false, - "reason": "ASR support and LFM2.5 Audio support are related only at a broad audio level, not the same change." + "reason": "Qwen3-ASR and EUPE are unrelated model requests." }, { - "left": "issue:43909", - "right": "issue:44205", + "left": "issue:43837", + "right": "issue:44121", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different model families." + "reason": "Qwen3-ASR support and weight-sparse transformer support are different concrete changes." }, { - "left": "issue:44121", - "right": "issue:45322", + "left": "issue:43909", + "right": "issue:44205", "accept": false, - "reason": "Sparse-transformer support and EUPE support are unrelated changes." + "reason": "LFM2.5 Audio and SAM3-LiteText are different model families/modalities." }, { - "left": "issue:43837", - "right": "issue:44121", + "left": "issue:44121", + "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR and the sparse-transformer request target different code paths." + "reason": "Weight-sparse transformer and EUPE are distinct architecture requests with no shared fix." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral-TTS are different concrete audio-model integrations." + "reason": "Qwen3-ASR and Voxtral TTS are different speech-model support requests and would not be merged into one PR/issue." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are different tasks and models." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are separate model support items." } ] }, @@ -15026,82 +14995,82 @@ "issue:43668|issue:44031", "issue:43668|issue:43813" ], - "prepared_review_unit_hash": "0ab1a9b3f48581e5ce6637fec5422564f2d1d4e7401a049d1bdd9149206412c1", + "prepared_review_unit_hash": "39c90b0d3d2e9bee4f22a56d19d6f7e4bef110e316bc856047d0adf799dffdf0", "component_budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6291, - "estimated_input_tokens": 1573, - "estimated_eval_tokens": 3402 + "serialized_chars": 6294, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 6291, - "estimated_input_tokens": 1573, - "estimated_eval_tokens": 3402 + "serialized_chars": 6294, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are all superficially similar only at the wording level, but they describe different bugs across unrelated components (norm/config typing, Mask2Former loss handling, tokenizer regex warnings, and a typo in a quantization attribute). I do not see a true duplicate cluster here.", - "confidence": 0.93, - "canonical_issue_reason": "No single issue clearly represents a duplicate set; the items span distinct subsystems and problem types, so there is no good canonical duplicate issue.", + "summary": "These issues are only superficially similar by wording; they describe different components and bug types, so this cluster should not be merged as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 43668 is the cleanest representative only in the sense that it is a narrowly scoped, concrete bug report, but it is not a cluster canonical because the other issues are unrelated.", + "best_issue_reason": "issue:43668 is the cleanest single bug report in the set: a concrete config type-hint problem with a specific target and no sign of overlapping with the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Different problems: Llama4TextL2Norm/RMS norm semantics vs ModernBERTConfig type-hint error. No shared code path or concrete duplicate evidence." + "reason": "Both mention normalization, but one asks whether a model layer is RMSNorm and the other reports an incorrect type hint in a config; different artifacts and different failure modes." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Mask2Former ignore_value handling and kwargs not reaching loss calculation are different bugs, even though both touch loss-related code." + "reason": "Mask2Former ignore_value handling is unrelated to kwargs not being forwarded into a loss function; different code paths and symptoms." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2Former loss ignore_value bug is unrelated to tokenizer regex warning behavior." + "reason": "One is a segmentation-model parameter plumbing issue, the other is a tokenizer regex warning regression; no shared underlying bug." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Different subsystems and failure modes: model loss parameter use vs a typo in an attribute name." + "reason": "An ignore_value usage bug is unrelated to a typo in a quantization operation name." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former loss issue and ModernBERT config type hint issue are unrelated." + "reason": "Different models and bug classes: ignored loss parameter versus a ModernBERT config annotation error." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Loss kwargs forwarding bug is not the same as an incorrect config type hint." + "reason": "Missing kwargs in loss calculation and an incorrect type hint are distinct issues with no common code-path evidence." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "ModernBERTConfig typing issue and tokenizer regex warning issue do not describe the same underlying defect." + "reason": "A config type-hint bug is not the same as a tokenizer warning regression." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "A type-hint mistake in ModernBERTConfig is unrelated to a typo in quantization-operation naming." + "reason": "The ModernBERT config type hint issue and the quantization-operation typo are separate, unrelated defects." } ] }, @@ -15131,7 +15100,7 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "d32cdff39c4e1d0b88ebeb119b69fc9c0518059ad9a72492dbebb8435786e533", + "prepared_review_unit_hash": "107b7b76c27297bc4ce8e5f2d998daf451bf0504a600f9809cd3b80b15a9a423", "component_budget": { "node_count": 2, "item_count": 2, @@ -15153,18 +15122,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are unrelated: one requests Block Decoding support, the other requests DeepSeek v3.2 support. There is no shared codepath, filename, or explicit linkage suggesting duplication.", + "summary": "The two issues are unrelated: one asks for block decoding support, the other requests DeepSeek v3.2 support. They should remain separate; no duplicate/merge candidate here.", "confidence": 0.97, - "canonical_issue_reason": "Issue 41084 is the more concrete, feature-oriented request (\u201cSet Block Decoding\u201d) and is not dependent on a model-specific support request.", + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 41084 is the better canonical anchor because it names a specific functionality, whereas 43037 is a broader model-support request.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different underlying requests: block decoding vs DeepSeek v3.2 support. The similarity is only superficial and not enough to treat them as duplicates." + "reason": "Different underlying requests with no shared code path or explicit linkage: block decoding vs DeepSeek v3.2 model support." } ] }, @@ -15194,12 +15163,12 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "4103cf883f3a49d9259d06e6f1f5bc9e9d370f32c18573bdf5bd7eede20b75a0", + "prepared_review_unit_hash": "1c1f2cc8a8db9e9428b951e2acfebf84eca26dcadf47fbb7fe54774df1887f16", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1582, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -15207,7 +15176,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1582, + "serialized_chars": 1583, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -15216,18 +15185,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are related only in that both mention custom trainers, but they describe different problems: one is about `remove_unused_column` behavior in `Trainer`, the other is about FLOPS estimation. They should not be treated as duplicates.", + "summary": "These two issues are only loosely related by custom trainer behavior, but they describe different problems: one about `remove_unused_column` customization, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", "confidence": 0.93, - "canonical_issue_reason": null, + "canonical_issue_reason": "No true canonical duplicate emerges: #43598 is about Trainer column-pruning customizability, while #43600 is about FLOPS estimation logic for custom trainers. Same general area, different underlying bug/request.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43598 is the more concrete Trainer behavior/customizability report, but it still does not subsume 43600; the latter is a separate FLOPS-accounting concern.", + "best_issue_reason": "If forced to pick a representative issue for the cluster theme, #43598 is broader and more directly about Trainer behavior customization, but it does not subsume #43600.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Same broad area (custom trainers), but different underlying bugs: `remove_unused_column` customizability versus FLOPS estimation. They would not be merged as one issue." + "reason": "Related only at a high level (custom trainers/Trainer internals). They target different code paths and fixes: `remove_unused_column` behavior versus FLOPS estimation." } ] }, @@ -15257,12 +15226,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "718ff7d60e2025fb7923a1f824c9e1fc0d0b79990e64c22f00360e466157782a", + "prepared_review_unit_hash": "eff1114ea095549cd1a4704486035f9f0c11076a09ef81b1b17cbcd3130872d9", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -15270,7 +15239,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -15279,18 +15248,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicates: one requests AutoModel/AutoModelForConditionalGeneration registration for Qwen3-Omni, while the other requests AutoModelForSequenceClassification support for Qwen3.5. They touch different model families and different auto-class entry points.", - "confidence": 0.94, - "canonical_issue_reason": "issue:43684 is the broader, more central registration request and best represents the cluster\u2019s general theme of adding missing AutoModel support.", + "summary": "These are related only at a high level: both request Qwen-family AutoModel support, but one is for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration and the other is for Qwen3.5 sequence classification. They are separate feature requests, not the same underlying bug/change.", + "confidence": 0.92, + "canonical_issue_reason": "issue:43684 is the broader, more central request (model registration for Qwen3-Omni across AutoModel and AutoModelForConditionalGeneration); issue:44405 targets a different config and task-specific AutoModelForSequenceClassification support.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43684 is the best representative because it is the broader core-model registration task; issue:44405 is a narrower, different head-specific support request.", + "best_issue_reason": "issue:43684 is the better representative because it is the more general model-registration request; issue:44405 is a distinct downstream capability request for sequence classification.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different model families and different features: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration vs Qwen3.5 sequence-classification support. Not the same underlying bug/change." + "reason": "Do not merge: different model variants and different auto-class targets (registration vs sequence classification support), so they do not look like the same underlying change." } ] }, @@ -15327,7 +15296,7 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "2e8c6528fa82edb81e5cb9233bd1f2fc53677184b19d347f9d3fc994a0e0a9d5", + "prepared_review_unit_hash": "a3ce3a6725b1b1fa003de6cdb94f784efc70f0bf7741ebeff54ff1d3674ff820", "component_budget": { "node_count": 4, "item_count": 4, @@ -15349,55 +15318,58 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster is mostly low-signal: three nearly empty issues titled \"huggingface\" and one unrelated short issue titled \"Fett\". The title match alone is not enough to confirm duplicate underlying bugs, so I would not merge any of the soft edges.", - "confidence": 0.58, - "canonical_issue_reason": "Issue 43845 is the earliest of the three near-identical \"huggingface\" reports, but the bodies are too sparse to establish a concrete duplicate target with confidence.", + "summary": "The three issues titled \"huggingface \" appear to be duplicates of the same report; the issue titled \"Fett\" looks unrelated. No PRs are present.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43845 is the earliest of the near-identical \"huggingface \" reports and is a reasonable canonical issue for the duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue must represent the cluster, 43845 is the safest pick because it is the earliest repeated title; however, the content is too thin to treat it as a strong canonical bug report.", + "best_issue_reason": "issue:43845 is the best issue anchor because it matches the repeated title and was opened first among the duplicate-looking reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no body/detail evidence tying them to the same underlying bug." + "reason": "Different title and no shared evidence beyond generic similarity; not enough to indicate the same underlying issue." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no concrete shared issue description beyond weak similarity." + "reason": "Different title and no concrete evidence of the same bug." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no explicit target or technical detail indicating the same problem." + "reason": "Different title and no concrete evidence of the same bug." }, { "left": "issue:43845", "right": "issue:43846", - "accept": false, - "reason": "Same short title, but the reports are too empty to confirm they describe the same concrete bug." + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; they look like duplicate reports of the same issue." }, { "left": "issue:43845", "right": "issue:43859", - "accept": false, - "reason": "Same short title, but there is not enough detail to safely treat these as duplicate reports of one bug." + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; likely the same underlying report." }, { "left": "issue:43846", "right": "issue:43859", - "accept": false, - "reason": "Same short title, but the available evidence is too thin to confirm a shared underlying issue." + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; likely duplicate issues." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall: the three closed issues share the exact same title and close timestamps are close enough to support a duplicate cluster, while issue:43835 has a clearly different title. The canonical choice of issue:43845 as the earliest duplicate-looking report is conservative. No PRs are present, and the soft-edge verdicts stay within the evidence available." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -15431,7 +15403,7 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "616eca8ac804bb082d3b2a9b45fd5f7259c8ef66031a1e7cb6561c55445a91d7", + "prepared_review_unit_hash": "5256e3b1eddb96349ae21c7880b824da21f5f3b99bdfd66bd5fafa36f8693bc3", "component_budget": { "node_count": 5, "item_count": 5, @@ -15453,66 +15425,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are low-information spam/documentation-like issues with no concrete shared bug or change, so they should not be deduplicated into one canonical thread.", - "confidence": 0.93, - "canonical_issue_reason": "No clear canonical duplicate exists: the issues do not describe the same problem, and the available text is too generic/incoherent to establish a single underlying report.", + "summary": "These are low-signal, likely unrelated spam/placeholder issues. The apparent similarity is only at the title level; there is no evidence of a shared bug or duplicate report.", + "confidence": 0.95, + "canonical_issue_reason": "Issue 44418 is the least spam-like and most readable title, but it is still a generic placeholder rather than a concrete defect; it is only a weak cluster representative.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a good global canonical because the cluster does not form a real duplicate set; the titles point to unrelated spam/marketing-style reports rather than one defect.", + "best_issue_reason": "44418 is the best available representative because it is the only non-spam-gibberish title, but none of these issues look like a real duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are generic spam markers, but there is no concrete shared bug or change to prove they are the same underlying issue." + "reason": "Both are titled \"\", but there is no body content or evidence they describe the same underlying issue." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Identical-looking spam titles are not enough to confirm a duplicate; no substantive issue details are present." + "reason": "Same generic spam title only; no concrete shared bug or change is visible." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is spam-like and the other is a documentation suggestion; they do not describe the same problem." + "reason": "A spam placeholder and a generic documentation suggestion are not the same underlying issue." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "The titles refer to different topics, and there is no evidence of a shared underlying defect or request." + "reason": "\"\" and \"Check out \\\"Google LLC\\\"\" are unrelated titles with no shared bug report." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "Documentation enhancement and spam are not the same user problem." + "reason": "Generic documentation suggestion versus spam placeholder; no duplicate semantics." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Documentation enhancement and spam are not the same user problem." + "reason": "Generic documentation suggestion versus spam placeholder; no evidence of the same problem." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are generic spam titles, but the content is too empty to safely treat them as the same underlying report." + "reason": "Identical spam titles alone are insufficient to treat them as the same underlying issue without any supporting details." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Spam and a Google LLC mention are different topics with no concrete shared bug/change." + "reason": "Spam placeholder and unrelated quoted-name title; no common bug/change is indicated." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Spam and a Google LLC mention are different topics with no concrete shared bug/change." + "reason": "Spam placeholder and unrelated quoted-name title; no duplicate relationship is supported." } ] }, @@ -15542,7 +15514,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "edc315d7efaaef003d33479330b683c205cf7fccfa2e14cedd97f52410a3c64c", + "prepared_review_unit_hash": "1e9e3d2bf1cbe557331adcf72457bf849b80df3671059f9d0ab7793605ebab5a", "component_budget": { "node_count": 2, "item_count": 2, @@ -15564,28 +15536,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues appear to be duplicates of the same request/problem: identical titles, nearly identical body sizes, and strong text similarity. Issue 45145 is the better canonical choice because it was created earlier and has an inbound reference.", - "confidence": 0.98, - "canonical_issue_reason": "Issue 45145 is the best canonical issue: same title/topic as 45151, created earlier, and it already has an inbound reference.", + "summary": "The pair looks highly similar on title and length, but the packet does not expose body content or any explicit linkage, so the duplicate call is not strong enough to accept with confidence. If they are duplicates, #45145 is the better representative because it was created first.", + "confidence": 0.68, + "canonical_issue_reason": "#45145 is the earliest of the two and would be the natural representative if they are later confirmed to be duplicates, but the current evidence is too thin to state that definitively.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45145 is the strongest representative of the duplicate set because it is earlier and has slightly more link activity.", + "best_issue_reason": "#45145 is the best issue to keep as the provisional representative because it is the first-created item in the pair.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", - "accept": true, - "reason": "Identical titles and very similar bodies strongly indicate the same underlying issue/request." + "accept": false, + "reason": "Identical titles and a soft similarity signal suggest possible duplication, but the packet does not show body text, explicit links, or other concrete evidence that they are the same underlying issue. Given the non-deterministic signal, this should be treated as inconclusive rather than accepted as a duplicate." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet: both issues have identical titles, similar body sizes, and the pair is marked with soft_similarity. Choosing 45145 as canonical is also supported by its earlier creation time and inbound reference. The soft-edge verdict is conservative enough given the evidence available." + "accept": false, + "feedback": "The duplicate judgment is a bit too strong for the evidence provided. The packet supports only identical titles, similar body lengths, and a soft similarity signal; it does not show body content directly or any explicit linkage. Given the pair is marked non-deterministic, the soft-edge verdict should be more conservative rather than stating it is safe to merge as duplicates." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -15662,40 +15634,40 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "48cb58b1df25447b25d967ba80fc36ec4ed4c1cd1fdcaff98a33700102be2c0c", + "prepared_review_unit_hash": "ce936d403433a7492cb509877fd6456a91c8c47dda352a825328762ed32f5134", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1669, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1669, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues both concern 4-bit inference regressions, but they describe different root causes and different failure paths: one is Gemma4/audio-module specific torch.finfo() on uint8 weights, the other is a BitsAndBytesConfig skip_modules regression that drops the default lm_head exclusion.", - "confidence": 0.95, - "canonical_issue_reason": "Issue 45674 is the better cluster representative because it describes a self-contained configuration regression with a clear trigger and failure mode; issue 45672 is a separate, model-specific Gemma4 bug.", + "summary": "These are two separate 4-bit inference bugs in bitsandbytes-related code paths: one is a Gemma4/audio-module uint8 + torch.finfo TypeError, the other is an llm_int8_skip_modules regression that drops the default lm_head exclusion and triggers an AssertionError. They should not be deduplicated.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue for this cluster: 45672 is a model-specific torch.finfo/type handling failure, while 45674 is a BitsAndBytesConfig regression affecting module exclusion logic.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45674 is the strongest standalone issue to keep because it is more general and actionable, while 45672 appears to be a narrower, unrelated model-specific defect.", + "best_issue_reason": "Issue 45674 is the better standalone representative because it describes a concrete configuration regression with a clear failure mode and broader applicability than the Gemma4-specific report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Reject: both mention 4-bit inference, but the symptoms and causes differ. 45672 is about torch.finfo() on uint8 weights in Gemma4 audio modules; 45674 is about BitsAndBytesConfig llm_int8_skip_modules clearing the default lm_head exclusion and triggering an AssertionError. These are not the same underlying bug." + "reason": "Different underlying bugs and code paths: 45672 is about uint8 weights in Gemma4 audio modules calling torch.finfo during NF4 inference, while 45674 is about llm_int8_skip_modules clearing the default lm_head exclusion and causing an AssertionError." } ] }, @@ -15725,40 +15697,40 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "984d883ec9e367730dd9ec8978243c3136573b858c38d78562b05b793fd65f76", + "prepared_review_unit_hash": "0c2e022a1fea4025c9bb75d69775e0d7f57885f386ef5912361f027f95452f26", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1476, - "estimated_input_tokens": 369, - "estimated_eval_tokens": 994 + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1476, - "estimated_input_tokens": 369, - "estimated_eval_tokens": 994 + "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": "The two closed issues are just person-name/date entries with no body or shared context, so they do not look like duplicates or the same underlying problem.", - "confidence": 0.99, - "canonical_issue_reason": null, + "summary": "Both items are empty, likely unrelated personal-name/date posts with no bug or feature context. They do not look like duplicates of the same underlying issue.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 45762 is the earlier of the two and can serve as the representative only by chronology, not content; neither issue has substantive triage value.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 45762 is marginally better as the cluster representative because it was created first, but there is no meaningful canonical issue here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Different names and dates, with no description, references, or shared context indicating the same underlying issue." + "reason": "Only superficial name/date similarity; no shared problem description, reproduction, or technical context indicating the same underlying issue." } ] }, @@ -15788,7 +15760,7 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "55ee6551cd9d69cb5bc35db67c2536ee0d9ce188d65a7369315c1bf7ddb30caf", + "prepared_review_unit_hash": "87a1c596742e6e92f12ffc701b249d1bfd84c611b5f33b2579a6b4d541931d87", "component_budget": { "node_count": 2, "item_count": 2, @@ -15810,18 +15782,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues are not duplicates or part of the same bug/change; they are unrelated title-only entries with different names and dates.", - "confidence": 0.99, + "summary": "The two closed issues look like unrelated placeholder/person-name entries with dates in the titles and no body or supporting evidence. There is no indication they describe the same underlying bug or change.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No issue is a good canonical representative here; both are unrelated, body-less placeholder-style issues, so neither should be preferred as a cluster canonical.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different personal names and dates, no body or shared problem description; not the same underlying issue." + "reason": "Different personal-name/date titles with no body, links, or shared context; they do not appear to describe the same issue." } ] }, @@ -15843,7 +15815,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-39895-13", + "cluster_id": "cluster-39895-12", "nodes": [ "pull_request:39895", "pull_request:41855", @@ -15856,8 +15828,7 @@ "pull_request:44827", "pull_request:45170", "pull_request:45213", - "pull_request:45655", - "pull_request:45770" + "pull_request:45655" ], "soft_pairs": [ "pull_request:39895|pull_request:43424", @@ -15868,10 +15839,8 @@ "pull_request:43251|pull_request:44827", "pull_request:43424|pull_request:44827", "pull_request:43424|pull_request:43823", - "pull_request:43823|pull_request:45770", "pull_request:43823|pull_request:45655", "pull_request:43823|pull_request:43995", - "pull_request:43995|pull_request:45770", "pull_request:43995|pull_request:45213", "pull_request:43995|pull_request:44827", "pull_request:44827|pull_request:45170", @@ -15879,182 +15848,156 @@ "pull_request:43823|pull_request:44827", "pull_request:44827|pull_request:45213", "pull_request:43823|pull_request:45213", - "pull_request:45655|pull_request:45770", - "pull_request:44827|pull_request:45770", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "25829ef5c0f193de131203f2bb21e58e34de0d326dc6ca44f01760b74bcb2194", + "prepared_review_unit_hash": "a4f45caad386ab705895e7d478f00493fe983edcbe54d3973209f4cef6a52dd8", "component_budget": { - "node_count": 13, - "item_count": 13, - "soft_pair_count": 22, - "serialized_chars": 36416, - "estimated_input_tokens": 9104, - "estimated_eval_tokens": 18464 + "node_count": 12, + "item_count": 12, + "soft_pair_count": 18, + "serialized_chars": 32870, + "estimated_input_tokens": 8218, + "estimated_eval_tokens": 16692 }, "budget": { - "node_count": 13, - "item_count": 13, - "soft_pair_count": 22, - "serialized_chars": 36416, - "estimated_input_tokens": 9104, - "estimated_eval_tokens": 18464 + "node_count": 12, + "item_count": 12, + "soft_pair_count": 18, + "serialized_chars": 32870, + "estimated_input_tokens": 8218, + "estimated_eval_tokens": 16692 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: model additions, auto/processing refactors, tests, and docs are mixed together. The only true duplicate-like pair is the two fixed_cross_entropy PRs (43251 and 43254); all other soft edges look unrelated and should stay separate.", - "confidence": 0.96, + "summary": "Only one soft-duplicate pair is a true match: the two `fixed_cross_entropy` PRs. The rest are same-subsystem or same-file-neighborhood overlaps, but they target different concrete changes and should stay separate.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 43251 is the best canonical representative: it is explicitly tied to issue 43240 and contains the same concrete `fixed_cross_entropy` change as the near-identical follow-up PR.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "PR 43251 is the strongest cluster anchor because it is issue-linked, directly fixes the underlying loss-kwargs bug, and has the fuller implementation context.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "VideoPrism model addition vs an Executorch dynamic-shape test; different code paths and goals." + "reason": "Different features: VideoPrism model/docs addition vs an ExecuTorch dynamic-shape test. No shared code-path bug." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Mistral tokenizer helper methods vs adding a new MobileLLM model; unrelated areas." + "reason": "Tokenizer utility methods for MistralCommon vs adding a new MobileLLM model. Unrelated changes." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Tokenizer API changes vs Executorch export test coverage; not the same bug or change." + "reason": "Tokenizer method additions vs export dynamic-shape test; same broad infra area only superficially." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Processor auto-loading robustness vs a new model implementation; separate problems." + "reason": "Processor auto-loading robustness vs adding MobileLLM. Different underlying change." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Cross-entropy kwargs fix vs MobileLLM model addition; no shared underlying change." + "reason": "Loss helper kwargs fix vs new model implementation. No shared bug or code-path." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Loss-function API fix vs Mistral4 test fixes; unrelated code paths." + "reason": "Cross-entropy helper fix vs Mistral4 test/model cleanup. Different subsystems and problems." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "Executorch export test vs Mistral4 tests; both are tests, but not the same concrete issue." + "reason": "ExecuTorch export test vs Mistral4 test fixes. Both are tests, but not the same change." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Executorch dynamic-shape export test vs model addition; unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45770", - "accept": false, - "reason": "New MobileLLM model files vs AutoModel text_config unwrapping; different subsystems." + "reason": "Dynamic-shape export test vs MobileLLM model addition. Unrelated." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "MobileLLM model addition vs config resolution-order fix; not the same change." + "reason": "New model scaffold vs class config-resolution ordering fix. Different code paths." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "New model implementation vs Falcon output-collection refactor; unrelated." - }, - { - "left": "pull_request:43995", - "right": "pull_request:45770", - "accept": false, - "reason": "Falcon model refactor vs AutoFactory config unwrapping; different code paths." + "reason": "New MobileLLM model vs Falcon output-interface refactor. Not the same bug/change." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon refactor vs model-creation skill documentation; clearly unrelated." + "reason": "Falcon output-collection refactor vs model-creation skill draft. Completely different purpose." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon refactor vs Mistral4 test fixes; not the same bug or patch." + "reason": "Falcon refactor vs Mistral4 test fix; no concrete shared fix." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fixes vs a layernorm typo rename sweep; unrelated changes." + "reason": "Mistral4 test/model changes vs `layrnorm` renaming in CLIP-like models. Different underlying defect." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4 tests vs cls.config resolution order fix; different issues." + "reason": "Mistral4 tests/model work vs config-resolution ordering fix. No same code-path overlap." }, { "left": "pull_request:43823", "right": "pull_request:44827", "accept": false, - "reason": "MobileLLM model addition vs Mistral4 test fixes; unrelated." + "reason": "New MobileLLM model vs Mistral4 test fixes. Shared auto/model infra is too broad to merge." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "Mistral4 tests vs AI skill/docs for model creation; not a duplicate." + "reason": "Mistral4 fix PR vs draft model-creation skill. Not comparable as a duplicate change." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "Model implementation PR vs workflow skill/documentation PR; unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45770", - "accept": false, - "reason": "MobileLLM model addition vs AutoFactory text_config unwrap; different areas." + "reason": "MobileLLM model addition vs a tooling/skill PR. Different artifact types and goals." }, { - "left": "pull_request:43995", + "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Falcon refactor vs Python 3.14 config-resolution fix; separate concerns." - }, - { - "left": "pull_request:45655", - "right": "pull_request:45770", - "accept": false, - "reason": "Both touch config handling, but one fixes __init_subclass__ resolution order and the other unwraps text_config in AutoModelFor.from_config; not the same underlying bug." + "reason": "Mistral4-specific PR vs generic config-resolution ordering fix. Different concrete issues." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both PRs patch fixed_cross_entropy to forward supported nn.functional.cross_entropy kwargs (weight/label_smoothing) in the same function and code path; they are effectively the same fix." + "reason": "Both PRs change `fixed_cross_entropy` the same way to pass supported kwargs through to `nn.functional.cross_entropy`; they are the same underlying fix." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the cluster is heterogeneous, and the only clearly duplicate-like pair is 43251/43254, which both patch `fixed_cross_entropy` in the same file to forward the same supported `nn.functional.cross_entropy` kwargs. The other soft-edge rejections are conservative and do not overclaim a shared bug or mergeable single-PR fix." + "feedback": "The cluster judgment is conservative and grounded in the packet. The only clear duplicate pair is PR 43251/43254: both modify `src/transformers/loss/loss_utils.py` in the same `fixed_cross_entropy` direction, and both are tied to issue 43240. The other pairs are reasonably rejected as different features or different concrete changes, with no overclaim that they are mergeable into a single bugfix PR. One minor precision note: PR 43254 is also issue-linked, so 43251 is better justified as canonical mainly by having fuller context, not by being uniquely issue-linked." }, "evaluator_used": true, "retried": false, @@ -16081,12 +16024,12 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "5aa00e3b995dedc48a6f9df50c5a085ee1f1dc814b444063a0e42ad708a976e3", + "prepared_review_unit_hash": "59b603c896446edebf869252aa97513be67f9f0441447324a5af3b4807e07bf4", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7226, + "serialized_chars": 7225, "estimated_input_tokens": 1807, "estimated_eval_tokens": 3870 }, @@ -16094,7 +16037,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7226, + "serialized_chars": 7225, "estimated_input_tokens": 1807, "estimated_eval_tokens": 3870 }, @@ -16103,18 +16046,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related VibeVoice PRs, but they are not duplicates: PR 40546 is the base VibeVoice integration, while PR 42781 adds VibeVoice Realtime plus an acoustic tokenizer and other realtime-specific changes. The overlap is mostly model-family plumbing, not the same concrete fix/change.", - "confidence": 0.95, + "summary": "Both PRs are VibeVoice-related, but they target different deliverables: one is the base VibeVoice integration, the other is a draft realtime extension with an acoustic tokenizer. The overlap is broad, but the underlying change is not the same.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 40546 is the better canonical PR because it is the main, broader VibeVoice implementation and has substantial review activity; PR 42781 is a separate realtime-focused extension with additional components and a different scope.", + "canonical_pr_reason": "PR #40546 is the better canonical PR because it is the more complete, non-draft base implementation of VibeVoice and has substantial review activity. It represents the core integration rather than a derivative realtime extension.", "best_issue_reason": null, - "best_pr_reason": "PR 40546 is the best representative of the cluster since it appears to be the foundational VibeVoice addition. PR 42781 is draft and targets a different feature set, so it should not be treated as the same PR.", + "best_pr_reason": "PR #40546 is the best overall representative because it is the main feature landing path, while #42781 is a draft add-on focused on realtime support and an extra tokenizer component.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "Shared VibeVoice files indicate the same model family, but the changes are not the same underlying code-path problem. 40546 implements VibeVoice generally; 42781 adds realtime support and an acoustic tokenizer, which is a separate extension and not plausibly mergeable as one duplicate PR." + "reason": "Shared VibeVoice files are not enough to treat these as duplicates. #40546 is the base VibeVoice implementation, while #42781 is a draft for VibeVoice Realtime with additional acoustic-tokenizer and generation changes. They address different feature scopes and would not plausibly be merged into one PR as the same change." } ] }, @@ -16144,12 +16087,12 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "29ac0ff1a0017db82c84920c2cfc934dcc5d17751ac02c8874fddd77931d5134", + "prepared_review_unit_hash": "16c609975c1dd9165250ffca7dcc4646c77a05306e3ebf8dceb21901cd66523f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4675, + "serialized_chars": 4676, "estimated_input_tokens": 1169, "estimated_eval_tokens": 2594 }, @@ -16157,7 +16100,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4675, + "serialized_chars": 4676, "estimated_input_tokens": 1169, "estimated_eval_tokens": 2594 }, @@ -16166,18 +16109,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests target completely different models and code paths: TimesFM covariate forecasting vs. DINOv3 image classification support. They should not be clustered as duplicates.", + "summary": "The two PRs are unrelated feature additions for different model families: TimesFM covariates forecasting vs. DINOv3 image-classification support. They do not appear to fix the same underlying bug or implement the same change.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "[TimesFM] Add support for forecasting with covariates is the broader, more self-contained PR in this cluster and is the better representative if a single PR must be chosen.", + "canonical_pr_reason": "No canonical PR: the PRs target different code paths, different models, and different user-facing features.", "best_issue_reason": null, - "best_pr_reason": "[TimesFM] Add support for forecasting with covariates is the best overall representative because it has the larger, more substantive model change and clearer feature scope.", + "best_pr_reason": "No single best PR for deduping this cluster; each is a distinct feature request with no meaningful overlap.", "soft_edge_verdicts": [ { - "left": "[TimesFM] Add support for forecasting with covariates", - "right": "Add DINOv3ViTForImageClassification support", + "left": "pull_request:40755", + "right": "pull_request:41224", "accept": false, - "reason": "Different model families, different tasks, and different implementation paths; they do not fix the same underlying bug or add the same feature." + "reason": "Different subsystems and model families (TimesFM forecasting with covariates vs. DINOv3 image classification); they are not the same change and would not be merged into one PR." } ] }, @@ -16262,48 +16205,56 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-41797-2", + "cluster_id": "cluster-41797-3", "nodes": [ "pull_request:41797", - "pull_request:45186" + "pull_request:45186", + "pull_request:45630" ], "soft_pairs": [ - "pull_request:41797|pull_request:45186" + "pull_request:41797|pull_request:45186", + "pull_request:45186|pull_request:45630" ], - "prepared_review_unit_hash": "dd632b20348ef0297b4280709455a2d6ca01c645f51e1ca897e30a61eae44643", + "prepared_review_unit_hash": "e2e1599f247a2a47c0e49535575ac6ea7da26485a3675ab257915d685be72daf", "component_budget": { - "node_count": 2, - "item_count": 2, - "soft_pair_count": 1, - "serialized_chars": 6722, - "estimated_input_tokens": 1681, - "estimated_eval_tokens": 3618 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10427, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 }, "budget": { - "node_count": 2, - "item_count": 2, - "soft_pair_count": 1, - "serialized_chars": 6722, - "estimated_input_tokens": 1681, - "estimated_eval_tokens": 3618 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10427, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate model-addition PRs (DeepSeek OCR vs Isaac) that share framework registration plumbing but do not fix the same code-path or bug.", - "confidence": 0.97, + "summary": "These are three unrelated new-model onboarding PRs (DeepSeek OCR, Isaac, and Kimi2-6). They share common integration boilerplate and registry/docs changes, but they do not fix the same underlying bug or implement the same change, so none should be clustered as duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical PR: the PRs introduce different model families with distinct codepaths and files, so there is no single duplicate target.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a duplicate of the other. If forced to pick one representative, 45186 is the cleaner model-addition PR, but it is not a canonical merge target for this cluster because the changes are unrelated.", + "best_pr_reason": "No best PR for duplicate triage: each PR is a separate model addition, not a mergeable duplicate of the others. If forced to pick a representative, none is clearly canonical over the rest.", "soft_edge_verdicts": [ { "left": "pull_request:41797", "right": "pull_request:45186", "accept": false, - "reason": "Different model families and separate integrations. The overlap is only in shared Transformers scaffolding (auto registries, docs, repo checks), not the same concrete bug or change, so they should not be merged as duplicates." + "reason": "Different model additions (DeepSeek OCR vs Isaac); shared auto-registry/docs plumbing is boilerplate, not the same concrete change." + }, + { + "left": "pull_request:45186", + "right": "pull_request:45630", + "accept": false, + "reason": "Different model additions (Isaac vs Kimi2-6); overlap is only broad model-integration infrastructure, so they are not duplicates." } ] }, @@ -16333,7 +16284,7 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "4134162635f963596c7da92e60ed426224de5a5084e030abfa8d2d12d1a4b273", + "prepared_review_unit_hash": "df2697848dc835f3c7d8d56f35d77438b90ec1c5b8040eb50d02afb96c4a62e3", "component_budget": { "node_count": 2, "item_count": 2, @@ -16355,18 +16306,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two workflow-maintenance PRs overlap in several GitHub Actions files, but they make different upgrades: PR 42975 is a broad Node 24 compatibility bump for core actions, while PR 42976 is a separate latest-version update centered on docker/build-push-action and related workflows.", + "summary": "Both PRs are broad GitHub Actions dependency updates, but they address different action families and different maintenance goals, so they are not duplicates.", "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the clearest representative of the broader GitHub Actions compatibility upgrade; PR 42976 is a separate maintenance change focused on different action versions, so they are not duplicates.", + "canonical_pr_reason": "PR 42975 is the better canonical representative because it is the more focused, clearly scoped workflow-actions compatibility update; PR 42976 is a separate upgrade pass for different action versions.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the better representative for the cluster because it has a clear, specific rationale (Node 24 compatibility) and the broader workflow surface; PR 42976 is a distinct action-version bump rather than the same change.", + "best_pr_reason": "PR 42975 is the best overall representative of the cluster since it covers the more central Node 24 compatibility theme across many workflows.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Shared workflow filenames alone are not enough: 42975 updates actions for Node 24 compatibility, while 42976 upgrades docker/build-push-action and other actions to newer versions. Different concrete maintenance changes, so not the same underlying PR." + "reason": "Do not treat as duplicates: 42975 upgrades core GitHub Actions for Node 24 compatibility (checkout/cache/upload-artifact), while 42976 bumps docker/build-push-action and related workflow actions to newer releases. They overlap only at the broad 'workflow maintenance' level, not the same underlying change or code-path problem." } ] }, @@ -16388,13 +16339,12 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-42978-5", + "cluster_id": "cluster-42978-4", "nodes": [ "pull_request:42978", "pull_request:43448", "pull_request:43451", - "pull_request:45490", - "pull_request:45630" + "pull_request:45490" ], "soft_pairs": [ "pull_request:42978|pull_request:43448", @@ -16402,33 +16352,32 @@ "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:45490|pull_request:45630" + "pull_request:43451|pull_request:45490" ], - "prepared_review_unit_hash": "5ef69bba952317c3a888a426673ef7d4d9f91a1826e2112882f73ebb15d3da71", + "prepared_review_unit_hash": "0cf06da84e6ce17eb5573f3679a6b75c21e123d895510d8f9c43efca4e4d54f8", "component_budget": { - "node_count": 5, - "item_count": 5, - "soft_pair_count": 7, - "serialized_chars": 17616, - "estimated_input_tokens": 4404, - "estimated_eval_tokens": 9064 + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14079, + "estimated_input_tokens": 3520, + "estimated_eval_tokens": 7296 }, "budget": { - "node_count": 5, - "item_count": 5, - "soft_pair_count": 7, - "serialized_chars": 17616, - "estimated_input_tokens": 4404, - "estimated_eval_tokens": 9064 + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14079, + "estimated_input_tokens": 3520, + "estimated_eval_tokens": 7296 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are separate model-addition PRs, not duplicates. They only overlap on shared boilerplate areas like docs and auto-registry updates, but each targets a different model/code path (ViT NEPA, Molmo, Molmo2, CTSM, Kimi2-6).", - "confidence": 0.99, + "summary": "All four items are separate model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM). They share a common template/scaffold pattern, but they target different models, code paths, and docs, so they are not duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, @@ -16438,43 +16387,37 @@ "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model additions (ViT NEPA vs Molmo); shared docs/auto wiring is boilerplate, not the same fix." + "reason": "Different model families (ViT NEPA vs Molmo) with different files and implementations; shared docs/auto-registration scaffolding is not enough to treat them as the same change." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "ViT NEPA and Molmo2 are distinct architectures with different files and tests; not the same underlying change." + "reason": "ViT NEPA and Molmo2 are unrelated model additions; they do not fix the same code-path problem and could not be merged into one PR." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "These add different models (ViT NEPA vs CTSM) and do not fix one shared bug or implement one shared feature." + "reason": "ViT NEPA and CTSM are distinct model integrations with different modeling and conversion code, so this is not a duplicate." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are separate model families; overlap is limited to common integration scaffolding." + "reason": "Molmo and Molmo2 are separate model variants with different implementation files and capabilities; similar naming and shared auto-mapping/docs are not sufficient to merge as duplicates." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo and CTSM are unrelated model additions; similar registry/docs edits are not enough to merge." + "reason": "Molmo and CTSM are unrelated model additions; overlapping registry/docs edits are incidental and do not indicate the same underlying bug or change." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 and CTSM touch different model implementations and test suites, so they are not duplicates." - }, - { - "left": "pull_request:45490", - "right": "pull_request:45630", - "accept": false, - "reason": "CTSM and Kimi2-6 are different model additions with different code paths; they should remain separate PRs." + "reason": "Molmo2 and CTSM are different model PRs with different code paths and test suites; they are not the same underlying fix or feature." } ] }, @@ -16512,7 +16455,7 @@ "pull_request:43665|pull_request:45749", "pull_request:44815|pull_request:45747" ], - "prepared_review_unit_hash": "71667fba30a134eafec6910aed568191e2ede3cd51942216d6011056408ec2eb", + "prepared_review_unit_hash": "a08203b030e6cfd4ba626e78e6254c1a5e22fb2864ddafecfa5c40dde8b67bbd", "component_budget": { "node_count": 5, "item_count": 5, @@ -16534,48 +16477,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are only superficially similar (mostly small fixes with generic titles). They do not appear to address the same underlying bug or change, so the soft links should be rejected. Among them, the FP8/dequant work is the strongest standalone code fix.", - "confidence": 0.9, + "summary": "These PRs are mostly unrelated: one substantive FP8/dequant code fix, one Whisper test/tokenizer fix, one test-only placeholder patch, one vague batch-size fix with no diff shown, and one docstring typo. No issue is present; the strongest PR is the FP8/dequant fix.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44815 is the most substantive, concrete code fix in the set: it changes loading and FP8 integration logic across multiple files, with tests, and has a clear bug-fix scope compared with the others.", + "canonical_pr_reason": "PR 44815 is the clearest substantive bug fix, with real code-path changes in model loading and FP8 quantization integration; the other PRs are test-only, vague, or documentation-only.", "best_issue_reason": null, - "best_pr_reason": "PR #44815 is the best representative of the cluster because it has the clearest technical problem statement and the most meaningful implementation, while the other PRs are either test-only, cosmetic, or too underspecified.", + "best_pr_reason": "PR 44815 is the best standalone change to keep because it addresses an actual runtime/dequantization path rather than test scaffolding or wording changes.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Different subsystems and different behavior: Whisper tokenizer decoding vs CLIP/ViT tests. Similarity is only in being small fixes, not the same bug." + "reason": "Whisper tokenizer/test fix vs. CLIP/ViT test placeholders; different models and different kinds of changes." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Unrelated changes: Whisper decoding logic/tests versus FP8 dequant/loading code. No shared code path or concrete defect." + "reason": "Whisper decoding/tests are unrelated to FP8 model-loading/dequantization changes." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "CLIP/ViT test edits are not the same underlying issue as FP8 loading/dequant changes. Shared 'fix' wording is too weak to merge." + "reason": "Test-only assertions in vision tests are not the same underlying bug as the FP8/dequant code fix." }, { "left": "pull_request:43665", "right": "pull_request:45747", "accept": false, - "reason": "No evidence of a shared bug or code path; one is test-file changes, the other is an unspecified batch-size fix." + "reason": "Test stubs in CLIP/ViT are unrelated to an unspecified split-batch-size fix." }, { "left": "pull_request:43665", "right": "pull_request:45749", "accept": false, - "reason": "Test edits in vision models and a docstring spelling correction are clearly unrelated changes." + "reason": "Vision test placeholders do not match a continuous_batching docstring spelling correction." }, { "left": "pull_request:44815", "right": "pull_request:45747", "accept": false, - "reason": "FP8 loading/dequant changes and a split-batch-size fix are different problems with no concrete overlap." + "reason": "FP8/dequant logic and split-batch-size behavior are different code paths with no shared concrete bug." } ] }, @@ -16605,7 +16548,7 @@ "soft_pairs": [ "pull_request:43488|pull_request:43532" ], - "prepared_review_unit_hash": "bc87674bcfacdf3a6004bbc7ad3e7a28853a54b7424532c30efc671742a872c0", + "prepared_review_unit_hash": "fce4bd253549e3d2542bcb04d5b82834f4a4aaac7cda7fdbc14b9410fbc18502", "component_budget": { "node_count": 2, "item_count": 2, @@ -16627,18 +16570,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are not duplicates: #43488 is a small bot-format/versioning test change around setup.py and Vit test scaffolding, while #43532 rewires several GitHub workflows plus helper/notification code. The shared Vit test filename is incidental, not evidence of the same underlying fix.", - "confidence": 0.97, + "summary": "These two PRs are both test/bot-oriented, but they change different areas and do not appear to fix the same concrete problem. The shared Vit test file looks incidental rather than evidence of duplication.", + "confidence": 0.79, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs target different code paths and different maintenance purposes, so they should not be merged into one duplicate cluster.", + "canonical_pr_reason": "No clear duplicate canonical PR here; the two PRs target different bot/test concerns. If one must be chosen as the cluster representative, PR #43532 is the better fit because it contains the more substantive workflow/bot behavior changes, while PR #43488 is mainly a narrow repo-bot formatting check.", "best_issue_reason": null, - "best_pr_reason": "If one PR must represent the cluster, #43532 is the broader and more substantive change, but it is still not a duplicate of #43488.", + "best_pr_reason": "PR #43532 is the stronger representative for the cluster because it touches the CI/comment/diff-display machinery directly and has broader behavioral impact. PR #43488 is a small synthetic setup/version tweak and is less representative of the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "Different underlying changes: #43488 is a setup/versioning and test-bot check, while #43532 changes CI/workflow behavior and notification utilities. They do not fix the same concrete problem and would not be merged as one PR." + "reason": "Reject: the diffs are unrelated in substance. #43488 changes setup/versioning and a vit test as a repo-bot formatting check, while #43532 changes GitHub workflows, testing utilities, and notification code for diff/comment behavior. Shared test-file overlap is incidental, not the same underlying fix." } ] }, @@ -16668,12 +16611,12 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "650d45821801a3666ffaa3cfe5b0bb77aad4f96932887409a164d4257e43d21b", + "prepared_review_unit_hash": "a052fb4292f19ede9cfa7770fa2917f918e9733c6c0719e15d47db3c9f487336", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6083, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -16681,7 +16624,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6083, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -16690,18 +16633,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs add similarly named but distinct pipelines: one for promptable concept segmentation and one for promptable visual segmentation. They share repository scaffolding, but they target different feature sets and model integrations, so they should not be treated as duplicates.", + "summary": "These are two distinct feature PRs: one adds Promptable Concept Segmentation, the other Promptable Visual Segmentation. They share scaffolding and docs patterns, but not the same underlying pipeline or code-path change.", "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the two PRs implement different pipelines with different model backends and task docs, not the same underlying change.", + "canonical_pr_reason": "No canonical PR: the PRs implement different segmentation pipelines with different model/test coverage and should not be collapsed into one duplicate target.", "best_issue_reason": null, - "best_pr_reason": "No single best PR for deduplication; each PR is the primary candidate for its own distinct feature area.", + "best_pr_reason": "No single best PR for this cluster; both are valid but separate feature additions, so neither should represent the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Reject: these are separate pipeline additions. One introduces Promptable Concept Segmentation (SAM3-focused), the other Promptable Visual Segmentation (SAM/SAM2/EdgeTam/tracker-focused). Shared docs/registry edits do not make them the same code-path fix or something that could be merged into one PR." + "reason": "Different features and implementations: concept segmentation vs visual segmentation, with different pipeline files, model integrations, and tests. Shared docs/registry scaffolding is not enough to treat them as the same change." } ] }, @@ -16731,7 +16674,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "e69588df2d965c49574fbf07583655a432c49b0a7145660ae1d99720ff7458ba", + "prepared_review_unit_hash": "48e6916cb0f2bd039725ebc5d1d0fd2633f12073bbf52e64417ba298a7c82304", "component_budget": { "node_count": 2, "item_count": 2, @@ -16753,18 +16696,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items are open PRs in `trainer.py`, but they address different Trainer extensibility points: one changes gradient-accumulation loss scaling, the other adds an override hook for signature-column filtering. They are not the same fix and should not be clustered as duplicates.", - "confidence": 0.96, + "summary": "Both PRs touch `src/transformers/trainer.py`, but they implement different Trainer extension hooks for different code paths: one for gradient-accumulation loss scaling, the other for signature/column filtering. They are related by subsystem, not the same change or bug.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR stands out: `#43651` and `#43663` modify different code paths and solve unrelated Trainer customization problems.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "Neither PR is a good global representative for the other; each targets a separate behavior change in `Trainer` rather than a shared underlying bug.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Reject: same file, but different functionality and code paths. `#43651` is about loss normalization under gradient accumulation; `#43663` is about extracting signature columns for filtering. They are not plausibly merged as one fix." + "reason": "Different underlying Trainer behaviors: `_loss_is_scaled_for_ga` changes loss scaling during training, while `_get_signature_columns` changes input column filtering based on model signature. Shared file alone is not enough to treat them as duplicates." } ] }, @@ -16796,12 +16739,12 @@ "pull_request:43656|pull_request:43842", "pull_request:43836|pull_request:43842" ], - "prepared_review_unit_hash": "bc99fb03682d89a03d7b4e6e342bd0d78c4215353840669a655441578aacfafc", + "prepared_review_unit_hash": "b377c268cc62d0fe86c8200158b61474990b70a20461a7f419286242d5a97559", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 5992, + "serialized_chars": 5991, "estimated_input_tokens": 1498, "estimated_eval_tokens": 3252 }, @@ -16809,7 +16752,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 5992, + "serialized_chars": 5991, "estimated_input_tokens": 1498, "estimated_eval_tokens": 3252 }, @@ -16818,30 +16761,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three PRs target the same `src/transformers/cli/serve.py` runtime annotation/TypeAdapter NameError in the CLI server. PR 43842 is the cleanest, most targeted fix; the other two are broader or alternate annotation-workaround variants.", - "confidence": 0.86, + "summary": "All three PRs point at the same runtime-annotation problem in `src/transformers/cli/serve.py`, but `43842` is the most direct and minimal fix. The other two appear to be broader annotation-cleanup variants of the same bug, not separate changes.", + "confidence": 0.8, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43842 is the most precise fix: it guards `TypeAdapter` behind `TYPE_CHECKING` and quotes the affected annotation, directly addressing the reported pydantic-missing NameError with minimal scope.", + "canonical_pr_reason": "PR 43842 is the clearest representative: it directly fixes the `TypeAdapter` NameError by moving the import under `TYPE_CHECKING` and stringifying the annotation, with a focused diff tied to the reported CLI crash.", "best_issue_reason": null, - "best_pr_reason": "PR 43842 best matches the concrete crash path and has the smallest, clearest patch while still fixing the underlying runtime NameError.", + "best_pr_reason": "43842 is the best overall PR to keep because it is the most targeted fix for the concrete `TypeAdapter` NameError and is less entangled with broader typing refactors than the others.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": true, - "reason": "Both are fixes for the same CLI server annotation NameError around `TypeAdapter`/forward references in `serve.py`; they address the same underlying runtime problem and could be merged into one PR." + "reason": "Both PRs fix the same `serve.py` annotation-evaluation crash in the CLI; `43656` is a broader typing-postponement variant of the same runtime NameError fix." }, { "left": "pull_request:43836", "right": "pull_request:43842", "accept": true, - "reason": "Both patch the same `serve.py` `TypeAdapter` NameError path with annotation-only changes; they are alternative implementations of the same bug fix and could plausibly be combined." + "reason": "Both target the same `TypeAdapter`/annotation NameError in `src/transformers/cli/serve.py` and could plausibly be combined into one patch for the same code path." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: all three PRs touch `src/transformers/cli/serve.py` and are aimed at preventing annotation-related `NameError`s, with PR 43842 being the narrowest fix. The soft-edge pair verdicts are also conservative enough: 43836 and 43842 clearly address the same `TypeAdapter` annotation issue, and 43656 is plausibly another implementation of the same underlying forward-reference/runtime annotation problem in the same file." + "feedback": "The summary and PR selection are grounded in the packet: 43842 is indeed the most targeted fix, and the pairwise merges are plausibly the same serve.py annotation/NameError bug family and look mergeable. The wording is slightly broad for 43656, but not enough to reject." }, "evaluator_used": true, "retried": false, @@ -16977,16 +16920,16 @@ "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:44116", "pull_request:44018|pull_request:44086", "pull_request:44018|pull_request:44085", "pull_request:44068|pull_request:44086", "pull_request:44071|pull_request:44085", "pull_request:44059|pull_request:44086", "pull_request:44085|pull_request:44086", - "pull_request:44098|pull_request:44101", "pull_request:44059|pull_request:44722", + "pull_request:44098|pull_request:44101", "pull_request:44018|pull_request:44071", "pull_request:44017|pull_request:44161", "pull_request:44017|pull_request:44722", @@ -17031,20 +16974,20 @@ "pull_request:44044|pull_request:44154", "pull_request:44024|pull_request:44076" ], - "prepared_review_unit_hash": "7ae479abfc365a6763527dbfc1f0619b7ea1781d5a75fa633416ca225bff432b", + "prepared_review_unit_hash": "d77b78cb77b3f77a64dce8d49d1a8d4bf0d55fe7821c0996f56eeab5ed8aa1da", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84521, - "estimated_input_tokens": 21131, - "estimated_eval_tokens": 42518 + "serialized_chars": 84519, + "estimated_input_tokens": 21130, + "estimated_eval_tokens": 42516 }, "budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 72, - "serialized_chars": 76955, + "serialized_chars": 76953, "estimated_input_tokens": 19239, "estimated_eval_tokens": 38734 }, @@ -17053,318 +16996,372 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of independent model-specific output-tracing refactors tied to the same tracking issue, not one duplicate bug. The only strong duplicate/superseding pairs are within the same model family and concrete code path (notably GPT-Neo, and GPT-J/CodeGen).", - "confidence": 0.9, + "summary": "This cluster is mostly a broad wave of model-by-model output-tracing refactors around the same umbrella tracking issue, but most PRs are not duplicates of each other because they target different model implementations. The only clear duplicate subclusters are the GPT-J/CodeGen refactors and the GPT-Neo pair.", + "confidence": 0.83, "canonical_issue_reason": null, - "canonical_pr_reason": "44722 is the best canonical PR for the cluster\u2019s main theme: it is a later, concrete GPT-J/CodeGen output-tracing refactor with the same issue target and the clearest, most complete code-path change among the overlapping GPT-J work.", + "canonical_pr_reason": "PR 44722 is the strongest representative of the overlapping GPT-J/CodeGen refactor set: it explicitly targets the tracking issue, touches the same GPT-J code path as the narrower GPT-J-only PRs, and is the most complete version of that concrete change.", "best_issue_reason": null, - "best_pr_reason": "44722 is the strongest representative PR: it has explicit issue linkage, touches the actual implementation files, and most clearly reflects the standardized output-tracing refactor pattern seen across the cluster.", + "best_pr_reason": "44722 is the best single PR anchor because it most fully captures the duplicate GPT-J refactor family and subsumes the narrower GPT-J-only variants.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different models/code paths (CVT/FNet vs GPT-J/CodeGen); same tracking issue is not enough." + "reason": "Different model families and different code paths; sharing the umbrella output-tracing issue is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different model families and different implementation paths; only the output-tracing theme overlaps." + "reason": "CVT/FNet vs DeBERTa-v2 are separate implementations of the same refactor theme, not the same change." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Same GPT-J output-tracing refactor in the same file family; one is effectively a variant/superset of the other." + "reason": "Both are GPT-J output-tracing refactors on the same file; 44066 is just a broader version that also includes CodeGen." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "ResNet vs EfficientNet are separate model implementations, so not the same underlying change." + "reason": "RegNet/ResNet-style CV refactor vs EfficientNet; same umbrella topic, different underlying change." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "Different models and code paths; shared issue target alone does not make them duplicates." + "reason": "EfficientNet output tracing and GPT-J/CodeGen output tracing are unrelated code paths." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen refactor vs EfficientNet refactor; different underlying change." + "reason": "GPT-J/CodeGen vs EfficientNet; only the general refactor theme overlaps." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 vs DeBERTa V2 are unrelated implementations." + "reason": "MobileNetV2 and DeBERTa-v2 are separate model implementations, not duplicate fixes." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen output tracing vs MGP-STR output tracing; different code paths." + "reason": "GPT-J/CodeGen refactor vs MGP-STR refactor; same pattern, different model code." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both are GPT-Neo output-tracing refactors over the same file and tests; 44068 looks like a later superseding version." + "reason": "Same GPT-Neo output-tracing refactor on the same file/test area; this looks like a rebased or revised duplicate." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen vs MPT; separate model implementations." + "reason": "GPT-J/CodeGen and MPT are unrelated concrete implementations." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "Different models (GPT-J/CodeGen vs GPT-Neo), so not the same concrete change." + "reason": "GPT-J/CodeGen and GPT-Neo are different model changes." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "Different model families; only the output-tracing migration pattern is shared." + "reason": "GPT-Neo vs GPT-J/CodeGen; same umbrella issue only." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen output tracing." + "reason": "ResNet and GPT-J/CodeGen are different model-specific refactors." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR vs GPT-J/CodeGen are different model-specific changes." + "reason": "MGP-STR and GPT-J/CodeGen are unrelated code paths." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet vs GPT-J; separate code paths." + "reason": "ResNet and GPT-J are distinct model implementations." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated model implementations." + "reason": "GPT-Neo and ResNet are separate refactors, not the same bug fix." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet vs MPT; different underlying refactors." + "reason": "ResNet and MPT are different model code paths." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo vs Flaubert/XLM output tracing are different model changes." + "reason": "GPT-Neo and Flaubert are unrelated model changes." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo is a different model/code path from GPT-J/CodeGen." + "reason": "GPT-Neo and GPT-J/CodeGen are different concrete implementations." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo vs Flaubert/XLM are unrelated implementations." - }, - { - "left": "pull_request:44071", - "right": "pull_request:44722", - "accept": false, - "reason": "MPT vs GPT-J/CodeGen are different concrete changes." + "reason": "GPT-Neo and Flaubert are not the same underlying fix." }, { "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo vs MGP-STR; different model code paths." + "reason": "GPT-Neo and MGP-STR are separate model refactors." }, { "left": "pull_request:44018", "right": "pull_request:44085", "accept": false, - "reason": "GPT-Neo vs GPT-J/CodeGen are not the same underlying change." + "reason": "GPT-Neo and GPT-J are different code paths." }, { "left": "pull_request:44068", "right": "pull_request:44086", "accept": false, - "reason": "GPT-Neo and MGP-STR are different model-specific refactors." + "reason": "GPT-Neo and MGP-STR are unrelated model implementations." }, { "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 vs MGP-STR; same output-tracing theme but different code paths." + "reason": "GPT-2 and MGP-STR are different models; shared output-tracing theme is too broad." }, { "left": "pull_request:44085", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J/CodeGen vs MGP-STR are unrelated model changes." + "reason": "GPT-J and MGP-STR do not fix the same concrete code path." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-2 and GPT-J/CodeGen are separate model refactors." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT vs Flaubert/XLM are different model families." + "reason": "ViLT and Flaubert are different model implementations." }, { - "left": "pull_request:44059", + "left": "pull_request:44018", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated concrete changes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are distinct model code paths." + }, + { + "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 refactor is separate from GPT-J/CodeGen." + "reason": "SegFormer and GPT-J/CodeGen are different underlying changes." }, { - "left": "pull_request:44071", - "right": "pull_request:44085", + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "SegFormer and VisualBert are separate refactors." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are not the same model bug or change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "RegNet and VisualBert are unrelated concrete implementations." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBert and Flaubert are different model refactors." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are separate model code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-2 and GPT-Neo are different model-specific changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert are unrelated concrete changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", "accept": false, - "reason": "MPT vs GPT-J; different underlying changes." + "reason": "MPNet and LongT5 are different model implementations." }, { "left": "pull_request:43997", "right": "pull_request:44017", "accept": false, - "reason": "RegNet vs Segformer are unrelated model implementations." + "reason": "RegNet and SegFormer are separate CV model refactors, not duplicates." }, { "left": "pull_request:43997", - "right": "44056", + "right": "pull_request:44056", "accept": false, - "reason": "RegNet vs MPNet are separate code paths." + "reason": "RegNet and MPNet are distinct code paths." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT vs ImageGPT are different models; shared output-tracing pattern is too broad." + "reason": "MPT and ImageGPT are different model families." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet vs Flaubert/XLM are unrelated changes." + "reason": "TextNet and Flaubert are unrelated model changes." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet vs GPT-Neo are different model-specific refactors." + "reason": "RegNet and GPT-Neo are separate implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBERT and MPT are unrelated model code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBERT and VisualBert are different models." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo and SpeechT5 are not the same underlying change." }, { "left": "pull_request:44071", "right": "pull_request:44129", "accept": false, - "reason": "MPT vs SpeechT5 are separate code paths." + "reason": "MPT and SpeechT5 are unrelated refactors." }, { "left": "pull_request:43996", "right": "pull_request:44161", "accept": false, - "reason": "CVT/FNet vs LongT5 are unrelated model changes." + "reason": "CVT/FNet and LongT5 are completely different model code paths." }, { "left": "pull_request:43996", "right": "pull_request:44074", "accept": false, - "reason": "CVT/FNet vs TextNet are different underlying changes." + "reason": "CVT/FNet and TextNet are separate model refactors." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBert vs DeBERTa V2 are separate model implementations." + "reason": "SqueezeBERT and DeBERTa-v2 are different model implementations." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa V2 vs TextNet are unrelated changes." + "reason": "DeBERTa-v2 and TextNet are unrelated changes." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are different concrete changes." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J/CodeGen output-tracing refactor family; 44722 is a broader later version of the same change." + "reason": "Same GPT-J refactor family; 44722 is the broader successor that also includes CodeGen." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both are GPT-J output-tracing refactors on the same code path; 44722 appears to supersede/extend the earlier PR." + "reason": "Both touch the same GPT-J output-tracing code path; 44722 subsumes the narrower GPT-J-only PR." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J vs SpeechT5 are unrelated model changes." + "reason": "GPT-J and SpeechT5 are unrelated model changes." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "CVT/FNet vs ResNet are different model-specific refactors." + "reason": "CVT/FNet and RegNet/ResNet are different model-specific output-tracing refactors." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "CVT/FNet vs SpeechT5 are different underlying changes." + "reason": "CVT/FNet and SpeechT5 do not share the same underlying change." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa V2 vs SpeechT5 are unrelated model changes." + "reason": "DeBERTa-v2 and SpeechT5 are separate model refactors." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa V2 vs Vits are separate model-specific refactors." - }, - { - "left": "pull_request:44024", - "right": "pull_request:44076", - "accept": false, - "reason": "FocalNet vs ImageGPT are different models; same refactor style does not make them duplicates." - }, - { - "left": "pull_request:44017", - "right": "pull_request:44161", - "accept": false, - "reason": "Segformer vs LongT5 are unrelated code paths." - }, - { - "left": "pull_request:44019", - "right": "pull_request:44101", - "accept": false, - "reason": "ResNet vs Flaubert/XLM are different model implementations." - }, - { - "left": "pull_request:44018", - "right": "pull_request:44068", - "accept": true, - "reason": "Same GPT-Neo output-tracing refactor with the same file/test surface; likely a later duplicate/superseding PR." + "reason": "DeBERTa-v2 and ViTS are different code paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the conservative theme distinction is appropriate: most PRs are separate model-specific output-tracing refactors sharing only the tracking issue. The accepted soft-edge pairs are limited to same-family, same-code-path overlaps (GPT-Neo and GPT-J/CodeGen), which do look mergeable as one bugfix PR. I did not see any clear overstatement of evidence." + "feedback": "The summary is conservative and grounded in the packet. The accepted soft-edge pairs are plausible duplicates/revisions of the same model-specific output-tracing refactors, and the rejected pairs correctly stay within separate model code paths rather than overclaiming duplication. I don\u2019t see evidence of an unsupported mergeability claim." }, "evaluator_used": true, "retried": false, @@ -17399,14 +17396,14 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "52fdf146c1522d186152b3265164bfc207c5c6a49734db5fae454e7c6b8a1ab2", + "prepared_review_unit_hash": "f901b03a01f140e3364665fc7e8eca313849048875bd8d84cf87e919f494a772", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84521, - "estimated_input_tokens": 21131, - "estimated_eval_tokens": 42518 + "serialized_chars": 84519, + "estimated_input_tokens": 21130, + "estimated_eval_tokens": 42516 }, "budget": { "node_count": 6, @@ -17421,42 +17418,42 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "A cluster of similar-sounding but model-specific refactors to migrate output tracing/capture across different Transformer implementations. They share a pattern, not the same concrete bug or change.", - "confidence": 0.69, + "summary": "These are model-specific PRs all refactoring output tracing/capture to the new standardized decorator pattern, but they target different architectures and code paths, so they are not duplicates of one another.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44129 is the best representative: it has an explicit issue link, a clear title, and a substantive output-capture refactor in one model.", + "canonical_pr_reason": "PR 44129 is the strongest representative: it has an explicit issue target and a concrete SpeechT5 output-tracing refactor, making it the clearest anchor for this thematic cluster.", "best_issue_reason": null, - "best_pr_reason": "PR 44129 is the strongest single PR to represent the cluster because it is the most concretely scoped and best tracked, but it is still model-specific rather than duplicate of the others.", + "best_pr_reason": "PR 44129 is the best overall representative because it is the most concretely scoped and the only one tied to a tracking issue; however, it is still model-specific rather than a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Both are output-tracing refactors, but for different models (VisualBert vs TextNet) and different code paths; not the same change." + "reason": "Both are output-tracing refactors, but one is VisualBert and the other is TextNet; they touch different model implementations and are not the same underlying fix." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are separate implementations with independent decorator migrations, not one shared fix." + "reason": "Same broad refactor theme, but MPNet and VisualBert are distinct code paths with model-specific changes, so they should not be merged as duplicates." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "MPNet vs TextNet are unrelated model code paths; this is only a thematic similarity." + "reason": "These are separate MPNet and TextNet refactors of the standardized output-capture pattern, not one concrete bug or change." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert refactor different model internals; they do not fix the same concrete issue." + "reason": "RegNet and SqueezeBert are unrelated model implementations; the similar decorator migration is only thematic, not a duplicate fix." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "These touch different models and output APIs; they are separate migrations, not mergeable duplicate PRs." + "reason": "Both modernize output tracing, but SqueezeBert and SpeechT5 modify different model-specific forward paths and are not plausibly one combined PR." } ] }, @@ -17536,274 +17533,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "c3919d02f3fbf8f474c9c504e44a56d4d295f4af20d359637347970b2e3071b6", + "prepared_review_unit_hash": "229b8e7c3e703507881ab74464325b48b5245d26c4b31ce334e72385559bd30e", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40161, - "estimated_input_tokens": 10041, - "estimated_eval_tokens": 20338 + "serialized_chars": 40174, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40161, - "estimated_input_tokens": 10041, - "estimated_eval_tokens": 20338 + "serialized_chars": 40174, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a set of similar-but-separate output-tracing refactors across different model families. They share a theme, but they are not duplicates of the same underlying bug/change.", + "summary": "These PRs are all part of a broad output-tracing refactor initiative, but they target different model families and code paths. They are thematically similar, not duplicates of the same fix/change.", "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44025 is the cleanest representative of the cluster: it applies the same output-tracing refactor across two closely related Depth Anything files and is a focused, coherent change.", + "canonical_pr_reason": "PR 44025 is a good cluster representative because it makes the output-tracing transition explicit across two closely related model files and shows the common can_return_tuple pattern seen throughout the set.", "best_issue_reason": null, - "best_pr_reason": "PR 44025 is the best overall representative for triage because it is narrow, self-contained, and clearly exemplifies the cluster\u2019s output-tracing modernization.", + "best_pr_reason": "PR 44025 is the best single representative of the cluster\u2019s recurring refactor pattern, with a concrete multi-file implementation and clear output-tracing scope.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model families and files (timm_backbone vs upernet); same refactor pattern, not the same concrete change." + "reason": "Same broad refactor theme, but different model families and separate code paths; not one mergeable change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different model families (timm_backbone vs vision_text_dual_encoder); only the output-tracing theme matches." + "reason": "Both are output-tracing refactors, but they touch unrelated model implementations and would not be merged as one fix." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Different architectures and files; both are refactors of return handling, not one underlying fix." + "reason": "Shared refactor pattern only; timm_backbone and univnet are distinct code paths." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "timm_backbone and mamba/falcon_mamba are unrelated code paths; shared refactor motif is too broad." + "reason": "Different model families and files; same initiative, not the same underlying bug/change." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Different models and execution paths (timm_backbone vs rwkv); not mergeable as one concrete PR." + "reason": "Both refactor model outputs, but the code paths are unrelated (timm_backbone vs rwkv)." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "MobileNetV1 and UnivNet are unrelated models; same tracing cleanup pattern only." + "reason": "Same output-tracing theme, but mobilenet_v1 and univnet are separate implementations." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different subsystems (MobileNetV1 vs UperNet); no shared bug or code path." + "reason": "Only a broad similarity in refactoring style; no shared concrete bug or code path." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "MobileNetV1 and Depth Anything are separate model families; both refactor outputs but do not fix the same thing." + "reason": "Different model families and target files; not the same change." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "MobileNetV1 and VisionEncoderDecoder are unrelated code paths; only the refactor style is similar." + "reason": "These are parallel refactors in different models, not a single fix." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "MobileNetV1 and SpeechEncoderDecoder are different architectures; not the same concrete change." + "reason": "Same general tracing refactor, but speech_encoder_decoder and mobilenet_v1 are unrelated." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Vision-Text Dual Encoder and UnivNet are unrelated models; same output-tracing modernization, not a duplicate." + "reason": "Same refactor pattern, but vision_text_dual_encoder and univnet do not share the same code path." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model families and file scopes; no shared underlying bug." + "reason": "Different model implementations; similar output-tracing cleanup only." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "Vision-Text Dual Encoder vs Mamba/FalconMamba are distinct code paths; too broad to combine." + "reason": "These are separate model refactors with no shared concrete bug to consolidate." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Different model families and behaviors; only the tracing refactor is common." + "reason": "Different subsystems and files; the overlap is only in the tracing refactor pattern." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "UnivNet and UperNet are unrelated; same refactor theme does not make them duplicates." + "reason": "Similar output API cleanup, but univnet and upernet are distinct code paths." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "UnivNet and Mamba/FalconMamba fix different model code paths; not mergeable as one PR." + "reason": "Not the same underlying bug/change; only a thematic match." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "UnivNet and Depth Anything are separate model implementations; only the API cleanup is shared." + "reason": "Separate model families; would not be merged into one PR." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "UperNet and RWKV are unrelated architectures; not the same underlying issue." + "reason": "Both are refactors, but they affect unrelated models and code paths." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "UperNet and Mamba/FalconMamba are different code paths; shared pattern is insufficient." + "reason": "No shared concrete change beyond the generic tracing refactor effort." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "Mamba/FalconMamba and SuperPoint are unrelated model families; these are separate refactors." + "reason": "Different model targets and implementation details; not duplicates." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Mamba/FalconMamba and CodeGen are different implementations with different output-shape changes; not one concrete fix." + "reason": "Both are output-related PRs, but codegen and mamba are separate concrete changes." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Timm Backbone and CodeGen are unrelated model paths; same tracing cleanup only." + "reason": "Shared refactor style only; timm_backbone and codegen are unrelated code paths." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "UnivNet and CodeGen are different model types and outputs; not a duplicate change." + "reason": "Different model families and separate forward-path changes." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "UperNet and CodeGen are unrelated; the refactor pattern alone is not enough." + "reason": "No single underlying bug/change; just the same refactor theme." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Vision-Text Dual Encoder and CodeGen address different forward paths; no shared concrete bug." + "reason": "Vision-text dual encoder and codegen are unrelated implementations." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Depth Anything and VisionEncoderDecoder are unrelated model families; only the output-tracing refactor pattern overlaps." + "reason": "Both are tracing refactors, but depth_anything and vision_encoder_decoder are different model stacks." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Depth Anything and SpeechEncoderDecoder are different code paths; not the same underlying change." + "reason": "Different model families and files; only broadly similar cleanup." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Depth Anything and SuperPoint are distinct models; same modernized return handling, not a duplicate." + "reason": "Same initiative, but superpoint and depth_anything are distinct code paths." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Depth Anything and RWKV are unrelated implementations; too broad to merge as one PR." + "reason": "These are model-specific refactors, not one shared fix." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Depth Anything and DPR are separate model families; same refactor theme only." + "reason": "Different targets and implementations; not mergeable as one PR." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "VisionEncoderDecoder and RWKV are unrelated model paths; not the same concrete problem." + "reason": "Output-tracing changes in separate model families; not duplicates." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are different wrapper models; no shared bug." + "reason": "Similar refactor pattern only, with no shared concrete bug." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "VisionEncoderDecoder and SuperPoint are unrelated; both just adopt the same tracing API changes." + "reason": "Distinct model code paths; broad theme overlap is insufficient." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "VisionEncoderDecoder and DPR are distinct code paths; not duplicates." + "reason": "No common underlying change beyond the tracing refactor series." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "SpeechEncoderDecoder and SuperPoint are unrelated model families; same output-tracing theme only." + "reason": "Different model stacks and separate implementation changes." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "SpeechEncoderDecoder and RWKV are different architectures; not one underlying change." + "reason": "Same high-level initiative, but not the same code path or fix." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "SpeechEncoderDecoder and DPR are unrelated; no shared concrete fix." + "reason": "They are parallel refactors in unrelated models." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are different model paths; not the same bug or change." + "reason": "Different models and forward paths; not one duplicate change." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "SuperPoint and DPR are unrelated implementations; only the refactor style matches." + "reason": "Only the refactor theme matches; the concrete code changes do not." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR are different model families; same output-tracing motif is too broad to treat as duplicate." + "reason": "Separate model implementations and output-handling changes; not the same PR." } ] }, @@ -17836,7 +17833,7 @@ "pull_request:44662|pull_request:45075", "pull_request:45075|pull_request:45401" ], - "prepared_review_unit_hash": "9ed0a4d6e72151d80cc757f8aabedde7367ea672ef04932995acd51146e3c693", + "prepared_review_unit_hash": "0ef1716f4f515e04d3fc6f43456eb0c9b75546874b8cdb33cb962020bd724bab", "component_budget": { "node_count": 3, "item_count": 3, @@ -17858,30 +17855,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate model-addition PRs (PenguinVL, DeepSeek-OCR-2, Voxtral TTS). They share common Transformers scaffolding and docs/auto-registry edits, but they do not address the same model or concrete code-path bug, so they are not duplicates.", - "confidence": 0.97, + "summary": "These are three separate model-addition pull requests (PenguinVL, DeepSeek-OCR-2, and Voxtral TTS). They share some boilerplate model-registration files, but they do not appear to fix the same underlying bug or implement the same concrete change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical PR: each PR introduces a different model family with distinct code, configs, tests, and docs, so they are not duplicates.", "best_issue_reason": null, - "best_pr_reason": "No single PR is a duplicate anchor here; each introduces a different model implementation with its own files and scope.", + "best_pr_reason": "No single PR is a good global representative for duplicate triage here because the PRs are unrelated model implementations rather than alternate fixes for the same change.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different models and different code paths; only generic model-registration/docs overlap." + "reason": "Different models and different implementation targets: PenguinVL vs Voxtral TTS. Shared auto-registration/docs files are expected for independent model additions, not evidence of the same change." }, { "left": "pull_request:44662", "right": "pull_request:45075", "accept": false, - "reason": "Separate model implementations (PenguinVL vs DeepSeek-OCR-2); shared auto/docs plumbing is not enough to merge as one PR." + "reason": "Both are separate model implementations (PenguinVL vs DeepSeek-OCR-2) with distinct configs, model code, and tests; they are not the same code-path fix." }, { "left": "pull_request:45075", "right": "pull_request:45401", "accept": false, - "reason": "Different model families and functionality; overlap is limited to shared Transformers boilerplate." + "reason": "DeepSeek-OCR-2 and Voxtral TTS are unrelated model additions. Overlap in transformer registry files is boilerplate, not a shared underlying bug or mergeable joint PR." } ] }, @@ -17911,12 +17908,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "f6c5b2dbc9d5006d4549d38c3eef47bad39c7fc3b50f36d1d92a952477e0a08a", + "prepared_review_unit_hash": "6203f6a537273ab9502635ac0724c819ef76bc63e87dc5823f6aff76acb99cf2", "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 }, @@ -17924,7 +17921,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4408, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -17933,18 +17930,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related conceptually around module/kernel fusion, but they implement different layers of the stack: #44979 adds a new generic Module Fusion API, while #45363 adds KernelConfig-based n-to-1 kernel fusion plumbing and an example. They are not the same underlying change and should stay separate.", - "confidence": 0.86, + "summary": "These two PRs are related by the broad theme of fusion, but they target different layers and should not be deduplicated: one adds a generic module-fusion API, the other integrates KernelConfig-driven n-to-1 kernel fusion into model loading.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "#44979 is the better canonical PR because it is the more foundational, self-contained introduction of the fusion API and tests; #45363 is a broader, more specific follow-on around KernelConfig integration and example usage.", + "canonical_pr_reason": "Not a duplicate pair; 44979 is the core module_fusion API addition, while 45363 is a separate KernelConfig/integration feature built on top of fusion concepts.", "best_issue_reason": null, - "best_pr_reason": "#44979 is the best PR representative for the cluster since it defines the core abstraction being introduced. #45363 is a downstream implementation/integration PR rather than the same concrete code-path fix or mergeable change set.", + "best_pr_reason": "45363 is the more complete end-user feature PR, with a concrete loading/integration path and end-to-end example; 44979 is foundational but narrower.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Related topic, but not the same change: #44979 introduces the generic Module Fusion API, while #45363 wires in KernelConfig-based n-to-1 kernel fusion and example code. They touch different code paths and would not plausibly be merged as one PR." + "reason": "Reject: both are fusion-related, but 44979 introduces a generic module-fusion API, whereas 45363 implements a separate KernelConfig-based n-to-1 kernel fusion integration. They do not appear to fix the same concrete code-path problem and would not be merged as one PR." } ] }, @@ -17974,12 +17971,12 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "07996f7c637634ea9322ce3656527843f98c8799c76a290ca3866d560dd48483", + "prepared_review_unit_hash": "d66176afe9feeb6175f0d4c6b2e86bba9812f4dc84103dde03ec0098c12888ee", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4413, + "serialized_chars": 4414, "estimated_input_tokens": 1104, "estimated_eval_tokens": 2464 }, @@ -17987,7 +17984,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4413, + "serialized_chars": 4414, "estimated_input_tokens": 1104, "estimated_eval_tokens": 2464 }, @@ -17996,18 +17993,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests address the same Mistral regex AttributeError in `_patch_mistral_regex`; 45317 is the more complete version because it includes the fix plus tests, so it is the best representative of the cluster.", - "confidence": 0.97, + "summary": "Both pull requests patch the same Mistral regex pre-tokenizer AttributeError in the same file, and 45317 is the more complete version with test coverage.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45317 is the stronger canonical PR: it implements the same core code-path fix as 45086 and adds test coverage, making it the more complete representative of the duplicate pair.", + "canonical_pr_reason": "pull_request:45317 is the better canonical PR: it applies the same code fix as 45086 and also adds a test, making it the more complete representative.", "best_issue_reason": null, - "best_pr_reason": "PR 45317 is the best PR to keep because it contains the same bug fix as 45086 plus validation, so it is the most useful single record for this cluster.", + "best_pr_reason": "pull_request:45317 is the strongest PR candidate overall because it addresses the same concrete bug with additional regression coverage.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying AttributeError fix in `_patch_mistral_regex` on the same file and shared issue target; the second PR appears to be a superset with tests, so they are mergeable duplicates." + "reason": "Same underlying AttributeError fix in _patch_mistral_regex, same target file, and the diffs overlap materially; 45317 just adds tests." } ] }, @@ -18037,7 +18034,7 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "3e7e75e904d3a0d100ab4c19578907fbba5e573c92ee02f7e68e716d82a02cd1", + "prepared_review_unit_hash": "6bb8333098594d16bb03bb518604c583eb79aa2799f9e2cbc23cd59563f4c984", "component_budget": { "node_count": 2, "item_count": 2, @@ -18059,18 +18056,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related NemotronH refactors, but they are not the same change: one is a narrow inheritance/mapping refactor, the other is a broader model split introducing NemotronHDense/NemotronHSparse plus docs and tests.", - "confidence": 0.89, + "summary": "Two NemotronH-related PRs overlap in a few core files, but they address different levels of change: one is a narrow refactor/conversion-mapping update, the other is a broader architectural split into NemotronHDense/NemotronHSparse with docs and tests. They should not be deduplicated.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:45569 is the broader, more complete implementation: it adds the dedicated dense/sparse model packages, updates auto-mapping/docs, and includes tests. pull_request:45115 is a narrower support refactor for conversion/mapping and inheritance.", + "canonical_pr_reason": "PR #45569 is the broader, more complete change: it introduces the dense/sparse model split, updates auto-mappings/docs, and adds tests, so it better represents the overall NemotronH architecture work.", "best_issue_reason": null, - "best_pr_reason": "pull_request:45569 is the best representative PR because it captures the full architectural change and user-visible packaging/docs updates, whereas pull_request:45115 is only a smaller internal refactor.", + "best_pr_reason": "PR #45569 is the best representative because it covers the finalized end-state of the feature, while #45115 is a narrower preparatory refactor focused on conversion/key renaming.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Shared NemotronH filenames do not imply duplicate scope: 45115 focuses on conversion mapping and inheritance refactoring, while 45569 introduces separate dense/sparse model implementations, docs, auto-mapping, and tests. They address related but different code paths and would not plausibly collapse into one PR." + "reason": "They overlap on NemotronH internals, but #45115 is a targeted refactor/conversion-mapping change, while #45569 is a broader model-split and documentation/test update. Same subsystem, but not the same concrete code-path problem." } ] }, @@ -18100,7 +18097,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "7351ab2a7b013fcc34fe7972ac6f0c83bba09ce692e7b20e3ca17b177b19f36d", + "prepared_review_unit_hash": "3f57fcf67c74a7c5c860151ab4d6dc4a8e3abfb0c3c0e451800a02a8f817963a", "component_budget": { "node_count": 2, "item_count": 2, @@ -18122,18 +18119,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related but distinct PRs: one adds per-layer heterogeneous config plumbing, the other adds broader heterogeneous model/runtime support. They are not the same concrete change, so the soft edge should be rejected.", - "confidence": 0.92, + "summary": "Two related but distinct PRs for heterogeneous-model work: one adds per-layer config support, the other adds broader heterogeneous model/runtime support. They share foundation files, but they are not the same concrete change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "No true canonical duplicate here; PR 45333 is the more foundational per-layer config piece, while PR 45332 builds additional modeling/runtime support on top of that theme.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 45333 is the cleaner representative for the shared heterogeneity/configuration thread because it introduces the core per-layer config abstraction used by the broader support work.", + "best_pr_reason": "PR #45332 is the broader, more complete feature implementation and best represents the cluster\u2019s end state, but it is not a duplicate of #45333.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "Related but not duplicate: 45333 is config-layer plumbing, while 45332 is a larger heterogeneous model/runtime change. They touch the same area but do not appear to fix the same concrete code-path problem, and would not naturally collapse into one PR." + "reason": "Reject: #45333 is config-layer support only, while #45332 extends heterogeneous behavior into cache/masking/modeling paths. They are complementary feature pieces, not the same fix and not mergeable as a single duplicate PR." } ] }, @@ -18163,40 +18160,40 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "088834af3a3151c52c465b2346ff99f7e3d7474af9b58464bc4a4702a7dc72a4", + "prepared_review_unit_hash": "1e54711d0f4a1398b817573d6bbe2b36041f6047591588775cfde631d5d09977", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5512, - "estimated_input_tokens": 1378, - "estimated_eval_tokens": 3012 + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5512, - "estimated_input_tokens": 1378, - "estimated_eval_tokens": 3012 + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are type-checking related, but they address different changes: #45415 is a broad pass adding type checking and suppressing many type-ignore warnings across numerous files, while #45425 introduces typing protocols/helpers and adjusts a smaller set of call sites. They are not the same underlying fix and should not be merged as duplicates.", - "confidence": 0.94, + "summary": "These are related typing-focused pull requests, but they are not the same underlying change: one is a broad repo-wide type-checking sweep, the other adds specific typing protocols and helpers around modeling_utils/peft integration.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "#45425 is the better canonical PR for this cluster because it is narrower and more concrete, centered on new typing abstractions and targeted call-site updates rather than a wide repository-wide type-check sweep.", + "canonical_pr_reason": "PR 45425 is the more focused, coherent typing change; PR 45415 is a broader type-checking cleanup spanning many files and config. They should not be collapsed as duplicates.", "best_issue_reason": null, - "best_pr_reason": "#45425 is the better representative PR: it has a clearer, more specific change set and a tighter code-path focus than #45415's broad type-checking cleanup.", + "best_pr_reason": "PR 45425 is the better representative of the cluster because it has a narrower, concrete code-path goal, while PR 45415 is a sweeping type-checking pass with many unrelated edits.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "They share a typing theme and a few filenames, but the changes are substantively different: #45415 is a broad type-checking/config cleanup across many modules, while #45425 adds typing protocols and targeted adjustments for specific interfaces. Not the same concrete bug/change." + "reason": "Both touch typing-related files, but 45415 is a broad type-checking enablement/cleanup across many modules, while 45425 adds specific typing interfaces and adjustments in modeling_utils/peft. They are not the same concrete fix and would not plausibly be merged as one PR." } ] }, @@ -18226,12 +18223,12 @@ "soft_pairs": [ "pull_request:45487|pull_request:45615" ], - "prepared_review_unit_hash": "4d8f8e6fc38d36dde1c43d759efc5af5b24d0e92cb6567d7d9803e6468bbc7d8", + "prepared_review_unit_hash": "818f0eea79fc7a4216c0c167f3f6ff657c251d47199b6bc17680f42d195f4fca", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5160, + "serialized_chars": 5159, "estimated_input_tokens": 1290, "estimated_eval_tokens": 2836 }, @@ -18239,7 +18236,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5160, + "serialized_chars": 5159, "estimated_input_tokens": 1290, "estimated_eval_tokens": 2836 }, @@ -18248,18 +18245,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are unrelated: one fixes a real device-placement bug in model code for AltCLIP/ChineseClip, while the other only updates Qianfan OCR XPU test expectations.", - "confidence": 0.97, + "summary": "These two pull requests are unrelated: one fixes a real model-parallel device placement bug in AltCLIP/ChineseCLIP code, while the other only updates Qianfan OCR XPU test expectations.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:45487 is the canonical PR because it changes the actual forward path/device handling in model implementation files across AltCLIP and ChineseClip, addressing a concrete runtime bug.", + "canonical_pr_reason": "PR #45487 is the stronger canonical PR because it changes the actual model code to move buffered token type IDs onto the correct device, fixing a concrete runtime/device-placement bug across multiple model implementations.", "best_issue_reason": null, - "best_pr_reason": "pull_request:45487 is the best PR candidate for the cluster since it contains the substantive code fix; pull_request:45615 is test-only and does not fix the underlying code-path problem.", + "best_pr_reason": "PR #45487 is the best PR in the cluster since it addresses a substantive code-path defect; PR #45615 is test-data-only and does not appear to fix the underlying implementation.", "soft_edge_verdicts": [ { "left": "pull_request:45487", "right": "pull_request:45615", "accept": false, - "reason": "Reject: these do not fix the same bug. 45487 patches model code to move token_type_ids to the correct device; 45615 only adds XPU test expectations for a different model." + "reason": "Reject: they target different models and problems. #45487 fixes AltCLIP/ChineseCLIP model-parallel device handling in core code; #45615 only adds XPU test expectations for qianfan_ocr." } ] }, @@ -18289,12 +18286,12 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "01b3cde4d662b289184fb928740237ba651cce03d593a2190748ec164bf777ae", + "prepared_review_unit_hash": "a35e6384f6aafadf2ec48b4de3e8e5353eab46614d90ae5741dffb7039228700", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -18302,7 +18299,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -18311,18 +18308,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are about expert parallelism, but they fix different concrete failure modes: one is EP + ZeRO-3/accelerate config loading, the other is EP + FSDP2 expert parameters being overwritten during sharded loading. They are related, not duplicates.", - "confidence": 0.92, + "summary": "Both PRs concern expert-parallel loading behavior, but they target different failure modes and code paths: one is about EP + DeepSpeed ZeRO-3/accelerate launch config handling, the other about expert tensors being overwritten during FSDP2-style distributed loading.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45662 is the better representative because it targets the concrete expert-overwrite path in sharded loading/runtime, with broader EP/FSDP2 handling changes. PR 45548 is a separate accelerate/ZeRO-3 loading/config serialization fix.", + "canonical_pr_reason": "PR 45662 is the stronger representative because it names a concrete bug ('experts silently overwritten by rank-0 broadcast') and the diff shows a focused fix in the loading path, including distributed post-shard wrapping and EP-related integration changes.", "best_issue_reason": null, - "best_pr_reason": "PR 45662 is the strongest standalone candidate: it addresses a specific silent corruption bug for experts and includes the relevant loading and DTensor/sharding changes. PR 45548 looks like a different EP-related loading bug.", + "best_pr_reason": "PR 45662 is the best overall candidate in this cluster: it addresses a specific, concrete expert-parallel corruption issue with a targeted code change, whereas 45548 is a separate loading/configuration fix.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Different bug paths and fixes: 45548 is about EP + DeepSpeed ZeRO-3 loading via accelerate launch and config serialization; 45662 is about EP + FSDP2 experts being overwritten by rank-0 broadcast during sharded loading. Shared EP context is not enough to merge them as one PR." + "reason": "Do not merge: they share the broad EP/loading subsystem, but the bugs differ. 45548 fixes EP + ZeRO-3 loading via accelerate launch and config serialization; 45662 fixes expert tensors being overwritten during distributed/FSDP2 loading. Not the same underlying change." } ] }, @@ -18355,12 +18352,12 @@ "pull_request:45645|pull_request:45778", "pull_request:45639|pull_request:45778" ], - "prepared_review_unit_hash": "0e6802102eca3044722f28bf7c0c1ac2815887d2f6f30ab0f23d6410c4412efb", + "prepared_review_unit_hash": "2b893865bf30eb52deb5c59e15534f5a27f8b6a58ed4029ccf492982476d6e80", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7470, + "serialized_chars": 7469, "estimated_input_tokens": 1868, "estimated_eval_tokens": 3992 }, @@ -18368,7 +18365,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7470, + "serialized_chars": 7469, "estimated_input_tokens": 1868, "estimated_eval_tokens": 3992 }, @@ -18377,40 +18374,40 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three PRs address the same pytest-xdist/captured_info collision in testing_utils, with PR 45645 providing the broadest end-to-end fix because it also updates CI artifact collection. The relationships are best treated as overlapping fixes for the same bug, not as a strict superset claim.", - "confidence": 0.89, + "summary": "PR 45645 is still the strongest representative of the cluster because it covers the xdist-safe captured_info fix and the CI-side handling for multiple captured_info files. PR 45639 and 45778 look like related but not clearly mergeable variants of the same bugfix, so the 45639\u219445778 edge should be rejected.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the best canonical PR because it fixes the captured_info collision in the test helper and also updates the CI workflow that consumes those logs, making it the most complete standalone change in the cluster.", + "canonical_pr_reason": "PR 45645 is the best canonical PR because it combines the core xdist-safe captured_info fix with the workflow change needed to consume per-worker artifacts, making it the most complete end-to-end implementation.", "best_issue_reason": null, - "best_pr_reason": "PR 45645 is the strongest representative of the cluster because it covers the core xdist-safe logging fix plus the downstream artifact handling needed for CI, whereas the others are narrower variants of the same bugfix.", + "best_pr_reason": "PR 45645 is the best representative for the cluster because it addresses the same testing_utils xdist collision and also preserves CI debug log collection across multiple captured_info files.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying xdist captured_info collision in the same testing code path; 45645 is an overlapping, more complete fix that also adjusts CI log collection." + "reason": "Both change the same testing_utils code path to avoid captured_info collisions under pytest-xdist; 45645 is a broader version that adds the CI artifact handling, but it is still the same underlying fix." }, { "left": "pull_request:45645", "right": "pull_request:45778", "accept": true, - "reason": "Both target the same captured_info xdist race in testing_utils; they use different implementation details, but they are still overlapping fixes for the same bug." + "reason": "45778 is a narrower source-only variant of the same worker-specific captured_info fix that 45645 implements, with the same concrete bug and code path." }, { "left": "pull_request:45639", "right": "pull_request:45778", - "accept": true, - "reason": "Both PRs fix the same concrete captured_info collision under pytest-xdist in src/transformers/testing_utils.py, so they look like alternative implementations for the same bug." + "accept": false, + "reason": "They share the same issue target and general intent, but the diffs suggest different implementations of the xdist fix rather than a clearly mergeable single change; that is not strong enough to treat them as duplicates." } ] }, "evaluator_result": { "accept": false, - "feedback": "The cluster is plausible, but one soft-edge verdict is too strong for the evidence. In particular, describing PR 45645 as a \"superset\" of PR 45639 is not directly supported by the packet: both touch the same xdist/captured_info area, but they appear to use different implementation details, so the relationship should be stated more conservatively as \"overlapping fix for the same bug.\"" + "feedback": "The cluster summary is mostly grounded, but one soft-edge verdict is too strong: 45639 vs 45778 should not be accepted as mergeable based only on a shared issue target and similar intent. The packet marks that pair as deterministic_accept=false, and the diffs suggest potentially different implementations of the same bugfix rather than clearly mergeable changes. Please keep that verdict conservative or reject the pair. The other pairwise links to 45645 are better supported." }, "evaluator_used": true, "retried": true, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -18422,8 +18419,8 @@ "review_unit_count": 1, "status": "reviewed", "reason": null, - "source": "llm", - "cache_hit": false, + "source": "cache", + "cache_hit": true, "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-45683-2", "nodes": [ diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 2ac77f6ccca999f208d0ce8cf74d5cbe38bccc5c..cdf29e32343c7606fad7086ffdccd471c74be0a8 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260505t060044z", + "analysis_id": "hybrid-model-20260505t120036z", "archived_artifacts": { - "hybrid": "snapshots/20260505T060044Z/analysis-runs/hybrid-model-20260505t060044z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260505T060044Z/analysis-runs/hybrid-model-20260505t060044z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-05-05T06:09:30Z", + "published_at": "2026-05-05T12:09:10Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260505T060044Z", + "snapshot_id": "20260505T120036Z", "variant": "hybrid" } diff --git a/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..1ed87795637b20a556e5dae03eba1632c04a9117 --- /dev/null +++ b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json @@ -0,0 +1,2281 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260505T120036Z", + "generated_at": "2026-05-05T12:09:07Z", + "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-43656-4", + "summary": "Cluster of 1 issues and 3 PRs centered on issue #43824.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43824, + "canonical_pr_number": 43656, + "issue_numbers": [ + 43824 + ], + "pr_numbers": [ + 43656, + 43836, + 43842 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43656, + "right_pr_number": 43836, + "code_similarity": 0.36, + "size_similarity": 0.176, + "file_overlap": 1.0, + "area_overlap": 0.051, + "patch_similarity": 0.048, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ] + } + ] + }, + { + "left_pr_number": 43656, + "right_pr_number": 43842, + "code_similarity": 0.405, + "size_similarity": 0.5, + "file_overlap": 1.0, + "area_overlap": 0.0, + "patch_similarity": 0.036, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 30, + 36 + ], + [ + 315, + 323 + ], + [ + 665, + 671 + ], + [ + 931, + 937 + ], + [ + 1843, + 1849 + ], + [ + 1868, + 1874 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + }, + { + "left_pr_number": 43836, + "right_pr_number": 43842, + "code_similarity": 0.332, + "size_similarity": 0.088, + "file_overlap": 1.0, + "area_overlap": 0.033, + "patch_similarity": 0.017, + "shared_filenames": [ + "src/transformers/cli/serve.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/cli/serve.py", + "left_ranges": [ + [ + 11, + 18 + ], + [ + 359, + 429 + ], + [ + 584, + 590 + ], + [ + 1892, + 1910 + ], + [ + 1917, + 1923 + ] + ], + "right_ranges": [ + [ + 54, + 61 + ], + [ + 587, + 593 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45561-4", + "summary": "Cluster of 1 issues and 3 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, + 45778 + ], + "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 + ] + ] + } + ] + }, + { + "left_pr_number": 45639, + "right_pr_number": 45778, + "code_similarity": 0.341, + "size_similarity": 0.141, + "file_overlap": 0.5, + "area_overlap": 0.386, + "patch_similarity": 0.181, + "shared_filenames": [ + "src/transformers/testing_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3529, + 3536 + ], + [ + 3753, + 3787 + ] + ], + "right_ranges": [ + [ + 3529, + 3537 + ], + [ + 3762, + 3770 + ] + ] + } + ] + }, + { + "left_pr_number": 45645, + "right_pr_number": 45778, + "code_similarity": 0.175, + "size_similarity": 0.087, + "file_overlap": 0.25, + "area_overlap": 0.18, + "patch_similarity": 0.128, + "shared_filenames": [ + "src/transformers/testing_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ], + "right_ranges": [ + [ + 3529, + 3537 + ], + [ + 3762, + 3770 + ] + ] + } + ] + } + ] + }, + { + "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-45081-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45081.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45081, + "canonical_pr_number": 45317, + "issue_numbers": [ + 45081 + ], + "pr_numbers": [ + 45086, + 45317 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45086, + "right_pr_number": 45317, + "code_similarity": 0.589, + "size_similarity": 0.34, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.136, + "shared_filenames": [ + "src/transformers/tokenization_utils_tokenizers.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/tokenization_utils_tokenizers.py", + "left_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ], + "right_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-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-43656-4", + "canonical_pr_number": 43656, + "duplicate_pr_numbers": [ + 43836, + 43842 + ], + "target_issue_number": 43824, + "reason": "PRs in cluster-43656-4 are treated as duplicates because they converge on issue #43824 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-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-45561-4", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639, + 45778 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-4 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." + } + ], + "best_issue": { + "cluster_id": "cluster-45561-4", + "issue_number": 45561, + "reason": "Issue #45561 is the strongest global issue candidate because it is open, belongs to a cluster with 4 artifacts, and carries 2 discussion comments plus 8 inbound references.", + "score": 152.0 + }, + "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/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..0e8ab030133f0a0f7c553615df23ebeeb71a5faf --- /dev/null +++ b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,18544 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260505T120036Z", + "generated_at": "2026-05-05T12:09:07Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 16, + "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": "1fdeafe9f49fea0220d1477225d72dc9a7221709cfe69fa8056ae6f21684cb04", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The three issues do not look like duplicates: one is about Tapas tokenization, one about a bad auth/download error message, and one about a generic model-loading failure for bert-base-uncased. The soft similarities are too broad to merge them.", + "confidence": 0.88, + "canonical_issue_reason": "Issue #16618 is the broadest and most central of the set, describing a concrete model-loading failure, but it does not appear to subsume the Tapas tokenization or private-download error reports.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 is the best standalone representative because it is the most general user-facing failure report and the other two issues are clearly about different code paths.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tapas tokenization differences and a bert-base-uncased load failure are different bugs in different code paths; only superficial NLP/model similarity." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "#15980 is specifically about a misleading error when downloading a private model while logged out, while #16618 is a generic inability to load a public model; related subsystem, but 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": 2, + "component_count": 16, + "review_unit_index": 1, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44661", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45070", + "issue:45081", + "issue:45084", + "issue:45161", + "issue:45237", + "issue:45362", + "issue:45464", + "issue:45507", + "issue:45685" + ], + "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:43638|issue:44805", + "issue:43596|issue:44805", + "issue:44805|issue:45161", + "issue:44805|issue:44918", + "issue:44661|issue:44805", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43975|issue:44457", + "issue:43039|issue:45161", + "issue:44292|issue:44457", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:39290|issue:45081", + "issue:43828|issue:45161", + "issue:43425|issue:44292", + "issue: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:44560|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:43039|issue:44805", + "issue:34928|issue:36010", + "issue:41762|issue:43596", + "issue:43901|issue:44512", + "issue:39290|issue:41720", + "issue:43039|issue:44560", + "issue:43866|issue:44863", + "issue:43854|issue:43866", + "issue:39290|issue:43531", + "issue:41093|issue:43638", + "issue:43425|issue:45070", + "issue:45362|issue:45464", + "issue:43792|issue:45070", + "issue:43404|issue:43866", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:36010|issue:39290", + "issue:44841|issue:45084", + "issue:41093|issue:41762" + ], + "prepared_review_unit_hash": "3f26833fc075bb4cba37195a72705af19d0575681da582ba6dae0ca3d30d34a4", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 67, + "serialized_chars": 47115, + "estimated_input_tokens": 11779, + "estimated_eval_tokens": 23814 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated issue reports, with one clear duplicate pair around DeepSpeed ZeRO-3 + Bert index-out-of-bounds. I would reject the docs-only pair as too different from the titles alone, and leave the rest unmerged as separate bugs.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43596 is the best canonical issue anchor because its title gives the clearest generic label for the DeepSpeed ZeRO-3/Bert index-out-of-bounds bug, and it has the strongest near-duplicate in issue:43638.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43596 is the strongest overall issue candidate: it is specific enough to anchor the DeepSpeed/Bert out-of-bounds cluster and is the one side of the only accepted duplicate-like pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Same model family name, but one is a config attribute error and the other is a streaming chat/completions failure." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different backends and different failures: AMD GPU loading versus MPS histogram op support." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Image-preprocessor dtype mismatch is not the same bug as AMD GPU model loading failure." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Autocast dtype mismatch and AMD GPU incompatibility are different issues." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Only loosely related by MoE; the concrete failures are different." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Qwen NVFP4 runtime error is a different problem from AMD GPU support." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch and ModernBERT ONNX export are unrelated bugs." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ONNX export failure and FineGrainedFP8Config failure are different model/runtime issues." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "ModernBERT export and Qwen3 device-mapping CUDA assert do not share the same code-path problem." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are v5 docs cleanups, but the removed topics differ: pipeline tasks versus the transformers run command." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ONNX export and GenerationMixin import error are distinct regressions." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both are index errors, but the titles point to different shapes and different model/setup contexts." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Same broad error class, but the titles indicate different bugs and different setups." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Mask-shape mismatch and GPT-OSS tensor-parallel failure are different issues." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Tokenizer-mapping utility failure is unrelated to the mask/tensor shape indexing bug." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice preset handling and ModernBERT ONNX export are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Both involve DeepSpeed, but the titles indicate different concrete failures." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "The titles describe the same DeepSpeed ZeRO-3/Bert index-out-of-bounds failure, with the second report only narrowing the Bert setup." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Different symptoms and models: detokenization bug versus LoRA save/reload mismatch." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and GPT-OSS tensor-parallel failure are not the same issue." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen NVFP4 runtime failure and LoRA merge/save inconsistency are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Activation-checkpointing/FSDP recomputation mismatch is a different bug from the mask/tensor index error." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "Both are DeepSpeed-related, but the titles point to different failures." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Bert ZeRO-3 index error and GPT-OSS tensor-parallel failure do not match." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "FSDP checkpointing mismatch and custom trainer signature change are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "DeepSpeed+Bert index error and GPT-OSS tensor-parallel issue are different code paths." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Image-preprocessor dtype mismatch is unrelated to the GraniteMoE invalid-method failure." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 config attribute issue and Mistral tokenizer regex patch crash are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:45161", + "accept": false, + "reason": "Autocast dtype mismatch and MoE tensor-parallel failure are not the same underlying defect." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen NVFP4 runtime failure are different problems." + }, + { + "left": "issue:44292", + "right": "issue:44918", + "accept": false, + "reason": "Different model families and different failure modes; no shared concrete bug." + }, + { + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and autocast dtype mismatch are not duplicates." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and trainer kwarg mismatch are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Kernel selection issue and indexing shape mismatch are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Checkpointing/FSDP issue and GenerationMixin import error are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Both are zero-3/loading errors, but Gemma3 and Bert failures are different concrete bugs." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Different docs topics: return_all_scores versus the removed run command." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 config attribute issue and Qwen3 device-mapping CUDA assert are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and Qwen3-vl video StopIteration are different issues." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Checkpoint corruption and NemotronH checkpoint loading failure are not the same bug." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Unit-test model loading failure and corrupted checkpoint report different root causes." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding-window-like behavior, but the model/config contexts are different." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Mask/index shape mismatch and Bert ZeRO-3 index error are different failures." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch compatibility and pydantic PretrainedConfig field regression are unrelated." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Both are Qwen3.5 chat-related, but the crash and streaming API failure are different bugs." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper runtime failure and pydantic model-field breakage are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "lm_head tying bug and checkpoint corruption are different issues." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Qwen3.5 streaming failure and MPS histogram unsupported error are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Torch Dynamo grouped_mm tracing error and MPS histogram issue are different code paths." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing failure and autocast dtype mismatch are different runtime bugs." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "GenerationMixin import error and Gemma3 sliding-window config error are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure for Voxtral and non-template-node compile error are different problems." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Mask-shape mismatch and Gemma3 ZeRO-3 index error are not the same bug." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "Mostly conservative, but not fully grounded in the packet. The canonical/best-issue reasons rely on body-level claims like \u201cconcrete, reproducible\u201d and \u201cnarrower reproduction,\u201d which are not supported by the metadata shown here. The accepted 44509/44512 docs pair is also only weakly justified from the titles alone, since they describe different removed docs topics rather than an obviously identical change." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 2, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36331", + "issue:39186", + "issue:39290", + "issue:39692", + "issue:40990", + "issue:41093", + "issue:41762", + "issue:42915", + "issue:43054", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43742", + "issue:43828", + "issue:43866", + "issue:43873", + "issue:43950", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44403", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44534", + "issue:44589", + "issue:44863", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45070", + "issue:45161", + "issue:45399", + "issue:45464", + "issue:45507", + "issue:45632", + "issue:45698" + ], + "soft_pairs": [ + "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:43531|issue:45070", + "issue:39290|issue:43742", + "issue:43541|issue:44589", + "issue:41093|issue:43596", + "issue:43054|issue:43994", + "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:44863|issue:45005", + "issue:42915|issue:45399", + "issue:43873|issue:45005", + "issue:44360|issue:44512", + "issue:44492|issue:44512", + "issue:44918|issue:45507", + "issue:43975|issue:44292", + "issue:43950|issue:44534", + "issue:43644|issue:43950", + "issue:44292|issue:45507", + "issue:39692|issue:43054", + "issue:43828|issue:44928", + "issue:43866|issue:44403", + "issue:36331|issue:41762" + ], + "prepared_review_unit_hash": "62f08e0d122c65f4a427d1f9d66f92a3ff0221169ff87ec37ca974fb80ba22f9", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37278, + "estimated_input_tokens": 9320, + "estimated_eval_tokens": 18896 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Transformers/model-specific bugs, but there are a few real duplicate pairs around the same root regressions: non-persistent buffer corruption, SigLIP2 output degradation, and trust_remote_code cache/module mixups.", + "confidence": 0.85, + "canonical_issue_reason": "issue:43644 is the clearest root report for the non-persistent-buffer regression and sits at the center of the strongest duplicate chain with issue:43950 and issue:44534.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the best single canonical issue because it describes the underlying bug most generally and has direct duplicate support from later reports with near-identical symptoms.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "Different problems: PaddleOCR-VL/vLLM load failure vs GGUF support for gpt-oss architecture." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Both are config-related v5 regressions, but they affect different fields/code paths and are not the same bug." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Unrelated failures: int4 CUDA memory/OOM vs tied-weights translation-model regression." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Both involve MoE/MoE-like models, but one is a torch-dynamo grouped_mm tracing error and the other is a tensor-parallel limitation." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Different symptoms and code paths: autocast dtype mismatch vs streaming chat/completions failure." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a distinct initialization bug from the pydantic PretrainedConfig break." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "No meaningful overlap beyond being model-specific issues; one is GGUF architecture support, the other is streaming inference." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Sliding-window handling for Qwen3-MoE is a different regression from the PretrainedConfig/pydantic field issue." + }, + { + "left": "issue:39290", + "right": "issue:43742", + "accept": false, + "reason": "Both are loading-time errors, but the missing sliding_window_pattern on Gemma3TextConfig is unrelated to the MobileLLM key error." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Tracing/grouped_mm failure is unrelated to the missing Float8 storage serialization issue." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are IndexErrors, but one is a mask/tensor shape mismatch and the other is a zero-sized dimension during DeepSpeed ZeRO-3 init." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": true, + "reason": "Both report the same SigLIP2 regression: text embeddings / AutoModel outputs are badly degraded or nonsensical." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Different model families and failure modes: high perplexity on gpt-oss-20b vs a GLM5 issue." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Different bugs: Mistral-3 image preprocessor dtype mismatch vs Qwen3.5 embedding unpacking failure in TRL SFT trainer." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP 'weight must be 2-D' is unrelated to the SFT trainer embedding unpacking error." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": true, + "reason": "Both point to the same trust_remote_code/custom-module cache collision, where the wrong local module is loaded after saving or when models share a leaf directory name." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Mistral3 tied-weight loading bug is different from the translation-model tied-weights regression described in issue:45005." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Jupyter custom-model initialization and flash-attn fallback blocking are unrelated regressions." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache on Qwen3.5 is a different runtime bug from tied weights in translation models." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is unrelated to translation tied-weights behavior." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3 MoE + FineGrainedFP8 failure is unrelated to flash-attn fallback gating." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization offloading behavior is a different problem from tied weights in translation models." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "DSA indexer/ReLU discussion is unrelated to the removed transformers run command docs issue." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A typo in cache strategies and outdated CLI docs are separate documentation issues." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Qwen3.5 embedding unpacking in TRL is unrelated to GraniteMoEHybrid invalid method calls." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Incorrect detokenization and Qwen-3-8B-NVFP4 runtime errors are different bugs." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying non-persistent-buffer corruption regression in Transformers 5.x, described with very similar wording." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same bug: from_pretrained corrupts non-persistent buffers / fills them with junk in Transformers 5.x." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime error and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "SigLIP2 doc example errors are not the same as the broader SigLIP2 embedding quality regression." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different Qwen/Phi model training/runtime failures with different mechanisms." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Corrupted Ovis2 checkpoint is unrelated to the loading-noise behavior issue." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "CustomTrainer unexpected argument and Gemma3 ZeRO-3 out-of-bounds are unrelated failures despite both involving training/loading." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The canonical issue choice is well grounded: issue:43644 is clearly supported by the duplicate chain to 43950 and 44534, and the cluster summary stays appropriately cautious about the rest being mostly unrelated model-specific bugs. The accepted soft-edge pairs are also conservative given the packet, and I don't see any overstatement that would warrant rejection." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 3, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:38175", + "issue:41553", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43257", + "issue:43404", + "issue:43493", + "issue:43643", + "issue:43742", + "issue:43782", + "issue:43792", + "issue:43824", + "issue:43827", + "issue:43854", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43940", + "issue:43950", + "issue:43975", + "issue:44220", + "issue:44291", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44509", + "issue:44589", + "issue:44661", + "issue:44863", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45237", + "issue:45305", + "issue:45313", + "issue:45356", + "issue:45399", + "issue:45520", + "issue:45701" + ], + "soft_pairs": [ + "issue:41553|issue:43883", + "issue:43872|issue:44589", + "issue:43054|issue:45701", + "issue:38175|issue:43054", + "issue:41553|issue:43881", + "issue:42947|issue:44387", + "issue:43054|issue:43493", + "issue:44403|issue:44451", + "issue:43643|issue:45020", + "issue:43881|issue:43883", + "issue:43975|issue:45520", + "issue:43975|issue:45237", + "issue:44403|issue:44488", + "issue:42947|issue:43856", + "issue:41762|issue:42915", + "issue:42491|issue:44368", + "issue:43782|issue:44960", + "issue:43975|issue:44451", + "issue:43824|issue:43872", + "issue:44661|issue:45020", + "issue:44991|issue:45399", + "issue:45020|issue:45356", + "issue:45081|issue:45399", + "issue:43827|issue:44509", + "issue:43792|issue:44220", + "issue:43404|issue:43950", + "issue:43742|issue:45399", + "issue:42915|issue:45020", + "issue:43257|issue:43866", + "issue:43742|issue:45020", + "issue:41553|issue:44991", + "issue:43872|issue:44291", + "issue:43940|issue:45313", + "issue:42915|issue:43872", + "issue:43854|issue:44863", + "issue:43257|issue:45305" + ], + "prepared_review_unit_hash": "3a45aa316acf5c5e5d54d3730c425dc93bbf57944a5da7632eb90c389a130a5b", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 36, + "serialized_chars": 37873, + "estimated_input_tokens": 9469, + "estimated_eval_tokens": 19194 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous set of Transformers issues with only a few clear duplicate pairs: SigLIP2 implementation mismatch, remote_code/config loading regressions, docs about removed pipelines, and a bitsandbytes/_is_hf_initialized loading regression. Most other soft edges are different model-specific failures and should stay separate.", + "confidence": 0.58, + "canonical_issue_reason": "issue:45020 is the broadest open umbrella for the recent remote_code / config-loading regression subset, even though the cluster overall is mixed.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the best global issue candidate because it is open and captures the widest shared regression theme among the closest matches.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Different models and failures: Voxtral tokenizer error vs Molmo tied-weights attribute error." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Both are loading/runtime errors, but the root causes differ: bitsandbytes init arg vs Float8 storage lookup." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "One is a SigLIP2 quality complaint; the other is a tokenizer/version regression. Not the same bug." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both concern SigLIP2, but one is zero probabilities and the other is lower embedding quality; too broad to merge." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Unrelated model-loading failures for different models and error paths." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Different areas: LoRA gradient checkpointing vs int4 quantization memory/OOM behavior." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": true, + "reason": "Both point to the same SigLIP2 implementation discrepancy causing degraded text embeddings." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic loading-noise report vs a specific ScandiBERT load failure; not the same underlying bug." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": true, + "reason": "Both describe recent remote_code/config-loading breakage; 43643 looks like a specific manifestation of the broader regression." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Different models and different failure modes." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Tokenizer detokenization regression vs flash_attn import failure; unrelated code paths." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Different problems: tokenizer behavior vs AMD GPU incompatibility." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Both are load-related, but they are model-specific complaints with no clear shared root cause." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Training memory issues, but one is gradient checkpointing with LoRA and the other is general Qwen3 MoE memory usage." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different models and configurations: Gemma3 ZeRO-3 indexing error vs Qwen3 MoE FP8 load failure." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "Both mention Qwen3 and LoRA, but one is an hf4.x/hf5.x compatibility break and the other is a tie_word_embeddings warning." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different model families and unrelated failure descriptions." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Import error for a missing class vs bitsandbytes parameter incompatibility; different underlying issues." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "A tokenizer mapping utility bug is not the same as the remote_code loading regression." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading failure vs flash-attn fallback logic; unrelated." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Remote_code regression and Kimi-K2.5 tokenizer codec regression are different bugs." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Mistral regex patch crash vs flash-attn fallback gating; separate code paths." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs regression: removed pipeline tasks still referenced in v5 documentation." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Whisper cannot run is too broad to equate confidently with the specific fbank feature-extraction issue." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Tied lm_head weights vs non-persistent buffer corruption; distinct model state bugs." + }, + { + "left": "issue:43742", + "right": "issue:45399", + "accept": false, + "reason": "MobileLLM loading failure and flash-attn fallback gating are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:45020", + "accept": false, + "reason": "Qwen3 MoE FP8 loading failure is model/config specific, not the remote_code regression." + }, + { + "left": "issue:43257", + "right": "issue:43866", + "accept": false, + "reason": "Qwen3 MoE accelerate/deepspeed weight conversion vs corrupted Ovis2 checkpoint; different causes." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "Different model-loading bugs with no shared concrete failure path." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer/loading failures for different models." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Both center on the unexpected `_is_hf_initialized` argument during model init / init_empty_weights loading." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": true, + "reason": "Same DeepSpeed ZeRO-3 weight-loading failure pattern across Qwen3 variants." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model families and different load failures." + }, + { + "left": "issue:44403", + "right": "issue:45020", + "accept": false, + "reason": "Generic loading noise and remote_code breakage are not the same bug." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "FP8 config loading failure vs bitsandbytes Int8Params argument incompatibility." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-specific checkpoint/load problems." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Different Qwen3 training/deepspeed issues: weight conversion vs gradient averaging." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative overall. The accepted duplicate pairs are plausible from the issue titles and themes, and the rejections generally avoid overclaiming shared root cause. The summary correctly characterizes the cluster as mostly heterogeneous with a few clear duplicate themes. One small omission is the Qwen3 ZeRO-3 pair, but that does not undermine the overall evaluation." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 4, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:41093", + "issue:41553", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43824", + "issue:43828", + "issue:43872", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:45030", + "issue:45313", + "issue:45325", + "issue:45357", + "issue:45538", + "issue:45584", + "issue:45588" + ], + "soft_pairs": [ + "issue:44928|issue:45588", + "issue:36331|issue:41093", + "issue:41553|issue:44488", + "issue:43872|issue:45538", + "issue:41093|issue:44805", + "issue:44530|issue:44863", + "issue:44479|issue:44560", + "issue:24643|issue:30064", + "issue:43366|issue:43828", + "issue:44155|issue:45584", + "issue:43366|issue:43531", + "issue:43550|issue:44910", + "issue:43824|issue:43957", + "issue:43645|issue:43646", + "issue:43366|issue:44589", + "issue:42915|issue:43957", + "issue:43950|issue:45325", + "issue:44560|issue:44918", + "issue:42617|issue:43366", + "issue:36010|issue:42915", + "issue:38175|issue:43994", + "issue:43646|issue:43950", + "issue:44560|issue:44805", + "issue:44877|issue:45030", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:43317|issue:44756", + "issue:43644|issue:44534", + "issue:38175|issue:39692", + "issue:45313|issue:45357" + ], + "prepared_review_unit_hash": "f49ee4c7b46a9b62ed03917a01b7498a973f7f8322ab87ff1f013b1e1fa3baaf", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36228, + "estimated_input_tokens": 9057, + "estimated_eval_tokens": 18370 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a mix of unrelated issue reports across different models and subsystems. Only a few soft pairs look like true duplicates: the Transformers 5 custom-model initialization regressions, the non-persistent buffer regression, and the SigLIP2 output regression.", + "confidence": 0.89, + "canonical_issue_reason": "No single canonical issue fits the whole cluster: the items span unrelated bugs and regressions, with only a handful of local duplicate pairs.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue: there is no coherent umbrella bug in this set, so any single issue would misrepresent the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Different failures: Qwen3.5 RLHF/SDPA BF16 gradient instability vs flash_attention crash on sink-less models." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Unrelated symptoms and code paths: trainer API keyword mismatch vs mask/tensor shape IndexError." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Both are loading problems, but for different models and different failure modes; not the same bug." + }, + { + "left": "issue:43872", + "right": "issue:45538", + "accept": false, + "reason": "Bitsandbytes constructor incompatibility is unrelated to the CLIPTokenizer model_max_length issue." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": false, + "reason": "Same error family, but no evidence they come from the same concrete bug or code path." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "Distinct model-specific problems: PagedAttentionCache linear_attention crash vs NemotronH checkpoint loading failure." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Both involve Qwen video flows, but one is a broad regression and the other is a specific embedding StopIteration failure." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "Completely different subsystems: DeepSpeed training RuntimeError vs image processor void segmentation map handling." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "GGUF gpt-oss support and torch.autocast dtype mismatch are unrelated." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 batched leakage and Whisper empty-transcription failure are different bugs." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "Different Qwen/GGUF topics with no shared concrete failure path." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different model families and different bugs: torch.compile/SDPA vs FlashAttention 3D position_ids handling." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Importing a specific class and meta-device loading regressions are separate issues." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same custom-model initialization regression in Transformers 5.0.0; one report is just more specific (Jupyter notebooks)." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "Unrelated: GGUF architecture support vs missing Float8 storage object." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Different failures: FineGrainedFP8Config on Qwen3Moe vs meta-device model loading regression." + }, + { + "left": "issue:43950", + "right": "issue:45325", + "accept": false, + "reason": "Non-persistent buffer corruption is unrelated to Qwen2.5-VL rope/index scaling." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Different bugs: video StopIteration in Qwen3-vl-embedding vs TRL SFT unpacking failure." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to GGUF gpt-oss support." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "GenerationMixin import breakage and Qwen3Moe FP8 loading are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 model regression causing incorrect/zero probabilities and nonsensical outputs; the second is the broader symptom report." + }, + { + "left": "issue:43646", + "right": "issue:43950", + "accept": false, + "reason": "Custom model init regression and non-persistent buffer corruption are separate Transformers 5.x bugs." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Video-generation StopIteration and generic mask shape mismatch are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both are config validation issues, but they affect different models and are not the same bug." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "DeepSpeed ZeRO-3 BertModel initialization failure is unrelated to tokenizer-mapping add-new-model-like behavior." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Different code paths and symptoms: ZeRO-3 index error vs tokenizer mapping failure." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "device_map/offload loading failure is unrelated to disabling mmap for Strix Halo OOM." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same regression about non-persistent buffers being filled with junk in Transformers v5; the titles are effectively the same bug." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Same model family, but the documented example errors are not clearly the same root cause as the zero-probabilities regression." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3.5 ZeRO-3 weight loading and incorrect visual encoder save keys are different bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster-level summary is grounded in the packet: the items span many unrelated model/subsystem failures, and the three accepted soft pairs are plausible near-duplicates based on their titles. The rejected pairs are phrased conservatively and do not overclaim equivalence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 5, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43335", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43720", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44488", + "issue:44661", + "issue:44683", + "issue:44756", + "issue:44910", + "issue:44960", + "issue:44977", + "issue:44991", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45362", + "issue:45406", + "issue:45588", + "issue:45663", + "issue:45704", + "issue:45759" + ], + "soft_pairs": [ + "issue:44960|issue:44977", + "issue:44451|issue:44488", + "issue:42491|issue:43720", + "issue:41720|issue:42915", + "issue:43708|issue:44075", + "issue:43854|issue:43931", + "issue:42491|issue:45276", + "issue:43856|issue:43873", + "issue:43257|issue:43854", + "issue:42915|issue:44756", + "issue:39692|issue:43994", + "issue:44488|issue:44991", + "issue:42491|issue:43931", + "issue:43856|issue:44387", + "issue:44661|issue:44683", + "issue:43644|issue:43645", + "issue:45362|issue:45406", + "issue:44910|issue:45588", + "issue:42491|issue:45292", + "issue:43404|issue:43454", + "issue:36010|issue:39404", + "issue:45276|issue:45335", + "issue:44164|issue:44291", + "issue:41720|issue:44155", + "issue:44402|issue:44488", + "issue:39692|issue:44960", + "issue:43425|issue:43957", + "issue:42947|issue:45704", + "issue:43873|issue:44387", + "issue:43335|issue:45663", + "issue:45663|issue:45759", + "issue:44387|issue:44756", + "issue:44291|issue:44756", + "issue:43854|issue:45216" + ], + "prepared_review_unit_hash": "24566ff4e2209aa9feed2401135e79bd80b14eea04a6dff752a1750ac0004a9f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 34, + "serialized_chars": 37373, + "estimated_input_tokens": 9344, + "estimated_eval_tokens": 18944 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly heterogeneous; there is one likely duplicate subgroup around `resize_token_embeddings` not propagating to nested embeddings in Gemma/T5Gemma variants, but the rest of the items look unrelated.", + "confidence": 0.34, + "canonical_issue_reason": "issue:45276 is the broadest description of the `resize_token_embeddings` regression, covering nested embedding/output weight updates rather than a single narrow symptom.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45276 is the best representative of the only plausible duplicate cluster: it states the underlying embedding-resize bug more generally than the paired report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Both are model-specific generation failures, but one is a generic GLM5 report and the other is a Qwen3.5 flash-attention crash; not the same bug." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are tokenizer/model loading complaints, but they target different checkpoints and likely different model metadata problems." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Different code paths: one is LoRA compatibility across HF versions, the other is BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Same broad family (Qwen3 MoE) but different failures: device-mapping CUDA assert vs FineGrainedFP8Config loading." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint math and optimizer argument handling are unrelated bugs." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different model-loading failures on different architectures; no shared code-path is evident." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "LoRA/Qwen3-MoE compatibility is unrelated to token embedding resizing in Gemma4." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Both mention memory/resource issues, but one is Qwen3 MoE training memory usage and the other is quantization offloading behavior." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate+deepspeed is unrelated to GLM-4.7-Flash loading." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3Moe FP8 loading and Strix Halo mmap OOM are different subsystems and failure modes." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Both are SigLIP2-related, but one is a docs example with model/processor mismatch and the other is a specific checkpoint producing bad outputs; not clearly the same bug." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both involve tokenizer loading, but they affect different models and no common underlying regression is shown." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3-MoE LoRA incompatibility is unrelated to Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Different quantization issues: training memory inefficiency vs int4 reserved-memory/OOM regression." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Tokenizer-mapping registration and compiled flex_attention are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "One is junk-filled non-persistent buffers, the other is notebook custom-model initialization; both may be v5 regressions but not the same bug." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Qwen3.5 chat crashes and Gemma4Processor missing `_tokenizer` are different model-specific failures." + }, + { + "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 in sink-less models." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "LoRA/Qwen3-MoE compatibility is not the same bug as generic `resize_token_embeddings` not updating output embeddings." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both involve untied `lm_head`, but they are different model classes with different generation paths; too model-specific to merge as one duplicate." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "A GenerationMixin import error and Whisper `return_language` pipeline regression are unrelated." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Both report the same underlying regression: `resize_token_embeddings` does not propagate to nested embeddings/output weights in composite Gemma/T5Gemma models." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Saving/loading extra_state and `init_empty_weights` `_is_hf_initialized` are distinct serialization/init bugs." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Qwen3 device-mapping CUDA assert and AudioFlamingo3 batched inference token leak are unrelated." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "One is a tokenizer vocab-size mismatch for a specific model, the other is a separate model loading failure." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "SigLIP2 docs errors and a GLM5 report are unrelated model families." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch 2.10 compatibility and meta-device loading failures are different regressions." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing with PEFT LoRA and T5 using apex FusedRMSNorm are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Quantization offloading expectations and int4 reserved-memory OOM are not the same concrete issue." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Sparse-layer config bug and Gemma-4 FSDP2 shared_kv_states error are different model internals." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Gemma-4 FSDP2 attention state rebuilding and Qwen 3.5/3.6 `from_config` text_config unwrapping are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Int4 quantization OOM and Strix Halo mmap OOM are different memory problems with different triggers." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "`init_empty_weights` loading error and Strix Halo mmap OOM are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading and Qwen3.5 save_pretrained regression are different model/version problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet and stays appropriately conservative. The one accepted duplicate pair (45276/45335) is well supported by the titles and the shared `resize_token_embeddings` propagation bug. The rest of the soft-edge rejections avoid overclaiming and do not appear to merge unrelated issues." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 6, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:39404", + "issue:42947", + "issue:43257", + "issue:43299", + "issue:43381", + "issue:43425", + "issue:43611", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43828", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44292", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45084", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45701" + ], + "soft_pairs": [ + "issue:44292|issue:44912", + "issue:44743|issue:44960", + "issue:43957|issue:44756", + "issue:42947|issue:43381", + "issue:45684|issue:45698", + "issue:45276|issue:45292", + "issue:43381|issue:44928", + "issue:44164|issue:44756", + "issue:45588|issue:45663", + "issue:44479|issue:44877", + "issue:43646|issue:43708", + "issue:45292|issue:45335", + "issue:44402|issue:44451", + "issue:43645|issue:44164", + "issue:43257|issue:44863", + "issue:43828|issue:43957", + "issue:43425|issue:43828", + "issue:39404|issue:45701", + "issue:44589|issue:44756", + "issue:44479|issue:45663", + "issue:43931|issue:44863", + "issue:43906|issue:44530", + "issue:43611|issue:43645", + "issue:44530|issue:45084", + "issue:43425|issue:43716", + "issue:39404|issue:44912", + "issue:43901|issue:44509", + "issue:44863|issue:45663", + "issue:44265|issue:44898", + "issue:43299|issue:43931", + "issue:44479|issue:44805", + "issue:43994|issue:44079", + "issue:44671|issue:44991", + "issue:39404|issue:44671", + "issue:43425|issue:44918" + ], + "prepared_review_unit_hash": "cecc7395e8c2b83be9c11081938997f6b4564fd76e2f057caa650afd76bedb99", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38104, + "estimated_input_tokens": 9526, + "estimated_eval_tokens": 19308 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are distinct regressions affecting different models, code paths, or behaviors. The only clear duplicate cluster here is the resize_token_embeddings propagation bug around output_embeddings / decoder.embed_tokens, where the Gemma4 and T5Gemma reports describe the same underlying failure pattern.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45292 is the cleanest general representative of the resize_token_embeddings regression because it states the core symptom without model-specific noise, and it matches the shared output_embeddings failure described in the nearby model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45292 is the best standalone anchor for this cluster: it is broad, concise, and captures the shared API-level bug that the other resize_token_embeddings reports appear to instantiate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Same broad quantization/loading theme, but different models and different failure modes; not the same underlying bug." + }, + { + "left": "issue:44743", + "right": "issue:44960", + "accept": false, + "reason": "Different model families and different runtime behavior; no shared concrete code-path bug." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "One is a meta-device loading regression, the other is an mmap/OOM workaround issue; unrelated." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is about PEFT LoRA ineffectiveness and the other about eval-mode prohibition; different bugs." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Both involve custom-model save/load flows, but one is a permissions propagation issue and the other is the wrong module being loaded." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Both describe the same resize_token_embeddings propagation failure; 45276 is the Gemma4-specific manifestation of the broader output_embeddings bug." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Different problems: eval-mode checkpointing restriction versus RLHF NaN/SDPA fallback instability." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Extra_state serialization failure is unrelated to the mmap/OOM issue." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Different subsystems and failure points: flash attention aux handling versus Gemma4 FSDP2 shared_kv_states construction." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "One is a video-input regression in Qwen models; the other is strict config handling for granite_speech." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization in notebooks and Trainer resume/max_steps calculation are unrelated." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Both report the same resize_token_embeddings bug on model embeddings; 45335 is the T5Gemma-specific variant of the broader output_embeddings regression." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab-size mismatch and inability to load a specific model are different issues." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Notebook-defined custom model initialization and extra_state save/load handling are distinct regressions." + }, + { + "left": "issue:43257", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for different architectures; not the same code-path bug." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch and meta-device loading failure are separate bugs." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch version compatibility and an autocast dtype mismatch are not the same issue." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to tokenizer version-dependent behavior." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap/OOM problem." + }, + { + "left": "issue:44479", + "right": "issue:45663", + "accept": false, + "reason": "Video-input regression in Qwen models and Gemma4 FSDP2 state handling are different bugs." + }, + { + "left": "issue:43931", + "right": "issue:44863", + "accept": false, + "reason": "Different model families and distinct checkpoint-loading failures." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "Insufficient evidence of a shared underlying bug; one is only an isolated reproduction reference." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5 custom-model regressions, but they affect different mechanisms and should not be merged." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and template-node compilation TypeError are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch compatibility and a model/preprocessor dtype mismatch are different problems." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline behavior and MXFP4 quantization loading are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but they cover different pipeline task deprecations and different pages." + }, + { + "left": "issue:44863", + "right": "issue:45663", + "accept": false, + "reason": "NemotronH checkpoint loading and Gemma4 FSDP2 shared_kv_states rebuilding are unrelated." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure with torch_compilable_check and Perceiver non-default-resolution failure are different code paths." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3VL MoE loading regression and Qwen3-VL weight-shape mismatch are distinct loading issues." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression and mask shape IndexError are different failures." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Incorrect SigLIP2 outputs and ModelOutput key assignment are unrelated." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT MLM prediction regression and tokenizer loading failure are separate issues." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline return_language and CamemBERT MLM predictions are unrelated bugs." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility and Qwen3.5 embedding unpacking in TRL are different regressions." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays appropriately conservative. The only soft-edge accepts (45276\u219445292 and 45292\u219445335) are well-supported by the near-identical resize_token_embeddings/output_embeddings wording and the high similarity scores. The non-duplicates are mostly correctly rejected, and the choice of 45292 as the canonical anchor is reasonable because it is the broadest formulation of the shared bug." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 7, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42915", + "issue:42947", + "issue:43317", + "issue:43404", + "issue:43421", + "issue:43425", + "issue:43452", + "issue:43482", + "issue:43541", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43856", + "issue:43883", + "issue:43957", + "issue:44164", + "issue:44206", + "issue:44220", + "issue:44451", + "issue:44466", + "issue:44488", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45507", + "issue:45584", + "issue:45685", + "issue:45759" + ], + "soft_pairs": [ + "issue:43856|issue:44756", + "issue:44589|issue:45084", + "issue:43596|issue:44683", + "issue:40990|issue:42947", + "issue:43425|issue:45685", + "issue:44164|issue:45759", + "issue:36032|issue:43452", + "issue:32090|issue:39404", + "issue:44991|issue:45081", + "issue:43957|issue:44164", + "issue:43582|issue:43632", + "issue:44488|issue:45081", + "issue:40990|issue:43421", + "issue:43883|issue:44488", + "issue:43611|issue:43644", + "issue:44488|issue:44987", + "issue:44206|issue:44220", + "issue:43716|issue:45084", + "issue:34567|issue:40444", + "issue:43643|issue:44530", + "issue:44811|issue:45584", + "issue:43644|issue:43646", + "issue:36032|issue:42915", + "issue:43883|issue:44991", + "issue:44451|issue:44987", + "issue:43541|issue:45084", + "issue:43452|issue:43482", + "issue:42915|issue:44530", + "issue:43317|issue:43482", + "issue:44849|issue:45325", + "issue:36032|issue:43645", + "issue:43404|issue:44466", + "issue:44987|issue:44991", + "issue:45084|issue:45507" + ], + "prepared_review_unit_hash": "e255bb86267d6131dedc9646957504987dc0c2696dc169b2b386156846c5cbbf", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37846, + "estimated_input_tokens": 9462, + "estimated_eval_tokens": 19180 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Highly heterogeneous issue-only cluster with many unrelated Transformers regressions and model-specific bugs. None of the proposed soft edges look like true duplicates; they mostly share broad themes (v5 loading, Whisper, Qwen, tokenizer/runtime issues) without the same concrete failure mode.", + "confidence": 0.88, + "canonical_issue_reason": "No single canonical issue: the set spans unrelated bugs across training, tokenizers, multimodal loading, serialization, and version regressions.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the broadest representative of the Transformers v5 model-loading regression reports, so it is the closest anchor despite the cluster not being a true duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Different problems: Qwen3 MoE memory usage during training vs an mmap/OOM workaround for Strix Halo." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Unrelated failures: missing Float8 storage object vs a template-compilation error." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "Different code paths: DeepSpeed ZeRO3/BERT init error vs compiled flex_attention on newer torch." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Different symptoms and subsystems: high perplexity on gpt-oss vs gradient checkpointing with PEFT LoRA." + }, + { + "left": "issue:43425", + "right": "issue:45685", + "accept": false, + "reason": "Torch version compatibility issue vs an MPS MoE histogram kernel limitation; not the same bug." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "save/from_pretrained extra_state handling is unrelated to unwrapping text_config in composite configs." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "Tokenizer special-token method conflict vs gguf_file loading regression; different failure modes." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error vs Whisper pipeline return_language regression." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer loading failure for EMBEDDIA/est-roberta vs a Mistral regex patch crash; not the same defect." + }, + { + "left": "issue:43957", + "right": "issue:44164", + "accept": false, + "reason": "Meta-device loading breakage is different from extra_state serialization handling." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError vs the _is_hf_initialized flag regression." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Two different Mistral/tokenizer loading issues with different root causes." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Perplexity regression on a model/data setup vs runtime special-token post-processor updates." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing all_tied_weights_keys on Molmo is unrelated to loading sleng-bert." + }, + { + "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 serialization; different bugs." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Different model-loading regressions affecting different checkpoints and failure points." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Unsupported center argument in LasrFeatureExtractor is not the same as _torch_extract_fbank_features behavior." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral image-preprocessor dtype mismatch vs a template-node compilation error." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Trainer token-count tracking bug vs Qwen2.5-VL iterable dataset with multiple images." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "trust_remote_code missing fields in AutoConfig is unrelated to PagedAttentionCache linear_attention failures." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Whisper batch_decode skip_special_tokens handling is a different issue from empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both involve v5 custom-model behavior, but one is non-persistent buffer serialization and the other is custom model initialization; not the same concrete bug." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer add_special_tokens conflict is unrelated to Qwen3 MoE FineGrainedFP8Config failure." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Molmo tied-weights attribute error vs tokenizer loading failure for a specific model." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "ScandiBERT loading failure and physical-intelligence/fast loading failure are separate model-specific regressions." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Mixtral torch-dynamo tracing error at grouped_mm is unrelated to a template compilation error." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "gguf_file breaking from_pretrained calls is different from Qwen2.5-GGUF failing under transformers v5." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3 MoE FP8 training issue vs PagedAttentionCache linear_attention crash; different concrete problems." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "device_map=auto dequantized offload failure is unrelated to Qwen2.5-GGUF loading under v5." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Qwen3.5 output_hidden_states bug is separate from Qwen2.5-VL rope-index scaling." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method conflict and notebook custom-model initialization are different regressions." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "Untied lm_head in Mistral3 is not the same as device-dependent inconsistent serialization of lm_head.weight." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Two different model-loading failures with different affected checkpoints and failure points." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template-node compilation error is unrelated to GraniteMoEHybrid calling an invalid method." + } + ] + }, + "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": 16, + "review_unit_index": 8, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43531", + "issue:43540", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43819", + "issue:43856", + "issue:43994", + "issue:44062", + "issue:44079", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44479", + "issue:44534", + "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" + ], + "soft_pairs": [ + "issue:43262|issue:43377", + "issue:43646|issue:44534", + "issue:43994|issue:44610", + "issue:43645|issue:44534", + "issue:44403|issue:45081", + "issue:43322|issue:43540", + "issue:43377|issue:43819", + "issue:44220|issue:44683", + "issue:44451|issue:45081", + "issue:45216|issue:45357", + "issue:42915|issue:45084", + "issue:44265|issue:44610", + "issue:44987|issue:45081", + "issue:43404|issue:45127", + "issue:43632|issue:44756", + "issue:41553|issue:43582", + "issue:44062|issue:44987", + "issue:42915|issue:44291", + "issue:44479|issue:45325", + "issue:43296|issue:43531", + "issue:36032|issue:43742", + "issue:43531|issue:44877", + "issue:43611|issue:44534", + "issue:42915|issue:44589", + "issue:44079|issue:45663", + "issue:40444|issue:45325", + "issue:40990|issue:44811", + "issue:43317|issue:43856", + "issue:42175|issue:45542", + "issue:36032|issue:41720" + ], + "prepared_review_unit_hash": "bd83c715d17aa61f4c0a0468d316c3010607f05318b731b727f1f4f9e6a1c396", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36679, + "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 heterogeneous: it mixes unrelated tokenizer-loading, multimodal, quantization, device-mapping, and config-regression reports. None of the soft pairs look like the same concrete bug or change, so no duplicates should be merged here.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue emerges; the items cover many unrelated failure modes across different models and subsystems.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a strong global representative because the cluster is thematically fragmented rather than one bug family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting is unrelated to MIMI batched-vs-single padding-mask behavior." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Custom model initialization breakage is a different issue from non-persistent buffer corruption." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 nonsensical outputs and OmDet-Turbo input-size mismatch are separate model-specific bugs." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Jupyter/custom-model init problems do not match the non-persistent buffer junk-data regression." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Loading noise vs tokenizer patch crash are distinct symptoms and code paths." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Llava Next loading segfault and Qwen3OmniMoe video ValueError are different multimodal failures." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "MIMI padding-mask mismatch is unrelated to DAC forward/from_latents STE behavior." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Feature-extraction issues in audio helpers are not the same as compiled flex_attention failures." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "ScandiBERT loading failure and Mistral regex patch crash are separate tokenizer/model-loading bugs." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 regressions, but one is checkpoint saving correctness and the other is visual-encoder key naming; not the same fix." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure is unrelated to template compilation errors." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export with torch_compilable_check and OmDet-Turbo processor size mismatch are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Loading physical-intelligence/fast and Mistral tokenizer backend access are different loading paths." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Untied lm_head weights in Mistral3 are not the same bug as LoRA merge collapse with extended vocab." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "_is_hf_initialized flag regression and Strix Halo mmap OOM are unrelated runtime issues." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad tokenizer error messaging for Voxtral is unrelated to Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken special-argument duplication and physical-intelligence/fast loading failure are different bugs." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3Moe FP8 configuration failure is not the same as init_empty_weights/_is_hf_initialized argument breakage." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Video-input regression in Qwen3.5/Qwen2.5-VL is different from still-image rope-index scaling." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window issue are unrelated model-loading versus attention-window bugs." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "T5Tokenizer add_special_tokens conflict and MobileLLM key error are separate tokenizer/model init problems." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and granite_speech strict-config loading are not the same underlying issue." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "base_model_prefix loading regression and junk non-persistent buffers are different Transformers v5 breakages." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Qwen3Moe FP8 config failure is unrelated to missing Float8 storage handling." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment bug and Gemma4 shared_kv_states FSDP2 KeyError are unrelated internal state issues." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "IterableDataset multi-image finetuning failure is not the same as Qwen2.5-VL rope-index scaling." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and Whisper batch_decode skip_special_tokens are unrelated evaluation/decoding bugs." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "device_map auto dequantized offload failure and Qwen3 MoE memory usage are different memory-management problems." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "pip extra dependency resolution and tensorboard-without-TF backend error are not the same packaging bug." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "T5 tokenizer loading conflict and Qwen3 auto-device-map cudaErrorAssert are unrelated failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 9, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43296", + "issue:43377", + "issue:43421", + "issue:43425", + "issue:43454", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43646", + "issue:43749", + "issue:43792", + "issue:43818", + "issue:43828", + "issue:43881", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44466", + "issue:44488", + "issue:44534", + "issue:44568", + "issue:44589", + "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" + ], + "soft_pairs": [ + "issue:43425|issue:44403", + "issue:43632|issue:44534", + "issue:43818|issue:45357", + "issue:43421|issue:44568", + "issue:44683|issue:45020", + "issue:43792|issue:44589", + "issue:44403|issue:44991", + "issue:43482|issue:45538", + "issue:33666|issue:40444", + "issue:44829|issue:45479", + "issue:44079|issue:44534", + "issue:35707|issue:40444", + "issue:43482|issue:43525", + "issue:44291|issue:44589", + "issue:43377|issue:45491", + "issue:43749|issue:44079", + "issue:43644|issue:44079", + "issue:41720|issue:45081", + "issue:44265|issue:45072", + "issue:43531|issue:44291", + "issue:40444|issue:45381", + "issue:43632|issue:45072", + "issue:42175|issue:43296", + "issue:45325|issue:45381", + "issue:43611|issue:43646", + "issue:43994|issue:44898", + "issue:43454|issue:44466", + "issue:43881|issue:44488", + "issue:43828|issue:44683", + "issue:42915|issue:43482", + "issue:43792|issue:43957" + ], + "prepared_review_unit_hash": "d76ec877ff01877395ffcef9444d017fb713fb3258ce7dad8ab225ab28460442", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36533, + "estimated_input_tokens": 9134, + "estimated_eval_tokens": 18524 + }, + "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 and model-specific loading/training bugs, not a tight duplicate set. A few items are thematically adjacent, but none of the soft-edge pairs look like the same concrete defect.", + "confidence": 0.61, + "canonical_issue_reason": "issue:43632 is the best concrete representative of the recurring transformers v5 model-initialization/loading regressions in this set, and it is narrower/more actionable than the broad umbrella reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the broadest open umbrella for recent transformers-version breakages, but it is too generic to be a true duplicate canonical; if forced to pick one issue as the cluster's overall anchor, it best reflects the cross-cutting regression theme.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "One is a Torch 2.10 compatibility complaint; the other is a loading-noise report. Different bugs and different failure modes." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Both involve transformers v5 initialization behavior, but one is about `_is_hf_initialized` and the other about non-persistent buffers; these are distinct code-path regressions." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Different models and different symptoms: Video-LLaVA temporal attention/weight-sharing vs Qwen3.5 visual encoder key serialization." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer post-processor update logic at runtime is not the same as missing BOS/EOS insertion for a specific tokenizer." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "A torch>=2.9 flex_attention compile failure is unrelated to the broad remote_code loading regressions." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Different runtime errors on different models; one is a Whisper load failure, the other is a missing Float8 storage type." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Generic loading-noise issue vs a specific tokenizer-loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "GGUF loading failure for Qwen2.5 is unrelated to CLIPTokenizer's model_max_length sentinel value." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen VL training issues, but one is multi-GPU training and the other is IterableDataset multi-image finetuning; not the same bug." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both mention degenerate classification behavior, but the root causes differ: flash_attention_3 training pathology vs single-label/num_labels=1 loss handling." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer serialization are unrelated." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is a different generation path than VL finetuning with multiple images." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Different loading errors for different models: GGUF/transformers v5 vs missing pad_token_id on Llama4Config." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized argument handling is distinct from the Float8 storage lookup failure." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both are batching-related, but MIMI padding masks and Gemma3 sliding-window/all-padding NaNs are separate model-specific bugs." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "FSDP CPU RAM efficient loading is unrelated to ModelOutput key bookkeeping." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer serialization junk is not the same defect as ModelOutput key assignment." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "CUDA assert on Qwen3 auto device mapping is unrelated to tokenizer patching crashing on missing backend_tokenizer." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export with torch_compilable_check and bfloat16 dtype mismatches are different failure modes." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is not the same as init_empty_weights argument incompatibility." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Multi-image finetuning failure and Qwen2.5-VL video position_ids bug are separate VL preprocessing issues." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "_is_hf_initialized regression and bfloat16 dtype mismatches are different transformers v5 issues." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Packaging/backend inclusion for pip extras is unrelated to PaddleOCR-VL loading in vLLM." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL position-id bugs, but one is still-image temporal scaling and the other is video vision_position_ids; not the same bug." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 model-loading regressions, but one is about base_model_prefix handling and the other about custom model initialization." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different vision-model failures with different inputs and symptoms." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "AyaVision lm_head tying causing garbage generation is not the same as device-dependent lm_head.weight serialization inconsistency." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Different model-loading failures with no shared concrete code path." + }, + { + "left": "issue:43828", + "right": "issue:44683", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to compiled flex_attention failures on newer torch." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and Qwen2.5-GGUF loading failure are distinct model/config issues." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper-large-v2 load failure and meta-device loading breakage are separate problems." + } + ] + }, + "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": 16, + "review_unit_index": 10, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:43262", + "issue:43296", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43632", + "issue:43701", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43792", + "issue:43819", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44387", + "issue:44466", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44683", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45325", + "issue:45357", + "issue:45563", + "issue:45584", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45759" + ], + "soft_pairs": [ + "issue:44387|issue:44617", + "issue:43701|issue:45325", + "issue:44291|issue:44877", + "issue:43526|issue:43883", + "issue:40990|issue:45584", + "issue:43262|issue:43746", + "issue:43606|issue:44265", + "issue:43317|issue:43873", + "issue:34634|issue:35707", + "issue:43632|issue:45563", + "issue:43873|issue:45701", + "issue:44186|issue:44265", + "issue:43881|issue:44991", + "issue:35707|issue:41720", + "issue:43606|issue:43819", + "issue:44610|issue:45325", + "issue:43454|issue:43550", + "issue:43377|issue:44610", + "issue:43454|issue:45072", + "issue:43572|issue:45759", + "issue:43377|issue:44898", + "issue:39692|issue:40444", + "issue:43749|issue:44534", + "issue:44466|issue:45357", + "issue:43792|issue:44683", + "issue:43296|issue:44991", + "issue:43296|issue:44877", + "issue:43454|issue:43720", + "issue:41720|issue:43322", + "issue:41720|issue:44977", + "issue:45632|issue:45684" + ], + "prepared_review_unit_hash": "ca84ac56a22ffb4ed9b48ff6885e79d5da8949fffd6a1ab081734e031800413f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36985, + "estimated_input_tokens": 9247, + "estimated_eval_tokens": 18750 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a noisy cluster of mostly unrelated Transformers issues, with a loose theme around v5-era loading/init/offload regressions but no safe duplicate pairs.", + "confidence": 0.42, + "canonical_issue_reason": "issue:43632 is the broadest platform-level regression here: a core v5 `_is_hf_initialized` breakage that can affect multiple loading paths, making it a better umbrella than the model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43632 is the best global representative because it captures the widest shared loading/init regression theme in the set; the other issues are mostly narrower, model-specific variants or unrelated bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both mention memory, but one is an int4 quantization regression and the other is a video-model OOM; different failure modes and code paths." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL rope/position-id scaling." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "`_is_hf_initialized` constructor incompatibility and strict config loading are separate regression surfaces." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "A Fast image processor label-reduction bug is unrelated to a model attribute missing during tied-weight handling." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "Extremely high perplexity and Whisper empty-transcription generation failure are different model/runtime problems." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sample-rate defaulting is not the same bug as PEFT local checkpoint loading." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch and torch.export/export failure via `torch_compilable_check` are distinct issues." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Dequantized-model loading with auto device map is related to offloading only broadly; the concrete bug and path differ from the quantization offload report." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice preset handling and progressive generation with `inputs_embeds`/`past_key_values` are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:45563", + "accept": false, + "reason": "Core init-flag regression is unrelated to a stale warning around `num_return_sequences`." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offloading behavior and tokenization changes are not the same bug." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crashes are unrelated to export-time `torch_compilable_check` failures." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "A model loading failure for glm-4v-9b is not the same as tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation state handling and Qwen3 device-map CUDA asserts are different bugs." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "CPU offload device mismatch and DAC latent reconstruction mismatch have no shared code-path." + }, + { + "left": "issue:44610", + "right": "issue:45325", + "accept": false, + "reason": "Processor image-size mismatch and rope-index temporal scaling are unrelated model-preprocessing bugs." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "Weight tying failure and torch.compile SDPA failure are different problems." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "MIMI batching/padding-mask output drift is unrelated to OmDet processor resolution mismatch." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "AyaVision tied-weight serialization and dtype mismatch in inference are not the same defect." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Missing `pad_token_idx` and failure to unwrap `text_config` are separate config-loading regressions." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Batched-vs-single encoder output mismatch is not the same as Perceiver interpolation failing at non-default resolution." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Documentation/model-processor mismatch for SigLIP2 and Qwen2.5-VL iterable-dataset multi-image finetuning are different issues." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are separate v5 regressions." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Device-dependent lm_head serialization and incorrect visual encoder keys are different serialization bugs." + }, + { + "left": "issue:43792", + "right": "issue:44683", + "accept": false, + "reason": "Whisper inference failure and compiled flex_attention failure on newer torch are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:44991", + "accept": false, + "reason": "PaddleOCR-VL load failure in vLLM and tokenizer loading failure for est-roberta are different artifact types and paths." + }, + { + "left": "issue:43296", + "right": "issue:44877", + "accept": false, + "reason": "PaddleOCR-VL model loading and strict config rejection of granite_speech are not the same underlying bug." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "AyaVision tied weights and BitNet packed-weight unpacking during accelerate loading are distinct model-loading issues." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3 auto device-map CUDA assert and Llava Next segmentation fault are different crashes." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "CUDA assert on device mapping and flash-attention generation issues for Qwen3.5 are not the same code path." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Cache-path collisions for `trust_remote_code` and permission propagation in `save_pretrained` are unrelated filesystem behaviors." + } + ] + }, + "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": 16, + "review_unit_index": 11, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43388", + "issue:43404", + "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:44292", + "issue:44361", + "issue:44387", + "issue:44423", + "issue:44479", + "issue:44610", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44918", + "issue:44991", + "issue:45059", + "issue:45072", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704" + ], + "soft_pairs": [ + "issue:43404|issue:45357", + "issue:40990|issue:45701", + "issue:43873|issue:44756", + "issue:44186|issue:45072", + "issue:44423|issue:45406", + "issue:42617|issue:42915", + "issue:43577|issue:43708", + "issue:44387|issue:45704", + "issue:44877|issue:44991", + "issue:43526|issue:43881", + "issue:36331|issue:42617", + "issue:36331|issue:42915", + "issue:43450|issue:45701", + "issue:43550|issue:43720", + "issue:44610|issue:44898", + "issue:44479|issue:44918", + "issue:44186|issue:44610", + "issue:43316|issue:44913", + "issue:42915|issue:43317", + "issue:43550|issue:43606", + "issue:43295|issue:45701", + "issue:41720|issue:42617", + "issue:44671|issue:44743", + "issue:41720|issue:45161", + "issue:43531|issue:43572", + "issue:44292|issue:44479", + "issue:36010|issue:42617", + "issue:44186|issue:44361", + "issue:43540|issue:43782", + "issue:43388|issue:45059", + "issue:41720|issue:43873", + "issue:42915|issue:43322", + "issue:45406|issue:45464", + "issue:36331|issue:41720" + ], + "prepared_review_unit_hash": "05eb5f197e9da3f9d27c98613ec833a41ef35085b41c73ab110e2b1d9457b087", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37800, + "estimated_input_tokens": 9450, + "estimated_eval_tokens": 19156 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is heterogeneous: most items are unrelated model-, tokenizer-, training-, or device-mapping bugs. None of the proposed soft edges look like the same underlying defect, so I would not merge any pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43873 is the broadest and most general issue in the set (offloading/quantization with multiple references), so it is the best representative if one must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 has the widest scope and strongest signal among otherwise unrelated issues, making it the best global issue representative.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Both involve multimodal/model key handling, but they are different models and different bugs: lm_head tying vs incorrect saved visual encoder keys." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "A perplexity regression on gpt-oss is not the same as a version-dependent tokenization change." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Both mention memory/offload, but one is quantization offloading behavior and the other is a Strix Halo mmap OOM workaround." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crashes on NER/batched padding are unrelated to dtype mismatches in bfloat16 inference." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve/multimodal crashes, but the failing code paths differ: continuous batching with string tensors vs Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "3d_parallel.py startup failure and Qwen3Moe FP8 failure are unrelated model/runtime issues." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "BLIP2 dtype loading behavior and Trainer checkpoint resume step accounting are unrelated." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Both are memory-related, but one is int4 reserved-memory growth in Transformers 5.x and the other is a T5 apex RMSNorm leak." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config loading for granite_speech and tokenizer loading for est-roberta are different failure modes in different subsystems." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "A BeitImageProcessorFast label-reduction bug is unrelated to a glm-4v-9b model loading failure." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "CustomTrainer.compute_loss signature breakage is unrelated to 3d_parallel.py failing to run." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "A trainer API mismatch is not the same bug as Qwen3Moe failing with FineGrainedFP8Config." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batched-shape regression and tokenization-version changes are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba and packed-weight unpacking in BitNet are different code paths." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both involve image size mismatches, but one is an OmDet processor/model expectation mismatch and the other is Perceiver interpolation logic." + }, + { + "left": "issue:44479", + "right": "issue:44918", + "accept": false, + "reason": "A video-input regression in Qwen VL models is unrelated to TRL SFT embedding unpacking." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "LayoutLMv2 tokenizer NER/padding crashes and OmDet input-size mismatch are different bugs." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy and GPTNeoXConfig rotary_pct reload behavior are unrelated config bugs." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Qwen3Moe FP8 failure and device_map auto dequantized-model offload failure are different model-loading problems." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and Bark CPU-offload device mismatch do not share the same underlying bug." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "A regression around processor.tokenizer/images in v4.57.5 is unrelated to general tokenization changes across versions." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping cuda asserts and a 3d_parallel script failure are not the same issue." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked-LM prediction errors and Qwen3.5 recurrent-state reset in modular code are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both are model-parallelism related, but one is an A800 cudaErrorAssert during auto device mapping and the other is GPT-OSS TP not working." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "A Qwen3-MoE sliding-window bug and a StableLmConfig pad_token_idx regression are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44479", + "accept": false, + "reason": "Qwen-3-8B-NVFP4 runtime errors are unrelated to the video-input regression in Qwen VL models." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import failure and 3d_parallel.py runtime failure are completely different issues." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "These are different tokenizer bugs affecting different tokenizers and tasks." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Qwen3OmniMoe video processing and Qwen3VL weight_only loading are different failure paths." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Metric gathering dropping tuple labels is unrelated to SAM3 text+box behavior." + }, + { + "left": "issue:41720", + "right": "issue:43873", + "accept": false, + "reason": "Both involve device/offload themes, but one is a Qwen3 cuda assert and the other is a broader quantization/offloading complaint." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor _tokenizer crash and Qwen3.5 streaming inference failure are different models and different bugs." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "Trainer compute_loss API breakage is unrelated to Qwen3 device-mapping crashes." + } + ] + }, + "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": 16, + "review_unit_index": 12, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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:44492", + "issue:44509", + "issue:44514", + "issue:44545", + "issue:44589", + "issue:44610", + "issue:44625", + "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:45701", + "issue:45704", + "issue:45759" + ], + "soft_pairs": [ + "issue:43421|issue:45701", + "issue:30064|issue:34634", + "issue:33357|issue:35707", + "issue:44492|issue:44509", + "issue:41720|issue:43746", + "issue:43606|issue:43720", + "issue:43873|issue:45704", + "issue:44514|issue:44545", + "issue:43550|issue:43653", + "issue:44625|issue:45479", + "issue:43531|issue:45759", + "issue:43606|issue:43653", + "issue:43421|issue:43927", + "issue:34634|issue:41720", + "issue:34567|issue:35707", + "issue:44368|issue:45357", + "issue:44451|issue:44991", + "issue:44898|issue:45072", + "issue:43782|issue:45381", + "issue:44077|issue:44625", + "issue:44062|issue:44451", + "issue:43421|issue:43653", + "issue:44987|issue:45042", + "issue:42886|issue:44843", + "issue:45588|issue:45759", + "issue:43531|issue:44589", + "issue:43421|issue:44186", + "issue:44918|issue:44936", + "issue:43381|issue:43844", + "issue:39692|issue:44610" + ], + "prepared_review_unit_hash": "9b18ae9013403b2979de2471d0adc1b13e4b3467a0d5c8372c5df6096f4c1c36", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36272, + "estimated_input_tokens": 9068, + "estimated_eval_tokens": 18392 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are unrelated; the only clear duplicate is the Qwen2_5_VLProcessor batched padding=False crash. No PRs are present.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44514 is the earliest report of the Qwen2_5_VLProcessor.apply_chat_template batched-input crash, and issue:44545 appears to be the same bug reported later.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44514 is the best canonical issue candidate because it is the original report for the only true duplicate pair and describes the same concrete failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43421", + "right": "issue:45701", + "accept": false, + "reason": "Both concern tokenization-related behavior, but one is about runtime special-token updates and the other about tokenizer output changing across versions; not the same bug." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Different subsystems and failures: image processor segmentation maps vs BarkProcessor voice_preset handling." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "Unrelated errors: MacOS bus error with a CLIP model vs progressive generation with inputs_embeds/past_key_values." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "Both are documentation/style-related, but they target different doc issues and are not the same change." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Different problems: Qwen3 auto device mapping CUDA assert vs PEFT adapter loading from local checkpoints." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Both are CI/model-loading failures, but one is CPU offload device mismatch and the other is unpacking packed BitNet weights during accelerate loading." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization/offloading vs T5 apex RMSNorm memory leak; different code paths and fixes." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same title and same concrete failure: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Different model-specific bugs: torch.compile/SDPA for Bamba vs BigBirdTokenizer special-token registration." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "One is Qwen3.5 config propagation for num_labels; the other is a sequence-classification loss bug across models." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Both involve Qwen3.x configs, but one is a sliding_window issue in MoE models and the other is from_config unwrapping text_config; not the same bug." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Distinct issues: CPU offload device mismatch vs tokenizer mask token registration/empty decode output." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Runtime special-token/post-processor behavior is different 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 are unrelated." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token counting and progressive generation with cache are unrelated bugs." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "Different Qwen3.5 problems: a training warning about tie_word_embeddings vs incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading complaints, but they concern different models and likely different root causes." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Perceiver image resolution handling and dtype mismatch in SwitchTransformers/TimmWrapperModel are different failures." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3VL weight_only loading and qwen2.5-vl video vision_position_ids are unrelated multimodal bugs." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "PatchTSMixer post_init policy is unrelated to Qwen3.5 num_labels propagation." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken constructor argument conflict is different from a model-loading failure for ScandiBERT." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Special-token runtime updates and BigBirdTokenizer mask-token registration are related to tokenizers in general, but not the same underlying defect." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Different issues: loading physical-intelligence/fast on newer transformers vs PIL image processors requiring torchvision." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Both involve offline mode, but one is tokenizer cache loading and the other is an unconditional model_info() call in mistral regex patching." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Flash attention sink-less model crash and Qwen multimodal from_config handling are different code paths." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding_window behavior for Qwen3-MoE is unrelated to missing Float8 storage lookup." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Special-token runtime updates and LayoutLMv2Tokenizer NER/batched padding crashes are different tokenizer bugs." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "TRL SFT input-embedding unpacking and trainer.evaluate() after train() are different training workflow issues." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Gradient checkpointing in eval mode and DeepSpeed ZeRO-3 gradient growth are unrelated training bugs." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 docs example errors and OmDet-Turbo processor/model resolution mismatch are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The analysis is grounded in the packet. The duplicate call on issue:44514/44545 is well supported by the matching titles and chronology, and the rest of the soft-edge rejections stay conservative rather than overclaiming equivalence. The note that no PRs are present is also consistent with the packet." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 13, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:42915", + "issue:43322", + "issue:43335", + "issue:43381", + "issue:43421", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43716", + "issue:43720", + "issue:43746", + "issue:43756", + "issue:43819", + "issue:43873", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44488", + "issue:44661", + "issue:44671", + "issue:44756", + "issue:44811", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:45072", + "issue:45081", + "issue:45183", + "issue:45356", + "issue:45381", + "issue:45479", + "issue:45685", + "issue:45704", + "issue:45759" + ], + "soft_pairs": [ + "issue:43994|issue:44671", + "issue:43525|issue:43531", + "issue:45081|issue:45356", + "issue:43596|issue:44220", + "issue:41720|issue:45704", + "issue:43381|issue:45704", + "issue:39692|issue:43873", + "issue:43746|issue:43873", + "issue:40444|issue:43746", + "issue:42915|issue:45759", + "issue:43746|issue:45381", + "issue:44220|issue:44661", + "issue:45183|issue:45479", + "issue:43618|issue:43697", + "issue:43540|issue:44008", + "issue:43720|issue:43819", + "issue:30064|issue:36010", + "issue:44756|issue:45759", + "issue:43716|issue:44008", + "issue:43335|issue:43756", + "issue:43653|issue:43720", + "issue:44062|issue:44488", + "issue:43322|issue:45685", + "issue:43550|issue:45072", + "issue:44186|issue:44898", + "issue:43550|issue:43819", + "issue:44843|issue:45081", + "issue:43421|issue:44811", + "issue:43643|issue:44877" + ], + "prepared_review_unit_hash": "7efa06bd9315f4cb57142f25ab1f85e4ac08b1efb5f49745a22a1a716210983f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36117, + "estimated_input_tokens": 9030, + "estimated_eval_tokens": 18316 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous cluster. issue:45081 is the best anchor for the Mistral tokenizer regression family, but the Mistral-related pair with issue:45356 should stay rejected: they share a theme and keyword, not enough evidence that they are the same concrete bug.", + "confidence": 0.72, + "canonical_issue_reason": "issue:45081 is the best canonical issue because it is open, directly names the failing `_patch_mistral_regex` path, and is the clearest actionable report in the Mistral tokenizer area.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45081 is the strongest representative issue: it is open, specific, and best captures the shared Mistral regex/tokenizer problem without over-merging distinct regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and failures: SigLIP2 output/pipeline corruption is not the same bug as CamemBERT masked-LM regression." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Separate Qwen3 issues with different symptoms: missing pad_token_id vs sliding_window behavior." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Related Mistral tokenizer theme, but only title-level similarity; no shared files, targets, or concrete evidence of the same underlying bug." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Unrelated subsystems: DeepSpeed ZeRO-3/BERT init failure vs audio fbank extraction." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "Different code paths and failure modes: Qwen3 device-map CUDA assert vs T5 apex RMSNorm memory leak." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing eval-mode behavior is unrelated to the T5 RMSNorm memory leak." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "Doc-example/model mismatch and quantization failure are not the same as generic offloading under quantization." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Both involve loading/quantization, but the concrete bugs are different." + }, + { + "left": "issue:40444", + "right": "issue:43746", + "accept": false, + "reason": "Different multimodal loading failures: iterable dataset with multiple images vs GraniteSpeech checkpoint loading." + }, + { + "left": "issue:42915", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3Moe FP8 failure and composite config unwrapping are distinct bugs." + }, + { + "left": "issue:43746", + "right": "issue:45381", + "accept": false, + "reason": "Different models and failures: GraniteSpeech checkpoint loading vs Qwen2.5-VL video position IDs." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction and tokenizer mapping-name registration are unrelated." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Generic input-validation messaging is distinct from degenerate zero-loss classification behavior." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIP attentions assignment and RTDetrV2 output drift are separate regressions." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Unrelated failures: Qwen3OmniMoe video processing vs Gemma3n variable collision." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Packed-weight loading in BitNet and DAC forward/from_latents mismatch are different concrete bugs." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Void segmentation-map processing has nothing to do with a GenerationMixin import error." + }, + { + "left": "issue:44756", + "right": "issue:45759", + "accept": false, + "reason": "Strix Halo mmap/OOM and Qwen multimodal config unwrapping are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Mistral image-preprocessor dtype mismatch and Gemma3n audio-mask collision are different failures." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Sparse-layer creation in SwitchTransformers and Smollm3 RoPE-layer dropping are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight unpacking are separate issues." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken constructor duplication and model-loading failure for cjvt/sleng-bert are not the same bug." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Llava Next segfault and MPS histogram-on-Int error are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "torch.compile SDPA failure in Bamba is not the same as dtype mismatches in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer NER/padding crash and Perceiver non-default-resolution failure are different subsystems." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Different model/runtime failures with no shared concrete code path." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both touch `_patch_mistral_regex`, but one is HF offline handling and the other is a backend-attribute crash; not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Runtime special-token post-processor update bug is distinct from Whisper batch_decode skip_special_tokens handling." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, but it overstates one key soft-edge verdict. The packet\u2019s only Mistral-related pair, issue:45081 vs issue:45356, is marked deterministic_accept=false and is supported only by soft title similarity; there are no shared filenames, shared targets, or direct evidence that they are the same concrete bug. Treating that pair as accepted is not conservative enough. Everything else looks appropriately cautious, but this one verdict should remain rejected or at most tentative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 14, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:30064", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:41720", + "issue:42175", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43643", + "issue:43644", + "issue:43650", + "issue:43653", + "issue:43720", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43906", + "issue:43994", + "issue:44060", + "issue:44186", + "issue:44291", + "issue:44368", + "issue:44442", + "issue:44448", + "issue:44485", + "issue:44509", + "issue:44561", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44821", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45381", + "issue:45406", + "issue:45701" + ], + "soft_pairs": [ + "issue:43577|issue:43818", + "issue:43531|issue:43643", + "issue:43540|issue:45381", + "issue:44448|issue:44671", + "issue:43653|issue:43819", + "issue:43782|issue:44977", + "issue:44561|issue:45020", + "issue:43644|issue:43749", + "issue:43825|issue:43994", + "issue:43577|issue:43582", + "issue:44821|issue:45084", + "issue:41720|issue:44991", + "issue:30064|issue:36331", + "issue:44186|issue:44442", + "issue:43825|issue:43827", + "issue:44485|issue:44509", + "issue:44671|issue:45701", + "issue:44805|issue:44936", + "issue:43606|issue:45072", + "issue:34634|issue:36010", + "issue:42175|issue:45081", + "issue:44610|issue:45072", + "issue:43782|issue:45325", + "issue:45081|issue:45406", + "issue:44060|issue:44368", + "issue:43720|issue:45072", + "issue:43650|issue:43906", + "issue:43749|issue:44291" + ], + "prepared_review_unit_hash": "7fa13227828298125442195d6d93d96133d991c3daa9d7b41d8ed9aed9a65b1a", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36241, + "estimated_input_tokens": 9061, + "estimated_eval_tokens": 18378 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a heterogeneous mix of Transformers bug reports that only share broad themes like v5 regressions, model loading, tokenization, or processor issues. None of the soft-similarity pairs look like true duplicates of the same underlying bug/change, so I rejected all of them. The broadest, most representative issue is the open remote_code regression tracker.", + "confidence": 0.62, + "canonical_issue_reason": "issue:45020 is the broadest umbrella in the set: it captures recent-version regressions affecting remote_code models and is open with active discussion, making it the best representative issue for this noisy cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the most generally useful issue to anchor the cluster because it is broad, current, and close to the recurring theme across several reports about recent Transformers breakage.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "Different models and different failure modes: dtype loading for BLIP-2 vs missing temporal attention/shared weights in Video-LLaVA." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Unrelated bugs: Qwen3-MoE sliding_window behavior vs AutoConfig missing fields with trust_remote_code." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Both involve video inputs, but they target different model families and distinct bugs (Qwen3OmniMoe ValueError vs Qwen2.5-VL position-id scaling)." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Both are v5 output regressions, but they affect different models and tasks, so they do not look like the same underlying bug." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Completely different code paths: tokenizer special-token registration vs DAC latent decoding/STE mismatch." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Different Qwen variants and different stages of failure: loading with weight_only vs generation with flash-attention." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Related theme, but not the same bug: one is a specific is_torch_fx_available removal breakage, the other is a broad remote_code regression report." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Both are loading regressions, but one is non-persistent buffer initialization and the other is FSDP CPU RAM efficient loading; too different to merge." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Different issues: an incorrect pipeline error message vs a model producing nonsensical outputs through AutoModel/pipeline." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "No shared underlying bug: BLIP-2 dtype handling vs Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "Both are runtime errors, but one is AutoImageProcessor-from-URL loading and the other is template compilation; unrelated." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Different symptoms and likely causes: Qwen3 auto device mapping CUDA assert vs tokenizer loading regression for est-roberta." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image processor segmentation-map failure and CustomTrainer compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "Tokenizer crashes on NER/batched padding vs AutoTokenizer failing to load a specific tokenizer are not the same bug." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "Both mention removed pipeline tasks, but one is an error-message issue and the other is stale docs; not the same defect." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion is unrelated to docs still mentioning removed pipeline tasks." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "A specific CamemBERT masked-LM regression is not the same as a general tokenization-change report." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: mask/index shape mismatch vs trainer.evaluate failing after training." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Different model/device issues: CPU offload device mismatch vs bfloat16 dtype mismatches." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "BarkProcessor voice_preset behavior and a GenerationMixin import error are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Packaging/backend install issue vs Mistral tokenizer regex patch crash; no shared underlying bug." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor image-size mismatch and bfloat16 dtype mismatch are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen VL bugs: weight_only load error vs incorrect still-image temporal position_ids." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Both are AttributeErrors during loading/serve, but they involve different objects and causes, so they are not duplicates." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings, but one is a model-internal incorrect warning and the other is a LoRA fine-tuning warning; not the same bug." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "BitNet packed-weight loading failure is unrelated to bfloat16 dtype mismatch issues." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "A placeholder/data-request issue and an isolated reproduction report are not the same bug or change." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "Two distinct loading regressions: FSDP CPU RAM efficient loading vs init_empty_weights TypeError from _is_hf_initialized." + } + ] + }, + "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": 16, + "review_unit_index": 15, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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:43643", + "issue:43645", + "issue:43653", + "issue:43696", + "issue:43697", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43819", + "issue:43854", + "issue:43927", + "issue:43931", + "issue:44479", + "issue:44554", + "issue:44568", + "issue:44589", + "issue:44610", + "issue:44756", + "issue:44811", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45042", + "issue:45084", + "issue:45356", + "issue:45488", + "issue:45538", + "issue:45600", + "issue:45759" + ], + "soft_pairs": [ + "issue:43582|issue:44857", + "issue:43526|issue:43697", + "issue:43262|issue:43819", + "issue:43526|issue:44554", + "issue:45084|issue:45538", + "issue:34634|issue:36331", + "issue:43329|issue:45600", + "issue:43653|issue:44912", + "issue:43452|issue:44991", + "issue:43638|issue:44936", + "issue:43257|issue:43329", + "issue:43746|issue:44811", + "issue:44589|issue:45042", + "issue:38175|issue:44610", + "issue:43454|issue:43577", + "issue:43854|issue:44977", + "issue:43257|issue:43696", + "issue:36010|issue:43645", + "issue:44568|issue:45488", + "issue:39692|issue:44912", + "issue:43819|issue:43927", + "issue:43720|issue:44610", + "issue:43550|issue:44912", + "issue:43749|issue:44756", + "issue:44568|issue:45356", + "issue:43262|issue:44811", + "issue:43531|issue:45042", + "issue:36010|issue:36331", + "issue:43643|issue:45759", + "issue:43329|issue:44479", + "issue:43257|issue:43931" + ], + "prepared_review_unit_hash": "758bc0f2b7333327127fd3cc7fa456c93205223bf6a69c2f1454a9124b927d7c", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36672, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it spans unrelated bugs in tokenizers, audio/video processors, model loading, training APIs, quantization, and hardware-specific failures. None of the soft pairs look like the same underlying defect, so there are no duplicate merges here.", + "confidence": 0.98, + "canonical_issue_reason": "No true canonical issue: the items are not one bug cluster, but a mix of unrelated regressions across different subsystems and model families.", + "canonical_pr_reason": null, + "best_issue_reason": "If one representative must be chosen, issue:43749 is the broadest and most externally visible loading regression in the set, but it should not be treated as a duplicate canonical for the rest.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Different failures: Apple Silicon allocator warmup TypeError vs CUDA float16 AMP crash in LwDetrImageLoss." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to RTDetrV2 output drift in v5." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat template sampling-rate default bug is unrelated to DAC.from_latents / missing STE mismatch." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Image processor label handling is unrelated to the MPS attention correctness issue." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Non-template node compilation error and CLIPTokenizer max_length bug are different code paths." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Bark voice_preset failure is unrelated to CustomTrainer.compute_loss signature changes." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Video multimodal token counting bug is unrelated to stale auto_mappings CI references." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBird special-token decode bug is unrelated to MXFP4 quantization loading fallback." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "GGUF from_pretrained breakage is not the same as the EMBEDDIA tokenizer load regression." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Zero3 non-pretrained BERT crash is a different problem from evaluate-after-train failure." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed is unrelated to video token helper undefined names." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "PEFT adapter local checkpoint loading and Whisper batch_decode skip_special_tokens are distinct bugs." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Float8 storage lookup TypeError is unrelated to PIL backend torchvision requirement regressions." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 zero-probability outputs and OmDet 640px vs 224px processor mismatch are different issues." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "AyaVision lm_head tying failure is unrelated to Blip2 dtype propagation." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "GLM-4.7-Flash load test failure is not the same as Qwen3.5 flash-attention generation regression." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "MoE weight conversion/loading bug is unrelated to GPT-oss out-of-memory." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import error and Jupyter custom-model init regression are separate API breakages." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "mdeberta add_special_tokens BOS/EOS regression is different from LlamaTokenizer ByteLevel-to-Metaspace override." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "SigLIP2 documentation example problems are unrelated to git-oss-20b MXFP4 quantization loading." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent/forward mismatch is unrelated to DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to OmDet processor size mismatch." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Bamba torch.compile + SDPA failure is unrelated to quantization fallback in git-oss-20b." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different loading/hardware issues." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer regressions: missing BOS/EOS addition vs codec handling / warning behavior in Kimi-K2.5." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "apply_chat_template sampling-rate default bug is unrelated to Whisper batch_decode special-token handling." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to PIL backend torchvision dependency issues." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Importing GenerationMixin and compute_loss keyword-argument mismatch are unrelated failures." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "trust_remote_code missing fields in AutoConfig is not the same as unwrapping text_config in AutoModelForCausalLM.from_config." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Both involve video paths, but one is an internal undefined-name bug in token counting while the other is a model-specific video regression; not enough to treat as the same defect." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "MoE loading/conversion under accelerate+deepspeed is unrelated to Qwen3-VL weight-shape mismatch." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 16, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:28282", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:38175", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43329", + "issue:43450", + "issue:43452", + "issue:43526", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43742", + "issue:43756", + "issue:43824", + "issue:43906", + "issue:43927", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44560", + "issue:44737", + "issue:44821", + "issue:44849", + "issue:44898", + "issue:44912", + "issue:44938", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45137", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45479", + "issue:45538", + "issue:45588", + "issue:45701" + ], + "soft_pairs": [ + "issue:34567|issue:40990", + "issue:43329|issue:43450", + "issue:44849|issue:45701", + "issue:43257|issue:43756", + "issue:43697|issue:45357", + "issue:44987|issue:45538", + "issue:43452|issue:43582", + "issue:41720|issue:43540", + "issue:42915|issue:43742", + "issue:44291|issue:45042", + "issue:44479|issue:45381", + "issue:33357|issue:34567", + "issue:43577|issue:45072", + "issue:42915|issue:44821", + "issue:42915|issue:45538", + "issue:42175|issue:43824", + "issue:43317|issue:45538", + "issue:43540|issue:44560", + "issue:43927|issue:44912", + "issue:43540|issue:44479", + "issue:44062|issue:44991", + "issue:42175|issue:44488", + "issue:42175|issue:44938", + "issue:45137|issue:45405", + "issue:43526|issue:45479", + "issue:38175|issue:44898", + "issue:44737|issue:45588", + "issue:43906|issue:45381", + "issue:28282|issue:34689" + ], + "prepared_review_unit_hash": "0c5a27ae8180f760e8178ac5d7c2ce828fe7346ce3f9e7d265c32461f0c76611", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35926, + "estimated_input_tokens": 8982, + "estimated_eval_tokens": 18220 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous transformer bug reports with only one plausible duplicate subcluster around Qwen video-input regressions. Most soft pairs are not the same underlying issue.", + "confidence": 0.33, + "canonical_issue_reason": "Issue 44479 is the best umbrella candidate because it describes the broad v5.3.0 video-input regression affecting several Qwen video models, and it can subsume the narrower Qwen2.5-VL report in 45381.", + "canonical_pr_reason": null, + "best_issue_reason": "44479 is the most representative issue here: it is broad, regression-focused, and closest to the only plausible duplicate subcluster in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counting and GPT-OSS perplexity are unrelated bugs in different code paths." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined names in multimodal token counting vs batched video processor shape handling are different failures." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Qwen3.5 hidden-states behavior and version-dependent tokenization changes are not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed and Smollm3 RoPE-layer dropping are unrelated model issues." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "RTDetrV2 output differences and Qwen3.5 save_pretrained key persistence affect different models and code paths." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Loading a specific model fails for one reason, while CLIPTokenizer max length is a separate tokenizer constant bug." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf_file loading failures and Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and Qwen3OmniMoe video processing ValueError are different failures." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and MobileLLM key error do not share the same underlying code path." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "init_empty_weights argument handling and PIL backend torchvision dependency are separate regressions." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Both report the same class of v5.3.0 Qwen video-input regression; 45381 looks like a narrower manifestation of the broader 44479 bug." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS CLIP bus error and TrainerState token accounting are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "BLIP2 dtype persistence and SwitchTransformers/TimmWrapperModel bf16 mismatch are different bugs." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FP8 loading and AutoImageProcessor-from-URL are not the same issue." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Qwen3Moe FP8 configuration and CLIPTokenizer max_length are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "TensorFlow backend packaging and missing Qwen2.5-VL import are different package/import problems." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "device_map=auto offload failure and CLIPTokenizer model_max_length are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both involve video, but they are different models and different error modes, so they are not clearly the same bug." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig save/load token IDs and MXFP4 loading fallback are different configuration bugs." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Qwen3OmniMoe video ValueError and the broader Qwen video-input regression are too different in model family and symptom to merge confidently." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "AddedToken special-argument error and tokenizer loading failure for a specific model are distinct problems." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "TensorFlow backend packaging and model loading for cjvt/sleng-bert are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "TensorFlow backend packaging and Python 3.14 loading failure are not the same bug." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error and PEFT version bump are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and zero-loss sequence classification are different bugs." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Siglip2 zero probabilities and Perceiver non-default resolution failure are separate model-specific issues." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU tensor allocation and flash_attention sink-less AttributeError are unrelated." + }, + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "A reproduction of another issue and a Qwen video-position-id regression are not the same underlying bug." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Missing PyTorch import error and Llama 3.2 vision model loading regression are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster-level summary is grounded in the packet, and the only accepted soft pair (44479/45381) is a conservative, well-supported duplicate call based on the shared Qwen video-input regression theme. The rejected pairs are generally cautious and do not overclaim sameness across clearly different model/code paths. " + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 17, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:34567", + "issue:36331", + "issue:41762", + "issue:43452", + "issue:43493", + "issue:43531", + "issue:43577", + "issue:43582", + "issue:43643", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43994", + "issue:44062", + "issue:44164", + "issue:44291", + "issue:44466", + "issue:44484", + "issue:44488", + "issue:44514", + "issue:44561", + "issue:44589", + "issue:44625", + "issue:44704", + "issue:44792", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44936", + "issue:44964", + "issue:45005", + "issue:45042", + "issue:45081", + "issue:45216", + "issue:45290", + "issue:45325", + "issue:45375", + "issue:45381", + "issue:45405", + "issue:45600" + ], + "soft_pairs": [ + "issue:43749|issue:44164", + "issue:43742|issue:44291", + "issue:28282|issue:29127", + "issue:43756|issue:45325", + "issue:43582|issue:43782", + "issue:43493|issue:43994", + "issue:44062|issue:45081", + "issue:34567|issue:36331", + "issue:43643|issue:45375", + "issue:44291|issue:44821", + "issue:43452|issue:44488", + "issue:44792|issue:44964", + "issue:45042|issue:45081", + "issue:44484|issue:44704", + "issue:45405|issue:45600", + "issue:43742|issue:44589", + "issue:43531|issue:44821", + "issue:43688|issue:43749", + "issue:43697|issue:45216", + "issue:44514|issue:45381", + "issue:43643|issue:44561", + "issue:44625|issue:44877", + "issue:43452|issue:43577", + "issue:43720|issue:44898", + "issue:44514|issue:45290", + "issue:43452|issue:44843", + "issue:41762|issue:44936", + "issue:24643|issue:29127", + "issue:44466|issue:45005" + ], + "prepared_review_unit_hash": "12cb4a235a328e662df6943e95021d5a14e16ecd09ebc1f171cad0268a3f6e5e", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36669, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. Nearly all soft pairs are adjacent but not true duplicates; only the two SigLIP2 reports look like the same underlying model-implementation discrepancy.", + "confidence": 0.89, + "canonical_issue_reason": "issue:43493 is the clearest, most specific report of the underlying SigLIP2 correctness problem and is the better anchor for the only plausible duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43493 has the strongest reproducible bug description and best matches the likely shared root cause for the SigLIP2 pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43749", + "right": "issue:44164", + "accept": false, + "reason": "Both involve loading/saving paths, but one is FSDP CPU RAM-efficient loading and the other is extra_state serialization; different bugs and code paths." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Both fail during model loading, but MobileLLM key errors and init_empty_weights/_is_hf_initialized errors are distinct causes." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Completely different problems: missing PyTorch import vs. a LayoutLMv3 box-validation error-message issue." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both mention RoPE/position handling, but they affect different models and different logic; not the same concrete bug." + }, + { + "left": "issue:43582", + "right": "issue:43782", + "accept": false, + "reason": "Different failures in unrelated areas: Apple Silicon allocator warmup vs. Qwen3VL weight-only loading." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both describe SigLIP2 producing incorrect outputs relative to the expected implementation; one is the JAX-vs-HF discrepancy and the other is the same symptom in model/pipeline use." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer-related, but one is an AddedToken constructor conflict and the other is a Mistral regex patch crash; different root causes." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Both are trainer-related, but num_input_tokens_seen not updating and compute_loss signature mismatch are separate bugs." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both concern config handling, but one is generic trust_remote_code field loss and the other is a specific config field dropped by strict schema; not the same bug." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "Different loading paths and failures: init_empty_weights with an unexpected argument vs. AutoImageProcessor loading from URL." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading complaints, but gguf_file support and a specific sleng-bert load failure are not the same underlying issue." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Different multimodal model failures; a failing janus test case is not the same as loading Phi-4 multimodal instruct." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Unrelated bugs: PIL image processors wrongly requiring torchvision vs. a tokenizer regex crash." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "Save-sharding defaults and AutoProcessor kwargs forwarding are unrelated." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "Unreleased PEFT version pinning and removed auto_mappings config references are separate release/CI issues." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Different load-time errors with different causes and affected code paths." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to AutoImageProcessor URL loading." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Both touch training/loading infrastructure, but auxiliary-loss normalization and FSDP loading are unrelated bugs." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "Different regressions in different models: RTDetrV2 output changes vs. Qwen3.5 checkpoint save_pretrained behavior." + }, + { + "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 IDs; not the same bug." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both affect trust_remote_code workflows, but missing config fields and removal of is_torch_fx_available are different failure modes." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config propagation/strictness, but they affect different models and different mechanisms." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "gguf_file loading failures and Blip2 dtype propagation are unrelated." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "BitNet packed-weight loading and Perceiver interpolation behavior are unrelated code paths." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Same public API, but the crashes are on different inputs and edge cases; too broad to treat as one bug." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "gguf_file loading issues and offline mistral-regex patch behavior are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Both are training/loading related, but Gemma3 DeepSpeed ZeRO-3 loading and trainer.evaluate() after train() are distinct problems." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "Completely different issues: DeepSpeed weight-shape runtime error vs. LayoutLMv3 box-validation messaging." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both concern tied weights, but one is inconsistent serialization by device and the other is translation-model tied-weight issues; not enough evidence for same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge calls stay conservative. The only accepted pair, issue:43493 with issue:43994, is a reasonable same-bug match based on the shared SigLIP2 output-discrepancy symptom. The rejected pairs are mostly justified by distinct titles/bug classes without overstating evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 18, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:28282", + "issue:30064", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:34567", + "issue:35141", + "issue:38175", + "issue:41720", + "issue:43122", + "issue:43295", + "issue:43329", + "issue:43452", + "issue:43504", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44361", + "issue:44442", + "issue:44451", + "issue:44464", + "issue:44479", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44625", + "issue:44743", + "issue:44811", + "issue:44843", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45325", + "issue:45356", + "issue:45440", + "issue:45588", + "issue:45701", + "issue:45704" + ], + "soft_pairs": [ + "issue:43295|issue:45356", + "issue:43819|issue:44811", + "issue:43653|issue:44568", + "issue:44464|issue:44743", + "issue:43688|issue:45440", + "issue:43688|issue:45704", + "issue:41720|issue:43906", + "issue:38175|issue:43295", + "issue:44361|issue:44442", + "issue:43653|issue:44361", + "issue:30064|issue:33666", + "issue:28282|issue:32090", + "issue:43646|issue:43749", + "issue:43577|issue:43697", + "issue:43688|issue:45701", + "issue:44479|issue:44936", + "issue:28282|issue:33357", + "issue:44843|issue:44991", + "issue:28282|issue:35141", + "issue:44977|issue:45356", + "issue:43577|issue:45588", + "issue:43531|issue:43742", + "issue:43653|issue:43927", + "issue:43329|issue:45325", + "issue:43122|issue:43906", + "issue:44464|issue:44977", + "issue:43577|issue:44554", + "issue:43550|issue:43927", + "issue:44479|issue:44625", + "issue:43550|issue:44361", + "issue:43504|issue:43720", + "issue:32090|issue:34567", + "issue:43504|issue:43550", + "issue:41720|issue:44464", + "issue:32090|issue:33666", + "issue:43504|issue:44610", + "issue:43452|issue:44451" + ], + "prepared_review_unit_hash": "8293159639519b8ead9728c9615676121b2713b88a74606f95eefcfc08ff2841", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 37, + "serialized_chars": 38916, + "estimated_input_tokens": 9729, + "estimated_eval_tokens": 19714 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is not a single duplicate cluster; it mixes many unrelated regressions across tokenization, loading, training, vision, and compilation. If one issue must represent the cluster, the broad tokenizer-regression report is the best umbrella, but the soft pairs do not look like true duplicates.", + "confidence": 0.31, + "canonical_issue_reason": "No strong duplicate anchor exists. Issue 45701 is the broadest representative of the recurring version-dependent tokenization complaints and has the most cross-references, so it is the best umbrella issue if one must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "45701 is the most general and central issue in this mixed set, so it is the best single triage representative even though it is not a confirmed duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different behaviors and code paths: processor/tokenizer access vs codec/regex handling." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC latent decoding and Whisper batch decoding are unrelated features with different failure modes." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both mention special-token behavior, but one is BigBird decode registration and the other is mDeBERTa BOS/EOS insertion." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Chunked generation with compiled forward is a different bug from recurrent-state reset in modular_qwen3_5." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Aux-loss normalization and DeepseekV3MoE divergence are different model/math bugs." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Aux-loss normalization is unrelated to the T5 apex RMSNorm memory leak." + }, + { + "left": "issue:41720", + "right": "issue:43906", + "accept": false, + "reason": "Qwen3 auto device mapping cuda asserts are not the same as the isolated reproduction issue linked to another bug." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Siglip2 zero probabilities and processor.tokenizer regression are different problems." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "MLukeTokenizer task errors and AutoTokenizer failing on FastSpeech2Conformer are distinct loading/attribute issues." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBird special-token decode and MLukeTokenizer AttributeError are unrelated tokenizer bugs." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Void segmentation maps in an image processor and Qwen2-VL multi-GPU training are different subsystems." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "Missing PyTorch import error and Trainer _gpu_broadcast_one NoneType error are unrelated runtime failures." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Custom model initialization breakage and FSDP CPU RAM efficient loading are different initialization/loading regressions." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "Blip2 dtype propagation and RTDetrV2 output drift are separate model-specific bugs." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Only a broad regression theme overlaps; aux-loss normalization is not the same as tokenization changing across versions." + }, + { + "left": "issue:44479", + "right": "issue:44936", + "accept": false, + "reason": "Video-input regression and trainer.evaluate after trainer.train affect different execution paths." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "PyTorch missing import and MacOS bus error are different environment/runtime issues." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "HF_HUB_OFFLINE patch behavior and tokenizer loading for a specific model are different tokenizer-loading bugs." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "ImportError for PyTorch and post_init reinitializing embeddings are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45356", + "accept": false, + "reason": "Flash-attention generation issues and Kimi tokenizer codec handling are different regressions." + }, + { + "left": "issue:43577", + "right": "issue:45588", + "accept": false, + "reason": "Blip2 dtype mismatch and flash_attention.py crashing on s_aux=None are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and a MobileLLM key error are different bugs." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "BigBird special-token decode and DiaConfig losing custom token IDs are both token-related, but the root causes differ." + }, + { + "left": "issue:43329", + "right": "issue:45325", + "accept": false, + "reason": "Undefined video helper names and rope index temporal scaling are separate video-VLM bugs." + }, + { + "left": "issue:43122", + "right": "issue:43906", + "accept": false, + "reason": "Tokenization differences across versions and an isolated reproduction of another issue are not the same bug." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled-forward chunked generation and flash-attention generation are different failures." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Blip2 dtype handling and an MPS attention correctness issue are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Bamba torch.compile SDPA failure and DiaConfig token ID loss are different code paths." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Video-input regression and Qwen3.5 num_labels propagation are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Bamba compile/S D P A issues and MLukeTokenizer AttributeError are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Beit legacy-field loading and BitNet packed-weight unpacking are different loading bugs." + }, + { + "left": "issue:32090", + "right": "issue:34567", + "accept": false, + "reason": "Trainer broadcast NoneType and TrainerState token counter not updating are different trainer issues." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Beit pretrained loading and Bamba compile/S D P A are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "Qwen3 auto device mapping asserts and chunked generation with compiled forward are different Qwen3 bugs." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Trainer NoneType broadcast and Qwen2-VL multi-GPU training are different domains." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Legacy-field loading in Beit and OmDet processor size mismatch are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "gguf_file breaking tokenizer/model loading and inability to load ScandiBERT are both loading failures, but not the same mechanism or model-specific bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 19, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:33357", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:43257", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43540", + "issue:43572", + "issue:43606", + "issue:43646", + "issue:43688", + "issue:43716", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43824", + "issue:43825", + "issue:43844", + "issue:43927", + "issue:44361", + "issue:44423", + "issue:44466", + "issue:44488", + "issue:44589", + "issue:44625", + "issue:44734", + "issue:44821", + "issue:44843", + "issue:44849", + "issue:44913", + "issue:44964", + "issue:44977", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45357", + "issue:45405", + "issue:45520", + "issue:45584", + "issue:45759" + ], + "soft_pairs": [ + "issue:43278|issue:43688", + "issue:43742|issue:44821", + "issue:43756|issue:44913", + "issue:43525|issue:43572", + "issue:33666|issue:36010", + "issue:43493|issue:43825", + "issue:44849|issue:45357", + "issue:43262|issue:43295", + "issue:43540|issue:43716", + "issue:43720|issue:43927", + "issue:43606|issue:43927", + "issue:43334|issue:43572", + "issue:43278|issue:43844", + "issue:33666|issue:36331", + "issue:43493|issue:43844", + "issue:39692|issue:43493", + "issue:43334|issue:45759", + "issue:43720|issue:44361", + "issue:43257|issue:43299", + "issue:44964|issue:45405", + "issue:43540|issue:44977", + "issue:44625|issue:45584", + "issue:44488|issue:44843", + "issue:43756|issue:45083", + "issue:45200|issue:45229", + "issue:43824|issue:44589", + "issue:43646|issue:44466", + "issue:44589|issue:45520", + "issue:39692|issue:43295", + "issue:33357|issue:33666", + "issue:43504|issue:43606", + "issue:44423|issue:44734" + ], + "prepared_review_unit_hash": "c4c876c1389763b07ff3e169931712d8c220928ef53513b6bf12360af585ce64", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37450, + "estimated_input_tokens": 9363, + "estimated_eval_tokens": 18982 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes many unrelated Transformers bugs across different models, configs, serving paths, and docs. None of the soft-edge pairs look like true duplicates, so I would not merge any of them.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out: the cluster spans unrelated regressions and model-specific bugs, so none is a safe representative.", + "canonical_pr_reason": null, + "best_issue_reason": "No clear best issue for deduplication; several Qwen-related reports recur, but they target different code paths and failure modes.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Different bugs: dtype drift in evaluation vs auxiliary-loss normalization." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "Unrelated failures: MobileLLM load key error vs loading AutoImageProcessor from URL." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "Both mention RoPE, but one is a model design bug and the other is config reload serialization." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Same broad theme of missing token IDs, but different models and fields; not the same bug." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training issue and GenerationMixin import error are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy vs pipeline translation error messaging; different issues." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Hidden-states bug and save_pretrained visual-key bug are separate regressions." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio chat-template sampling-rate default and processor/tokenizer regression are different code paths." + }, + { + "left": "issue:43540", + "right": "issue:43716", + "accept": false, + "reason": "Video-input processing failure vs image preprocessor/model dtype mismatch; not duplicates." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Packed-weight unpacking during accelerate loading is unrelated to DiaConfig token-ID loss." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch and config token-ID loss are different problems." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Different models and different missing config fields; not the same defect." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype change and ZeRO-3 gradient growth are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Training setup issue vs trainer signature incompatibility; distinct bugs." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 discrepancy vs DeepSpeed gradient abnormality are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Docs/example errors are not the same as a core HF-vs-JAX implementation discrepancy." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Both concern Qwen multimodal configs, but one is missing pad_token_id on load and the other is from_config unwrapping text_config." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet packed-weight loading and MLukeTokenizer AttributeError are unrelated." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both are Qwen3 MoE-related, but one is accelerate/deepspeed weight conversion and the other is dev-version model loading." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Model load failure and a dependency version pin issue are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Video-input processing bug vs flash-attention generation bug; different failures." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "num_labels propagation and Whisper empty-transcription generation are unrelated." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Model loading failure and offline-mode hub lookup bug are separate issues." + }, + { + "left": "issue:43756", + "right": "issue:45083", + "accept": false, + "reason": "RoPE-layer bug and helper length computation bug are different code paths." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Text-only token-type default and multi-GPU OOM are not the same defect." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Import error for Qwen2.5-VL and float8 storage lookup error are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization regression and lm_head serialization inconsistency are different bugs." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup and flash_attn KeyError on Python 3.13 are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Documentation example problems do not match the processor/tokenizer regression." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS CLIP bus error and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field load failure and bark CPU-offload device mismatch are distinct bugs." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Serve multimodal crash from string tensor handling and KV-cache continuation indexing bug are 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": 2, + "component_count": 16, + "review_unit_index": 20, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:15354", + "issue:29942", + "issue:30333", + "issue:37428", + "issue:38175", + "issue:39401", + "issue:43232", + "issue:43278", + "issue:43295", + "issue:43493", + "issue:43504", + "issue:43540", + "issue:43572", + "issue:43575", + "issue:43582", + "issue:43606", + "issue:43653", + "issue:43701", + "issue:43844", + "issue:43927", + "issue:43957", + "issue:43994", + "issue:44112", + "issue:44291", + "issue:44315", + "issue:44361", + "issue:44464", + "issue:44493", + "issue:44521", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44792", + "issue:44849", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45084", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45325", + "issue:45356", + "issue:45405", + "issue:45446", + "issue:45520", + "issue:45584", + "issue:45701" + ], + "soft_pairs": [ + "issue:43572|issue:44291", + "issue:43927|issue:44361", + "issue:43504|issue:43653", + "issue:43575|issue:45405", + "issue:44464|issue:45584", + "issue:43994|issue:44493", + "issue:43278|issue:43701", + "issue:44792|issue:45405", + "issue:43606|issue:44112", + "issue:44521|issue:45290", + "issue:29942|issue:30333", + "issue:43844|issue:44964", + "issue:45084|issue:45520", + "issue:44554|issue:45446", + "issue:44568|issue:45701", + "issue:15354|issue:38175", + "issue:43540|issue:45200", + "issue:43582|issue:44977", + "issue:45072|issue:45198", + "issue:43232|issue:44849", + "issue:37428|issue:43957", + "issue:39401|issue:44568", + "issue:44315|issue:44849", + "issue:44554|issue:45356", + "issue:43493|issue:43701", + "issue:43295|issue:44610", + "issue:45200|issue:45325" + ], + "prepared_review_unit_hash": "4be067b8df5b74014bed75d52f77247e9f4c83cc1cf62c91e30507cbb7b2b96a", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35863, + "estimated_input_tokens": 8966, + "estimated_eval_tokens": 18188 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: tokenizer regressions, generation bugs, loading/config issues, device/dtype mismatches, and unrelated CI failures are all mixed together. I do not see a single underlying bug/change unifying the items, so the soft edges should all be rejected.", + "confidence": 0.14, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No true canonical issue stands out because the cluster is noisy; if a placeholder is needed, issue:44521 is the most discussion-heavy and broadly representative of a multimodal chat-template regression, but it is not a real duplicate hub for the rest of the items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Both are transformer 5.x loading regressions, but one is a StableLmConfig padding-field issue and the other is an init_empty_weights/_is_hf_initialized constructor mismatch; different concrete code paths." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "One is a DiaConfig save/load regression causing generation IndexError, the other is an MLukeTokenizer AttributeError; both are tokenizer/config problems but not the same bug." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "BEiT semantic segmentation loading with a legacy field and BigBirdTokenizer special-token registration/empty decode are unrelated failures." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "Qwen2 tensor-parallel OOM and a PEFT minimum-version bump are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and Whisper empty-transcription failure after align_special_tokens are different generation bugs." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 nonsense outputs and a generic position-id key warning are not the same underlying issue." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "BF16-to-FP32 embedding dtype drift in eval is unrelated to resume_from_checkpoint key mismatches." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "CPU-offload device mismatch in bark-small and a stale GraniteSpeech CI test both mention device behavior, but they are different code paths and failure modes." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal inputs and the other is a crash on tool-call assistant messages with no content; not the same concrete bug." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures and MLflow job-reporting after failed training are unrelated." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient growth and Phi-4-multimodal loading failure are different model/runtime issues." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Non-template node compile errors and a Python 3.13 flash_attn import KeyError are unrelated." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "An MPS attention correctness bug and a PyTorch version check for flex_attention imports are distinct issues." + }, + { + "left": "issue:44568", + "right": "issue:45701", + "accept": false, + "reason": "Specific BOS/EOS token omission for mdeberta-v3-base and a broad tokenization-change report are too general/different to be duplicates." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure and SigLIP2 zero probabilities are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma 4 mm_token_type_ids defaults are different multimodal model bugs." + }, + { + "left": "issue:43582", + "right": "issue:44977", + "accept": false, + "reason": "Apple Silicon caching_allocator_warmup TypeError and Qwen3.5 flash-attention generation issues are unrelated." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "BF16 dtype mismatches in inference and Wav2Vec2 save/tokenization failures do not share the same root cause." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "sync_gpus/_update_model_kwargs_for_generation behavior and Qwen3.5 output_hidden_states regression are different generation-path bugs." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Flash attention import failure and meta-device loading breakage are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Wrong offset_mapping in Qwen3 tokenizer and add_special_tokens behavior for mdeberta-v3-base are different tokenizer bugs." + }, + { + "left": "issue:44315", + "right": "issue:44849", + "accept": false, + "reason": "Liger Kernel application with model_init and output_hidden_states regression are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Processor/tokenizer regression in v4.57.5 and OmDet-Turbo input-size mismatch are different multimodal issues." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Gemma 4 mm_token_type_ids defaults and Qwen2.5-VL rope-index scaling are unrelated multimodal bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 21, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:37428", + "issue:38175", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43388", + "issue:43425", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43526", + "issue:43572", + "issue:43575", + "issue:43653", + "issue:43819", + "issue:43824", + "issue:43827", + "issue:43844", + "issue:43856", + "issue:43873", + "issue:43901", + "issue:43957", + "issue:43975", + "issue:43994", + "issue:44112", + "issue:44206", + "issue:44360", + "issue:44442", + "issue:44464", + "issue:44466", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44792", + "issue:44857", + "issue:44898", + "issue:44936", + "issue:44987", + "issue:45083", + "issue:45127", + "issue:45198", + "issue:45245", + "issue:45412", + "issue:45563", + "issue:45584" + ], + "soft_pairs": [ + "issue:43334|issue:43525", + "issue:44898|issue:45412", + "issue:44466|issue:45127", + "issue:43873|issue:44464", + "issue:37428|issue:43824", + "issue:43572|issue:44987", + "issue:44792|issue:45563", + "issue:43844|issue:44936", + "issue:43572|issue:43957", + "issue:43388|issue:43526", + "issue:43824|issue:43975", + "issue:43295|issue:43450", + "issue:43295|issue:43873", + "issue:38175|issue:45412", + "issue:44514|issue:44521", + "issue:42915|issue:43824", + "issue:43278|issue:43856", + "issue:44442|issue:45198", + "issue:43295|issue:43994", + "issue:38175|issue:43493", + "issue:43827|issue:43901", + "issue:43653|issue:44112", + "issue:44206|issue:45584", + "issue:43450|issue:43493", + "issue:44360|issue:44561", + "issue:43450|issue:43575", + "issue:45127|issue:45245", + "issue:43819|issue:45083", + "issue:43262|issue:43450", + "issue:43819|issue:44112", + "issue:37428|issue:43425", + "issue:44610|issue:44857" + ], + "prepared_review_unit_hash": "f3ab976e8bee4b5fe23ff457308298160668759d7f6717760d8585acc14b6e9d", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37401, + "estimated_input_tokens": 9351, + "estimated_eval_tokens": 18958 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are clearly unrelated and should stay separate. The only plausible duplicate-like pair is the SigLIP2 implementation/output discrepancy; everything else differs by model, subsystem, or failure mode. No PRs are present.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43493 is the broadest and most diagnostic SigLIP2 report; it reads like the underlying implementation mismatch that could also explain the zero-probabilities symptom in issue:38175.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43493 best represents the cluster\u2019s only believable duplicate theme because it describes the core SigLIP2 correctness discrepancy rather than a single downstream symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Same general error shape ('missing pad_token_id'), but different model families and config classes; not the same bug." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver interpolation failure vs RT-DETR memory release issue; unrelated subsystems and symptoms." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "One is serialization instability by device, the other is LoRA merge collapse with tied embeddings; different failure paths." + }, + { + "left": "issue:43873", + "right": "issue:44464", + "accept": false, + "reason": "Quantization/offloading behavior and compiled chunked generation are separate code paths." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different missing symbols in different modules; not the same underlying breakage." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "Both involve model loading after v5 changes, but the reported failures are different and not clearly one bug." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "Janus test failure vs stale warning in paged generation; no shared code-path bug." + }, + { + "left": "issue:43844", + "right": "issue:44936", + "accept": false, + "reason": "Gradient explosion under ZeRO-3 is unrelated to trainer.evaluate() failing after train()." + }, + { + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "Missing pad_token_idx in one config is not the same as meta-device loading breakage in others." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both touch label handling, but one is gather_for_metrics batching and the other is BeitImageProcessorFast reduction logic." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Import failure for a VL class vs DeepSeek detokenization regression; unrelated problems." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression and batched video shape bug affect different processor behaviors." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Custom processor API regression is unrelated to quantization/offloading behavior." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 inference issue and RT-DETR memory leak are unrelated." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both use apply_chat_template, but one crashes on batched padding=False input while the other returns wrong assistant masks; different bugs." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "Qwen3MoE FP8 config failure and a Qwen2.5-VL import error are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Training/eval dtype mismatch is not the same as Qwen3 MoE memory inefficiency." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizers and different failure modes; no shared duplicate basis." + }, + { + "left": "issue:43295", + "right": "issue:43994", + "accept": false, + "reason": "Processor API regression is unrelated to SigLIP2 wrong outputs." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": true, + "reason": "Both are SigLIP2 correctness complaints on the same model; 38175\u2019s zero probabilities plausibly stem from the broader HF-vs-JAX discrepancy in 43493." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but they concern different pipeline APIs and different changes." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Tokenizer special-token bug and CI stale-device test are unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Feature extractor center-arg crash and Whisper empty-transcription generation bug are separate issues." + }, + { + "left": "issue:43450", + "right": "issue:43493", + "accept": false, + "reason": "Batched video processor shape bug is unrelated to SigLIP2 implementation discrepancy." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA indexer discussion and trust_remote_code breakage have no shared code path." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video batching shape bug and Qwen2-57B tp OOM are unrelated." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/tied-weights collapse and category-limit runtime error are different model behaviors." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC latent/STE mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio apply_chat_template sample-rate defaulting and video batch-shape bugs are unrelated." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent math issue and GraniteSpeech CI test flake are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Specific import error for a missing symbol and a general Torch 2.10 incompatibility report are not the same bug." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss AMP crash are different issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The judgments are generally grounded in the packet and stay conservative: nearly all soft pairs are clearly separated by model, subsystem, or failure mode, and the single accepted SigLIP2 pair is plausibly the same theme without overstating certainty. The summary also matches the packet: all items are issues and there are no PRs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 22, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:32090", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:43278", + "issue:43335", + "issue:43404", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43720", + "issue:43819", + "issue:43825", + "issue:43931", + "issue:44112", + "issue:44292", + "issue:44351", + "issue:44448", + "issue:44462", + "issue:44493", + "issue:44561", + "issue:44568", + "issue:44743", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45290", + "issue:45310", + "issue:45381", + "issue:45459", + "issue:45478", + "issue:45491", + "issue:45520", + "issue:45588", + "issue:45593", + "issue:45712" + ], + "soft_pairs": [ + "issue:43278|issue:43825", + "issue:44857|issue:45593", + "issue:45290|issue:45381", + "issue:43606|issue:45083", + "issue:44448|issue:44743", + "issue:45200|issue:45588", + "issue:43335|issue:43550", + "issue:29942|issue:32090", + "issue:37428|issue:44351", + "issue:44561|issue:44779", + "issue:44462|issue:44561", + "issue:43611|issue:43931", + "issue:39692|issue:43526", + "issue:43819|issue:44493", + "issue:45310|issue:45478", + "issue:29127|issue:32090", + "issue:43335|issue:43653", + "issue:44292|issue:45520", + "issue:44855|issue:45083", + "issue:43404|issue:45200", + "issue:39692|issue:44493", + "issue:43550|issue:44112", + "issue:43335|issue:45083", + "issue:43335|issue:43504", + "issue:44448|issue:44779", + "issue:38175|issue:44493", + "issue:43335|issue:45072", + "issue:44977|issue:45229", + "issue:43720|issue:44112", + "issue:40444|issue:43526", + "issue:43335|issue:45593", + "issue:45459|issue:45712", + "issue:44448|issue:44977", + "issue:44568|issue:45491" + ], + "prepared_review_unit_hash": "e123cc3dafbfee7eee9dbf19dba3fbae467a5f517eb12c0cce6068f8527badde", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 34, + "serialized_chars": 37570, + "estimated_input_tokens": 9393, + "estimated_eval_tokens": 19042 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a mix of unrelated Transformers bug reports with broad thematic overlap around v5 regressions. Only one soft pair looks like a true duplicate: the two identical Qwen3.5 MoE from_pretrained error issues. Everything else should stay separate.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45310 is the earlier, clearer report of the exact Qwen3.5 MoE from_pretrained regression, and it matches the later duplicate wording in issue:45478.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the strongest representative because it is the original report for the only clear duplicate pair and describes the concrete failure mode succinctly.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Different bugs: dtype drift in train/eval vs a pipeline error-message regression." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Different models and failures: AMP float16 crash vs missing auxiliary losses when denoising is off." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Both involve chat/vision behavior, but they are different code paths and symptoms." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "Unrelated failures: CPU offload device mismatch vs a helper length-calculation bug." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Both mention output differences in v5-era Qwen/PEGASUS-like models, but they describe different mechanisms and model paths." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "Different feature areas: Gemma4 text-only mm_token_type_ids handling vs flash attention aux-state attribute crash." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "Different configuration bug and runtime compile bug; same broad subsystem is not enough." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Both are training/runtime test failures, but the underlying faults are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they are for different missing symbols and different modules." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 regressions affecting trust_remote_code/tokenization, but they are distinct API removals and tokenizer behavior issues." + }, + { + "left": "issue:44462", + "right": "issue:44561", + "accept": false, + "reason": "Tokenizer repo-loading bug vs removal of a helper function for trust_remote_code models." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Both are model loading regressions, but one is base_model_prefix handling and the other is Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "Both concern BEiT/SigLIP2 processing details, but one is a docs example failure and the other is a fast processor label bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different internal bugs: DAC latent reconstruction mismatch vs unexpected position-id keys." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same exact bug report: identical Qwen3.5 MoE from_pretrained error on transformers>=5.4.0." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "Different models and failures: LayoutLMv3 box validation messaging vs Trainer GPU broadcast NoneType error." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Both are config/token-related, but one is sparse-layer creation and the other is tokenizer special-token registration." + }, + { + "left": "issue:44292", + "right": "issue:45520", + "accept": false, + "reason": "Both mention runtime/import problems, but the Qwen NVFP4 error and Python 3.13 flash_attn KeyError are unrelated." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Different failures: Python parsing/import error in DebertaV2 vs feature-extract length behavior in qwen3_omni_moe." + }, + { + "left": "issue:43404", + "right": "issue:45200", + "accept": false, + "reason": "Different model families and bugs: tied weights in Mistral3 vs Gemma4 mm_token_type_ids defaults." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "Different issues despite both being model-related; one is docs/quantization, the other is position-id key handling." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/runtime failures, but one is torch.compile+SDPA and the other is a stale device override test." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "Different bugs in different models: sparse layer creation vs feature length helper behavior." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Config creation bug vs pretrained preset loading failure; not the same underlying defect." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 behavior regressions, but one is Pegasus output divergence and the other is Deepseek tokenizer corruption." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Different model symptoms: zero probabilities in SigLIP2 vs unexpected position-id keys." + }, + { + "left": "issue:43335", + "right": "issue:45072", + "accept": false, + "reason": "Different dtype/runtime issues: sparse-layer config bug vs bfloat16 inference mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Different models and symptoms: Qwen3.5 flash-attention generation bug vs Gemma4 multi-GPU CUDA OOM." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Different CI failures: BitNet packed-weight loading vs stale device override test." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Different bug classes: Qwen2.5-VL iterable dataset finetuning failure vs BEiT fast processor label reduction." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "Different model/config bugs: sparse-layer creation vs auxiliary-loss omission." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Different areas: tokenizer error hiding on missing protobuf vs dummy class cleanup and repo checks." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Both concern generation regressions, but they affect different models and code paths." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Different tokenizer/model issues: add_special_tokens behavior vs Gemma3 NaN embeddings from sliding-window attention." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The canonical pair is well-supported: issue:45310 and issue:45478 have identical titles and clearly refer to the same Qwen3.5 MoE from_pretrained regression. The remaining soft-edge rejections are conservative and stay within the evidence in the packet. No PR-pair concerns apply here." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 23, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:25251", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:41720", + "issue:43065", + "issue:43066", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43412", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43701", + "issue:43723", + "issue:43873", + "issue:43927", + "issue:43976", + "issue:43994", + "issue:44186", + "issue:44448", + "issue:44451", + "issue:44462", + "issue:44466", + "issue:44493", + "issue:44554", + "issue:44743", + "issue:44857", + "issue:44977", + "issue:44987", + "issue:45072", + "issue:45083", + "issue:45198", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45520", + "issue:45563", + "issue:45588", + "issue:45698" + ], + "soft_pairs": [ + "issue:34689|issue:35141", + "issue:43723|issue:44554", + "issue:43606|issue:44857", + "issue:44857|issue:45072", + "issue:45372|issue:45588", + "issue:37428|issue:44987", + "issue:43701|issue:45083", + "issue:43873|issue:44493", + "issue:44977|issue:45563", + "issue:43927|issue:45698", + "issue:43550|issue:45083", + "issue:29942|issue:35141", + "issue:43412|issue:45588", + "issue:32090|issue:33357", + "issue:43278|issue:43329", + "issue:43066|issue:44462", + "issue:43873|issue:45229", + "issue:45245|issue:45563", + "issue:43976|issue:45440", + "issue:45372|issue:45406", + "issue:41720|issue:45229", + "issue:43065|issue:43994", + "issue:45372|issue:45459", + "issue:44448|issue:44466", + "issue:43065|issue:45412", + "issue:25251|issue:32090", + "issue:43278|issue:43381", + "issue:44857|issue:45588", + "issue:44451|issue:45520", + "issue:43653|issue:44743", + "issue:44186|issue:45198", + "issue:43994|issue:45412" + ], + "prepared_review_unit_hash": "e9fbe6a0c2ced4c8cccfe16a696c4fb8f90beeabf029143c657be8571f5b3af3", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37273, + "estimated_input_tokens": 9319, + "estimated_eval_tokens": 18894 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is largely a false-positive similarity set: the issues span unrelated models, tokenizers, loading paths, attention/backends, and memory/offload problems. I would not merge any of the soft pairs.", + "confidence": 0.94, + "canonical_issue_reason": "No single issue is a strong canonical; the items are heterogeneous and do not describe one shared bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue here is a reliable global representative for duplicate triage, because the cluster breaks into many unrelated problem areas.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Different failures: Llama 3.2 Vision model loading vs embedding reinitialization after resize." + }, + { + "left": "issue:43723", + "right": "issue:44554", + "accept": false, + "reason": "Tokenizer loading regression in v5 vs an MPS attention correctness issue; different code paths and symptoms." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch in bark-small vs a float16 AMP crash in LwDetr loss." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Both mention dtype/precision, but the affected models and concrete bugs are different." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma 4 processor import failure vs a flash_attention.py AttributeError on sink-less models." + }, + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "Both are loading-related, but one is a missing internal flash-attention symbol and the other is a model-specific load failure." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "Checkpoint resume key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:43873", + "right": "issue:44493", + "accept": false, + "reason": "Offloading/quantization behavior is unrelated to unexpected position-id keys." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Flash-attention generation failure vs an unrelated stale warning in paged generate()." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Saving/loading custom token IDs is different from loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba is unrelated to qwen3_omni_moe length computation." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Failing Flash Attention 2 tests do not match the embedding reinit bug." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export guard/erfinv support issues are not the same as the flash_attention None crash." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer broadcast NoneType error and MacOS bus error on CLIP are different failures." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype drift in eval is unrelated to undefined video-branch variables in multimodal token counting." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are both tokenizer-related, but not the same defect." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization/offload mismatch is not the same as Gemma4 multi-GPU CUDA OOM." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-cardinality RuntimeError is unrelated to the stale num_return_sequences warning." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility and DeepseekV3MoE output divergence are unrelated issues." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Both touch Gemma 4, but one is a dependency import error and the other is a missing _tokenizer attribute in serve." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3 device-map cudaErrorAssert and Gemma4 OOM are different models and failure modes." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Sam3PixelDecoder dummy Conv2d is unrelated to SigLIP2 producing nonsensical outputs." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma4 processor import dependency break is not the same as protobuf-related tokenizer error masking." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Pegasus v4/v5 output differences are unrelated to lm_head.weight serialization inconsistency." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder does not match RT-DETR memory release problems." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting change is unrelated to _gpu_broadcast_one rejecting NoneType." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype changes in train/eval are unrelated to gradient checkpointing being disallowed in eval mode." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "Different layers: LwDetr AMP loss crash vs flash_attention sink-less model attribute error." + }, + { + "left": "issue:44451", + "right": "issue:45520", + "accept": false, + "reason": "ScandiBERT loading failure is unrelated to a Python 3.13 flash_attn import-utils KeyError." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer decode/special-token bug is unrelated to Qwen3 recurrent state reset behavior." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER padding/truncation crash and Wav2Vec2 save/tokenization failure are different tokenizer bugs." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 inference quality issue is unrelated to RT-DETR memory being retained after deletion." + } + ] + }, + "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": 16, + "review_unit_index": 24, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:35141", + "issue:36010", + "issue:39692", + "issue:42915", + "issue:42994", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43278", + "issue:43334", + "issue:43381", + "issue:43388", + "issue:43421", + "issue:43526", + "issue:43531", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43931", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44279", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44479", + "issue:44610", + "issue:44871", + "issue:44977", + "issue:45072", + "issue:45103", + "issue:45198", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45412", + "issue:45440", + "issue:45491", + "issue:45593" + ], + "soft_pairs": [ + "issue:44265|issue:45198", + "issue:44361|issue:45198", + "issue:44206|issue:44479", + "issue:42915|issue:42994", + "issue:43421|issue:45491", + "issue:44871|issue:44977", + "issue:43257|issue:43278", + "issue:43756|issue:45440", + "issue:29942|issue:30064", + "issue:33357|issue:39692", + "issue:29127|issue:34567", + "issue:32090|issue:33290", + "issue:44610|issue:45412", + "issue:45103|issue:45372", + "issue:43824|issue:44279", + "issue:45440|issue:45593", + "issue:43381|issue:43701", + "issue:45072|issue:45229", + "issue:43653|issue:44442", + "issue:43526|issue:44186", + "issue:29942|issue:36010", + "issue:29127|issue:33357", + "issue:43334|issue:43531", + "issue:44464|issue:45491", + "issue:44977|issue:45245", + "issue:29127|issue:35141", + "issue:42915|issue:43931", + "issue:43262|issue:44206", + "issue:43232|issue:43388", + "issue:44442|issue:45072" + ], + "prepared_review_unit_hash": "83d93b41f4ed85d2c1d5ddcdb5a9a82b014603e310ba6b8df73d5c78bed5a57f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36784, + "estimated_input_tokens": 9196, + "estimated_eval_tokens": 18648 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: the issues span unrelated bugs in tokenizers, generation, vision, training, quantization, and docs. None of the soft pairs look like true duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue stands out; the items cover unrelated root causes across multiple subsystems.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a strong global representative because the cluster is too diverse and the pairwise candidates do not share the same concrete bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "Torch export failure on torch_compilable_check is unrelated to Wav2Vec2 save_pretrained/tokenization failures." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError on tasks is a different tokenizer bug from Wav2Vec2 save/tokenization failures." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Unsupported center arg in LasrFeatureExtractor is unrelated to the Qwen VL video-input regression." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "FineGrainedFP8Config failure in Qwen3-MoE and quantized model saving are both quantization-related, but they are different concrete bugs." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime post-processor updates for special tokens are unrelated to Gemma3 NaNs from sliding-window all-padding windows." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id inconsistency and Qwen3.5 flash-attention generation failure are different issues." + }, + { + "left": "issue:43257", + "right": "issue:43278", + "accept": false, + "reason": "MOE weight conversion with accelerate+deepspeed is unrelated to BF16-to-FP32 embedding dtype drift in eval." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch and DeepseekV3MoE divergence concern different models and different implementation bugs." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to void segmentation map processing." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS CLIP bus error is a different failure mode from the SigLIP2 docs/model-processor mismatch and quantization issue." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and TrainerState num_input_tokens_seen not updating are unrelated bugs." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "NoneType broadcast in Trainer and Adafactor+deepspeed OOM are different training problems." + }, + { + "left": "issue:44610", + "right": "issue:45412", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch is unrelated to RT-DETR memory not being released." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotation handling and Gemma4 processor import failure are unrelated code paths." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "A specific Qwen2.5-VL import error is not the same as a generic dependency issue." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepseekV3MoE implementation divergence and D-FINE auxiliary-loss omission are unrelated model bugs." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction and resume_from_checkpoint key mismatch are different trainer behaviors." + }, + { + "left": "issue:45072", + "right": "issue:45229", + "accept": false, + "reason": "Inference dtype mismatches are unrelated to a multi-GPU CUDA OOM." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and AutoTokenizer loading FastSpeech2ConformerTokenizer are different tokenizer issues." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels and LayoutLMv2Tokenizer padding/NER crashes are unrelated preprocessing bugs." + }, + { + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "Flash Attention 2 tests and GenerationMixin import errors are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error messaging and CLIP bus error on MacOS have no shared root cause." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Qwen3-VL missing pad_token_id and Qwen3-MoE sliding_window behavior affect different model components." + }, + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is a different bug from Gemma3 NaNs caused by all-padding windows." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Flash-attention generation regression and category-count runtime limit are unrelated failures." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "LayoutLMv3 error clarity and token-embedding reinitialization in post_init are unrelated bugs." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3-MoE FP8 config failure and Qwen3-VL weight-shape mismatch are different model-loading problems." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and unsupported center arg in LasrFeatureExtractor are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs sync_gpus behavior and gather_for_metrics label dropping are different trainer/data-path bugs." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "AutoTokenizer load failure is unrelated to dtype mismatch in bfloat16 inference." + } + ] + }, + "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": 16, + "review_unit_index": 25, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:33290", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:39692", + "issue:42907", + "issue:43316", + "issue:43381", + "issue:43441", + "issue:43452", + "issue:43493", + "issue:43525", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43611", + "issue:43646", + "issue:43701", + "issue:43723", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44206", + "issue:44387", + "issue:44442", + "issue:44479", + "issue:44743", + "issue:44811", + "issue:44857", + "issue:44898", + "issue:44977", + "issue:45003", + "issue:45216", + "issue:45290", + "issue:45440", + "issue:45479", + "issue:45491", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45698", + "issue:45759" + ], + "soft_pairs": [ + "issue:43526|issue:44857", + "issue:43526|issue:43994", + "issue:43441|issue:45588", + "issue:44743|issue:44811", + "issue:43761|issue:44079", + "issue:43452|issue:45588", + "issue:43575|issue:43701", + "issue:45003|issue:45440", + "issue:43646|issue:44077", + "issue:43723|issue:45479", + "issue:43575|issue:44387", + "issue:44977|issue:45584", + "issue:43761|issue:44188", + "issue:45588|issue:45600", + "issue:43316|issue:43525", + "issue:44479|issue:45290", + "issue:43931|issue:45216", + "issue:43994|issue:44206", + "issue:42907|issue:45698", + "issue:33290|issue:34567", + "issue:34689|issue:43611", + "issue:44442|issue:44898", + "issue:39692|issue:45491", + "issue:33290|issue:35141", + "issue:33290|issue:34689", + "issue:43493|issue:43827", + "issue:43381|issue:43873", + "issue:43550|issue:44743", + "issue:44077|issue:44479", + "issue:43316|issue:45759" + ], + "prepared_review_unit_hash": "2fd93cadc358197ab422656a6b5619c01ff599e1d11dd641896f769469e8e1bc", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36167, + "estimated_input_tokens": 9042, + "estimated_eval_tokens": 18340 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is not a true duplicate cluster: the items are mostly unrelated bugs/regressions across different models, APIs, and subsystems, so no single canonical issue emerges. The proposed soft edges are all too weak to merge.", + "confidence": 0.96, + "canonical_issue_reason": "No clear canonical issue: the set is heterogeneous, and none of the issues cleanly subsumes the others as the same underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "No global best issue for deduping; the cluster spans multiple distinct failure modes, so picking one as representative would be misleading.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "Different failures: BeitImageProcessorFast label reduction bug vs AMP/CUDA loss crash in LwDetrImageLoss." + }, + { + "left": "issue:43526", + "right": "issue:43994", + "accept": false, + "reason": "Unrelated subsystems and symptoms: BeitImageProcessorFast label handling vs SigLIP2 model output quality." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both mention FlashAttention, but one is a Ministral-3 v5 RC regression and the other is an AttributeError on sink-less models; not the same concrete bug." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different code paths: Qwen recurrent-state reset vs Whisper processor.decode token skipping." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression is distinct from generic ModelOutput key assignment behavior." + }, + { + "left": "issue:43452", + "right": "issue:45588", + "accept": false, + "reason": "Tokenizer/model loading with gguf_file is unrelated to a flash_attention AttributeError." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "OOM during tensor-parallel model load vs resume_from_checkpoint key mismatch are different problems." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "sys.modules access bug in modeling_utils is unrelated to DeepseekV3MoE implementation divergence." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Custom model initialization regression is too broad to equate with a specific patchtsmixer post_init restriction." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "Tokenizer loading regression and zero-loss sequence-classification bug are unrelated." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both end in OOM, but one is tensor-parallel load OOM and the other is int4 quantization reserved-memory growth; different root causes." + }, + { + "left": "issue:44977", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 flash-attention generation regression is unrelated to Whisper empty-transcription handling." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "Hidden_states regression in CLIPVisionModel is unrelated to attention-kernel branching under torch.compile." + }, + { + "left": "issue:45588", + "right": "issue:45600", + "accept": false, + "reason": "FlashAttention crash and stale auto_mappings references are unrelated issues." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "API inconsistency in Gemma3TextConfig is not the same bug as Llama4Config missing pad_token_id." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "Video-input regression for Qwen variants is unrelated to chat template crashes on tool-call assistant messages." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Model weight-shape mismatch on load is different from a save_pretrained checkpoint regression." + }, + { + "left": "issue:43994", + "right": "issue:44206", + "accept": false, + "reason": "SigLIP2 bad outputs and LasrFeatureExtractor center-arg crash are separate issues." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Dequantized model save failure is different from loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "OOM with Adafactor/DeepSpeed has no relation to TrainerState.num_input_tokens_seen not updating." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Both are model-loading regressions, but they affect different mechanisms and model classes." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "FastSpeech2ConformerTokenizer loading and Perceiver interpolation behavior are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:45491", + "accept": false, + "reason": "Documentation example errors in SigLIP2 do not match the Gemma3 NaN embedding bug." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "Adafactor/DeepSpeed OOM is unrelated to embedding reinitialization after resize_token_embeddings." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "OOM in training is not the same underlying bug as a Llama model-loading regression." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation mismatch and docs still using pipeline() are different issue types and code paths." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient-checkpointing eval-mode limitation is unrelated to quantization offloading behavior." + }, + { + "left": "issue:43550", + "right": "issue:44743", + "accept": false, + "reason": "torch.compile+SDPA failure in Bamba is unrelated to Qwen recurrent state reset." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "A patchtsmixer post_init policy issue does not match a Qwen video-input regression." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy is unrelated to AutoModelForCausalLM.from_config unwrapping text_config for Qwen multimodal configs." + } + ] + }, + "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": 16, + "review_unit_index": 26, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:28282", + "issue:33290", + "issue:34689", + "issue:36010", + "issue:36331", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:43065", + "issue:43295", + "issue:43316", + "issue:43381", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43475", + "issue:43540", + "issue:43643", + "issue:43673", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43825", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44162", + "issue:44190", + "issue:44442", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44617", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44877", + "issue:44913", + "issue:45137", + "issue:45372", + "issue:45584" + ], + "soft_pairs": [ + "issue:41628|issue:45372", + "issue:43643|issue:44913", + "issue:43720|issue:44442", + "issue:43749|issue:43931", + "issue:44190|issue:44496", + "issue:43873|issue:44871", + "issue:33290|issue:36331", + "issue:34689|issue:36010", + "issue:43381|issue:43856", + "issue:44855|issue:45137", + "issue:43673|issue:44743", + "issue:43540|issue:44514", + "issue:43316|issue:44877", + "issue:43381|issue:43742", + "issue:42175|issue:43316", + "issue:43404|issue:43441", + "issue:25251|issue:28282", + "issue:44162|issue:44792", + "issue:41950|issue:43475", + "issue:43388|issue:43688", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43065|issue:44617", + "issue:44077|issue:44534", + "issue:44871|issue:45584" + ], + "prepared_review_unit_hash": "81a29fb1bcd5c811c2bfa0e89e535db594b0ccfbceb1395c895561e07d3f6ff0", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 25, + "serialized_chars": 34485, + "estimated_input_tokens": 8622, + "estimated_eval_tokens": 17500 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is not a duplicate set: the soft pairs are mostly broad-theme lookalikes (loading, configs, processors, cache, memory) but each refers to a distinct concrete bug or model-specific regression. None are safe to merge as the same underlying issue.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41628", + "right": "issue:45372", + "accept": false, + "reason": "Both are import/processor-loading failures, but they involve different missing symbols and different dependency chains, so they are not the same bug." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "Both concern config values not surviving load/reload, but one is a trust_remote_code field-loss issue and the other is a specific rotary_pct round-trip regression." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "Unrelated failures: BitNet accelerate-loading unpacking vs AutoTokenizer loading a custom tokenizer class." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "Both happen during model loading, but one is broken FSDP CPU RAM efficient loading and the other is a shape-mismatch for a specific Qwen3-VL model." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "These are different problems in different code paths: local dataset loading in a training script versus unrecognized model/config loading." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offloading behavior and eos_token_id configuration are separate issues with different root causes." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "One is an OOM in Adafactor+DeepSpeed; the other is a Trainer API signature mismatch." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "A model-loading regression for Llama 3.2 Vision is unrelated to the GenerationMixin import error." + }, + { + "left": "issue:43381", + "right": "issue:43856", + "accept": false, + "reason": "Gradient checkpointing in eval mode is not the same as inefficient MoE training memory usage." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 torch.jit parsing failure and DeepSpeed ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both mention cache, but one is missing GenerationMixin cache in chunked_prefill and the other is recurrent-state reset in a Qwen3 modular model." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Different multimodal processor bugs: video input handling in Qwen3OmniMoe versus batched chat template crashes in Qwen2.5-VL." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "One is an API discrepancy in Gemma3TextConfig, the other is strict config validation blocking granite_speech loading." + }, + { + "left": "issue:43381", + "right": "issue:43742", + "accept": false, + "reason": "Gradient checkpointing/eval-mode behavior does not match a key error while loading MobileLLM-125M." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "TensorFlow packaging omission and Gemma3TextConfig API mismatch are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Both are Mistral-family issues, but one is tied lm_head weights and the other is a FlashAttention failure; different code paths." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline top_k output-shape change and missing PyTorch import are entirely different issues." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage and a janus image-generation test failure are unrelated." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor selection and Sam3VisionEncoderOutput missing an attribute are different bugs." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "The former is a gather_for_metrics last-batch label truncation bug; the latter is auxiliary-loss normalization in MoE models." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Both are v5 regressions, but one is about processor.tokenizer/image-tokenizer behavior and the other is a pipeline error message about translation support." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "Model_max_length on DeBERTa-V3 and a missing transformers.onnx module are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as Sam3Video CUDA OOM." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "Optional post_init handling in patchtsmixer is unrelated to non-persistent buffers being filled with junk in v5." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma-3 eos_token_id inconsistency and Whisper empty-transcription generation failure are different model-specific bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 27, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:41720", + "issue:42617", + "issue:43065", + "issue:43295", + "issue:43452", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43701", + "issue:43742", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44190", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44496", + "issue:44521", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44792", + "issue:44871", + "issue:44913", + "issue:44933", + "issue:44936", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45563", + "issue:45584", + "issue:45600", + "issue:45698" + ], + "soft_pairs": [ + "issue:43761|issue:44479", + "issue:45127|issue:45200", + "issue:43550|issue:43701", + "issue:43606|issue:43701", + "issue:43295|issue:44871", + "issue:43452|issue:44496", + "issue:44734|issue:45406", + "issue:43575|issue:43742", + "issue:43646|issue:44913", + "issue:44936|issue:45698", + "issue:42617|issue:43575", + "issue:43761|issue:44242", + "issue:43931|issue:44060", + "issue:44617|issue:45584", + "issue:43575|issue:43873", + "issue:32090|issue:35141", + "issue:45563|issue:45584", + "issue:44521|issue:45381", + "issue:44479|issue:44734", + "issue:44464|issue:44871", + "issue:43065|issue:44610", + "issue:43906|issue:45290", + "issue:43827|issue:43994", + "issue:44933|issue:45362", + "issue:44792|issue:44977", + "issue:44534|issue:44913", + "issue:44464|issue:45563", + "issue:41720|issue:45245", + "issue:44190|issue:44964", + "issue:44792|issue:45600" + ], + "prepared_review_unit_hash": "a90eaf00195ad706af739756993f85d791ceeca7134784d223a358beed820ce9", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36551, + "estimated_input_tokens": 9138, + "estimated_eval_tokens": 18532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous bundle of unrelated issue reports across training, loading, multimodal, serving, and docs regressions. None of the soft-similarity pairs look like the same underlying bug/change, so all soft edges are rejected.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "Both are regression-style failures, but one is CLIPVision hidden_states handling and the other is video-input position ids in Qwen VL models; different code paths." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA/extended-vocab collapse with tied embeddings is unrelated to Gemma 4's mm_token_type_ids defaulting issue." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile SDPA failure for Bamba is unrelated to a resume_from_checkpoint key mismatch." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch and checkpoint key mismatch are different bugs in different subsystems." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor.tokenizer regression is not the same as inconsistent eos_token_id config on Gemma-3." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "gguf_file loading breakage and unrecognized model/config missing model_type are distinct model-loading failures." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "KV-cache continuation indexing in serve is unrelated to Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Tensor-parallel OOM on Qwen3-57B is not the same as a key error loading MobileLLM-125M." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage and GPTNeoXConfig rotary_pct reload behavior are different config/init bugs." + }, + { + "left": "issue:44936", + "right": "issue:45698", + "accept": false, + "reason": "trainer.evaluate() after train() failure is unrelated to save_pretrained loading the wrong custom module." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py runtime failure is not the same underlying issue as Qwen3 tensor-parallel OOM." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "CLIPVision hidden_states being None is unrelated to MoE load-balancing loss omission." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-Next tied-weights warning point to different model/config problems." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video CUDA OOM is unrelated to Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:43575", + "right": "issue:43873", + "accept": false, + "reason": "Tensor-parallel OOM and quantization offload not working are both memory-related but not the same code-path bug." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Trainer _gpu_broadcast_one NoneType error is unrelated to post_init reinitializing output embeddings after resizing." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "Paged-generate warning behavior is unrelated to Whisper's empty-transcription crash." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "All-zero assistant masks for multimodal inputs and wrong video vision_position_ids are separate multimodal preprocessing bugs." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Video input regression in Qwen VL models is unrelated to serve KV-cache continuation tensor indexing." + }, + { + "left": "issue:44464", + "right": "issue:44871", + "accept": false, + "reason": "Chunked generation inconsistency under compiled forward is not the same as Gemma eos_token_id config mismatch." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and wrong processor image size are different SAM3 problems." + }, + { + "left": "issue:43906", + "right": "issue:45290", + "accept": false, + "reason": "An isolated reproduction of issue 38071 does not match the specific tokenize=True crash with tool-call assistant messages." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still referencing pipeline() and nonsensical SigLIP2 outputs are unrelated." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Missing image_utils import and Qwen3.5 chat crash are separate failures." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation bug are not the same bug." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffers filled with junk is unrelated to rotary_pct not persisting on reload." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency and stale num_return_sequences warning are different issues." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device-mapping CUDA assert is unrelated to the 2^24 category limit error." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading in no-trainer script is unrelated to Phi-4 multimodal model loading failure." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Janus image-generation test failure and auto_mappings Sam3LiteText CI breakage are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 28, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36246", + "issue:38175", + "issue:38617", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:43065", + "issue:43116", + "issue:43295", + "issue:43335", + "issue:43388", + "issue:43404", + "issue:43479", + "issue:43540", + "issue:43550", + "issue:43644", + "issue:43650", + "issue:43761", + "issue:43976", + "issue:44016", + "issue:44077", + "issue:44079", + "issue:44112", + "issue:44186", + "issue:44190", + "issue:44220", + "issue:44279", + "issue:44466", + "issue:44561", + "issue:44683", + "issue:44734", + "issue:44743", + "issue:44855", + "issue:44913", + "issue:44938", + "issue:45127", + "issue:45216", + "issue:45245", + "issue:45276", + "issue:45357", + "issue:45375", + "issue:45446", + "issue:45542", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45632" + ], + "soft_pairs": [ + "issue:44855|issue:45588", + "issue:43388|issue:44079", + "issue:43761|issue:44743", + "issue:45127|issue:45357", + "issue:43335|issue:44855", + "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:45276|issue:45584", + "issue:44112|issue:45600", + "issue:43650|issue:44220", + "issue:43116|issue:44190", + "issue:43295|issue:45276", + "issue:43335|issue:44190", + "issue:41950|issue:44016", + "issue:36246|issue:41628", + "issue:45245|issue:45584", + "issue:43404|issue:45216", + "issue:43479|issue:45375", + "issue:36246|issue:38617", + "issue:44734|issue:45632", + "issue:42175|issue:44279", + "issue:44186|issue:44734", + "issue:38175|issue:43065", + "issue:44016|issue:44220", + "issue:44561|issue:45632", + "issue:44683|issue:45446", + "issue:44466|issue:44913" + ], + "prepared_review_unit_hash": "d75f4dbc967ee3c049ce5ee3c1cfeab9090add6b12b00b3ee84149020caa87e3", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36652, + "estimated_input_tokens": 9163, + "estimated_eval_tokens": 18582 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated issues rather than a duplicate set. None of the soft pairs look like the same underlying bug, so I would not merge any of them; if forced to pick an anchor, I\u2019d use the processor/tokenizer regression issue as the broadest representative.", + "confidence": 0.28, + "canonical_issue_reason": "issue:43295 is the most representative high-signal regression in the set: it is a concrete, user-facing break in core processor/tokenizer behavior, and it is broader than the many narrow model-specific or environment-specific bugs here.", + "canonical_pr_reason": null, + "best_issue_reason": "As a global representative, issue:43295 is the best anchor because it is a clear upstream API regression with practical impact and a cleaner triage target than the surrounding one-off model, config, or tooling issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Different failure modes: Python import parsing on 3.13 vs flash_attention crashing on s_aux=None." + }, + { + "left": "issue:43388", + "right": "issue:44079", + "accept": false, + "reason": "Different bugs: metric gathering drops tuple labels, while ModelOutput keys are assigned incorrectly when a field was previously None." + }, + { + "left": "issue:43761", + "right": "issue:44743", + "accept": false, + "reason": "Different models and code paths: CLIPVisionModel hidden_states regression vs Qwen3_5 recurrent-state reset with cache." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Both involve saving/weights, but one is LoRA merge with extended vocab collapse and the other is incorrect visual encoder keys in save_pretrained." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Unrelated: SwitchTransformers sparse-layer construction vs an import-time IndentationError in DebertaV2Model." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "One is a Python 3.14 load failure; the other is a TensorFlow-backend dependency/install issue." + }, + { + "left": "issue:44077", + "right": "issue:44079", + "accept": false, + "reason": "Different subsystems: patchtsmixer post_init handling vs ModelOutput key assignment logic." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Python version compatibility break vs an unrelated Deberta import parsing error." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk values and GPTNeoX rotary_pct reload are separate state/serialization bugs." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Bamba torch.compile SDPA failure is unrelated to the local-dataset loading issue in the image classification script." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Qwen3OmniMoe video-input handling and serve KV-cache continuation indexing are distinct code paths." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Gemma4 embedding-resize wiring and Whisper empty-transcription generation are unrelated bugs." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "CI test flakiness in GraniteSpeech and stale auto_mappings references to removed configs are different problems." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "Placeholder issue versus an actual _torch_extract_fbank_features() bug; no shared underlying defect." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Different example-script failures: empty multi-label classification results vs local dataset loading." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Both touch multimodal/embedding behavior, but one is a processor.tokenizer regression and the other is resize_token_embeddings not wiring outputs correctly." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers config initialization and a training-script dataset-loading failure are unrelated." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification pipeline bug and notebook syntax error are clearly different issues." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols and different API surfaces, so they are not the same bug." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Runtime category-limit error and Whisper empty-transcription failure are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "lm_head weight tying in Mistral3ForConditionalGeneration is not the same as Qwen3.5 save_pretrained writing wrong visual encoder keys." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "One is default config initialization from None, the other is a missing field being dropped by strict serialization." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different import failures: Qwen2_5_VLImageProcessor missing vs layer_type_validation missing from configuration_utils." + }, + { + "left": "issue:44734", + "right": "issue:45632", + "accept": false, + "reason": "Serve KV-cache indexing crash and local-model cache-path collision are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "Both mention dependencies, but they are too vague and do not share a concrete package/version or identical failure mode." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "Tokenizer padding/truncation crash in NER is unrelated to serve KV-cache continuation indexing." + }, + { + "left": "issue:38175", + "right": "issue:43065", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 and a dummy Conv2d in Sam3PixelDecoder are different model bugs." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "A notebook syntax error is unrelated to an fbank feature-extraction issue." + }, + { + "left": "issue:44561", + "right": "issue:45632", + "accept": false, + "reason": "Removal of is_torch_fx_available and cache-path collisions for local models are different regressions." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both involve flex_attention, but one is compiled flex_attention on torch>=2.9 and the other is a wrong version check for AuxRequest import; not the same concrete bug." + }, + { + "left": "issue:44466", + "right": "issue:44913", + "accept": false, + "reason": "Tied-weight serialization inconsistency and GPTNeoX rotary_pct reset on reload are separate config/serialization problems." + } + ] + }, + "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": 16, + "review_unit_index": 29, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:41950", + "issue:42907", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43577", + "issue:43643", + "issue:43673", + "issue:43824", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44112", + "issue:44242", + "issue:44246", + "issue:44279", + "issue:44336", + "issue:44351", + "issue:44368", + "issue:44496", + "issue:44521", + "issue:44933", + "issue:44964", + "issue:44987", + "issue:45003", + "issue:45020", + "issue:45042", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45325", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45406", + "issue:45584", + "issue:45593", + "issue:45600", + "issue:45632", + "issue:45759" + ], + "soft_pairs": [ + "issue:44246|issue:45003", + "issue:45406|issue:45632", + "issue:43643|issue:44496", + "issue:44964|issue:45600", + "issue:44016|issue:44062", + "issue:43295|issue:45335", + "issue:42907|issue:43441", + "issue:44112|issue:44964", + "issue:44496|issue:45042", + "issue:43854|issue:44964", + "issue:43673|issue:44336", + "issue:43976|issue:44987", + "issue:43299|issue:45216", + "issue:43531|issue:45600", + "issue:43824|issue:44933", + "issue:43526|issue:43901", + "issue:45335|issue:45584", + "issue:43976|issue:44279", + "issue:44075|issue:44368", + "issue:43381|issue:45593", + "issue:43299|issue:45357", + "issue:44496|issue:45759", + "issue:44521|issue:45325", + "issue:41950|issue:45020", + "issue:43901|issue:44496", + "issue:44242|issue:45200", + "issue:43577|issue:45216", + "issue:43976|issue:45405", + "issue:44351|issue:45278", + "issue:43479|issue:43525" + ], + "prepared_review_unit_hash": "94a3ef83fddb90a27760172557c17e33859e121aff98138d20d0237c8bf31512", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36669, + "estimated_input_tokens": 9168, + "estimated_eval_tokens": 18592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These issues cluster around Transformers regressions, but the specific failures are too different to treat any pair as duplicates; I would keep the broad open regression report as the cluster anchor.", + "confidence": 0.88, + "canonical_issue_reason": "Issue 45020 is the broadest open umbrella report in the set, covering recent-version breakage around remote-code/model loading.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45020 is the best representative because it is open, high-level, and can plausibly subsume several nearby regression reports without overfitting to one model.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Both touch import-time behavior, but one is intermittent slowness and the other is an unsafe sys.modules access bug; not the same defect." + }, + { + "left": "issue:45406", + "right": "issue:45632", + "accept": false, + "reason": "Gemma4Processor _tokenizer crash and trust_remote_code cache-path collision are unrelated failure modes." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "Both are config/loading problems, but one is missing fields with trust_remote_code and the other is an unrecognized model due to missing model_type; different root causes." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Phi-4 multimodal loading and stale auto_mappings references to Sam3LiteText are unrelated issues." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken keyword collision are clearly different bugs." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "They affect different code paths: processor/tokenizer regression versus resize_token_embeddings not propagating to decoder embeddings." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Saving dequantized models and FlashAttention loading for Ministral-3 are different problems." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "A stale CI test override in GraniteSpeech is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Model config loading and PIL image processor torchvision dependency are separate issues." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "GLM-4.7-Flash unit-test loading failure is not the same as Phi-4 multimodal loading." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Generation cache handling and ANSI output formatting are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python version compatibility and loading physical-intelligence/fast are different reports." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Both involve Qwen3.5-era regressions, but one is a loading failure and the other is a save_pretrained checkpoint serialization bug." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and removed Sam3LiteText config references are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but they refer to different missing symbols and different causes." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "reduce_labels behavior in BeitImageProcessorFast is unrelated to a docs mismatch in TextClassificationPipeline." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "t5gemma embedding resizing and Whisper empty-transcription generation are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "A Python support regression and a generic dependency issue are not the same underlying change." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Unused SGD optimizer args and a tie_word_embeddings warning in Qwen3.5 are unrelated." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary-loss handling are different code paths." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is a loading regression and the other is incorrect visual-encoder key saving; not the same bug." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "An unrecognized-model error and a composite-config unwrap issue are distinct config-handling bugs." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "Multimodal assistant masks and Qwen2.5-VL rope-index scaling are different multimodal behaviors." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification processor lookup and remote_code model breakage are not the same issue." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "A documentation discrepancy is not the same as a model-loading failure." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "Load-balancing loss omission and mm_token_type_ids defaulting are separate model-training bugs." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "BLIP2 dtype persistence and Qwen3.5 save_pretrained corruption are unrelated." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "45278 is an umbrella report about many import errors, while 44351 is one specific missing export; the root cause is not established as the same." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python compatibility and the unreleased PEFT minimum-version bump are unrelated." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Same as above: one specific import failure versus a broad update-breakage report is too loose to merge." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Phi4MultimodalConfig default initialization and Llama4Config missing pad_token_id are different config bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 30, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:33290", + "issue:38175", + "issue:39692", + "issue:41950", + "issue:42994", + "issue:43065", + "issue:43262", + "issue:43295", + "issue:43299", + "issue:43317", + "issue:43441", + "issue:43450", + "issue:43650", + "issue:43653", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43854", + "issue:43901", + "issue:44016", + "issue:44162", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44291", + "issue:44297", + "issue:44462", + "issue:44466", + "issue:44521", + "issue:44617", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:44898", + "issue:44987", + "issue:45081", + "issue:45276", + "issue:45290", + "issue:45397", + "issue:45405", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45656", + "issue:45712" + ], + "soft_pairs": [ + "issue:44683|issue:45542", + "issue:44462|issue:45081", + "issue:44279|issue:45405", + "issue:43299|issue:43854", + "issue:39692|issue:43901", + "issue:43295|issue:44779", + "issue:44297|issue:44462", + "issue:38175|issue:44617", + "issue:41950|issue:43650", + "issue:33290|issue:45656", + "issue:44462|issue:44779", + "issue:45600|issue:45712", + "issue:44016|issue:44291", + "issue:43450|issue:44279", + "issue:42994|issue:43317", + "issue:43065|issue:44898", + "issue:44466|issue:45276", + "issue:44261|issue:44263", + "issue:43441|issue:44683", + "issue:43650|issue:45290", + "issue:44623|issue:45397", + "issue:43723|issue:44162", + "issue:44623|issue:44749", + "issue:43784|issue:45542", + "issue:43262|issue:44521", + "issue:44671|issue:44779", + "issue:43673|issue:43761", + "issue:43653|issue:45593", + "issue:44987|issue:45542" + ], + "prepared_review_unit_hash": "3039bd251b5b247ff5466506d98b3119c3ab3ec5341b6323b91868fb24ea6172", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36367, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a set of unrelated Transformers issues that only share broad v5/regression vocabulary. The soft-similarity pairs are all false positives: they point to different subsystems, different failure modes, and usually different code paths.", + "confidence": 0.95, + "canonical_issue_reason": "No canonical issue: the items do not share one underlying bug, so picking any single issue would misrepresent the cluster.", + "canonical_pr_reason": "No pull requests are present in this cluster.", + "best_issue_reason": "No global best issue: the set is heterogeneous (tokenizers, processors, model loading, serialization, docs, deepspeed, attention, etc.), so there is no good representative duplicate target.", + "best_pr_reason": "No pull requests are present in this cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Different bugs: compiled flex_attention failure on torch>=2.9 vs tensorboard installed without TensorFlow causing an undefined tf backend error." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer paths: AutoTokenizer ignoring repository tokenizer.json vs _patch_mistral_regex crashing because Tokenizer lacks backend_tokenizer." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Unrelated: a generic transformers dependency issue vs MIN_PEFT_VERSION being bumped to an unreleased PEFT version." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Different failures: Qwen3VL MoE model loading breakage vs NameError importing sentence-transformers with latest transformers." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Both involve docs/examples, but not the same bug: SigLIP2 example/model-processor mismatch and quantization failure vs stale TextClassificationPipeline docs mentioning return_all_scores." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Different regressions: custom code accessing processor.tokenizer/passing images to tokenizer vs Deepseek tokenizer producing incorrect outputs in v5." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Related to tokenizer serialization, but distinct: tokenizer.save_pretrained writes a mismatched tokenizer_class, while AutoTokenizer ignores tokenizer.json from the repo." + }, + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "Different models and symptoms: unexpected zero probabilities in SigLIP2 vs Sam3Video CUDA out of memory." + }, + { + "left": "issue:41950", + "right": "issue:43650", + "accept": false, + "reason": "No shared bug: video-classification pipeline using image processors vs an unrelated 'ADD THE DATA' issue." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention deepspeed/optimizer, but the concrete problems differ: AdaFactor OOM vs optimizer.step being called twice." + }, + { + "left": "issue:44462", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer issues: ignoring tokenizer.json in AutoTokenizer vs Deepseek tokenizer producing incorrect results." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Different maintenance problems: stale auto_mappings.py entries vs leftover dummy classes leaking into dummy_pt_objects.py and dir(transformers)." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Unrelated: notebook syntax error in the intro notebook vs init_empty_weights breaking on an unexpected _is_hf_initialized argument." + }, + { + "left": "issue:43450", + "right": "issue:44279", + "accept": false, + "reason": "Different scopes: batched video processor shape bug vs a generic dependency issue with transformers." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both are quantization-related, but not the same bug: saving a quantized model fails vs dequantized model loading with device_map=auto and cpu/gpu offload." + }, + { + "left": "issue:43065", + "right": "issue:44898", + "accept": false, + "reason": "Different model components: a dummy Conv2d in Sam3PixelDecoder vs Perceiver image classification failing for non-default resolution despite interpolate_pos_encoding=True." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both touch weights/embeddings, but one is inconsistent lm_head serialization by device while the other is resize_token_embeddings not updating Gemma4 per-layer embeddings/output embeddings." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different code paths and symptoms: MLA q_a_layernorm eps precision issue vs torch.split return handling in GlmMoeDsaIndexer." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Both mention attention, but one is Ministral FlashAttention failure and the other is compiled flex_attention breaking on torch>=2.9; not the same fix." + }, + { + "left": "issue:43650", + "right": "issue:45290", + "accept": false, + "reason": "Completely unrelated: a placeholder 'ADD THE DATA' issue vs apply_chat_template crashing on assistant tool-call messages with no content." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Different loading/saving failures: processor.save_pretrained missing files vs gemma-4 zero3 from_pretrained failing." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Different models and failure modes: AutoTokenizer.from_pretrained loading issue vs ESM2 being broken for users." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Different problems: missing files from processor.save_pretrained vs a performance regression when filtering data after upgrading Transformers." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Unrelated runtime errors: importing sentence-transformers hits NameError: nn is not defined vs tensorboard-only installation causing an undefined tf backend error." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both involve chat templates, but the bugs differ: audio apply_chat_template defaults to 16kHz instead of config value vs multimodal assistant_masks becoming all zero." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Different tokenizer/model behavior: CamemBERT masked LM predictions wrong in v5 vs Deepseek tokenizer outputs incorrect results." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Different regressions: GenerationMixin cache missing during chunked_prefill vs CLIPVisionModel.forward returning hidden_states=None." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "No common bug: BigBirdTokenizer special-token/decode issue vs D-FINE not applying auxiliary losses when denoising is off." + }, + { + "left": "issue:44987", + "right": "issue:45542", + "accept": false, + "reason": "Different failures: loading physical-intelligence/fast on transformers>=5.1.0 vs tensorboard without TensorFlow causing backend errors." + } + ] + }, + "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": 16, + "review_unit_index": 31, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:42491", + "issue:42757", + "issue:43122", + "issue:43295", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43650", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:43994", + "issue:44016", + "issue:44188", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44351", + "issue:44361", + "issue:44373", + "issue:44462", + "issue:44485", + "issue:44496", + "issue:44589", + "issue:44617", + "issue:44661", + "issue:45003", + "issue:45250", + "issue:45356", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45593", + "issue:45759" + ], + "soft_pairs": [ + "issue:43122|issue:43650", + "issue:43901|issue:43994", + "issue:42175|issue:43976", + "issue:44361|issue:45593", + "issue:43475|issue:44016", + "issue:44297|issue:45397", + "issue:43824|issue:43976", + "issue:44188|issue:44336", + "issue:43525|issue:45759", + "issue:43525|issue:43881", + "issue:44589|issue:45362", + "issue:43295|issue:44617", + "issue:45561|issue:45563", + "issue:42491|issue:43299", + "issue:41628|issue:43976", + "issue:43901|issue:44246", + "issue:44336|issue:44373", + "issue:43352|issue:45250", + "issue:44242|issue:45593", + "issue:41950|issue:44220", + "issue:44295|issue:45397", + "issue:44661|issue:45542", + "issue:44246|issue:44496", + "issue:43531|issue:45362", + "issue:44462|issue:45356", + "issue:43526|issue:45003", + "issue:36246|issue:42757", + "issue:44336|issue:44485", + "issue:36246|issue:44351", + "issue:45003|issue:45561" + ], + "prepared_review_unit_hash": "71d952df8f26c6ae5b10b65f9b92daae6c7398c5c8b2929b8b749bc3c5837801", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35748, + "estimated_input_tokens": 8937, + "estimated_eval_tokens": 18130 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose similarity cluster of mostly unrelated Transformers issues. The only common thread is broad import/load/regression language, but none of the soft pairs appear to describe the same concrete bug or change.", + "confidence": 0.83, + "canonical_issue_reason": "issue:43824 is the most representative of the repeated import/export failure pattern in this set, with concrete runtime impact and relatively strong discussion/reference activity.", + "canonical_pr_reason": null, + "best_issue_reason": "Among the mixed issues here, issue:43824 is the clearest, most central example of the recurring model-import/load regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43122", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer behavior regression and a vague data request are unrelated; no shared code path." + }, + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Docs mismatch for a pipeline option is not the same bug as bad SigLIP2 outputs." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Missing TensorFlow from a torch install and a Python-version compatibility break are different issues." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "MLukeTokenizer task-time AttributeError and D-FINE auxiliary-loss gating are unrelated model bugs." + }, + { + "left": "issue:43475", + "right": "issue:44016", + "accept": false, + "reason": "SAM3 video encoder attribute failure and a notebook syntax error are not the same problem." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma-4 ZeRO-3 from_pretrained loading." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "An import/export symbol error and a Python 3.9/3.10 support issue are different failure modes." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "torch.compile attention-kernel branching and stray ANSI codes in a report helper are unrelated." + }, + { + "left": "issue:43525", + "right": "issue:45759", + "accept": false, + "reason": "Both are config-loading regressions, but they affect different models and different missing attributes/unwrapping paths." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Llama4Config pad_token_id and GLM-4V loading failure are distinct model-loading bugs." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Missing Float8 storage support and a Qwen3.5 chat crash are not the same underlying issue." + }, + { + "left": "issue:43295", + "right": "issue:44617", + "accept": false, + "reason": "Processor/tokenizer regression and SAM3Video OOM are different code-path problems." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "A pytest-xdist file race and a stale generate() warning are unrelated testing/runtime issues." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Qwen3-MoE LoRA compatibility on hf4.x and Qwen3VL MoE loading are different model-family regressions." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "AutoImageProcessor import failure and Python version incompatibility are not the same bug." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "A docs/behavior mismatch and intermittent slow import are unrelated." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "ANSI output in a helper and a wrong docstring are separate documentation/UX issues." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "Model-specific FA2 unsupported error and a generic Flash Attention 2.0 topic are not the same concrete fix." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Missing router load-balancing loss and missing D-FINE auxiliary losses happen in different models and flags." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification pipeline image-processor lookup and audio fbank extraction are unrelated pipeline bugs." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Reading a buffered position_ids tensor and Gemma-4 zero3 loading are different errors." + }, + { + "left": "issue:44661", + "right": "issue:45542", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES model-generation failure and missing TensorFlow backend when only tensorboard is installed are unrelated." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Intermittent import slowness and unrecognized model/config loading are different problems." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and a Qwen3.5 chat crash are not the same bug." + }, + { + "left": "issue:44462", + "right": "issue:45356", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json and a Kimi-K2.5 tokenizer regression are separate tokenizer issues." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and unsafe sys.modules access are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Different import-name failures for different symbols and modules; same symptom class only." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "ANSI-code leakage and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing exports and different underlying fixes." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Unsafe sys.modules access and a captured_info.txt race are unrelated implementation bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 32, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36246", + "issue:41628", + "issue:41950", + "issue:42886", + "issue:43065", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43408", + "issue:43441", + "issue:43475", + "issue:43540", + "issue:43596", + "issue:43653", + "issue:43673", + "issue:43901", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44368", + "issue:44466", + "issue:44496", + "issue:44561", + "issue:44749", + "issue:44779", + "issue:44857", + "issue:44871", + "issue:44877", + "issue:44987", + "issue:44995", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45459", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45701", + "issue:45759" + ], + "soft_pairs": [ + "issue:43673|issue:44561", + "issue:44246|issue:44871", + "issue:44242|issue:44857", + "issue:36246|issue:43596", + "issue:44162|issue:45200", + "issue:44295|issue:44749", + "issue:44779|issue:45292", + "issue:43441|issue:43540", + "issue:44877|issue:45375", + "issue:42886|issue:45229", + "issue:45292|issue:45584", + "issue:44987|issue:45459", + "issue:41628|issue:41950", + "issue:45292|issue:45701", + "issue:44779|issue:45701", + "issue:43901|issue:45003", + "issue:45375|issue:45759", + "issue:43232|issue:44336", + "issue:43065|issue:43408", + "issue:43257|issue:45137", + "issue:44995|issue:45200", + "issue:44496|issue:45003", + "issue:44060|issue:44077", + "issue:43262|issue:44246", + "issue:43475|issue:43540", + "issue:44466|issue:45335", + "issue:44995|issue:45588", + "issue:43653|issue:44162", + "issue:44368|issue:45563" + ], + "prepared_review_unit_hash": "7526e9249fe665baab816697c58e74c20dbcf7fed155c901ad8de904b8f3e91f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36020, + "estimated_input_tokens": 9005, + "estimated_eval_tokens": 18266 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a noisy, heterogeneous cluster of Transformers regressions. The items share broad library-loading/modeling themes, but none of the soft pairs appear to be the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "issue:41628 is the cleanest top-level user-facing import/export failure in the set and the best weak anchor for the cluster, though the cluster is still quite mixed.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41628 is the most representative broad issue among these, but only loosely; it does not truly cover the many unrelated model/config/generation bugs in the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Different regressions: chunked_prefill cache handling vs removal of is_torch_fx_available breaking trust_remote_code models." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowdown is unrelated to Gemma eos_token_id/config mismatch." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "MoE load-balancing loss bug is unrelated to an AMP/CUDA crash in LwDetrImageLoss." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "Qwen2.5-VL import export failure and a DeepSpeed ZeRO3 deque underflow are different bugs." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage is unrelated to Gemma4 mm_token_type_ids defaults." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Buffer/position_ids access bug is unrelated to a tokenization performance regression." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer correctness regression is different from resize_token_embeddings not propagating to output_embeddings." + }, + { + "left": "issue:43441", + "right": "issue:43540", + "accept": false, + "reason": "FlashAttention failure in Ministral and Qwen3OmniMoe video-input processing are separate problems." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both involve strict config handling, but they affect different models, fields, and failure modes." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading is unrelated to multi-GPU CUDA OOM." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize propagation and Whisper empty-transcription handling are different bugs." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Loading a specific repo under v5 is not the same as protobuf error masking in tokenizer decoding." + }, + { + "left": "issue:41628", + "right": "issue:41950", + "accept": false, + "reason": "A missing top-level AutoImageProcessor import is not the same as the video-classification pipeline choosing image processors." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "resize_token_embeddings propagation is unrelated to generic version-dependent tokenization changes." + }, + { + "left": "issue:44779", + "right": "issue:45701", + "accept": false, + "reason": "DeepSeek tokenizer regression and broad tokenization changes are not the same concrete bug." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Documentation mismatch is unrelated to unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Missing deepstack_visual_indexes and unwrapping text_config are different Qwen multimodal config bugs." + }, + { + "left": "issue:43232", + "right": "issue:44336", + "accept": false, + "reason": "Generation kwargs after sync_gpus is unrelated to ANSI codes in loading_report." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder is not the same as a sam3_video vs sam3_tracker config mismatch warning." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Qwen3 MoE weight conversion and ZeRO3 deque underflow are distinct DeepSpeed-related failures." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "GlmMoeDsa stale cache crash is unrelated to Gemma4 token-type id requirements." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config loading and sys.modules access are different loading-path bugs." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "Incorrect tied-weights warning in Qwen3-Next is unrelated to patchtsmixer post_init policy." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling rate defaults have nothing to do with import latency." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM 3 Video missing output attribute and Qwen3OmniMoe video-input validation are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Serialization of tied weights by device is different from resize_token_embeddings not reaching decoder.embed_tokens." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale forward-pass cache in GlmMoeDsa is unrelated to flash-attention s_aux handling." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are different issues." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Tie_word_embeddings warning noise is unrelated to a stale num_return_sequences warning in paged generate()." + } + ] + }, + "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": 16, + "review_unit_index": 33, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42994", + "issue:43232", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43704", + "issue:43873", + "issue:43931", + "issue:44038", + "issue:44162", + "issue:44188", + "issue:44297", + "issue:44361", + "issue:44368", + "issue:44623", + "issue:44661", + "issue:44734", + "issue:44749", + "issue:44869", + "issue:44933", + "issue:44945", + "issue:44977", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45200", + "issue:45230", + "issue:45278", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45381", + "issue:45397", + "issue:45459", + "issue:45478", + "issue:45600", + "issue:45656" + ], + "soft_pairs": [ + "issue:45397|issue:45656", + "issue:44368|issue:44977", + "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:43673|issue:45278", + "issue:44297|issue:44623", + "issue:43646|issue:44162", + "issue:44361|issue:44749", + "issue:44623|issue:45656", + "issue:28282|issue:30990", + "issue:42994|issue:43482", + "issue:44749|issue:45310", + "issue:43475|issue:43643", + "issue:44749|issue:45656", + "issue:45230|issue:45397", + "issue:44038|issue:44623", + "issue:42371|issue:45092", + "issue:43475|issue:45381", + "issue:45081|issue:45103", + "issue:41628|issue:44933", + "issue:45375|issue:45478", + "issue:43475|issue:44661", + "issue:42673|issue:43704", + "issue:42994|issue:43525", + "issue:45310|issue:45375", + "issue:43232|issue:44188", + "issue:43873|issue:44945", + "issue:42222|issue:45200" + ], + "prepared_review_unit_hash": "0c70e05ee33c862d58f307374d658269893435cd2597fad35f63e84947036817", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36557, + "estimated_input_tokens": 9140, + "estimated_eval_tokens": 18536 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of distinct Transformers issues, with a broad concentration of v5 regressions, model-loading failures, tokenizer/config breakages, and unrelated performance/runtime bugs. Only one soft pair looks like a true duplicate; the rest are thematically similar but not the same underlying bug.", + "confidence": 0.84, + "canonical_issue_reason": "issue:45278 is the broadest representative of the cluster\u2019s main theme: user-facing breakages after upgrading to Transformers v5, especially import/loading failures. It is a good umbrella issue, though not an exact duplicate of most others.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best anchor for the cluster because it is the most general and representative report of the recurring v5 regression/import problem class seen across multiple items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Different problems: one is a Gemma-4 Zero3 from_pretrained failure, the other is a DeepSpeed optimizer-step duplication bug." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen3.5, but one is a warning about tie_word_embeddings and the other is a flash-attention generation failure." + }, + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Both involve generation/cache paths, but chunked_prefill cache handling and /v1/responses KV-cache continuation are different code paths and failures." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "Completely different: DeepSpeed ZeRO3 deque pop versus auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch is unrelated to a Qwen3.5 embedding-tie warning." + }, + { + "left": "issue:44933", + "right": "issue:45459", + "accept": false, + "reason": "Both are import-related, but image_utils import breakage is unrelated to tokenizer protobuf decode-error masking." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding crash and Kimi-K2.5 tokenizer regression are different subsystems and symptoms." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both touch generation internals, but sync_gpus kwargs updating and chunked_prefill cache missing are not the same bug." + }, + { + "left": "issue:43673", + "right": "issue:45278", + "accept": false, + "reason": "Generation cache handling is unrelated to broad import errors after upgrade; different subsystems and failure modes." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both concern save_pretrained behavior, but tokenizer_config mismatch and missing processor files are distinct save-path bugs." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Both are v5 breakages, but custom model initialization and ESM2 breakage are not the same code-path problem." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "Tokenizer AttributeError on tasks is unrelated to a performance slowdown after upgrading Transformers." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save_pretrained file loss and DeepSpeed optimizer-step duplication are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "PyTorch-missing ImportError and Sentence Transformers loading hang are different issues." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving failure and GGUF loading failure are both model-format related but not the same concrete bug." + }, + { + "left": "issue:44749", + "right": "issue:45310", + "accept": false, + "reason": "A slowdown regression after upgrade is not the same as a Qwen3.5 Moe from_pretrained error." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "Missing Sam3 vision output attribute and missing fields in trust_remote_code AutoConfig are different object initialization bugs." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance regression and DeepSpeed optimizer-step duplication are different symptoms and code paths." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "Generic bug report and Gemma-4 Zero3 from_pretrained failure do not point to the same underlying bug." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe transformers v5 issue is unrelated to processor.save_pretrained missing files." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API-settings warning and InternVL2 meta-initialization incompatibility are not the same problem." + }, + { + "left": "issue:43475", + "right": "issue:45381", + "accept": false, + "reason": "SAM3 missing attribute and Qwen2.5-VL video position-id bug are separate model-specific regressions." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Tokenizer regex patch crash and auto-docstring annotation crash are unrelated failures." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but they affect different symbols and likely different export paths; not the same bug." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 Moe, but one is a missing config field and the other is a general from_pretrained error; too broad to treat as duplicates." + }, + { + "left": "issue:43475", + "right": "issue:44661", + "accept": false, + "reason": "SAM3 vision output attribute error is unrelated to add-new-model-like failing on TOKENIZER_MAPPING_NAMES." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Same exact title and same described bug: Qwen3ForCausalLM leaking VRAM across multiple dataloader threads." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model saving failure and Llama4Config missing pad_token_id are unrelated." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is a from_pretrained error and the other is a specific missing-field/strictness issue; not the same concrete bug." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Generation kwargs syncing and torch.compile attention-kernel divergence are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization offloading and pipeline-parallel output corruption are different runtime behaviors and fixes." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "VitPose model breakage and Gemma-4 mm_token_type_ids requirements are unrelated model-specific issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative: it identifies 45278 as a broad umbrella for the v5 import/regression cluster, and the soft-pair calls are appropriately cautious. The only accepted pair (42673/43704) is well supported by identical titles and matching bug descriptions; the rest are reasonably rejected as related-but-distinct issues." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 34, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42831", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43377", + "issue:43479", + "issue:43504", + "issue:43526", + "issue:43673", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44079", + "issue:44164", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44351", + "issue:44373", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44792", + "issue:44869", + "issue:44933", + "issue:44945", + "issue:44964", + "issue:45103", + "issue:45125", + "issue:45137", + "issue:45310", + "issue:45356", + "issue:45561", + "issue:45712", + "issue:45753" + ], + "soft_pairs": [ + "issue:42831|issue:44222", + "issue:41628|issue:44351", + "issue:44945|issue:45753", + "issue:44222|issue:45310", + "issue:42994|issue:44164", + "issue:42175|issue:45712", + "issue:43673|issue:44188", + "issue:43377|issue:44242", + "issue:36010|issue:41628", + "issue:44038|issue:45310", + "issue:43526|issue:45561", + "issue:43232|issue:44792", + "issue:43504|issue:45561", + "issue:44496|issue:45561", + "issue:44964|issue:45137", + "issue:42898|issue:44568", + "issue:43479|issue:44079", + "issue:30990|issue:35141", + "issue:43479|issue:44869", + "issue:42175|issue:44933", + "issue:43976|issue:44933", + "issue:44188|issue:44373", + "issue:43278|issue:45137", + "issue:43784|issue:43824", + "issue:44623|issue:45310", + "issue:30333|issue:33290", + "issue:44734|issue:45103", + "issue:42898|issue:45356", + "issue:43723|issue:44568", + "issue:42898|issue:44779", + "issue:44060|issue:45125" + ], + "prepared_review_unit_hash": "4d1d3a59fc177c1151b0933dcfe3caf48321b0436391f152eb37351bf8c23e94", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36841, + "estimated_input_tokens": 9211, + "estimated_eval_tokens": 18678 + }, + "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 issues across imports, tokenizers, generation, FP8, save/load, multimodal, and testing. I rejected every soft edge.", + "confidence": 0.81, + "canonical_issue_reason": "issue:45561 is the strongest standalone report by activity and repro detail, but it is not a real representative of the rest of this heterogeneous set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45561 has the highest external signal (inbound refs) and a concrete race-condition description, so it is the best single issue to keep if one must be chosen.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "Both mention FP8, but one is an accuracy regression and the other is a save_pretrained bug for MoE; different failure modes." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols (`AutoImageProcessor` vs `HybridCache`) and different code paths." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption and an image-triggered infinite loop in Qwen3_5 are unrelated bugs." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "Both involve Qwen3.5/MoE serialization/loading, but one is FP8 save_pretrained and the other is a from_pretrained regression on newer transformers." + }, + { + "left": "issue:42994", + "right": "issue:44164", + "accept": false, + "reason": "Both are save/load issues, but `quantized model saving` and `extra_state` handling are distinct serialization failures." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "TensorFlow backend packaging and leftover dummy class cleanup are different maintenance problems." + }, + { + "left": "issue:43673", + "right": "issue:44188", + "accept": false, + "reason": "Generation cache missing in chunked prefill and attention-kernel divergence under torch.compile are different internals and symptoms." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "Missing padding-mask support in MIMI encoder vs missing load-balancing loss when router logits are disabled are not the same bug." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are import failures, but they target different exported classes and different regression sources." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen3.5/VL MoE compatibility, but one is a broad transformers 5.0 model issue and the other is a specific newer-version from_pretrained error." + }, + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "ImageProcessor label reduction and a pytest-xdist captured_info race are completely unrelated." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "One is a generation-kwarg update bug after sync_gpus; the other is a janus image-generation test failure." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "BEiT pretrained loading with a legacy field is unrelated to a testing-utils file race." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading error and xdist capture-file race are different layers of the stack." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading and a DeepSpeed ZeRO3 deque underflow do not share a code path." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer behavior regressions in v5, but one is whitespace cleanup and the other is add_special_tokens/BOS-EOS handling." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Multimodal config defaulting and ModelOutput key assignment are unrelated object-initialization bugs." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-Transformers loading hang vs embedding resize reinitialization are unrelated model-loading vs model-shape issues." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Phi-4 multimodal config init and Whisper timestamp decoding crash are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging dependency selection and a missing image_utils import are different issues." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility and a nonexistent image_utils import are not the same defect." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "A torch.compile-dependent kernel branch and a wrong docstring are not duplicates." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift during evaluation and a ZeRO3 deque pop error are unrelated." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they involve different packages and different missing names (`nn` vs `Qwen2_5_VLForConditionalGeneration`)." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Processor save_pretrained missing files and Qwen3.5 MoE from_pretrained are distinct save/load problems." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLFlow reporting on failed training jobs and deepspeed OOM with Adafactor are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "KV-cache indexing in `transformers serve` and an auto-docstring future-annotations crash are different subsystems." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and different underlying codec/whitespace behavior." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer loading in v5 and add_special_tokens behavior for mdeberta are separate tokenizer regressions." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 tokenizer regressions, but one is cleanup_spaces behavior and the other is Deepseek codec/output correctness." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Tied-weights warning misassignment and missing `_tp_plan` for tensor parallelism are different Qwen3-related model bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 35, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:30990", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:38175", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42898", + "issue:43295", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43824", + "issue:43874", + "issue:44206", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44442", + "issue:44448", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44623", + "issue:44871", + "issue:44913", + "issue:44933", + "issue:44977", + "issue:44991", + "issue:45092", + "issue:45103", + "issue:45310", + "issue:45357", + "issue:45405", + "issue:45406", + "issue:45459", + "issue:45656", + "issue:45712" + ], + "soft_pairs": [ + "issue:45405|issue:45712", + "issue:43874|issue:45092", + "issue:42898|issue:44448", + "issue:44295|issue:45656", + "issue:42898|issue:43295", + "issue:44933|issue:45712", + "issue:41628|issue:45459", + "issue:44222|issue:44297", + "issue:44297|issue:45656", + "issue:43479|issue:43644", + "issue:44206|issue:45103", + "issue:42831|issue:44623", + "issue:44222|issue:44623", + "issue:22355|issue:28282", + "issue:30333|issue:36331", + "issue:41628|issue:42757", + "issue:30333|issue:34689", + "issue:43723|issue:44977", + "issue:44297|issue:45310", + "issue:28282|issue:30333", + "issue:30333|issue:35141", + "issue:44222|issue:44484", + "issue:44483|issue:45406", + "issue:44263|issue:44360", + "issue:44263|issue:45092", + "issue:44514|issue:45103", + "issue:44373|issue:44485", + "issue:44295|issue:44623", + "issue:43723|issue:44871", + "issue:43723|issue:44991", + "issue:45092|issue:45357", + "issue:43479|issue:44913", + "issue:44261|issue:45092", + "issue:43723|issue:44442", + "issue:30990|issue:38175", + "issue:41628|issue:43824" + ], + "prepared_review_unit_hash": "8f99af6788e864c0cac33c73f001b1683e88d66762a9fd98f70a12f3f576a3d9", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 36, + "serialized_chars": 37826, + "estimated_input_tokens": 9457, + "estimated_eval_tokens": 19170 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it mixes unrelated import errors, tokenizer/model-loading regressions, multimodal processor issues, config serialization bugs, and training/runtime problems. I found no true duplicate pairs.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a strong cluster anchor; the closest broad theme is tokenizer/model-loading regressions, but the items are still too diverse to treat as one duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Unrelated: PEFT version pin mismatch vs leftover dummy classes/import surface cleanup." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Different failures: missing image-processor method vs remote-code/meta-init incompatibility." + }, + { + "left": "issue:42898", + "right": "issue:44448", + "accept": false, + "reason": "Both are v5 behavior changes, but one is tokenization cleanup and the other is Pegasus output drift." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Different code paths: position_ids buffer access vs DeepSpeed double optimizer step." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Tokenizer cleanup change vs processor/tokenizer regression on batched image input." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Different cleanup areas: nonexistent image_utils import vs dummy_pt_objects leakage." + }, + { + "left": "issue:41628", + "right": "issue:45459", + "accept": false, + "reason": "Import errors in different modules with different causes: AutoImageProcessor vs protobuf decode handling." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Different features: FP8 save_pretrained for MoE vs tokenizer_class metadata mismatch." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save metadata bug vs optimizer stepping twice under DeepSpeed." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config default initialization bug vs non-persistent buffer corruption." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature-specific image extractor argument bug vs docstring/introspection crash." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "Accuracy regression in FP8 vs processor.save_pretrained missing files." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both involve saving, but FP8 MoE save bug is not the same as processor artifact omission." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Different missing-import problems: transformers.onnx vs PyTorch availability for AutoModel." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "MLflow reporting on failed training vs Trainer compute_loss signature change." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but in different packages and unrelated symbols." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training failure reporting vs a model-loading regression." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading failure vs flash-attention generation issue; different subsystems." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer config mismatch vs Qwen3.5 MoE from_pretrained regression." + }, + { + "left": "issue:28282", + "right": "issue:30333", + "accept": false, + "reason": "PyTorch missing import error vs MLflow status reporting bug." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "Training logging issue vs embedding resize/post_init reinitialization bug." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 save_pretrained bug is unrelated to max_shard_size behavior." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat-completions request handling vs Gemma4Processor missing _tokenizer attribute." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Different GLM/DSA indexer problems: split return handling vs missing ReLU." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Torch split return bug vs remote-code/meta-init incompatibility." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Batched chat-template crash vs auto-docstring annotation crash." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring wording issue vs RoPE implementation discussion." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Position_ids buffer read error vs processor.save_pretrained missing files." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Both concern tokenizers, but one is a general v5 load issue and the other is eos_token_id mismatch." + }, + { + "left": "issue:43723", + "right": "issue:44991", + "accept": false, + "reason": "Same broad area, but the concrete failures/models differ; not a clear duplicate." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Different multimodal regressions: meta initialization incompatibility vs incorrect visual encoder keys on save." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Different config persistence bugs: multimodal defaults vs rotary_pct reload reset." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "RMS norm epsilon precision/config issue vs remote-code meta-init incompatibility." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Batched processor crash vs future-annotations docstring crash; unrelated." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Different missing-class imports: AutoImageProcessor vs Qwen2_5_VLForConditionalGeneration." + } + ] + }, + "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": 16, + "review_unit_index": 36, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:29127", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43475", + "issue:43576", + "issue:43756", + "issue:43784", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44117", + "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:44829", + "issue:44945", + "issue:45200", + "issue:45245", + "issue:45278", + "issue:45290", + "issue:45310", + "issue:45593", + "issue:45600", + "issue:45753" + ], + "soft_pairs": [ + "issue:42886|issue:42994", + "issue:42831|issue:45310", + "issue:44263|issue:44373", + "issue:42831|issue:44038", + "issue:43232|issue:45593", + "issue:43756|issue:44060", + "issue:42757|issue:45278", + "issue:42831|issue:45753", + "issue:43576|issue:43784", + "issue:42886|issue:43066", + "issue:43122|issue:44521", + "issue:44117|issue:44704", + "issue:43232|issue:44743", + "issue:30990|issue:33357", + "issue:42371|issue:44263", + "issue:44704|issue:45290", + "issue:42907|issue:44945", + "issue:43122|issue:44514", + "issue:43976|issue:44945", + "issue:44295|issue:44297", + "issue:43232|issue:45245", + "issue:39401|issue:44488", + "issue:43408|issue:43475", + "issue:43784|issue:45600", + "issue:42907|issue:43937", + "issue:41628|issue:45278", + "issue:44261|issue:44373", + "issue:44360|issue:44485", + "issue:39401|issue:44521", + "issue:43116|issue:45200", + "issue:44625|issue:44829", + "issue:29127|issue:30990" + ], + "prepared_review_unit_hash": "62ac4993119e57a9d823013d127a85b6b1741ce920c15a6a1392508029e0f64f", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36856, + "estimated_input_tokens": 9214, + "estimated_eval_tokens": 18684 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it mixes unrelated Transformers bugs around imports, tokenization, multimodal processors, generation, quantization, and model-specific regressions. None of the soft pairs appear to be the same underlying defect, so there is no strong duplicate group here.", + "confidence": 0.38, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report in the set ('Many import errors after update from 4.57.0 to 5.5.0') and is the closest thing to a cluster representative, though the cluster is still too mixed to treat it as a true canonical duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best representative only because it captures a general v5 regression theme; it is not a strong canonical for the whole cluster since most other issues are model- or API-specific and unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache loading and quantized model saving failures are different problems with different code paths." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FineGrainedFP8 accuracy drift and Qwen3.5 from_pretrained failure are unrelated bugs." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "A torch.split return-value issue is not the same as a wrong docstring." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FP8 accuracy regression and Qwen3-VL-Moe bug are distinct model/runtime issues." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation kwargs handling and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Smollm3 RoPE layer count and Qwen3-Next tied-weights warning do not describe the same defect." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "Both are import-related, but one is a specific missing symbol and the other is a broad upgrade-wide import regression." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy issues and an infinite loop on a specific image are not the same bug." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "Broken env command and a NameError when importing sentence-transformers are separate failures." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading and decoder-type mismatch are unrelated tokenizer issues." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Tokenization differences across versions and assistant mask generation for multimodal inputs are different problems." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "A mapping-name None assumption bug and AutoProcessor kwargs forwarding are not the same code-path." + }, + { + "left": "issue:43232", + "right": "issue:44743", + "accept": false, + "reason": "General generation kwarg updates and Qwen3.5 recurrent-state reset are distinct generation bugs." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Sentence-Transformers loading stall and a MacOS bus error on CLIP are different failures." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 settings guidance and torch.split return values are unrelated." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor caching kwargs and chat-template crashes with tool calls are separate APIs and failure modes." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized models and incorrect pipeline-parallel outputs are not the same issue." + }, + { + "left": "issue:43122", + "right": "issue:44514", + "accept": false, + "reason": "Version-to-version tokenization drift and chat-template batching crashes are different bugs." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version support regression and pipeline-parallel output corruption are unrelated." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Reading a buffer-backed position_ids tensor and tokenizer_class mismatch on save are different tokenizer/model serialization issues." + }, + { + "left": "issue:43232", + "right": "issue:45245", + "accept": false, + "reason": "Generation kwargs handling and a category-count runtime limit are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 tokenizer offset mapping and failing to load a specific checkpoint are different issues." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "A model-type warning and a missing attribute on Sam3VisionEncoderOutput are not the same defect." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Missing nn import in sentence-transformers and stale auto_mappings references are separate import/maintenance problems." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Quantized model saving failure and invalid GenerationConfig are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Missing AutoImageProcessor import and a broad upgrade-wide import-error report are related only at a very high level, not the same bug." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "A missing rms_norm_eps configuration and a wrong docstring are not the same bug." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "A missing ReLU in the DSA indexer and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset mapping and assistant mask generation are different tokenizer/processor behaviors." + }, + { + "left": "issue:29127", + "right": "issue:30990", + "accept": false, + "reason": "LayoutLMv3 error messaging and Sentence Transformers loading are completely 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": 2, + "component_count": 16, + "review_unit_index": 37, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:33453", + "issue:36296", + "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:43452", + "issue:43493", + "issue:43673", + "issue:43784", + "issue:43824", + "issue:43825", + "issue:43874", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44485", + "issue:44704", + "issue:44779", + "issue:44936", + "issue:44945", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45216", + "issue:45278", + "issue:45446", + "issue:45478", + "issue:45563", + "issue:45753" + ], + "soft_pairs": [ + "issue:43122|issue:44779", + "issue:42907|issue:44964", + "issue:44038|issue:44297", + "issue:42907|issue:43452", + "issue:44038|issue:44295", + "issue:36296|issue:43116", + "issue:43824|issue:44351", + "issue:44038|issue:44222", + "issue:43066|issue:44779", + "issue:45216|issue:45478", + "issue:42757|issue:44351", + "issue:43278|issue:44945", + "issue:43122|issue:44977", + "issue:43992|issue:44704", + "issue:43493|issue:45753", + "issue:42907|issue:44222", + "issue:43784|issue:45446", + "issue:42673|issue:45072", + "issue:38617|issue:45278", + "issue:22355|issue:29127", + "issue:43937|issue:44485", + "issue:42222|issue:43441", + "issue:43064|issue:43673", + "issue:43116|issue:43493", + "issue:42371|issue:43874", + "issue:43976|issue:45446", + "issue:44075|issue:44936", + "issue:43116|issue:43825", + "issue:44222|issue:45478", + "issue:44038|issue:45478", + "issue:45083|issue:45446", + "issue:43937|issue:45563", + "issue:33453|issue:39401", + "issue:39401|issue:44779" + ], + "prepared_review_unit_hash": "d079340d6fd74993fc1b68ba38daeea6ab934b515166292703aa96b3808123e7", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37369, + "estimated_input_tokens": 9343, + "estimated_eval_tokens": 18942 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: mostly unrelated Transformers regression/import/tokenizer issues with no true duplicate pairs among the soft candidates. The broadest representative is the open import-regression report, but it is not a true canonical duplicate for the rest.", + "confidence": 0.78, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella for the repeated import-regression reports, though the cluster overall is too mixed to have a single true canonical duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best overall representative because it spans the widest common theme here: version-related import failures after upgrading Transformers.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and symptoms; no evidence of the same concrete bug." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "One is about saving dequantized models, the other about loading a multimodal model; different code paths." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe loading failure and tokenizer_config metadata mismatch are distinct issues." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Dequantized save issues and gguf AutoTokenizer/AutoModel loading failures are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "A model-specific loading bug and a position_ids buffer access bug are not the same defect." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training failure and empty multi-label classification outputs are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they reference different missing symbols from different modules." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Both mention MoE, but one is a model load bug and the other is an FP8 save_pretrained bug." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Tokenizer decoder-type mismatch and Deepseek tokenization regression are different concrete failures." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained checkpoint corruption and from_pretrained loading errors are separate directions with different symptoms." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports from different packages; only the generic class of error matches." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Embedding dtype drift in evaluate and wrong outputs under pipeline parallelism are not the same bug." + }, + { + "left": "issue:43122", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer regression and flash-attention generation failure are unrelated." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Missing embed_tokens.weight in one encoder and kwargs not forwarded in AutoProcessor are different code-path issues." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Implementation discrepancy for SigLIP2 and infinite-loop generation on one image are not duplicates." + }, + { + "left": "issue:42907", + "right": "issue:44222", + "accept": false, + "reason": "Both involve save_pretrained, but they target different models and formats with different failures." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Sentence-transformers import NameError and a flex_attention version-check bug are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leakage in dataloader threads and dtype mismatches in bfloat16 inference are different problems." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "A specific missing import and a broad import-error roundup overlap only loosely; not the same underlying bug." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "ONNX module import failure and a LayoutLMv3 error-message clarity issue are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Invalid GenerationConfig and GLM-5 RoPE discussion are distinct issues." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "vitpose breakage and Ministral-3 FlashAttention failure are unrelated model-specific bugs." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Optimizer-state corruption in training and missing GenerationMixin cache in chunked prefill are different code paths." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Classification-script empty results and SigLIP2 implementation mismatch are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 behavior settings and a missing image-patch-count method are not the same defect." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python compatibility/version support and an incorrect PyTorch version check are different issues." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Ignored SGD arguments and trainer.train()/evaluate() lifecycle failure are unrelated." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Empty multi-label outputs and a misleading pipeline translation-task error message are different bugs." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "Both are MoE-related, but one is save_pretrained FP8 and the other is from_pretrained loading; not mergeable as one fix." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen model families and different API paths; not the same concrete bug." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "qwen3_omni_moe length-helper behavior and a flex_attention import/version-check issue are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Invalid GenerationConfig and a stale warning in paged generate are different behaviors." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer-related, but one is loading regression and the other is wrong offset mapping; different defects." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Different models and different tokenizer failure modes; no shared underlying bug is evident." + } + ] + }, + "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": 16, + "review_unit_index": 38, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:42673", + "issue:42831", + "issue:43012", + "issue:43066", + "issue:43122", + "issue:43408", + "issue:43421", + "issue:43450", + "issue:43577", + "issue:43582", + "issue:43673", + "issue:43742", + "issue:43827", + "issue:43906", + "issue:43937", + "issue:43986", + "issue:44075", + "issue:44117", + "issue:44220", + "issue:44261", + "issue:44291", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44496", + "issue:44829", + "issue:44933", + "issue:44977", + "issue:45071", + "issue:45081", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45405", + "issue:45446", + "issue:45464", + "issue:45542", + "issue:45561", + "issue:45588", + "issue:45656" + ], + "soft_pairs": [ + "issue:44117|issue:45588", + "issue:43827|issue:45446", + "issue:43937|issue:44977", + "issue:43450|issue:43986", + "issue:44291|issue:45071", + "issue:42673|issue:43906", + "issue:45405|issue:45446", + "issue:45341|issue:45561", + "issue:39401|issue:43421", + "issue:44117|issue:45081", + "issue:38617|issue:41628", + "issue:44829|issue:45216", + "issue:43408|issue:43742", + "issue:42222|issue:45542", + "issue:44410|issue:45310", + "issue:42175|issue:42222", + "issue:42673|issue:43582", + "issue:44484|issue:45310", + "issue:42371|issue:44261", + "issue:42673|issue:43577", + "issue:43937|issue:44496", + "issue:38617|issue:44220", + "issue:42617|issue:45341", + "issue:44933|issue:45278", + "issue:44483|issue:45464", + "issue:44075|issue:45656", + "issue:43066|issue:43122", + "issue:42371|issue:42831", + "issue:43012|issue:43577", + "issue:43673|issue:45341" + ], + "prepared_review_unit_hash": "d2b66a3f4451818dfceb613bc68e499156b4c336bf6ceec0bd93949d0aa8db42", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36482, + "estimated_input_tokens": 9121, + "estimated_eval_tokens": 18498 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The candidates are mostly unrelated closed issues with only superficial lexical overlap; none of the soft pairs looks like the same concrete bug/change, so I reject all pairings. The closest umbrella topic is the open import-regression report, but it still does not duplicate the specific import-error tickets here.", + "confidence": 0.8, + "canonical_issue_reason": "issue:45278 is the broadest tracking-style issue in the set: it summarizes multiple import regressions after upgrading from 4.57.0 to 5.5.0 and is the most plausible umbrella representative.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best representative issue because it is general enough to subsume several of the import-related reports, while the others are narrower point bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "Different failure modes: tokenizer mapping None vs flash-attention s_aux=None crash in integration code." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs mentioning removed pipelines is unrelated to the flex_attention PyTorch version-check bug." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "One is GenerationConfig validation, the other is Qwen3.5 flash-attention generation behavior; not the same bug." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video-processor shape output vs AutoProcessor import crash without torchvision are distinct issues." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized TypeError and PretrainedConfig type-checking are separate compatibility problems." + }, + { + "left": "issue:42673", + "right": "issue:43906", + "accept": false, + "reason": "VRAM leak across dataloader threads is not the same as an isolated reproduction of a different issue." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "PEFT version pinning and flex_attention import-version gating are unrelated." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "A testing_utils bug and an xdist race in captured_info.txt are different code paths and failure causes." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Tokenizer offset_mapping correctness is separate from runtime special-token/post-processor update behavior." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping returning None and the Mistral regex backend_tokenizer AttributeError are not the same defect." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different missing symbols from different modules; they do not share a concrete code-path fix." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "FlashAttention-3 training degeneration and Qwen3.5 save_pretrained regression are unrelated model-behavior bugs." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "A model-type warning for sam3_tracker/sam3_video is not the same as a key error when loading MobileLLM." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "Broken vitpose models and an undefined TensorFlow backend when only tensorboard is installed are different problems." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing projections in qwen3next and a Qwen3.5 Moe from_pretrained error are distinct model-loading issues." + }, + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "Missing TensorFlow in the torch extra and broken vitpose models are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:43582", + "accept": false, + "reason": "A dataloader-thread VRAM leak is unrelated to the Apple Silicon caching_allocator_warmup TypeError." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "A question about max_shard_size defaults is not the same concrete bug as the Qwen3.5 Moe from_pretrained failure." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API-setting guidance and MLA rms_norm_eps precision issues are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:43577", + "accept": false, + "reason": "VRAM leakage and Blip2 dtype remaining float32 are different code paths and symptoms." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "GenerationConfig invalidation and unrecognized model/model_type loading are separate loader validation failures." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "A missing import from configuration_utils is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "Failure to run 3d_parallel.py and a testing_utils bug are not the same underlying issue." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "A specific missing import from image_utils is not the same concrete bug as a broad umbrella of import errors after upgrade." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Generic chat/completions request rejection and a Qwen3.5-0.8B streaming inference failure are not the same defect." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "SGD arguments being ignored and optimizer.step() being called twice under deepspeed are different optimizer bugs." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Wrong decoder type in v5 and changed tokenization between versions are related only broadly; not a single concrete bug." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 behavior warnings and FineGrainedFP8 accuracy problems are distinct precision-related issues." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "A PyTorch warning during bfloat16 compilation and Blip2 dtype not propagating are different dtype issues." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "GenerationMixin cache missing in chunked_prefill and a testing_utils bug do not share the same code path." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 39, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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: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:44493", + "issue:44704", + "issue:44829", + "issue:44987", + "issue:45003", + "issue:45071", + "issue:45103", + "issue:45245", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45459", + "issue:45478", + "issue:45542", + "issue:45588" + ], + "soft_pairs": [ + "issue:42222|issue:45459", + "issue:44410|issue:45478", + "issue:43097|issue:44493", + "issue:44484|issue:45478", + "issue:43874|issue:44261", + "issue:44263|issue:44485", + "issue:43012|issue:43408", + "issue:43441|issue:44829", + "issue:41628|issue:45071", + "issue:45071|issue:45542", + "issue:43931|issue:44410", + "issue:43986|issue:44484", + "issue:45003|issue:45341", + "issue:44117|issue:44987", + "issue:43097|issue:43452", + "issue:43986|issue:45341", + "issue:43931|issue:44038", + "issue:43335|issue:45071", + "issue:43408|issue:44368", + "issue:43874|issue:44263", + "issue:43986|issue:45372", + "issue:43531|issue:45341", + "issue:43976|issue:45071", + "issue:44829|issue:45245", + "issue:44261|issue:44485", + "issue:38617|issue:42757", + "issue:43064|issue:44315", + "issue:44410|issue:45375", + "issue:44062|issue:44117", + "issue:43097|issue:45588", + "issue:38617|issue:44351", + "issue:44704|issue:45103", + "issue:44038|issue:44410", + "issue:43986|issue:44117", + "issue:43502|issue:43992", + "issue:43097|issue:43873", + "issue:43824|issue:43986", + "issue:43066|issue:43906" + ], + "prepared_review_unit_hash": "306082c946f0dae028a391fd24e69da0b373af4e410c6715de88daa17f023fdc", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 38, + "serialized_chars": 39082, + "estimated_input_tokens": 9771, + "estimated_eval_tokens": 19798 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose collection of Transformers issues around v5 regressions, model-loading failures, and unrelated warnings; I do not see any true duplicate pairs in the provided soft edges.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43097 is the clearest representative of the broad v5-breaking-change theme: it is concrete, user-facing, and has inbound references, while most other issues are narrower model-specific bugs or unrelated import/type warnings.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43097 is the strongest overall issue to keep as the cluster representative because it captures a clear core API regression with wider impact than the more specialized reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42222", + "right": "issue:45459", + "accept": false, + "reason": "Different failures: vitpose model breakage vs tokenizer/protobuf error." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen-related, but one is missing projections in qwen3next and the other is a Qwen3.5 MoE from_pretrained load failure." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Different bugs: removed config attribute vs unexpected position-id key warning." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to a Qwen3.5 MoE loading error." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Unrelated code paths: missing image patch method vs MLA RMS norm epsilon precision issue." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "torch.split return-value issue in GLM MoE indexer vs RoPE implementation bug." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "Both mention warnings, but they are different model/compiler warning paths with no shared bug." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Different failures: Ministral FlashAttention crash vs degenerate training with flash_attention_3." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "ImportError for AutoImageProcessor is not the same as a PretrainedConfig type-checking regression." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Type-checking breakage and undefined tf backend error are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Both are Qwen family loading issues, but they affect different models and different concrete failures." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "AutoProcessor/torchvision crash is unrelated to the save_pretrained shard-size discussion." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafe sys.modules access is unrelated to the testing_utils bug." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES None handling is unrelated to the physical-intelligence/fast loading failure." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed tie_embeddings behavior is not the same as gguf_file breaking tokenizer/model loading." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "AutoProcessor without torchvision and a testing_utils bug are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Both are Qwen3-VL-related, but they are distinct model variants with different loading problems." + }, + { + "left": "issue:43335", + "right": "issue:45071", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to PretrainedConfig type checking." + }, + { + "left": "issue:43408", + "right": "issue:44368", + "accept": false, + "reason": "Different warnings on different models and settings; not the same underlying issue." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Missing image-patch helper and torch.split return handling are different bugs." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Torchvision absence crash is unrelated to the mistral_common ReasoningEffort import error." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to the testing_utils issue." + }, + { + "left": "issue:43976", + "right": "issue:45071", + "accept": false, + "reason": "Python-version compatibility problem is not the same as a config type-check regression." + }, + { + "left": "issue:44829", + "right": "issue:45245", + "accept": false, + "reason": "Degenerate training with flash_attention_3 and category-count overflow are unrelated." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "Missing rms_norm_eps in MLA q_a_layernorm is unrelated to GLM-5 RoPE implementation." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they are for different missing symbols and different modules." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "FSDP2 optimizer-state corruption is unrelated to Liger Kernel not being applied with model_init." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Different Qwen vision config bugs: missing projections vs missing deepstack_visual_indexes field." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer AddedToken argument collision is unrelated to TOKENIZER_MAPPING_NAMES returning None." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "Removed tie_embeddings support is unrelated to the flash_attention s_aux=None crash." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing exports and different code paths." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "Passing kwargs to cached_file is unrelated to the auto_docstring annotations crash." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Both involve Qwen variants, but the concrete bugs and affected components differ." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "AutoProcessor/torchvision failure is unrelated to tokenizer mapping None handling." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "local_files_only network leakage is unrelated to missing embed_tokens.weight in UMT5Encoder." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "Config removal and offloading/quantization behavior are different issues." + }, + { + "left": "issue:43824", + "right": "issue:43986", + "accept": false, + "reason": "Missing Qwen2.5_VLForConditionalGeneration import is unrelated to the torchvision-less AutoProcessor crash." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type in v5 is unrelated to the isolated reproduction of issue 38071." + } + ] + }, + "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": 16, + "review_unit_index": 40, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:38617", + "issue:42757", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43381", + "issue:43475", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "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:44220", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44655", + "issue:44683", + "issue:45216", + "issue:45290", + "issue:45381", + "issue:45479", + "issue:45753" + ], + "soft_pairs": [ + "issue:43097|issue:43901", + "issue:43825|issue:44655", + "issue:43352|issue:44380", + "issue:43992|issue:44484", + "issue:43012|issue:44368", + "issue:42886|issue:43381", + "issue:43012|issue:43550", + "issue:43208|issue:43550", + "issue:42907|issue:45753", + "issue:44380|issue:45479", + "issue:42907|issue:44655", + "issue:43352|issue:43976", + "issue:43064|issue:44380", + "issue:36683|issue:43582", + "issue:43208|issue:43526", + "issue:44483|issue:44514", + "issue:43618|issue:43761", + "issue:43618|issue:44380", + "issue:43012|issue:44683", + "issue:42757|issue:43824", + "issue:44483|issue:45290", + "issue:44038|issue:44230", + "issue:43299|issue:43867", + "issue:38617|issue:43502", + "issue:44483|issue:45381", + "issue:43756|issue:44485", + "issue:43901|issue:44655", + "issue:43299|issue:45753", + "issue:36683|issue:43475", + "issue:43867|issue:45216", + "issue:43352|issue:43937", + "issue:42994|issue:44655", + "issue:44077|issue:44380", + "issue:43066|issue:45381", + "issue:36683|issue:44077", + "issue:36010|issue:42757", + "issue:44220|issue:44655" + ], + "prepared_review_unit_hash": "7bf07544d8ff79df8fbc396bde8ea9516a6119e6ede37d5716db139263476e6c", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 37, + "serialized_chars": 37951, + "estimated_input_tokens": 9488, + "estimated_eval_tokens": 19232 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes unrelated Transformers v5 regressions, import errors, model-specific load/save bugs, and docs/behavior mismatches. There is no real duplicate group here, so all soft links should be rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue is a true canonical duplicate for this mixed cluster; issue 44483 is the least-bad representative because it is a broad, user-facing v5 API regression.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44483 is the best cluster representative because it affects a core public endpoint and has broader impact than the more model-specific or doc-only items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43097", + "right": "issue:43901", + "accept": false, + "reason": "Removed embedding-tie API vs a docs/behavior mismatch for text classification; unrelated problems." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline translation-task messaging vs pipeline save_pretrained persistence; different code paths and failures." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Nemotron flash-attn support gate vs GPT2 attention scaling under SDPA/FlashAttention; different models and bugs." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing encoder weight in load path vs default shard size choice in save_pretrained; not the same defect." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "PyTorch precision warning during compile vs tie_word_embeddings warning from config handling; only superficially similar." + }, + { + "left": "issue:42886", + "right": "issue:43381", + "accept": false, + "reason": "Offline cache loading regression vs gradient checkpointing eval-mode restriction; unrelated behavior." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "bfloat16 compile warning vs Bamba torch.compile SDPA failure; different symptom and root cause." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs vs Bamba compile/SDPA failure; distinct models and issues." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Saving dequantized models vs image-specific infinite-loop generation; no shared underlying bug." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention scaling ignored in attention backends vs zero-loss sequence classification bug; unrelated subsystems." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Quantized/dequantized model save failure vs saving Pipeline objects; both involve saving but not the same bug." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model-specific flash-attn support vs Python 3.9/3.10 compatibility break; different failure domains." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2 optimizer-state mismatch vs GPT2 attention scaling; completely different training vs attention logic." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Gemma3Config missing vocab_size vs Apple Silicon allocator TypeError; unrelated errors." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training bugs vs BeitImageProcessorFast label reduction bug; different components and symptoms." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "v1 chat/completions request rejection vs batched chat-template crash; both chat-related but not the same concrete bug." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "CLIP attentions assignment regression vs CLIPVisionModel hidden_states regression; related area but different outputs and code paths." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIP output regression vs GPT2 attention scaling; no meaningful overlap beyond model internals." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "Compile precision warning vs flex_attention failure on newer torch; different failure modes." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Missing hub helper import vs missing Transformers model import; both import errors but distinct symbols and fixes." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Chat/completions endpoint rejection vs assistant tool-call chat-template crash; different APIs and failure points." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Qwen3-VL-Moe loading bug vs fp8 support request; one is a regression, the other is capability support." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL-Moe loading break vs sorted-state_dict load error; both loading-related but not the same issue." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "Missing configuration_utils import vs local_files_only still making network requests; unrelated import vs I/O behavior." + }, + { + "left": "issue:44483", + "right": "issue:45381", + "accept": false, + "reason": "Chat/completions request rejection vs Qwen2.5-VL video vision_position_ids bug; different endpoints and model logic." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Smollm3 RoPE-layer mismatch vs GLM-5 RoPE implementation discussion; related topic but not the same concrete bug." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mention stale flag vs inability to save Pipeline objects; unrelated." + }, + { + "left": "issue:43299", + "right": "issue:45753", + "accept": false, + "reason": "Qwen3VL-Moe loading regression vs infinite-loop generation on a specific image; distinct symptoms." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Gemma3 config attribute missing vs SAM3 vision encoder output attribute missing; both missing attributes but different models and fields." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "General load-model error with sorted state_dict vs Qwen3.5 save_pretrained checkpoint corruption; different save/load sides." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Flash-attention support error vs invalid GenerationConfig for GLM-5; unrelated model/config issues." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized model save failure vs Pipeline save_pretrained failure; both saving-related but not the same bug." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "patchtsmixer post_init validation vs GPT2 attention scaling backend bug; unrelated." + }, + { + "left": "issue:43066", + "right": "issue:45381", + "accept": false, + "reason": "Wrong decoder type in v5 tokenizer handling vs video vision_position_ids bug; different components." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Gemma3 config missing vocab_size vs patchtsmixer post_init policy; no shared underlying issue." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Missing GenerationMixin import vs missing is_offline_mode import; both import errors but different modules and fixes." + }, + { + "left": "issue:44220", + "right": "issue:44655", + "accept": false, + "reason": "Audio fbank extraction bug vs Pipeline save_pretrained bug; unrelated 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": 2, + "component_count": 16, + "review_unit_index": 41, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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:43723", + "issue:43867", + "issue:43931", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44077", + "issue:44112", + "issue:44188", + "issue:44242", + "issue:44246", + "issue:44263", + "issue:44355", + "issue:44360", + "issue:44380", + "issue:44484", + "issue:44492", + "issue:44556", + "issue:44855", + "issue:44857", + "issue:44938", + "issue:44964", + "issue:45071", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45381", + "issue:45440", + "issue:45478", + "issue:45561", + "issue:45588", + "issue:45593", + "issue:45600" + ], + "soft_pairs": [ + "issue:44263|issue:45440", + "issue:43976|issue:44938", + "issue:36683|issue:43525", + "issue:43352|issue:45593", + "issue:43064|issue:44492", + "issue:44188|issue:44380", + "issue:43618|issue:45216", + "issue:36683|issue:44242", + "issue:43352|issue:43931", + "issue:43010|issue:45278", + "issue:41669|issue:44246", + "issue:43992|issue:45310", + "issue:43867|issue:45375", + "issue:36683|issue:43064", + "issue:44355|issue:44855", + "issue:42890|issue:45561", + "issue:44938|issue:45071", + "issue:43867|issue:45357", + "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:43618|issue:44857", + "issue:43618|issue:45588", + "issue:43867|issue:44484", + "issue:28282|issue:33453", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43502|issue:44062", + "issue:43723|issue:44556" + ], + "prepared_review_unit_hash": "4d35b1c31327a139888132bf4056ee5bb95888904a531de3476e081144c9919b", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37242, + "estimated_input_tokens": 9311, + "estimated_eval_tokens": 18878 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a grab bag of unrelated model, config, training, and test failures. The only clear duplicate-like soft edge is import-time slowness around transformers imports; most other pairs are too model- or code-path-specific to merge.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45278 is the broadest umbrella in the set: it describes upgrade-related import/load regressions affecting multiple models, making it the best anchor for the import-regression subset of this cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best issue to keep as the cluster representative because it is the most general user-facing regression report and can subsume nearby import/loading failures better than the narrower model-specific bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44263", + "right": "issue:45440", + "accept": false, + "reason": "Different problems in different components: DSA indexer behavior vs DeepSeekV3 MoE divergence." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both mention Python-version issues, but one is a v5.1.0 compatibility problem and the other is a Python 3.14 load failure; no same code path." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Different missing config attributes on different models; same general symptom, not the same bug." + }, + { + "left": "issue:43352", + "right": "issue:45593", + "accept": false, + "reason": "Unrelated model-specific bugs: FlashAttention support gating vs D-FINE auxiliary loss behavior." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Training/FSDP optimizer-state bug is unrelated to a cache-strategy typo." + }, + { + "left": "issue:44188", + "right": "issue:44380", + "accept": false, + "reason": "Both are attention-related, but they affect different mechanisms and code paths." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIP attentions assignment regression and Qwen3.5 save_pretrained checkpoint regression are different issues." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Different failures: missing Gemma3Config field vs MoE load-balancing loss omission." + }, + { + "left": "issue:43352", + "right": "issue:43931", + "accept": false, + "reason": "Different models and failures: FA2 support check vs Qwen3-VL weight-shape mismatch." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "No overlap in code path: no_grad cache update vs broad import errors after upgrading." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": true, + "reason": "Both describe import-time slowdown in transformers; 41669\u2019s import * regression plausibly explains the intermittent slow import reported in 44246." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both are from_pretrained failures, but they affect different models and different loading paths." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Generic state_dict ordering error vs missing Qwen3.5 vision config field; not the same defect." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Completely different areas: config attribute access vs distributed training optimizer state handling." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Different import/compile failures with different root causes; one is compiled Python files, the other is torch.jit parsing a commented decorator block." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Both hit tests, but one is seed setup flakiness and the other is xdist file races; different failures." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 load failure and PretrainedConfig type-checking breakage are unrelated regressions." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Both touch Qwen3.5 loading/saving, but one is a load error and the other is incorrect save keys; not the same concrete bug." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Different model-loading errors on different architectures and code paths." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "No relation between cache update behavior and Python-version compatibility." + }, + { + "left": "issue:44938", + "right": "issue:45381", + "accept": false, + "reason": "Python 3.14 import failure vs Qwen2.5-VL video position IDs are unrelated." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different model issues: CLIP output fields vs patchtsmixer post_init validation." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "Unrelated algorithmic bugs in different model components." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Both mention tensor parallelism, but one is a training bug and the other is an OOM on loading a specific model; too different to merge." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "CLIP attentions regression and float16 AMP crash are unrelated." + }, + { + "left": "issue:43618", + "right": "issue:45588", + "accept": false, + "reason": "Different subsystems: CLIP outputs vs flash_attention sink-less crash." + }, + { + "left": "issue:43867", + "right": "issue:44484", + "accept": false, + "reason": "Generic load failure vs save_pretrained shard-size question; not the same bug." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Both are loading-related, but one is a missing-PyTorch ImportError and the other is a tokenizer-loading regression." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Test seeding flakiness is unrelated to a multimodal model loading failure." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Different CI/test issues with different failure modes." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Unrelated test flake vs auto_mappings CI breakage." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Local-files-only API leakage and AddedToken duplicate-arg TypeError are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 loading regressions, but one is tokenizer loading and the other is checkpoint reload compatibility; not the same code path." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The representative choice and summary are grounded in the packet, and the lone accepted soft edge (41669 \u2194 44246) is a conservative, well-supported import-slowdown match. The other refusals stay appropriately cautious about differing models or code paths." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 42, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "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:43824", + "issue:43827", + "issue:43874", + "issue:43901", + "issue:44075", + "issue:44222", + "issue:44230", + "issue:44261", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44368", + "issue:44387", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44655", + "issue:44749", + "issue:44908", + "issue:44961", + "issue:45071", + "issue:45125", + "issue:45341", + "issue:45375", + "issue:45468", + "issue:45478" + ], + "soft_pairs": [ + "issue:42890|issue:45341", + "issue:43010|issue:44075", + "issue:44387|issue:44556", + "issue:42548|issue:44351", + "issue:44222|issue:45125", + "issue:43010|issue:43065", + "issue:43010|issue:43232", + "issue:43502|issue:44485", + "issue:44355|issue:44749", + "issue:45125|issue:45478", + "issue:30990|issue:31795", + "issue:45125|issue:45375", + "issue:30333|issue:33453", + "issue:43010|issue:44368", + "issue:43299|issue:44556", + "issue:43502|issue:44222", + "issue:44315|issue:44360", + "issue:43502|issue:44961", + "issue:42831|issue:43874", + "issue:44355|issue:44655", + "issue:30333|issue:36296", + "issue:44222|issue:44230", + "issue:44355|issue:45071", + "issue:43827|issue:44961", + "issue:33453|issue:34689", + "issue:42757|issue:43576", + "issue:33453|issue:36296", + "issue:43576|issue:45071", + "issue:43576|issue:43824", + "issue:43673|issue:44492", + "issue:44961|issue:45468", + "issue:31795|issue:38617", + "issue:41628|issue:42548", + "issue:38617|issue:43576", + "issue:43576|issue:44351", + "issue:44908|issue:44961", + "issue:44485|issue:44961", + "issue:36296|issue:43901", + "issue:41628|issue:44355", + "issue:36296|issue:43526", + "issue:44261|issue:44360" + ], + "prepared_review_unit_hash": "17bce9ea23288d16d79d8dc5be12e9af4c4da3661ff57b5e3cb42852471cd966", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 41, + "serialized_chars": 38887, + "estimated_input_tokens": 9722, + "estimated_eval_tokens": 19700 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly a loose bundle of unrelated Transformers v5 regressions, import errors, model-loading issues, and docs/test bugs. I do not see any soft pair that clearly refers to the same underlying bug or mergeable change.", + "confidence": 0.9, + "canonical_issue_reason": "issue:43576 is the broadest representative of the recurring v5 breakage theme and has the strongest discussion/inbound signal among the set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43576 is the best single representative issue here because it is broad and central, while the others are narrower, model-specific, or unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:45341", + "accept": false, + "reason": "Different scope: flaky SamHQ integration tests vs a generic testing_utils bug." + }, + { + "left": "issue:43010", + "right": "issue:44075", + "accept": false, + "reason": "Unrelated code paths: cache/layer no_grad behavior vs SGD optimizer args." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Both are v5 regressions, but they describe different failures: int4 memory growth vs checkpoint reload compatibility." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different exported symbols and likely different fixes." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "Different problems: FP8 save_pretrained behavior vs tensor-parallel plan missing on Qwen3.5 MoE." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Different models and bugs: cache update semantics vs Sam3PixelDecoder Conv2d stub." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Different generation-related issues with separate code paths and symptoms." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "Unrelated: offline/local-files-only networking behavior vs GLM-5 RoPE implementation." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Unrelated: compiled Python file errors vs performance slowdown after upgrade." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Related model family, but one is missing _tp_plan and the other is a from_pretrained failure; not clearly the same bug." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Different concerns: loading hang vs documentation wording." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Different Qwen3.5 MoE config/load issues involving separate missing fields." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "Different failures: MLFlow reporting on job failure vs tokenizer loading regression." + }, + { + "left": "issue:43010", + "right": "issue:44368", + "accept": false, + "reason": "Different issues: no_grad cache updates vs tied-embeddings warning in LoRA fine-tuning." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Both are version-related loading regressions, but for different models and failure modes." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Different bugs: local_files_only network leakage vs FP8 moe save_pretrained." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Completely different subsystems: Liger kernel application vs DSA indexer activation." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "No overlap beyond being issues; networking bug vs unrelated raccoon title." + }, + { + "left": "issue:42831", + "right": "issue:43874", + "accept": false, + "reason": "Different model behaviors: FP8 accuracy vs multimodal token counting error." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Different areas: compiled-file errors vs pipeline save_pretrained support." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "Unrelated training concerns: MLFlow reporting vs tensor-parallel training bug." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Related to FP8, but one is save_pretrained and the other is inference support; not the same bug." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Different failure types: compiled file errors vs PretrainedConfig type-checking regression." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs/pipeline removal issue vs unrelated issue title." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading regressions, but they target different models and symptoms." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Different symptoms: missing hub import vs broken env command." + }, + { + "left": "issue:33453", + "right": "issue:36296", + "accept": false, + "reason": "Different problems: tokenizer loading regression vs tensor-parallel training bug." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Different code paths: env CLI regression vs config type-checking." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Different bugs: env command broken vs missing Qwen2.5-VL export." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Cache missing in generation vs a typo in cache strategies." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "No apparent relation; the titles point to unrelated issues." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation clarification vs import error in configuration_utils." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different export fixes." + }, + { + "left": "issue:38617", + "right": "issue:43576", + "accept": false, + "reason": "Different layers: missing helper import vs env CLI breakage." + }, + { + "left": "issue:43576", + "right": "issue:44351", + "accept": false, + "reason": "Both mention import-related breakage, but the affected symbols and failure contexts differ." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwargs bug vs unrelated issue title." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "Unrelated model implementation discussion vs unrelated issue title." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug vs documentation mismatch for text classification pipeline." + }, + { + "left": "issue:41628", + "right": "issue:44355", + "accept": false, + "reason": "Import export regression vs compiled-file execution errors." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Different model/training bug vs BeitImageProcessorFast label handling bug." + }, + { + "left": "issue:44261", + "right": "issue:44360", + "accept": false, + "reason": "Precision/config discussion for MLA vs a DSA indexer activation issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 43, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:34689", + "issue:37428", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43502", + "issue:43519", + "issue:43576", + "issue:43673", + "issue:43704", + "issue:43827", + "issue:43976", + "issue:44038", + "issue:44077", + "issue:44246", + "issue:44263", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44655", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:44945", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45478", + "issue:45542", + "issue:45753" + ], + "soft_pairs": [ + "issue:44273|issue:45003", + "issue:44623|issue:45230", + "issue:44273|issue:45542", + "issue:44393|issue:45753", + "issue:43576|issue:44556", + "issue:41669|issue:43673", + "issue:44393|issue:45071", + "issue:44038|issue:45125", + "issue:44556|issue:45092", + "issue:42371|issue:44492", + "issue:44704|issue:44908", + "issue:42890|issue:44908", + "issue:44246|issue:44273", + "issue:43408|issue:44315", + "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:41669|issue:44492", + "issue:44393|issue:45478", + "issue:44315|issue:44829", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "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", + "issue:43519|issue:44485", + "issue:41669|issue:43704", + "issue:43519|issue:44908", + "issue:31795|issue:34689", + "issue:31795|issue:37428", + "issue:44315|issue:45092", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:42371|issue:43704", + "issue:43502|issue:43519", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:44263|issue:45468" + ], + "prepared_review_unit_hash": "a5e04012ed0db84461f5b5d5bbce1b389117dc2521d7f28e12ce30945b378553", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 48, + "serialized_chars": 41424, + "estimated_input_tokens": 10356, + "estimated_eval_tokens": 20968 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose Transformers issue cluster with repeated loading/import/model-initialization themes, but the items are mostly distinct regressions, docs, or model-specific bugs rather than true duplicates.", + "confidence": 0.52, + "canonical_issue_reason": "issue:44556 is the clearest umbrella-style loading regression: checkpoint reload failures after upgrading to v5.2/v5.3. It is broad enough to represent the cluster\u2019s general model-loading breakage theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45092 is the most actionable and specific report: it describes a concrete incompatibility for old InternVL2 remote-code checkpoints during v5 meta initialization, which makes it the strongest standalone issue in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Both touch core runtime behavior, but one is lazy loading and the other is unsafe sys.modules access; different bugs and fixes." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "44623 is a concrete processor.save_pretrained bug; 45230 is a generic bug report with no matching failure mode." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading regressions and a TensorFlow backend error are unrelated issues." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Both mention Qwen3-VL/Qwen3.5 image behavior, but one is bbox output hallucination and the other is an infinite loop on a specific image." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "One is the v5 env command, the other is checkpoint reload compatibility; same era, different functionality and root cause." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-star cleanup causing import slowdown is not the same bug as missing cache state during chunked prefill." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Qwen3-VL bbox output issues are unrelated to PretrainedConfig type-checking regressions." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both are Qwen3.5/Moe related, but one is a model support bug and the other is a missing tensor-parallel plan." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both concern v5 loading breakage, but they describe different incompatibilities and code paths; not the same concrete bug." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API guidance and a cache-strategy typo are unrelated." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and inverse_sqrt scheduler kwargs are different APIs with different failures." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Missing set_seed in tests is a flaky-test issue, not a scheduler configuration bug." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import-time slowdown and lazy-loading correctness are adjacent but not the same bug." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Model-type warning in sam3_tracker and Liger Kernel application with model_init are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version compatibility and Qwen3-VL bounding-box output are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention support gating for Nemotron and Qwen3-VL bbox hallucination are different model issues." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "The env command regression and scheduler kwargs regression do not share a concrete fix." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer class metadata mismatch and a generic bug report are not duplicates." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs handling and PretrainedConfig type checking are unrelated." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Position_ids buffer access error and a generic bug report are not the same underlying issue." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star cleanup and HybridCache import failure are different import-path problems." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Model import bloat and a testing_utils bug are unrelated." + }, + { + "left": "issue:44492", + "right": "issue:44945", + "accept": false, + "reason": "Cache-strategy typo and incorrect pipeline-parallel output are separate concerns." + }, + { + "left": "issue:41669", + "right": "issue:44492", + "accept": false, + "reason": "The import-star cleanup issue is not the same as a cache-strategy typo." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Qwen3-VL bbox errors and Qwen3.5 Moe from_pretrained failures are different model regressions." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not applied in model_init and flash_attention_3 degenerate training are not the same code-path failure." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "A post_init policy change for patchtsmixer is unrelated to an AttributeError in tied-weight key handling." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still referencing pipeline() and scheduler kwargs are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion about forward args and a tokenizer loading regression are different issues." + }, + { + "left": "issue:41669", + "right": "issue:44908", + "accept": false, + "reason": "Import-time bloat and inverse_sqrt scheduler kwargs are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env command and lazy-loading malfunction are separate v5 issues." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Saving pipeline objects and tied-weight key attribute handling do not share the same failure mode." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer activation logic and Gemma-4 audio positional encoding are unrelated model internals." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report and Qwen3.5 Moe from_pretrained error are not the same problem." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading and scheduler kwargs handling are unrelated." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are different model-specific bugs." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star cleanup does not match a VRAM leak in multithreaded dataloaders." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Timestamp math in Qwen3VL and inverse_sqrt scheduler kwargs are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Forward-argument docs and model-loading breakage for Llama 3.2 are not duplicates." + }, + { + "left": "issue:31795", + "right": "issue:37428", + "accept": false, + "reason": "Documentation wording and an import error from flash attention utils are unrelated." + }, + { + "left": "issue:44315", + "right": "issue:45092", + "accept": false, + "reason": "Both concern model initialization/loading, but one is Liger Kernel application and the other is remote-code checkpoint meta-init incompatibility." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leakage in dataloader threads and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "GLM-5 RoPE math and Gemma-4 audio positional encoding are different model implementations." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 API guidance and a VRAM leak are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "local_files_only network calls and Qwen3VL timestamp calculation are different bugs." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp calculation and processor.save_pretrained file omission are unrelated." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Qwen3VL timestamp math and dataloader-thread VRAM leakage are different issues." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split return handling in GlmMoeDsaIndexer and Gemma-4 audio positional encoding do not share the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 16, + "review_unit_index": 44, + "review_unit_count": 44, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-15354-384", + "nodes": [ + "issue:30333", + "issue:31515", + "issue:31795", + "issue:33453", + "issue:36296", + "issue:38617", + "issue:42913", + "issue:43066", + "issue:43298", + "issue:43525", + "issue:43572", + "issue:43606", + "issue:43618", + "issue:43704", + "issue:44230", + "issue:44261", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44998", + "issue:45290", + "issue:45310", + "issue:45341", + "issue:45468" + ], + "soft_pairs": [ + "issue:43606|issue:45468", + "issue:44261|issue:45468", + "issue:43704|issue:45341", + "issue:42913|issue:43066", + "issue:44908|issue:44998", + "issue:44861|issue:45290", + "issue:44492|issue:44998", + "issue:44998|issue:45341", + "issue:43618|issue:44998", + "issue:44556|issue:44998", + "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": "dde875dd58897301f324d18ceeed74a25e15e4f4b743ddf5e96aedc13fa37388", + "component_budget": { + "node_count": 384, + "item_count": 384, + "soft_pair_count": 1441, + "serialized_chars": 742673, + "estimated_input_tokens": 185669, + "estimated_eval_tokens": 371594 + }, + "budget": { + "node_count": 26, + "item_count": 26, + "soft_pair_count": 20, + "serialized_chars": 20575, + "estimated_input_tokens": 5144, + "estimated_eval_tokens": 10544 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly heterogeneous. A small subgroup concerns Transformers v5 migration/loading/tokenizer/config regressions, but several items are unrelated model-specific bugs and one is clearly off-topic, so there is no safe duplicate merge within the provided pairs.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 44556 is the broadest representative of the coherent v4\u2192v5 checkpoint/loading regression theme that overlaps with several other token/config issues; it is the best cluster anchor among the related reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44556 is the most representative issue for the strongest shared theme in this set (version-upgrade loading failures and related tokenizer/config regressions).", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "Different model-specific bugs: CPU offload device mismatch vs Gemma-4 audio positional encoding behavior." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Both mention precision/config issues, but they target different architectures and different code paths." + }, + { + "left": "issue:43704", + "right": "issue:45341", + "accept": false, + "reason": "VRAM leakage in Qwen3ForCausalLM and a testing_utils bug are unrelated." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both concern tokenizer behavior in v5, but one is a general behavior difference and the other is a specific decoder-type regression; not enough to treat as the same bug." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Scheduler keyword handling and an off-topic issue are unrelated." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "One is a tied-weights helper crash; the other is a chat-template/tool-call crash. Different subsystems and failure modes." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "A cache-strategy typo and an unrelated off-topic issue are not duplicates." + }, + { + "left": "issue:44998", + "right": "issue:45341", + "accept": false, + "reason": "Off-topic issue versus a testing_utils bug; no shared underlying defect." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the off-topic issue." + }, + { + "left": "issue:44556", + "right": "issue:44998", + "accept": false, + "reason": "Checkpoint reload regression and an unrelated off-topic issue are not duplicates." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config is a config attribute regression, not the tied-weights list crash." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is a different config regression than the tied-weights helper error." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Slow from_pretrained checkpoint loading and an ImportError in configuration_utils are distinct problems." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 model support issue and missing processor save files are unrelated." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow job reporting failure and slow checkpoint loading are not the same bug." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Checkpoint loading performance and tensor-parallel training bug are different code paths." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Checkpoint loading slowness and documentation confusion are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "Both involve Qwen3.5 MoE/fp8/versioning, but one is fp8 inference support and the other is a from_pretrained error; too different to merge as one bug." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Both mention loading, but one is slow checkpoint loading and the other is a tokenizer-loading regression; not the same concrete defect." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching audio-model support and fp8 inference support are distinct feature/bug areas." + } + ] + }, + "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": 16, + "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": "14d51eae4584217817909e37006e4234b1da0ffb1000c08a3b16207b10221ed3", + "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 two separate community-contribution feature requests: one for Flash Attention 2 support across more architectures, and one for GGUF support across more architectures. They share a loose pattern but not the same underlying change, so they should not be merged as duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 26350 has the higher discussion activity and is a clearer standalone tracker, but it is not a duplicate of 33260; the requests target different features/code paths.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Both are broad 'support more architectures' requests, but Flash Attention 2 and GGUF are different features with different implementation paths and goals, so they 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": 4, + "component_count": 16, + "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": "5875a0c81c38295cacd1e437c7a34f3fc542b44d509076bd7cb10ae564113cba", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4455, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4455, + "estimated_input_tokens": 1114, + "estimated_eval_tokens": 2484 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These look like separate docs-translation requests, not duplicates of one bug/change. Issue 27088 is specifically for Turkish, while the others are generic i18n template issues that likely refer to different languages.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical duplicate stands out: the only concrete language is Turkish in 27088, while the others are generic placeholders and likely distinct language requests.", + "canonical_pr_reason": null, + "best_issue_reason": "27088 is the best representative only because it is the most specific and earliest concrete request, but it should not absorb the others as duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Both are translation-request templates, but there is no evidence they target the same language or same underlying work item." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "27088 is specifically Turkish; the other issue is a generic placeholder likely for a different language, so they are not the same request." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Different concrete localization targets are implied; same template text is not enough to treat them as duplicates." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": false, + "reason": "These appear to be separate instances of the same issue template, likely for different languages rather than the same underlying task." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": false, + "reason": "No evidence they refer to the same language or translation branch; similarity is only from the template." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": false, + "reason": "They share the same placeholder title, but that indicates a template family, not a duplicate issue." + } + ] + }, + "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": 16, + "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-36817-14", + "nodes": [ + "issue:36817", + "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:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:37015|issue:42503", + "issue:37015|issue:45027", + "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:43837|issue:45177" + ], + "prepared_review_unit_hash": "4d59586a97e1a311b15f907290a154371d56d810423ed2e934ab37bb97d952f1", + "component_budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 28, + "serialized_chars": 17649, + "estimated_input_tokens": 4413, + "estimated_eval_tokens": 9082 + }, + "budget": { + "node_count": 14, + "item_count": 14, + "soft_pair_count": 28, + "serialized_chars": 17649, + "estimated_input_tokens": 4413, + "estimated_eval_tokens": 9082 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These items are all model-addition/support requests, but they target different architectures and modalities. The soft-similarity links are thematic, not duplicates of the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": "No true canonical duplicate: the cluster mixes distinct model requests (different models, modalities, and implementation paths), so there isn\u2019t one shared issue to collapse into.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:37015 is the broadest/high-activity open model-addition request, so it works best as a representative anchor if one must be chosen, though it is not a duplicate sink.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBert config support vs DEIMv2 support are different model additions with separate code paths." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config support and Qwen3-TTS support target unrelated model families/modalities." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are separate model requests; not the same concrete change." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT vs weight-sparse transformer/circuit-sparsity are unrelated model architectures." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT and LFM2.5 Audio are different model support requests, not duplicates." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT concern different model families and implementations." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are distinct model additions with no shared fix." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 and Qwen3-ASR are different models and code paths." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 support and circuit-sparsity support are unrelated requests." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 and ModernVBERT are separate model support items." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 and Voxtral TTS are different model-modalities and cannot be merged as one fix." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated model additions." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are different model-specific requests." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS target different architectures/modalities." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT support and BERT-like-with-RoPE support are not the same concrete change." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like models with RoPE and Voxtral TTS are unrelated additions." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR share a prefix but are different modalities and implementation paths." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS and Voxtral TTS are distinct TTS model requests, not one duplicate." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model support requests." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS support and circuit-sparsity support are different model/code changes." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS and LFM2.5 Audio are separate audio model requests, not one fix." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio are different model additions despite both being audio-related." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR and EUPE are unrelated model requests." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR support and weight-sparse transformer support are different concrete changes." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are different model families/modalities." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "Weight-sparse transformer and EUPE are distinct architecture requests with no shared fix." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different speech-model support requests and would not be merged into one PR/issue." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are separate model support items." + } + ] + }, + "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": 16, + "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": "39c90b0d3d2e9bee4f22a56d19d6f7e4bef110e316bc856047d0adf799dffdf0", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6294, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 6294, + "estimated_input_tokens": 1574, + "estimated_eval_tokens": 3404 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These issues are only superficially similar by wording; they describe different components and bug types, so this cluster should not be merged as duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:43668 is the cleanest single bug report in the set: a concrete config type-hint problem with a specific target and no sign of overlapping with the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Both mention normalization, but one asks whether a model layer is RMSNorm and the other reports an incorrect type hint in a config; different artifacts and different failure modes." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Mask2Former ignore_value handling is unrelated to kwargs not being forwarded into a loss function; different code paths and symptoms." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "One is a segmentation-model parameter plumbing issue, the other is a tokenizer regex warning regression; no shared underlying bug." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "An ignore_value usage bug is unrelated to a typo in a quantization operation name." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Different models and bug classes: ignored loss parameter versus a ModernBERT config annotation error." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Missing kwargs in loss calculation and an incorrect type hint are distinct issues with no common code-path evidence." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "A config type-hint bug is not the same as a tokenizer warning regression." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "The ModernBERT config type hint issue and the quantization-operation typo are separate, unrelated defects." + } + ] + }, + "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": 16, + "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": "107b7b76c27297bc4ce8e5f2d998daf451bf0504a600f9809cd3b80b15a9a423", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are unrelated: one asks for block decoding support, the other requests DeepSeek v3.2 support. They should remain separate; no duplicate/merge candidate here.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different underlying requests with no shared code path or explicit linkage: block decoding vs DeepSeek v3.2 model support." + } + ] + }, + "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": 16, + "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": "1c1f2cc8a8db9e9428b951e2acfebf84eca26dcadf47fbb7fe54774df1887f16", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1583, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are only loosely related by custom trainer behavior, but they describe different problems: one about `remove_unused_column` customization, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "No true canonical duplicate emerges: #43598 is about Trainer column-pruning customizability, while #43600 is about FLOPS estimation logic for custom trainers. Same general area, different underlying bug/request.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick a representative issue for the cluster theme, #43598 is broader and more directly about Trainer behavior customization, but it does not subsume #43600.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Related only at a high level (custom trainers/Trainer internals). They target different code paths and fixes: `remove_unused_column` behavior versus FLOPS estimation." + } + ] + }, + "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": 16, + "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": "eff1114ea095549cd1a4704486035f9f0c11076a09ef81b1b17cbcd3130872d9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related only at a high level: both request Qwen-family AutoModel support, but one is for Qwen3-Omni registration in AutoModel/AutoModelForConditionalGeneration and the other is for Qwen3.5 sequence classification. They are separate feature requests, not the same underlying bug/change.", + "confidence": 0.92, + "canonical_issue_reason": "issue:43684 is the broader, more central request (model registration for Qwen3-Omni across AutoModel and AutoModelForConditionalGeneration); issue:44405 targets a different config and task-specific AutoModelForSequenceClassification support.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43684 is the better representative because it is the more general model-registration request; issue:44405 is a distinct downstream capability request for sequence classification.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Do not merge: different model variants and different auto-class targets (registration vs sequence classification support), so they do not look like 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": 10, + "component_count": 16, + "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": "a3ce3a6725b1b1fa003de6cdb94f784efc70f0bf7741ebeff54ff1d3674ff820", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4264, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4264, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The three issues titled \"huggingface \" appear to be duplicates of the same report; the issue titled \"Fett\" looks unrelated. No PRs are present.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43845 is the earliest of the near-identical \"huggingface \" reports and is a reasonable canonical issue for the duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43845 is the best issue anchor because it matches the repeated title and was opened first among the duplicate-looking reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different title and no shared evidence beyond generic similarity; not enough to indicate the same underlying issue." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different title and no concrete evidence of the same bug." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different title and no concrete evidence of the same bug." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; they look like duplicate reports of the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; likely the same underlying report." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same exact title \"huggingface \" and close creation times; likely duplicate issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall: the three closed issues share the exact same title and close timestamps are close enough to support a duplicate cluster, while issue:43835 has a clearly different title. The canonical choice of issue:43845 as the earliest duplicate-looking report is conservative. No PRs are present, and the soft-edge verdicts stay within the evidence available." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 16, + "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": "5256e3b1eddb96349ae21c7880b824da21f5f3b99bdfd66bd5fafa36f8693bc3", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5788, + "estimated_input_tokens": 1447, + "estimated_eval_tokens": 3150 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are low-signal, likely unrelated spam/placeholder issues. The apparent similarity is only at the title level; there is no evidence of a shared bug or duplicate report.", + "confidence": 0.95, + "canonical_issue_reason": "Issue 44418 is the least spam-like and most readable title, but it is still a generic placeholder rather than a concrete defect; it is only a weak cluster representative.", + "canonical_pr_reason": null, + "best_issue_reason": "44418 is the best available representative because it is the only non-spam-gibberish title, but none of these issues look like a real duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are titled \"\", but there is no body content or evidence they describe the same underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same generic spam title only; no concrete shared bug or change is visible." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "A spam placeholder and a generic documentation suggestion are not the same underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "\"\" and \"Check out \\\"Google LLC\\\"\" are unrelated titles with no shared bug report." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Generic documentation suggestion versus spam placeholder; no duplicate semantics." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Generic documentation suggestion versus spam placeholder; no evidence of the same problem." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Identical spam titles alone are insufficient to treat them as the same underlying issue without any supporting details." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder and unrelated quoted-name title; no common bug/change is indicated." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "Spam placeholder and unrelated quoted-name title; no duplicate relationship is supported." + } + ] + }, + "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": 16, + "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": "1e9e3d2bf1cbe557331adcf72457bf849b80df3671059f9d0ab7793605ebab5a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The pair looks highly similar on title and length, but the packet does not expose body content or any explicit linkage, so the duplicate call is not strong enough to accept with confidence. If they are duplicates, #45145 is the better representative because it was created first.", + "confidence": 0.68, + "canonical_issue_reason": "#45145 is the earliest of the two and would be the natural representative if they are later confirmed to be duplicates, but the current evidence is too thin to state that definitively.", + "canonical_pr_reason": null, + "best_issue_reason": "#45145 is the best issue to keep as the provisional representative because it is the first-created item in the pair.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "Identical titles and a soft similarity signal suggest possible duplication, but the packet does not show body text, explicit links, or other concrete evidence that they are the same underlying issue. Given the non-deterministic signal, this should be treated as inconclusive rather than accepted as a duplicate." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The duplicate judgment is a bit too strong for the evidence provided. The packet supports only identical titles, similar body lengths, and a soft similarity signal; it does not show body content directly or any explicit linkage. Given the pair is marked non-deterministic, the soft-edge verdict should be more conservative rather than stating it is safe to merge as duplicates." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 16, + "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": 16, + "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": "ce936d403433a7492cb509877fd6456a91c8c47dda352a825328762ed32f5134", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1668, + "estimated_input_tokens": 417, + "estimated_eval_tokens": 1090 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two separate 4-bit inference bugs in bitsandbytes-related code paths: one is a Gemma4/audio-module uint8 + torch.finfo TypeError, the other is an llm_int8_skip_modules regression that drops the default lm_head exclusion and triggers an AssertionError. They should not be deduplicated.", + "confidence": 0.91, + "canonical_issue_reason": "No single canonical issue for this cluster: 45672 is a model-specific torch.finfo/type handling failure, while 45674 is a BitsAndBytesConfig regression affecting module exclusion logic.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45674 is the better standalone representative because it describes a concrete configuration regression with a clear failure mode and broader applicability than the Gemma4-specific report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Different underlying bugs and code paths: 45672 is about uint8 weights in Gemma4 audio modules calling torch.finfo during NF4 inference, while 45674 is about llm_int8_skip_modules clearing the default lm_head exclusion and causing an AssertionError." + } + ] + }, + "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": 16, + "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": "0c2e022a1fea4025c9bb75d69775e0d7f57885f386ef5912361f027f95452f26", + "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": "Both items are empty, likely unrelated personal-name/date posts with no bug or feature context. They do not look like duplicates of the same underlying issue.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 45762 is the earlier of the two and can serve as the representative only by chronology, not content; neither issue has substantive triage value.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45762 is marginally better as the cluster representative because it was created first, but there is no meaningful canonical issue here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Only superficial name/date similarity; no shared problem description, reproduction, or technical context indicating 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": 16, + "component_count": 16, + "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": "87a1c596742e6e92f12ffc701b249d1bfd84c611b5f33b2579a6b4d541931d87", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two closed issues look like unrelated placeholder/person-name entries with dates in the titles and no body or supporting evidence. There is no indication they describe the same underlying bug or change.", + "confidence": 0.98, + "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": "Different personal-name/date titles with no body, links, or shared context; they do not appear to describe the same issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 26, + "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-39895-12", + "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" + ], + "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:44827", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:45655", + "pull_request:43823|pull_request:44827", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "a4f45caad386ab705895e7d478f00493fe983edcbe54d3973209f4cef6a52dd8", + "component_budget": { + "node_count": 12, + "item_count": 12, + "soft_pair_count": 18, + "serialized_chars": 32870, + "estimated_input_tokens": 8218, + "estimated_eval_tokens": 16692 + }, + "budget": { + "node_count": 12, + "item_count": 12, + "soft_pair_count": 18, + "serialized_chars": 32870, + "estimated_input_tokens": 8218, + "estimated_eval_tokens": 16692 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Only one soft-duplicate pair is a true match: the two `fixed_cross_entropy` PRs. The rest are same-subsystem or same-file-neighborhood overlaps, but they target different concrete changes and should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43251 is the best canonical representative: it is explicitly tied to issue 43240 and contains the same concrete `fixed_cross_entropy` change as the near-identical follow-up PR.", + "best_issue_reason": null, + "best_pr_reason": "PR 43251 is the strongest cluster anchor because it is issue-linked, directly fixes the underlying loss-kwargs bug, and has the fuller implementation context.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "Different features: VideoPrism model/docs addition vs an ExecuTorch dynamic-shape test. No shared code-path bug." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Tokenizer utility methods for MistralCommon vs adding a new MobileLLM model. Unrelated changes." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Tokenizer method additions vs export dynamic-shape test; same broad infra area only superficially." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor auto-loading robustness vs adding MobileLLM. Different underlying change." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Loss helper kwargs fix vs new model implementation. No shared bug or code-path." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Cross-entropy helper fix vs Mistral4 test/model cleanup. Different subsystems and problems." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "ExecuTorch export test vs Mistral4 test fixes. Both are tests, but not the same change." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Dynamic-shape export test vs MobileLLM model addition. Unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "New model scaffold vs class config-resolution ordering fix. Different code paths." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "New MobileLLM model vs Falcon output-interface refactor. Not the same bug/change." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon output-collection refactor vs model-creation skill draft. Completely different purpose." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon refactor vs Mistral4 test fix; no concrete shared fix." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test/model changes vs `layrnorm` renaming in CLIP-like models. Different underlying defect." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 tests/model work vs config-resolution ordering fix. No same code-path overlap." + }, + { + "left": "pull_request:43823", + "right": "pull_request:44827", + "accept": false, + "reason": "New MobileLLM model vs Mistral4 test fixes. Shared auto/model infra is too broad to merge." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 fix PR vs draft model-creation skill. Not comparable as a duplicate change." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM model addition vs a tooling/skill PR. Different artifact types and goals." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4-specific PR vs generic config-resolution ordering fix. Different concrete issues." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both PRs change `fixed_cross_entropy` the same way to pass supported kwargs through to `nn.functional.cross_entropy`; they are the same underlying fix." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster judgment is conservative and grounded in the packet. The only clear duplicate pair is PR 43251/43254: both modify `src/transformers/loss/loss_utils.py` in the same `fixed_cross_entropy` direction, and both are tied to issue 43240. The other pairs are reasonably rejected as different features or different concrete changes, with no overclaim that they are mergeable into a single bugfix PR. One minor precision note: PR 43254 is also issue-linked, so 43251 is better justified as canonical mainly by having fuller context, not by being uniquely issue-linked." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 26, + "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": "59b603c896446edebf869252aa97513be67f9f0441447324a5af3b4807e07bf4", + "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 are VibeVoice-related, but they target different deliverables: one is the base VibeVoice integration, the other is a draft realtime extension with an acoustic tokenizer. The overlap is broad, but the underlying change is not the same.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #40546 is the better canonical PR because it is the more complete, non-draft base implementation of VibeVoice and has substantial review activity. It represents the core integration rather than a derivative realtime extension.", + "best_issue_reason": null, + "best_pr_reason": "PR #40546 is the best overall representative because it is the main feature landing path, while #42781 is a draft add-on focused on realtime support and an extra tokenizer component.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Shared VibeVoice files are not enough to treat these as duplicates. #40546 is the base VibeVoice implementation, while #42781 is a draft for VibeVoice Realtime with additional acoustic-tokenizer and generation changes. They address different feature scopes and would not plausibly be merged into one PR as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 26, + "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": "16c609975c1dd9165250ffca7dcc4646c77a05306e3ebf8dceb21901cd66523f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4676, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4676, + "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 PRs are unrelated feature additions for different model families: TimesFM covariates forecasting vs. DINOv3 image-classification support. They do not appear to fix the same underlying bug or implement the same change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the PRs target different code paths, different models, and different user-facing features.", + "best_issue_reason": null, + "best_pr_reason": "No single best PR for deduping this cluster; each is a distinct feature request with no meaningful overlap.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different subsystems and model families (TimesFM forecasting with covariates vs. DINOv3 image classification); they are not the same change and would not be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 26, + "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": 26, + "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": "e2e1599f247a2a47c0e49535575ac6ea7da26485a3675ab257915d685be72daf", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10427, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10427, + "estimated_input_tokens": 2607, + "estimated_eval_tokens": 5470 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three unrelated new-model onboarding PRs (DeepSeek OCR, Isaac, and Kimi2-6). They share common integration boilerplate and registry/docs changes, but they do not fix the same underlying bug or implement the same change, so none should be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the PRs introduce different model families with distinct codepaths and files, so there is no single duplicate target.", + "best_issue_reason": null, + "best_pr_reason": "No best PR for duplicate triage: each PR is a separate model addition, not a mergeable duplicate of the others. If forced to pick a representative, none is clearly canonical over the rest.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41797", + "right": "pull_request:45186", + "accept": false, + "reason": "Different model additions (DeepSeek OCR vs Isaac); shared auto-registry/docs plumbing is boilerplate, not the same concrete change." + }, + { + "left": "pull_request:45186", + "right": "pull_request:45630", + "accept": false, + "reason": "Different model additions (Isaac vs Kimi2-6); overlap is only broad model-integration infrastructure, 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": 6, + "component_count": 26, + "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": "df2697848dc835f3c7d8d56f35d77438b90ec1c5b8040eb50d02afb96c4a62e3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5935, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are broad GitHub Actions dependency updates, but they address different action families and different maintenance goals, so they are not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the better canonical representative because it is the more focused, clearly scoped workflow-actions compatibility update; PR 42976 is a separate upgrade pass for different action versions.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the best overall representative of the cluster since it covers the more central Node 24 compatibility theme across many workflows.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Do not treat as duplicates: 42975 upgrades core GitHub Actions for Node 24 compatibility (checkout/cache/upload-artifact), while 42976 bumps docker/build-push-action and related workflow actions to newer releases. They overlap only at the broad 'workflow maintenance' level, not the same underlying change or code-path problem." + } + ] + }, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-4", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:45490" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43451|pull_request:45490" + ], + "prepared_review_unit_hash": "0cf06da84e6ce17eb5573f3679a6b75c21e123d895510d8f9c43efca4e4d54f8", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14079, + "estimated_input_tokens": 3520, + "estimated_eval_tokens": 7296 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14079, + "estimated_input_tokens": 3520, + "estimated_eval_tokens": 7296 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All four items are separate model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM). They share a common template/scaffold pattern, but they target different models, code paths, and docs, so they are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model families (ViT NEPA vs Molmo) with different files and implementations; shared docs/auto-registration scaffolding is not enough to treat them as the same change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "ViT NEPA and Molmo2 are unrelated model additions; they do not fix the same code-path problem and could not be merged into one PR." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "ViT NEPA and CTSM are distinct model integrations with different modeling and conversion code, so this is not a duplicate." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are separate model variants with different implementation files and capabilities; similar naming and shared auto-mapping/docs are not sufficient to merge as duplicates." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are unrelated model additions; overlapping registry/docs edits are incidental and do not indicate the same underlying bug or change." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM are different model PRs with different code paths and test suites; they are not the same underlying fix or feature." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 26, + "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-43291-5", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815", + "pull_request:45747", + "pull_request:45749" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:45747", + "pull_request:43665|pull_request:45749", + "pull_request:44815|pull_request:45747" + ], + "prepared_review_unit_hash": "a08203b030e6cfd4ba626e78e6254c1a5e22fb2864ddafecfa5c40dde8b67bbd", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 6, + "serialized_chars": 10367, + "estimated_input_tokens": 2592, + "estimated_eval_tokens": 5440 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 6, + "serialized_chars": 10367, + "estimated_input_tokens": 2592, + "estimated_eval_tokens": 5440 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are mostly unrelated: one substantive FP8/dequant code fix, one Whisper test/tokenizer fix, one test-only placeholder patch, one vague batch-size fix with no diff shown, and one docstring typo. No issue is present; the strongest PR is the FP8/dequant fix.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44815 is the clearest substantive bug fix, with real code-path changes in model loading and FP8 quantization integration; the other PRs are test-only, vague, or documentation-only.", + "best_issue_reason": null, + "best_pr_reason": "PR 44815 is the best standalone change to keep because it addresses an actual runtime/dequantization path rather than test scaffolding or wording changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Whisper tokenizer/test fix vs. CLIP/ViT test placeholders; different models and different kinds of changes." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper decoding/tests are unrelated to FP8 model-loading/dequantization changes." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Test-only assertions in vision tests are not the same underlying bug as the FP8/dequant code fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:45747", + "accept": false, + "reason": "Test stubs in CLIP/ViT are unrelated to an unspecified split-batch-size fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:45749", + "accept": false, + "reason": "Vision test placeholders do not match a continuous_batching docstring spelling correction." + }, + { + "left": "pull_request:44815", + "right": "pull_request:45747", + "accept": false, + "reason": "FP8/dequant logic and split-batch-size behavior are different code paths with no shared concrete bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 26, + "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": "fce4bd253549e3d2542bcb04d5b82834f4a4aaac7cda7fdbc14b9410fbc18502", + "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": "These two PRs are both test/bot-oriented, but they change different areas and do not appear to fix the same concrete problem. The shared Vit test file looks incidental rather than evidence of duplication.", + "confidence": 0.79, + "canonical_issue_reason": null, + "canonical_pr_reason": "No clear duplicate canonical PR here; the two PRs target different bot/test concerns. If one must be chosen as the cluster representative, PR #43532 is the better fit because it contains the more substantive workflow/bot behavior changes, while PR #43488 is mainly a narrow repo-bot formatting check.", + "best_issue_reason": null, + "best_pr_reason": "PR #43532 is the stronger representative for the cluster because it touches the CI/comment/diff-display machinery directly and has broader behavioral impact. PR #43488 is a small synthetic setup/version tweak and is less representative of the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Reject: the diffs are unrelated in substance. #43488 changes setup/versioning and a vit test as a repo-bot formatting check, while #43532 changes GitHub workflows, testing utilities, and notification code for diff/comment behavior. Shared test-file overlap is incidental, not the same underlying 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": 10, + "component_count": 26, + "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": "a052fb4292f19ede9cfa7770fa2917f918e9733c6c0719e15d47db3c9f487336", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6083, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6083, + "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 distinct feature PRs: one adds Promptable Concept Segmentation, the other Promptable Visual Segmentation. They share scaffolding and docs patterns, but not the same underlying pipeline or code-path change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the PRs implement different segmentation pipelines with different model/test coverage and should not be collapsed into one duplicate target.", + "best_issue_reason": null, + "best_pr_reason": "No single best PR for this cluster; both are valid but separate feature additions, so neither should represent the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different features and implementations: concept segmentation vs visual segmentation, with different pipeline files, model integrations, and tests. Shared docs/registry scaffolding is not enough to treat them as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 26, + "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": "48e6916cb0f2bd039725ebc5d1d0fd2633f12073bbf52e64417ba298a7c82304", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4292, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4292, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch `src/transformers/trainer.py`, but they implement different Trainer extension hooks for different code paths: one for gradient-accumulation loss scaling, the other for signature/column filtering. They are related by subsystem, not the same change or bug.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different underlying Trainer behaviors: `_loss_is_scaled_for_ga` changes loss scaling during training, while `_get_signature_columns` changes input column filtering based on model signature. Shared file alone is not enough to treat them 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": 12, + "component_count": 26, + "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": "b377c268cc62d0fe86c8200158b61474990b70a20461a7f419286242d5a97559", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5991, + "estimated_input_tokens": 1498, + "estimated_eval_tokens": 3252 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 5991, + "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 point at the same runtime-annotation problem in `src/transformers/cli/serve.py`, but `43842` is the most direct and minimal fix. The other two appear to be broader annotation-cleanup variants of the same bug, not separate changes.", + "confidence": 0.8, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43842 is the clearest representative: it directly fixes the `TypeAdapter` NameError by moving the import under `TYPE_CHECKING` and stringifying the annotation, with a focused diff tied to the reported CLI crash.", + "best_issue_reason": null, + "best_pr_reason": "43842 is the best overall PR to keep because it is the most targeted fix for the concrete `TypeAdapter` NameError and is less entangled with broader typing refactors than the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": true, + "reason": "Both PRs fix the same `serve.py` annotation-evaluation crash in the CLI; `43656` is a broader typing-postponement variant of the same runtime NameError fix." + }, + { + "left": "pull_request:43836", + "right": "pull_request:43842", + "accept": true, + "reason": "Both target the same `TypeAdapter`/annotation NameError in `src/transformers/cli/serve.py` and could plausibly be combined into one patch for the same code path." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary and PR selection are grounded in the packet: 43842 is indeed the most targeted fix, and the pairwise merges are plausibly the same serve.py annotation/NameError bug family and look mergeable. The wording is slightly broad for 43656, but not enough to reject." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 26, + "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": 26, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44024", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44018|pull_request:44068", + "pull_request:44066|pull_request:44071", + "pull_request:44066|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44019|pull_request:44722", + "pull_request:44086|pull_request:44722", + "pull_request:44019|pull_request:44085", + "pull_request:44018|pull_request:44019", + "pull_request:44019|pull_request:44071", + "pull_request:44068|pull_request:44116", + "pull_request:44068|pull_request:44722", + "pull_request:44071|pull_request:44722", + "pull_request:44018|pull_request:44116", + "pull_request:44018|pull_request:44086", + "pull_request:44018|pull_request:44085", + "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44059|pull_request:44086", + "pull_request:44085|pull_request:44086", + "pull_request:44059|pull_request:44722", + "pull_request:44098|pull_request:44101", + "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:44059|pull_request:44068", + "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", + "pull_request:44073|pull_request:44161", + "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:43997|pull_request:44018", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44010|pull_request:44073", + "pull_request:44018|pull_request:44129", + "pull_request:43997|pull_request:44074", + "pull_request:43997|pull_request:44129", + "pull_request:44010|pull_request:44018", + "pull_request:44071|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44074", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154", + "pull_request:44024|pull_request:44076" + ], + "prepared_review_unit_hash": "d77b78cb77b3f77a64dce8d49d1a8d4bf0d55fe7821c0996f56eeab5ed8aa1da", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84519, + "estimated_input_tokens": 21130, + "estimated_eval_tokens": 42516 + }, + "budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 72, + "serialized_chars": 76953, + "estimated_input_tokens": 19239, + "estimated_eval_tokens": 38734 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a broad wave of model-by-model output-tracing refactors around the same umbrella tracking issue, but most PRs are not duplicates of each other because they target different model implementations. The only clear duplicate subclusters are the GPT-J/CodeGen refactors and the GPT-Neo pair.", + "confidence": 0.83, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44722 is the strongest representative of the overlapping GPT-J/CodeGen refactor set: it explicitly targets the tracking issue, touches the same GPT-J code path as the narrower GPT-J-only PRs, and is the most complete version of that concrete change.", + "best_issue_reason": null, + "best_pr_reason": "44722 is the best single PR anchor because it most fully captures the duplicate GPT-J refactor family and subsumes the narrower GPT-J-only variants.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families and different code paths; sharing the umbrella output-tracing issue is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "CVT/FNet vs DeBERTa-v2 are separate implementations of the same refactor theme, not the same change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both are GPT-J output-tracing refactors on the same file; 44066 is just a broader version that also includes CodeGen." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "RegNet/ResNet-style CV refactor vs EfficientNet; same umbrella topic, different underlying change." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "EfficientNet output tracing and GPT-J/CodeGen output tracing are unrelated code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen vs EfficientNet; only the general refactor theme overlaps." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTa-v2 are separate model implementations, not duplicate fixes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J/CodeGen refactor vs MGP-STR refactor; same pattern, different model code." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Same GPT-Neo output-tracing refactor on the same file/test area; this looks like a rebased or revised duplicate." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen and MPT are unrelated concrete implementations." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen and GPT-Neo are different model changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo vs GPT-J/CodeGen; same umbrella issue only." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet and GPT-J/CodeGen are different model-specific refactors." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are unrelated code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are distinct model implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are separate refactors, not the same bug fix." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different model code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are unrelated model changes." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are different concrete implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are not the same underlying fix." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are separate model refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated model implementations." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 and MGP-STR are different models; shared output-tracing theme is too broad." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J and MGP-STR do not fix the same concrete code path." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-2 and GPT-J/CodeGen are separate model refactors." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert are different model implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated concrete changes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are distinct model code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44722", + "accept": false, + "reason": "SegFormer and GPT-J/CodeGen are different underlying changes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "SegFormer and VisualBert are separate refactors." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are not the same model bug or change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "RegNet and VisualBert are unrelated concrete implementations." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBert and Flaubert are different model refactors." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are separate model code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-2 and GPT-Neo are different model-specific changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert are unrelated concrete changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are different model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet and SegFormer are separate CV model refactors, not duplicates." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are distinct code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are different model families." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert are unrelated model changes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are separate implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBERT and MPT are unrelated model code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBERT and VisualBert are different models." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo and SpeechT5 are not the same underlying change." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are unrelated refactors." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "CVT/FNet and LongT5 are completely different model code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44074", + "accept": false, + "reason": "CVT/FNet and TextNet are separate model refactors." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBERT and DeBERTa-v2 are different model implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa-v2 and TextNet are unrelated changes." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are different concrete changes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J refactor family; 44722 is the broader successor that also includes CodeGen." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both touch the same GPT-J output-tracing code path; 44722 subsumes the narrower GPT-J-only PR." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are unrelated model changes." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and RegNet/ResNet are different model-specific output-tracing refactors." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 do not share the same underlying change." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa-v2 and SpeechT5 are separate model refactors." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa-v2 and ViTS are different code paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is conservative and grounded in the packet. The accepted soft-edge pairs are plausible duplicates/revisions of the same model-specific output-tracing refactors, and the rejected pairs correctly stay within separate model code paths rather than overclaiming duplication. I don\u2019t see evidence of an unsupported mergeability claim." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 26, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44073|pull_request:44074", + "pull_request:44056|pull_request:44073", + "pull_request:44056|pull_request:44074", + "pull_request:43997|pull_request:44010", + "pull_request:44010|pull_request:44129" + ], + "prepared_review_unit_hash": "f901b03a01f140e3364665fc7e8eca313849048875bd8d84cf87e919f494a772", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84519, + "estimated_input_tokens": 21130, + "estimated_eval_tokens": 42516 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 5, + "serialized_chars": 13066, + "estimated_input_tokens": 3267, + "estimated_eval_tokens": 6790 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are model-specific PRs all refactoring output tracing/capture to the new standardized decorator pattern, but they target different architectures and code paths, so they are not duplicates of one another.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44129 is the strongest representative: it has an explicit issue target and a concrete SpeechT5 output-tracing refactor, making it the clearest anchor for this thematic cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 44129 is the best overall representative because it is the most concretely scoped and the only one tied to a tracking issue; however, it is still model-specific rather than a duplicate of the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Both are output-tracing refactors, but one is VisualBert and the other is TextNet; they touch different model implementations and are not the same underlying fix." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "Same broad refactor theme, but MPNet and VisualBert are distinct code paths with model-specific changes, so they should not be merged as duplicates." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "These are separate MPNet and TextNet refactors of the standardized output-capture pattern, not one concrete bug or change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated model implementations; the similar decorator migration is only thematic, not a duplicate fix." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "Both modernize output tracing, but SqueezeBert and SpeechT5 modify different model-specific forward paths and are not plausibly one combined 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": 15, + "component_count": 26, + "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": "229b8e7c3e703507881ab74464325b48b5245d26c4b31ce334e72385559bd30e", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40174, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40174, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are all part of a broad output-tracing refactor initiative, but they target different model families and code paths. They are thematically similar, not duplicates of the same fix/change.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44025 is a good cluster representative because it makes the output-tracing transition explicit across two closely related model files and shows the common can_return_tuple pattern seen throughout the set.", + "best_issue_reason": null, + "best_pr_reason": "PR 44025 is the best single representative of the cluster\u2019s recurring refactor pattern, with a concrete multi-file implementation and clear output-tracing scope.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Same broad refactor theme, but different model families and separate code paths; not one mergeable change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch unrelated model implementations and would not be merged as one fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Shared refactor pattern only; timm_backbone and univnet are distinct code paths." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families and files; same initiative, not the same underlying bug/change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Both refactor model outputs, but the code paths are unrelated (timm_backbone vs rwkv)." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "Same output-tracing theme, but mobilenet_v1 and univnet are separate implementations." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Only a broad similarity in refactoring style; no shared concrete bug or code path." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families and target files; not the same change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "These are parallel refactors in different models, not a single fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "Same general tracing refactor, but speech_encoder_decoder and mobilenet_v1 are unrelated." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Same refactor pattern, but vision_text_dual_encoder and univnet do not share the same code path." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model implementations; similar output-tracing cleanup only." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "These are separate model refactors with no shared concrete bug to consolidate." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different subsystems and files; the overlap is only in the tracing refactor pattern." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "Similar output API cleanup, but univnet and upernet are distinct code paths." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Not the same underlying bug/change; only a thematic match." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "Separate model families; would not be merged into one PR." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "Both are refactors, but they affect unrelated models and code paths." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "No shared concrete change beyond the generic tracing refactor effort." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Different model targets and implementation details; not duplicates." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Both are output-related PRs, but codegen and mamba are separate concrete changes." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "Shared refactor style only; timm_backbone and codegen are unrelated code paths." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "Different model families and separate forward-path changes." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "No single underlying bug/change; just the same refactor theme." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder and codegen are unrelated implementations." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Both are tracing refactors, but depth_anything and vision_encoder_decoder are different model stacks." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Different model families and files; only broadly similar cleanup." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Same initiative, but superpoint and depth_anything are distinct code paths." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "These are model-specific refactors, not one shared fix." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Different targets and implementations; not mergeable as one PR." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "Output-tracing changes in separate model families; not duplicates." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "Similar refactor pattern only, with no shared concrete bug." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "Distinct model code paths; broad theme overlap is insufficient." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "No common underlying change beyond the tracing refactor series." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "Different model stacks and separate implementation changes." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "Same high-level initiative, but not the same code path or fix." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "They are parallel refactors in unrelated models." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "Different models and forward paths; not one duplicate change." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "Only the refactor theme matches; the concrete code changes do not." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "Separate model implementations and output-handling changes; not 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": 16, + "component_count": 26, + "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": "0ef1716f4f515e04d3fc6f43456eb0c9b75546874b8cdb33cb962020bd724bab", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10914, + "estimated_input_tokens": 2729, + "estimated_eval_tokens": 5714 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 10914, + "estimated_input_tokens": 2729, + "estimated_eval_tokens": 5714 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-addition pull requests (PenguinVL, DeepSeek-OCR-2, and Voxtral TTS). They share some boilerplate model-registration files, but they do not appear to fix the same underlying bug or implement the same concrete change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: each PR introduces a different model family with distinct code, configs, tests, and docs, so they are not duplicates.", + "best_issue_reason": null, + "best_pr_reason": "No single PR is a good global representative for duplicate triage here because the PRs are unrelated model implementations rather than alternate fixes for the same change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Different models and different implementation targets: PenguinVL vs Voxtral TTS. Shared auto-registration/docs files are expected for independent model additions, not evidence of the same change." + }, + { + "left": "pull_request:44662", + "right": "pull_request:45075", + "accept": false, + "reason": "Both are separate model implementations (PenguinVL vs DeepSeek-OCR-2) with distinct configs, model code, and tests; they are not the same code-path fix." + }, + { + "left": "pull_request:45075", + "right": "pull_request:45401", + "accept": false, + "reason": "DeepSeek-OCR-2 and Voxtral TTS are unrelated model additions. Overlap in transformer registry files is boilerplate, not a shared underlying bug or mergeable joint 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": 26, + "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": "6203f6a537273ab9502635ac0724c819ef76bc63e87dc5823f6aff76acb99cf2", + "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": "These two PRs are related by the broad theme of fusion, but they target different layers and should not be deduplicated: one adds a generic module-fusion API, the other integrates KernelConfig-driven n-to-1 kernel fusion into model loading.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "Not a duplicate pair; 44979 is the core module_fusion API addition, while 45363 is a separate KernelConfig/integration feature built on top of fusion concepts.", + "best_issue_reason": null, + "best_pr_reason": "45363 is the more complete end-user feature PR, with a concrete loading/integration path and end-to-end example; 44979 is foundational but narrower.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Reject: both are fusion-related, but 44979 introduces a generic module-fusion API, whereas 45363 implements a separate KernelConfig-based n-to-1 kernel fusion integration. They do not appear to fix the same concrete code-path problem and would 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": 18, + "component_count": 26, + "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": "d66176afe9feeb6175f0d4c6b2e86bba9812f4dc84103dde03ec0098c12888ee", + "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 patch the same Mistral regex pre-tokenizer AttributeError in the same file, and 45317 is the more complete version with test coverage.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:45317 is the better canonical PR: it applies the same code fix as 45086 and also adds a test, making it the more complete representative.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:45317 is the strongest PR candidate overall because it addresses the same concrete bug with additional regression coverage.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying AttributeError fix in _patch_mistral_regex, same target file, and the diffs overlap materially; 45317 just adds 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": 26, + "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": "6bb8333098594d16bb03bb518604c583eb79aa2799f9e2cbc23cd59563f4c984", + "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": "Two NemotronH-related PRs overlap in a few core files, but they address different levels of change: one is a narrow refactor/conversion-mapping update, the other is a broader architectural split into NemotronHDense/NemotronHSparse with docs and tests. They should not be deduplicated.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45569 is the broader, more complete change: it introduces the dense/sparse model split, updates auto-mappings/docs, and adds tests, so it better represents the overall NemotronH architecture work.", + "best_issue_reason": null, + "best_pr_reason": "PR #45569 is the best representative because it covers the finalized end-state of the feature, while #45115 is a narrower preparatory refactor focused on conversion/key renaming.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "They overlap on NemotronH internals, but #45115 is a targeted refactor/conversion-mapping change, while #45569 is a broader model-split and documentation/test update. Same subsystem, but not the same concrete code-path problem." + } + ] + }, + "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": 26, + "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": "3f57fcf67c74a7c5c860151ab4d6dc4a8e3abfb0c3c0e451800a02a8f817963a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5398, + "estimated_input_tokens": 1350, + "estimated_eval_tokens": 2956 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two related but distinct PRs for heterogeneous-model work: one adds per-layer config support, the other adds broader heterogeneous model/runtime support. They share foundation files, but they are not the same concrete change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR #45332 is the broader, more complete feature implementation and best represents the cluster\u2019s end state, but it is not a duplicate of #45333.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Reject: #45333 is config-layer support only, while #45332 extends heterogeneous behavior into cache/masking/modeling paths. They are complementary feature pieces, not the same fix and not mergeable as a single 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": 21, + "component_count": 26, + "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": "1e54711d0f4a1398b817573d6bbe2b36041f6047591588775cfde631d5d09977", + "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": "These are related typing-focused pull requests, but they are not the same underlying change: one is a broad repo-wide type-checking sweep, the other adds specific typing protocols and helpers around modeling_utils/peft integration.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45425 is the more focused, coherent typing change; PR 45415 is a broader type-checking cleanup spanning many files and config. They should not be collapsed as duplicates.", + "best_issue_reason": null, + "best_pr_reason": "PR 45425 is the better representative of the cluster because it has a narrower, concrete code-path goal, while PR 45415 is a sweeping type-checking pass with many unrelated edits.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Both touch typing-related files, but 45415 is a broad type-checking enablement/cleanup across many modules, while 45425 adds specific typing interfaces and adjustments in modeling_utils/peft. They are not the same concrete fix and would not plausibly be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 26, + "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-45487-2", + "nodes": [ + "pull_request:45487", + "pull_request:45615" + ], + "soft_pairs": [ + "pull_request:45487|pull_request:45615" + ], + "prepared_review_unit_hash": "818f0eea79fc7a4216c0c167f3f6ff657c251d47199b6bc17680f42d195f4fca", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5159, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5159, + "estimated_input_tokens": 1290, + "estimated_eval_tokens": 2836 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two pull requests are unrelated: one fixes a real model-parallel device placement bug in AltCLIP/ChineseCLIP code, while the other only updates Qianfan OCR XPU test expectations.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45487 is the stronger canonical PR because it changes the actual model code to move buffered token type IDs onto the correct device, fixing a concrete runtime/device-placement bug across multiple model implementations.", + "best_issue_reason": null, + "best_pr_reason": "PR #45487 is the best PR in the cluster since it addresses a substantive code-path defect; PR #45615 is test-data-only and does not appear to fix the underlying implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45487", + "right": "pull_request:45615", + "accept": false, + "reason": "Reject: they target different models and problems. #45487 fixes AltCLIP/ChineseCLIP model-parallel device handling in core code; #45615 only adds XPU test expectations for qianfan_ocr." + } + ] + }, + "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": 26, + "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": "a35e6384f6aafadf2ec48b4de3e8e5353eab46614d90ae5741dffb7039228700", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs concern expert-parallel loading behavior, but they target different failure modes and code paths: one is about EP + DeepSpeed ZeRO-3/accelerate launch config handling, the other about expert tensors being overwritten during FSDP2-style distributed loading.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45662 is the stronger representative because it names a concrete bug ('experts silently overwritten by rank-0 broadcast') and the diff shows a focused fix in the loading path, including distributed post-shard wrapping and EP-related integration changes.", + "best_issue_reason": null, + "best_pr_reason": "PR 45662 is the best overall candidate in this cluster: it addresses a specific, concrete expert-parallel corruption issue with a targeted code change, whereas 45548 is a separate loading/configuration fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Do not merge: they share the broad EP/loading subsystem, but the bugs differ. 45548 fixes EP + ZeRO-3 loading via accelerate launch and config serialization; 45662 fixes expert tensors being overwritten during distributed/FSDP2 loading. 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": "pull_request", + "component_index": 24, + "component_count": 26, + "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-45639-3", + "nodes": [ + "pull_request:45639", + "pull_request:45645", + "pull_request:45778" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645", + "pull_request:45645|pull_request:45778", + "pull_request:45639|pull_request:45778" + ], + "prepared_review_unit_hash": "2b893865bf30eb52deb5c59e15534f5a27f8b6a58ed4029ccf492982476d6e80", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7469, + "estimated_input_tokens": 1868, + "estimated_eval_tokens": 3992 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "PR 45645 is still the strongest representative of the cluster because it covers the xdist-safe captured_info fix and the CI-side handling for multiple captured_info files. PR 45639 and 45778 look like related but not clearly mergeable variants of the same bugfix, so the 45639\u219445778 edge should be rejected.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the best canonical PR because it combines the core xdist-safe captured_info fix with the workflow change needed to consume per-worker artifacts, making it the most complete end-to-end implementation.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the best representative for the cluster because it addresses the same testing_utils xdist collision and also preserves CI debug log collection across multiple captured_info files.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both change the same testing_utils code path to avoid captured_info collisions under pytest-xdist; 45645 is a broader version that adds the CI artifact handling, but it is still the same underlying fix." + }, + { + "left": "pull_request:45645", + "right": "pull_request:45778", + "accept": true, + "reason": "45778 is a narrower source-only variant of the same worker-specific captured_info fix that 45645 implements, with the same concrete bug and code path." + }, + { + "left": "pull_request:45639", + "right": "pull_request:45778", + "accept": false, + "reason": "They share the same issue target and general intent, but the diffs suggest different implementations of the xdist fix rather than a clearly mergeable single change; that is not strong enough to treat them as duplicates." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The cluster summary is mostly grounded, but one soft-edge verdict is too strong: 45639 vs 45778 should not be accepted as mergeable based only on a shared issue target and similar intent. The packet marks that pair as deterministic_accept=false, and the diffs suggest potentially different implementations of the same bugfix rather than clearly mergeable changes. Please keep that verdict conservative or reject the pair. The other pairwise links to 45645 are better supported." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 25, + "component_count": 26, + "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": 26, + "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 + } + ] +} diff --git a/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/manifest.json b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..33a763ec5fe8791d3c6d1a1b7f32f2826544a12d --- /dev/null +++ b/snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260505t120036z", + "artifacts": { + "hybrid": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-05-05T12:09:10Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260505T120036Z", + "variant": "hybrid" +} diff --git a/snapshots/20260505T120036Z/manifest.json b/snapshots/20260505T120036Z/manifest.json index b32d5786bd350f26235309775a01c1b5ddd8a993..180895de6961d261ba9ab373ce75ebc5d212a92f 100644 --- a/snapshots/20260505T120036Z/manifest.json +++ b/snapshots/20260505T120036Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 1043 }, "extracted_at": "2026-05-05T12:00:36Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260505t120036z", + "runs": { + "hybrid-model-20260505t120036z": { + "analysis_id": "hybrid-model-20260505t120036z", + "artifacts": { + "hybrid": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260505T120036Z/analysis-runs/hybrid-model-20260505t120036z/manifest.json", + "model": null, + "published_at": "2026-05-05T12:09:10Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260505T120036Z", "watermark": {