diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 6fac4a6a6d780e540cc4582c35bec501ec42c378..b7be96a867e6d4449729f9e8cf68326db93bde07 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": "20260603T180028Z", - "generated_at": "2026-06-03T18:43:18Z", + "snapshot_id": "20260604T000026Z", + "generated_at": "2026-06-04T00:08:53Z", "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 4e7cad1621bd890116a92a6380d8ff861b332487..1f2c90c2172b4989be9f7d13e7e5de093368c16b 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": "20260603T180028Z", - "generated_at": "2026-06-03T18:43:18Z", + "snapshot_id": "20260604T000026Z", + "generated_at": "2026-06-04T00:08:53Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -106,15 +106,15 @@ "issue:45874|issue:46294", "issue:39290|issue:45081", "issue:45823|issue:45874", - "issue:34928|issue:41720", "issue:43425|issue:44292", + "issue:34928|issue:41720", "issue:43828|issue:45161", "issue:43541|issue:43716", "issue:43828|issue:44292", "issue:43716|issue:45685", "issue:44918|issue:45161", - "issue:44292|issue:44918", "issue:45507|issue:45685", + "issue:44292|issue:44918", "issue:43716|issue:44292", "issue:43039|issue:44918", "issue:41762|issue:43638", @@ -132,380 +132,392 @@ "issue:41762|issue:43596", "issue:39290|issue:41720", "issue:44589|issue:46304", - "issue:45823|issue:46294", "issue:43901|issue:44512", + "issue:45823|issue:46294", "issue:45237|issue:45901" ], - "prepared_review_unit_hash": "68d47e4436ea24818a6f29af9bf7c9d0ed64ee887781089e9c9e3e4f613d8b9d", + "prepared_review_unit_hash": "61dc05edbad2a00ea7b6d6364739455596e81ef987b902e44894774533998ea1", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 67, - "serialized_chars": 47832, - "estimated_input_tokens": 11958, - "estimated_eval_tokens": 24172 + "serialized_chars": 47823, + "estimated_input_tokens": 11956, + "estimated_eval_tokens": 24168 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous cluster of unrelated bug reports, with one clear duplicate pair around DeepSpeed ZeRO-3 loading failures for BERT models. No PRs were present.", - "confidence": 0.83, - "canonical_issue_reason": "issue:43638 is the cleanest representative of the ZeRO-3/BERT loading failure and has the strongest true-duplicate partner (issue:43596) with the same concrete out-of-bounds index0 error path.", + "summary": "Mostly heterogeneous bug reports across unrelated subsystems. The only clear duplicate subgroup is the ZeRO-3 \"index 0 is out of bounds\" loading failures around BERT/Gemma3.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43596 is the clearest representative of the shared DeepSpeed ZeRO-3 loading failure; issue:43638 is a near-duplicate reproduction, and issue:41762 is close enough to the same empty-tensor/indexing path.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43638 best anchors the only obvious duplicate set in this group and is specific enough to represent the shared bug without overgeneralizing to other ZeRO-3 issues.", + "best_issue_reason": "issue:43596 is the best single anchor for this cluster because it names the concrete failure mode and is the most generic of the near-duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different Qwen3.5 failures: one is a missing config attribute, the other is a streaming chat/completions inference issue." + "reason": "Different failures: Qwen3.5 config attribute error vs streaming chat/completions bug." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Both mention GPU/runtime problems, but they are different code paths: AMD GPU failure vs MPS histogram op unsupported for Int." + "reason": "Unrelated hardware/backend issues: AMD GPU failure vs MPS histogram int limitation." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different models and symptoms: image-preprocessor dtype mismatch vs AMD GPU runtime incompatibility." + "reason": "Different code paths: image preprocessor dtype mismatch vs AMD GPU runtime failure." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Both are dtype/device related broadly, but the concrete failures and model paths differ." + "reason": "Autocast dtype mismatch is not the same bug as the AMD GPU assertion failure." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different issue types: AMD GPU inference failure vs GraniteMoEHybrid calling an invalid method." + "reason": "Different model/backend problem; not the same underlying failure." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Distinct models and errors; NVFP4 runtime failure is not the same as AMD GPU incompatibility." + "reason": "Qwen-3-8B NVFP4 runtime error is distinct from the AMD GPU issue." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Completely different subsystems: DeepSpeed collective mismatch vs ONNX export error." + "reason": "DeepSpeed collective mismatch and ONNX export error are unrelated." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Unrelated model/export bugs; no shared concrete code path." + "reason": "ONNX export failure vs FineGrainedFP8Config failure; different bugs." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ONNX export error and Qwen3 auto device mapping assert are unrelated failures." + "reason": "Export-time ONNX issue and auto device mapping cuda assert are different problems." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "One is a task crash, the other is outdated RAG documentation; not the same bug/change." + "reason": "Table QA crash and outdated RAG examples are not the same issue." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are stale v5 docs, but they cover different removed topics (pipeline tasks vs run command)." + "reason": "Both are stale docs, but they refer to different removed v5 documentation items." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different problems: ModernBERT ONNX export vs GenerationMixin import breakage." + "reason": "Model export error vs GenerationMixin import error; different code paths." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "TRl SFT embedding unpacking failure is unrelated to resume-training with trust_remote_code." + "reason": "TRL SFT embeddings issue is unrelated to resuming training with trust_remote_code." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Mask shape mismatch during indexing is not the same bug as Gemma4 device_map forward mismatch." + "reason": "Mask shape mismatch and Gemma4 PLE device mismatch are different failures." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Different errors and code paths; one is ZeRO-3/BERT initialization, the other is a mask/tensor shape mismatch." + "reason": "Both are IndexErrors, but the reported shapes and loading paths differ enough to avoid merging." }, { "left": "issue:43638", "right": "issue:46165", "accept": false, - "reason": "No substantive overlap: ZeRO-3 BERT init error vs model resume/trust_remote_code issue." + "reason": "ZeRO-3 index error is unrelated to resuming training with a local model." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Both are tensor/indexing failures, but one is specifically ZeRO-3 BERT model init and the other is a mask shape mismatch." + "reason": "Different symptom and code path: ZeRO-3 empty tensor indexing vs mask/tensor shape mismatch." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Different modalities and failure modes: generic mask shape mismatch vs Qwen3.5 video chat-template token/feature mismatch." + "reason": "Mask indexing error and Qwen3.5 video token mismatch are unrelated." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Different model/runtime failures; no shared underlying bug." + "reason": "Different bug classes: tensor mask mismatch vs GPT-OSS tensor-parallel issue." }, { "left": "issue:44560", "right": "issue:46165", "accept": false, - "reason": "Video apply_chat_template StopIteration is unrelated to training resume with local model and remote code." + "reason": "Video StopIteration bug is unrelated to the training resume problem." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "ZeRO-3/BERT init failure and training-resume issue are unrelated." + "reason": "ZeRO-3 Bert loading failure and resume-training issue are unrelated." }, { "left": "issue:43572", "right": "issue:46165", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig is unrelated to resume-training behavior." + "reason": "Missing pad_token_idx in config is unrelated to model-resume behavior." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice preset bug is unrelated to ModernBERT ONNX export." + "reason": "BarkProcessor voice preset bug is unrelated to ONNX export." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both involve distributed training, but one is collective mismatch in Jamba-v01 while the other is BERT ZeRO-3 init/indexing." + "reason": "Collective mismatch and ZeRO-3 index error are different distributed-training bugs." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same underlying ZeRO-3/BERT loading bug with the same out-of-bounds index0 failure signature; these look like true duplicates." + "reason": "Same underlying ZeRO-3 loading/index-out-of-bounds failure; 43638 looks like another reproduction on a Bert variant." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger kernel cross-entropy call and GPT-OSS TP-only issue are different failures." + "reason": "Liger Kernel cross_entropy call is unrelated to GPT-OSS MoE tensor-parallel failure." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Different model-specific tokenization/save-load issue vs LoRA merge/save consistency problem." + "reason": "Detokenization bug and LoRA merge/save inconsistency are different issues." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "NVFP4 runtime failure is unrelated to LoRA merge/save inconsistency." + "reason": "Qwen NVFP4 runtime error is unrelated to LoRA merge/save mismatch." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Activation checkpointing/FSDP recompute mismatch is not the same as a mask shape indexing error." + "reason": "Activation checkpointing/FSDP recompute mismatch is not the same as the mask shape error." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "Different DeepSpeed-related errors: weight dimensionality vs collective mismatch." + "reason": "DeepSpeed weight-dimension runtime error and collective mismatch are distinct failures." }, { "left": "issue:45859", "right": "issue:45902", "accept": false, - "reason": "Same class name, but different concrete bugs: CPU offload incompatibility vs uninitialized memory after meta-to-empty materialization." + "reason": "Same model family, but different concrete bugs: CPU offload incompatibility vs uninitialized inv_freq after materialization." }, { - "left": "issue:43638", - "right": "issue:45161", + "left": "issue:39290", + "right": "issue:45081", "accept": false, - "reason": "ZeRO-3 BERT init bug and GPT-OSS tensor-parallel bug are unrelated." + "reason": "Gemma3 config attribute missing and Mistral regex patch crash are unrelated." }, { - "left": "issue:34928", - "right": "issue:36331", + "left": "issue:45823", + "right": "issue:45874", "accept": false, - "reason": "Activation checkpointing/FSDP issue is unrelated to CustomTrainer compute_loss signature change." + "reason": "Device_map auto PLE device mismatch is not the same as reverse embedding lookup overhead." }, { - "left": "issue:43596", - "right": "issue:45161", + "left": "issue:43425", + "right": "issue:44292", "accept": false, - "reason": "Different models and code paths; not the same bug." + "reason": "Torch version incompatibility and Qwen NVFP4 runtime failure are different problems." }, { - "left": "issue:43716", - "right": "issue:45507", + "left": "issue:43828", + "right": "issue:45685", "accept": false, - "reason": "Dtype mismatch in Mistral-3 preprocessor vs invalid method call in GraniteMoEHybrid are unrelated." + "reason": "Autocast dtype mismatch and MPS int histogram limitation are unrelated." }, { - "left": "issue:45874", - "right": "issue:46294", + "left": "issue:43541", + "right": "issue:43716", "accept": false, - "reason": "Both involve Gemma/Qwen multimodal paths, but one is reverse embedding lookup and the other is video feature/token mismatch." + "reason": "Mixtral grouped_mm tracing error and Mistral-3 dtype mismatch are distinct." }, { - "left": "issue:39290", - "right": "issue:45081", + "left": "issue:43828", + "right": "issue:44292", "accept": false, - "reason": "Gemma3 vLLM config attribute error and Mistral regex patch crash are unrelated." + "reason": "Different failures: autocast dtype mismatch vs Qwen NVFP4 runtime issue." }, { - "left": "issue:45823", - "right": "issue:45874", + "left": "issue:43716", + "right": "issue:45685", "accept": false, - "reason": "Device mismatch in Gemma4 forward is different from the expensive reverse embedding lookup bug." + "reason": "Mistral-3 dtype mismatch is unrelated to the MPS histogram limitation." }, { - "left": "issue:34928", - "right": "issue:41720", + "left": "issue:44918", + "right": "issue:45161", "accept": false, - "reason": "FSDP checkpointing tensor recompute issue is unrelated to Qwen3 auto device mapping CUDA assert." + "reason": "TRL input-embedding unpacking bug is unrelated to GPT-OSS MoE tensor-parallel failure." }, { - "left": "issue:43425", - "right": "issue:44292", + "left": "issue:45507", + "right": "issue:45685", "accept": false, - "reason": "Torch/transformers compatibility issue is unrelated to Qwen-3 NVFP4 runtime errors." + "reason": "GraniteMoEHybrid invalid method call and MPS histogram issue are unrelated." }, { - "left": "issue:43828", - "right": "issue:45161", + "left": "issue:43827", + "right": "issue:44512", "accept": false, - "reason": "Autocast dtype mismatch and GPT-OSS TP-only failure are different bugs." + "reason": "Both are docs regressions, but they cover different removed v5 content and should not be merged." }, { - "left": "issue:43541", - "right": "issue:43716", + "left": "issue:43716", + "right": "issue:44560", "accept": false, - "reason": "Grouped_mm tracing error in Mixtral is unrelated to Mistral-3 image preprocessor dtype mismatch." + "reason": "dtype mismatch in Mistral-3 is unrelated to the Qwen3-vl video StopIteration bug." }, { "left": "issue:43828", - "right": "issue:44292", + "right": "issue:45685", "accept": false, - "reason": "Different model/runtime failures; no shared concrete code path." + "reason": "Autocast dtype mismatch and MPS histogram int issue are different bugs." }, { - "left": "issue:43716", + "left": "issue:44918", "right": "issue:45685", "accept": false, - "reason": "Mistral-3 dtype mismatch is unrelated to MPS histogram support failure." + "reason": "SFT trainer embedding issue is unrelated to the MPS backend error." }, { - "left": "issue:44918", - "right": "issue:45161", + "left": "issue:43039", + "right": "issue:43828", "accept": false, - "reason": "Qwen3.5 SFT unpacking error is unrelated to GPT-OSS TP-only failure." + "reason": "Liger kernel cross_entropy call and autocast dtype mismatch are different issues." }, { - "left": "issue:44292", - "right": "issue:44918", + "left": "issue:30277", + "right": "issue:36331", "accept": false, - "reason": "Different model families and different failure modes." + "reason": "DeepSpeed collective mismatch and compute_loss signature error are unrelated." }, { - "left": "issue:45507", - "right": "issue:45685", + "left": "issue:34928", + "right": "issue:36010", "accept": false, - "reason": "Invalid method call in GraniteMoEHybrid is unrelated to the MPS Int histogram error." + "reason": "Activation checkpointing/FSDP mismatch is unrelated to GenerationMixin import failure." }, { - "left": "issue:43716", - "right": "issue:44292", + "left": "issue:45237", + "right": "issue:46015", "accept": false, - "reason": "Different bug signatures and different model paths." + "reason": "AMD GPU runtime issue and outdated RAG examples are unrelated." }, { "left": "issue:43039", - "right": "issue:44918", + "right": "issue:44805", "accept": false, - "reason": "Liger kernel cross-entropy routing issue is unrelated to Qwen3.5 SFT embedding unpacking." + "reason": "Liger kernel cross_entropy usage bug is unrelated to the mask/tensor shape error." }, { - "left": "issue:41762", - "right": "issue:43638", + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Torch 2.4.1 compatibility issue and USE_DISTRIBUTED=0 breakage are distinct versioning problems." + }, + { + "left": "issue:43296", + "right": "issue:46304", "accept": false, - "reason": "Both are ZeRO-3-related, but Gemma3 loading index error and BERT zero3 init failure are different concrete bugs." + "reason": "PaddleOCR-VL/vLLM load failure and distributed-off breakage are different issues." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": true, + "reason": "Both report the same ZeRO-3 index-out-of-bounds loading failure with size-0 tensors; 41762 is another model-specific reproduction." }, { "left": "issue:43827", "right": "issue:44512", "accept": false, - "reason": "Both are docs regressions, but they cover different removed APIs and commands." + "reason": "Different documentation regressions about different removed v5 features." }, { - "left": "issue:43716", - "right": "issue:44560", + "left": "issue:39290", + "right": "issue:41720", "accept": false, - "reason": "Different multimodal issues: dtype mismatch vs video template StopIteration." + "reason": "Gemma3TextConfig missing attribute and Qwen3 auto-device-map cuda assert are unrelated." }, { "left": "issue:44589", "right": "issue:46304", "accept": false, - "reason": "Float8 storage lookup error is unrelated to USE_DISTRIBUTED=0 compatibility." + "reason": "Float8 storage lookup error and USE_DISTRIBUTED=0 support breakage are different problems." }, { - "left": "issue:45823", - "right": "issue:46294", + "left": "issue:43901", + "right": "issue:44512", "accept": false, - "reason": "Gemma4 device mismatch and Qwen3.5 video token/feature mismatch are not the same bug." + "reason": "TextClassificationPipeline docs and transformers run-command docs are separate stale-doc issues." }, { - "left": "issue:43901", - "right": "issue:44512", + "left": "issue:45823", + "right": "issue:46294", "accept": false, - "reason": "TextClassificationPipeline docs mismatch and removed run-command docs are separate documentation issues." + "reason": "Gemma4 device mismatch and Qwen3.5 video token mismatch are unrelated." }, { "left": "issue:45237", - "right": "issue:46015", + "right": "issue:45901", "accept": false, - "reason": "AMD GPU support bug is unrelated to outdated RAG examples." + "reason": "AMD GPU failure and table-question-answering crash are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: it conservatively identifies one clear duplicate pair (issue:43596 and issue:43638) and describes the rest of the cluster as heterogeneous without overreaching. The soft-edge verdicts are appropriately cautious, and the note that there were no PRs is consistent with the packet." + "feedback": "The cluster summary is grounded in the packet and stays conservative. The ZeRO-3 zero-sized tensor / index-out-of-bounds subgroup is the only clearly coherent duplicate set, and the accepted soft-edge pairs (43596\u219443638, 41762\u219443596) are well supported by the issue titles. The remaining non-duplicate verdicts are generally cautious and do not overstate evidence." }, "evaluator_used": true, "retried": false, @@ -580,8 +592,8 @@ "issue:43975|issue:46015", "issue:45812|issue:45920", "issue:44863|issue:46015", - "issue:43866|issue:44863", "issue:45020|issue:46304", + "issue:43866|issue:44863", "issue:43872|issue:46304", "issue:44292|issue:46015", "issue:39290|issue:43531", @@ -618,14 +630,14 @@ "issue:44479|issue:46294", "issue:41093|issue:43596" ], - "prepared_review_unit_hash": "e8d82e56b221099ee5e70eb7947decd8811f8209cb302544cd0ea0892dae9aa9", + "prepared_review_unit_hash": "6ab904eb1c28d9efa5315d0f7c04bbbcdee754913769e20a07a93bb1d448bbe3", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, @@ -640,273 +652,273 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a mix of unrelated model/regression reports across different subsystems. I only see two likely true duplicates: the broad AutoTokenizer token-ID regression pair and the Qwen3.5 video-input regression pair.", - "confidence": 0.86, - "canonical_issue_reason": "issue:45920 is the broadest canonical report here: it describes the v5 AutoTokenizer token-ID regression across multiple model families, so it best serves as the umbrella issue.", + "summary": "This cluster is a set of several small duplicate subgroups rather than one unified bug. The clearest grouping is the AutoTokenizer regression family, with issue:45920 as the broad representative; the video-input regression pair (issue:44479 / issue:46294) and the init-empty-weights TypeError family also look coherent. Most other soft matches are only superficial, and issue:41093 vs issue:43638 should not be accepted without stronger shared evidence.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45920 is the best canonical issue for the tokenizer-regression subgroup: it covers multiple affected model families, describes the broadest concrete symptom, and has inbound references that make it the most representative report in this set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the strongest single issue to keep as the representative umbrella for the tokenizer-regression subset of this noisy cluster.", + "best_issue_reason": "issue:45920 is the strongest overall issue candidate because it captures a broad, concrete regression affecting several model families and is more representative than the narrower single-model reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Different bugs: Voxtral processor load failure vs Gemma4 PLE device-mismatch during forward." + "reason": "Different model families and different failure modes; both are multimodal, but not the same bug." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Unrelated subsystems: Liger-kernel cross_entropy call vs Qwen3.5 video StopIteration." + "reason": "Unrelated topics: Liger Kernel cross-entropy call vs Qwen video-template failure." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "Model detokenization bug vs documentation/example cleanup; not the same change." + "reason": "Different subsystems: detokenization bug vs outdated RAG docs/examples." }, { "left": "issue:45812", "right": "issue:45920", "accept": true, - "reason": "Same v5 AutoTokenizer token-ID regression reported on different model families; 45920 is the broader umbrella." + "reason": "Same underlying AutoTokenizer wrong-token-IDs regression; issue:45812 is a narrower Granite instance of the broader multi-model issue." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "NemotronH checkpoint loading failure vs outdated RAG examples; unrelated." + "reason": "Checkpoint-loading bug vs documentation issue; no shared code-path." }, { - "left": "issue:43866", - "right": "issue:44863", + "left": "issue:45020", + "right": "issue:46304", "accept": false, - "reason": "Both are model-loading failures, but the reported root causes are different (corrupted Ovis2 checkpoint vs NemotronH loading support)." + "reason": "Both are broad breakages, but they point to different compatibility problems and different code paths." }, { - "left": "issue:45020", - "right": "issue:46304", + "left": "issue:43866", + "right": "issue:44863", "accept": false, - "reason": "Remote-code breakage vs USE_DISTRIBUTED=0/PyTorch support regression; different compatibility problems." + "reason": "Both are load failures, but for different models and likely unrelated causes." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "bitsandbytes Int8Params constructor mismatch vs PyTorch distributed-off compatibility; unrelated." + "reason": "The _is_hf_initialized TypeError is a specific init-empty-weights/bitsandbytes compatibility issue, not the USE_DISTRIBUTED=0 breakage." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "init_empty_weights TypeError vs documentation issue; unrelated." + "reason": "Unrelated: model-loading TypeError vs docs/examples issue." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Both mention sliding_window, but Gemma3/vLLM config attr failure and Qwen3-MoE behavior are different bugs." + "reason": "Both mention sliding_window, but they affect different models and likely different root causes." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "GLM-4.7-Flash loading failure vs Ovis2 checkpoint corruption; not the same underlying issue." + "reason": "Different models and different load failures; not the same bug." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Different zero-3 failures: mask/tensor shape mismatch vs index-out-of-bounds during Bert loading." + "reason": "Both are ZeRO-3/BERT-related indexing failures, but the signatures differ (mask-shape mismatch vs out-of-bounds with size 0) and there is no shared issue text or code evidence here." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper model runtime failure vs table-question-answering task crash; unrelated." + "reason": "Different tasks and failures: Whisper model loading vs table-question-answering crash." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch-version incompatibility vs a pydantic PretrainedConfig regression; unrelated." + "reason": "Tensor/torch version incompatibility vs pydantic PretrainedConfig regression; unrelated." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "Zero-3/Bert index error vs torch 2.4.1 and transformers 5.8 incompatibility; different problems." + "reason": "Different issues: ZeRO-3/BERT index error vs torch/transformers version incompatibility." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Detokenization bug vs table-QA crash; unrelated." + "reason": "Detokenization bug is unrelated to table-question-answering crashes." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper runtime failure vs pydantic model regression; unrelated." + "reason": "Whisper load failure and PretrainedConfig pydantic regression are different problems." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Different Qwen3.5 failures and stacks: chat crash vs streaming chat/completions API failure." + "reason": "Same model family, but one is transformers chat and the other is vLLM streaming API; not enough evidence of one underlying bug." }, { "left": "issue:44292", "right": "issue:45901", "accept": false, - "reason": "init_empty_weights TypeError vs table-QA crash; unrelated." + "reason": "Different failures and code paths; no shared underlying bug." }, { "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Gemma3 sliding_window_pattern/vLLM issue vs Gemma4 PLE device mismatch; different code paths." + "reason": "Unrelated model/config issues; only superficial similarity in being config-related." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Tied-weights bug in Mistral3 vs corrupted Ovis2 checkpoint loading; unrelated." + "reason": "Different model families and symptoms; not a duplicate." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Qwen3.5 streaming API failure vs MPS histogram_mps unsupported-for-Int error; unrelated." + "reason": "Chat/streaming failure vs MPS histogram-on-int error; unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Grouped_mm tracing failure vs MPS histogram error; different subsystems." + "reason": "Different failures in different execution paths: torch dynamo grouped_mm vs MPS histogram." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Torch-dynamo grouped_mm tracing error vs autocast dtype mismatch; not the same bug." + "reason": "Both involve dtype-ish symptoms, but they are different models and distinct code paths." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Voxtral processor load failure vs non-template-node compile error; unrelated." + "reason": "Different subsystems: processor loading vs compile/template-node error." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "GenerationMixin import error vs Gemma3 sliding_window_pattern/vLLM issue; unrelated." + "reason": "ImportError for GenerationMixin is unrelated to the Gemma3 sliding_window config issue." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers version incompatibility vs Gemma4 device mismatch; unrelated." + "reason": "Version incompatibility vs model-device mismatch; not the same bug." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are zero-3-related IndexErrors, but the exact failures and model contexts differ." + "reason": "Both are DeepSpeed-related indexing failures, but the error signatures and affected paths differ." }, { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "PaddleOCR-VL loading failure vs GGUF gpt-oss support request; unrelated." + "reason": "Different model load/use cases: PaddleOCR-VL load failure vs GGUF gpt-oss support request." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig vs pydantic PretrainedConfig regression; unrelated." + "reason": "Missing pad_token_idx in StableLmConfig is separate from the PretrainedConfig pydantic regression." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Int4 quantization OOM regression vs tied-weights translation-model issue; different bugs." + "reason": "Int4 OOM regression and tied-weights issues are different problems." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Mask-shape IndexError vs Qwen3.5 embedding unpacking with trl SFT trainer; unrelated." + "reason": "Similar words around shape errors, but these are different failures in different training/loading flows." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Mixtral grouped_mm tracing failure vs GPT-OSS TP-only issue; different problems." + "reason": "Different model families and unrelated MoE/TP behavior." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Phi MoE autocast dtype mismatch vs Qwen3.5 streaming chat/completions failure; unrelated." + "reason": "Autocast dtype mismatch and chat-streaming API failure are unrelated." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized TypeError vs pydantic PretrainedConfig field regression; unrelated." + "reason": "Unexpected _is_hf_initialized argument is a model-loading compatibility bug, not the pydantic PretrainedConfig regression." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF gpt-oss support vs Qwen3.5 streaming inference failure; unrelated." + "reason": "GGUF gpt-oss support request vs Qwen3.5 streaming failure; unrelated." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding_window issue vs pydantic PretrainedConfig regression; unrelated." + "reason": "Sliding_window model config issue is unrelated to the pydantic PretrainedConfig regression." }, { "left": "issue:43716", "right": "issue:46294", "accept": false, - "reason": "Mistral-3 image preprocessor dtype mismatch vs Qwen3.5 video token mismatch; different model paths." + "reason": "Different multimodal models and different symptoms; not enough evidence of a shared root cause." }, { "left": "issue:45020", "right": "issue:45800", "accept": false, - "reason": "Remote-code breakage vs torch 2.4.1/transformers 5.8 incompatibility; different compatibility issues." + "reason": "Both mention compatibility regressions, but they target different version conflicts and code paths." }, { "left": "issue:43541", "right": "issue:44589", "accept": false, - "reason": "Grouped_mm tracing failure vs Float8 storage lookup error; unrelated." + "reason": "Grouped_mm tracing error and Float8 storage lookup error are different runtime failures." }, { "left": "issue:44479", "right": "issue:46294", "accept": true, - "reason": "Both describe the same Qwen3.5/Qwen VL video-input regression where video features and video tokens mismatch in v5.3.0." + "reason": "Same Qwen video-input regression family; issue:46294 is a specific manifestation of the broader v5.3.0 video token/features mismatch." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Mask-shape mismatch IndexError vs zero-3 Bert out-of-bounds IndexError; different failure modes and fixes." + "reason": "Both are indexing errors, but one is a mask-shape mismatch and the other is a zero-sized tensor path; not clearly the same bug." } ] }, "evaluator_result": { - "accept": true, - "feedback": "Grounded and conservative overall. The two accepted duplicate pairs are supported by the packet, the umbrella choice of issue:45920 is reasonable, and the remaining rejections do not overclaim shared root causes." + "accept": false, + "feedback": "Most of the clustering is conservative and grounded, but one soft-edge verdict overstates the evidence: issue:41093 vs issue:43638 are both ZeRO-3/BERT-related indexing failures, yet the titles describe different signatures (mask-shape mismatch vs out-of-bounds with size 0). With no shared issue text or code evidence in the packet, treating them as the same underlying bug is too strong. The rest of the summary looks reasonable, but this pair should stay tentative or negative." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1007,235 +1019,235 @@ "issue:44910|issue:45859", "issue:42947|issue:44387" ], - "prepared_review_unit_hash": "f8086dc5f2b95424f9298be277b65ba578f44dad2687b46fc22de50d7042ec4b", + "prepared_review_unit_hash": "09765f6647f8495553c7c9393aa3466cf3f174e5ed882ba7712e08da394200be", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37143, - "estimated_input_tokens": 9286, - "estimated_eval_tokens": 18828 + "serialized_chars": 37146, + "estimated_input_tokens": 9287, + "estimated_eval_tokens": 18830 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed Transformers-v5 bug cluster. The only strong duplicate chain is the non-persistent buffer corruption regression; most other pairs are related only at a broad subsystem/model level and should stay separate.", - "confidence": 0.84, - "canonical_issue_reason": "43950 is the clearest umbrella report: it describes the non-persistent-buffer corruption regression directly and has the strongest duplicate-style overlap with nearby reports about junk-filled buffers.", + "summary": "The strongest duplicate cluster is the Transformers v5 non-persistent-buffer regression around 43950/43644/44534. The SigLIP2 pair looks like a plausible but tentative duplicate based on titles alone. Most other candidate pairs are unrelated.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 43950 is the clearest anchor for the non-persistent-buffer regression cluster because its title explicitly names the v5 regression, and it matches the related buffer-corruption reports.", "canonical_pr_reason": null, - "best_issue_reason": "Best representative for triage because it anchors the most convincing duplicate chain and states the underlying failure mode in the most general way.", + "best_issue_reason": "43950 is the best representative issue overall because it most explicitly states the underlying v5 non-persistent-buffer regression and aligns with the related buffer-corruption reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43054", "right": "issue:43994", "accept": true, - "reason": "Both report SigLIP2 producing broken outputs on the same model family; one manifests as poor text embeddings, the other as nonsensical AutoModel/pipeline results." + "reason": "Tentative duplicate: both titles describe SigLIP2 base-patch16-224 producing bad or nonsensical outputs, suggesting the same underlying model issue, though this is title-level evidence only." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Both involve Qwen3.5-related loading behavior, but one is a config-unwrapping issue and the other is uninitialized memory in rotary embeddings; different bugs." + "reason": "Different bugs: composite multimodal config unwrapping versus Qwen3_5MoeTextRotaryEmbedding uninitialized-memory behavior after materialization." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "Different problems: missing TensorFlow backend handling vs a torch/transformers version incompatibility." + "reason": "Unrelated failure modes: TensorBoard/TensorFlow backend detection versus a torch/transformers version incompatibility." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP weight-shape loading failure and Qwen3.5 embedding unpacking with TRL are unrelated code paths." + "reason": "Different concrete code paths and symptoms: FSDP weight-shape error versus TRL input-embedding unpacking failure." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Different symptoms and model paths: image preprocessor dtype mismatch vs Qwen3.5 embedding unpacking failure." + "reason": "Different model families and failures: Mistral-3 dtype mismatch versus Qwen3.5 trainer/unpacking behavior." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "Different models and failures; perplexity regression on GPT-OSS is not the same as a GLM5 model issue." + "reason": "No shared underlying bug is evident from the titles; they concern GPT-OSS perplexity and GLM5 separately." }, { "left": "issue:45632", "right": "issue:45698", "accept": false, - "reason": "Both touch custom-model loading, but one is a local cache-path collision and the other is wrong module restoration after save/load." + "reason": "Both involve loading custom code, but the concrete bugs differ: cache-path collision versus wrong custom module loaded after save/load." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Unrelated: notebook custom-model initialization regression vs flash-attn2 fallback gating." + "reason": "Unrelated: Jupyter custom-model initialization issue versus flash-attn2 fallback gating." }, { "left": "issue:43404", "right": "issue:45005", - "accept": true, - "reason": "Same tied-weights regression family: 43404 is a specific Mistral3 manifestation of the broader v5 tied-weight handling problem described in 45005." + "accept": false, + "reason": "Both mention tied weights, but one is a specific Mistral3 lm_head-tie bug while the other is a broader translation-model v5 report." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8 failure and flash-attn2 fallback gating are different bugs." + "reason": "Different issues: FP8 configuration failure versus flash-attn2 fallback checks." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash is unrelated to tied-weight handling." + "reason": "PagedAttentionCache linear_attention crash is unrelated to tied-weight behavior." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization/offloading behavior is a different bug class from tied weights." + "reason": "Quantization/offloading behavior is not the same bug as tied weights in translation models." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "Checkpoint loading failure in NemotronH is not the same as tied-weight regressions." + "reason": "NemotronH checkpoint loading failure is unrelated to tied-weight issues." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "A DSA indexer discussion and a docs typo are unrelated." + "reason": "Different areas: DSA indexer behavior versus documentation about a removed command." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Different model/component failures; Qwen3.5 embedding unpacking vs GraniteMoEHybrid method dispatch." + "reason": "Different concrete bugs: TRL unpacking failure versus GraniteMoEHybrid invalid-method call." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "Separate documentation/cosmetic issues." + "reason": "A cache-strategy typo is not the same as outdated run-command documentation." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Different problems: detokenization regression vs an NVFP4 runtime error." + "reason": "Detokenization regression and Qwen-3-8B-NVFP4 runtime error are unrelated." }, { "left": "issue:44292", "right": "issue:45507", "accept": false, - "reason": "Different model/runtime failures with no clear shared underlying bug." + "reason": "Different models and failure modes; no shared code-path bug is evident." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Near-identical non-persistent-buffer regression reports; both say v5 fills or corrupts register_buffer(..., persistent=False) values with junk." + "reason": "Same Transformers 5.x non-persistent-buffer corruption regression, described from slightly different angles." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Both are SigLIP2-related, but one is a docs/example mismatch and quantization failure while the other is a model-quality regression." + "reason": "Docs-example errors and model-quality regression are not the same underlying bug." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Same regression: non-persistent buffers are being populated with junk/corrupted values in Transformers v5." + "reason": "Same Transformers 5.x non-persistent-buffer corruption regression." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Both mention dtype problems, but one is autocast mismatch in Phi-tiny-MoE and the other is Qwen3.5 RLHF instability from 3D position_ids/SDPA behavior." + "reason": "Both mention dtype/position-related issues in Qwen-family models, but the actual failures are distinct." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Unexpected compute_loss keyword argument and ZeRO-3 Gemma3 loading IndexError are unrelated." + "reason": "Trainer signature mismatch and ZeRO-3 loading IndexError are unrelated." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Different quantization/storage compatibility failures; no shared concrete bug path." + "reason": "Different incompatibilities: bitsandbytes constructor breakage versus missing Float8 storage type." }, { "left": "issue:43950", "right": "issue:45902", - "accept": true, - "reason": "Both point to the same underlying non-persistent-buffer/materialization problem: uninitialized or junk values after meta/to_empty-style loading." + "accept": false, + "reason": "Both involve loading/initialization, but one is the broad non-persistent-buffer regression and the other is a specific Qwen rotary-embedding memory bug." }, { "left": "issue:43866", "right": "issue:44403", "accept": false, - "reason": "Checkpoint corruption and noisy load warnings are not the same bug." + "reason": "Checkpoint corruption and loading noise are not the same bug." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "Both are model/tokenization related in a broad sense, but there is no concrete shared failure mode." + "reason": "SigLIP2 output-quality regression is unrelated to tokenization changes in a later transformers version." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Same model family, but the reported symptoms differ too much to be confidently the same underlying bug." + "reason": "Both are SigLIP/SigLIP2-related, but one is zero probabilities and the other is degraded text embeddings; not the same bug." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Bad AutoTokenizer error messaging and missing all_tied_weights_keys are unrelated." + "reason": "Different model-loading problems: bad AutoTokenizer error for Voxtral versus missing all_tied_weights_keys on MolmoForCausalLM." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Gemma4 FSDP2 shared_kv_states rebuilding and Qwen3.5 rotary-buffer corruption are different bugs." + "reason": "Gemma4 FSDP2 shared_kv_states rebuilding and Qwen rotary-embedding uninitialized memory are unrelated." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Both involve Qwen3.5, but one is flash-attention position_ids handling and the other is CPU-offload compatibility in rotary embedding." + "reason": "Both involve Qwen3.5 and position/rotary behavior, but one is FlashAttention position_ids handling and the other is CPU offload incompatibility." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Different memory/training regressions: LoRA checkpointing ineffectiveness vs int4 CUDA reserved-memory growth." + "reason": "Gradient checkpointing with PEFT LoRA is unrelated to int4 quantization memory growth/OOM." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The clustering is grounded in the packet, and the accepted soft-edge pairs are stated conservatively enough. The non-persistent-buffer chain is well supported by issue titles and descriptions, and the tied-weights and SigLIP2 pairs are framed as related regressions rather than overclaimed duplicates. No mergeability issues apply here since these are issue pairs, not PRs." + "accept": false, + "feedback": "The output is mostly conservative, but one key reason overstates what is supported by the packet: `best_issue_reason` claims 43950 \u201chas a precise reproduction,\u201d yet the packet does not include issue bodies, only titles and metadata. That makes the claim unsupported. The SigLIP2 pair can be treated as a tentative duplicate, but it should stay explicitly tentative rather than implying stronger evidence than the packet provides." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -1315,8 +1327,8 @@ "issue:45685|issue:45823", "issue:44991|issue:45399", "issue:43824|issue:43872", - "issue:42491|issue:44368", "issue:43881|issue:43883", + "issue:42491|issue:44368", "issue:45020|issue:45823", "issue:44661|issue:45020", "issue:43828|issue:45823", @@ -1331,24 +1343,24 @@ "issue:42915|issue:43872", "issue:43792|issue:44220", "issue:43257|issue:45305", - "issue:43940|issue:45313", "issue:36331|issue:41093", + "issue:43940|issue:45313", "issue:42947|issue:45923" ], - "prepared_review_unit_hash": "63986c59a7ba4ffded2a1efba3674e9dacc4aec1fdb9037df5b92165c2cfd262", + "prepared_review_unit_hash": "e0f72825791ccfe2607c306a615d230d84b9eefdf17e2788697cb44a2ec849e9", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37210, + "serialized_chars": 37209, "estimated_input_tokens": 9303, "estimated_eval_tokens": 18862 }, @@ -1357,204 +1369,204 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: tokenizer/doc regressions, model-loading failures, memory/checkpointing issues, and a few model-specific bugs. Only two soft pairs look like true duplicates; the rest are merely same-subsystem or same-release noise.", - "confidence": 0.79, - "canonical_issue_reason": "issue:45020 is the broadest representative of the largest overlapping subgroup here (recent transformers / remote_code loading regressions). The cluster does not collapse into one duplicate bug, so this is an umbrella choice rather than a strict duplicate target.", + "summary": "This cluster is mostly heterogeneous, with a few clear duplicate subgroups: SigLIP2 implementation mismatch, docs still referencing removed pipeline tasks, the `_is_hf_initialized`/`init_empty_weights` compatibility regression, and a Qwen3 ZeRO-3 weight-loading failure. No single issue cleanly represents the whole set.", + "confidence": 0.69, + "canonical_issue_reason": "The cluster is too mixed for one true global canonical; 43872 is the clearest concrete regression anchor because 44291 describes the same `_is_hf_initialized`/`init_empty_weights` failure.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the best standalone representative of this mixed cluster, but it should not be treated as a sink for the unrelated reports. It captures the most general, recurring regression theme.", + "best_issue_reason": "43872 is the strongest representative issue: precise, model-agnostic, and backed by an almost identical duplicate (44291).", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43054", "right": "issue:43493", - "accept": false, - "reason": "Different problems: SigLIP2 text-embedding quality vs HF/JAX implementation discrepancy." + "accept": true, + "reason": "Same SigLIP2 problem: HF implementation diverges from the original JAX version and yields degraded embeddings." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "Generic noisy logging vs a specific model loading failure; not the same bug." + "reason": "Unrelated: generic loading noise vs a specific model-loading failure." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Both are loading errors, but for different models and different failure modes." + "reason": "Different models and different failure modes; only both are load errors." }, { "left": "issue:43975", "right": "issue:45237", "accept": false, - "reason": "Tokenizer detokenization regression vs an AMD GPU runtime compatibility issue." + "reason": "Completely different areas: tokenizer detokenization vs AMD GPU support." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "One is about load-time noise, the other is a concrete model loading failure." + "reason": "Not the same bug; one is a generic noise issue, the other is a model-specific load failure." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Both mention trust_remote_code, but one is missing fields in AutoConfig output while the other is broader remote_code breakage." + "reason": "45020 is a broad remote_code regression umbrella, while 43643 is a specific missing-fields config bug; not the same concrete issue." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Related training area, but gradient checkpointing ineffective and memory usage are not the same concrete bug." + "reason": "Different training problems: gradient checkpointing behavior vs MoE memory usage." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Different model families and distinct failure signatures." + "reason": "Different model families and different ZeRO/FP8 failure paths." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Unrelated model-specific load error vs a separate GLM5 report." + "reason": "Different model-loading bugs with no clear shared code-path." }, { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "MPS histogram kernel limitation vs Gemma4 device-map mismatch; not the same issue." + "reason": "Different device/backend issues; no shared underlying bug." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer load regression vs flash-attn fallback gating; same release era, different bug." + "reason": "Tokenizer loading regression vs flash-attn fallback gating; unrelated." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Import/export naming issue vs bitsandbytes constructor regression." + "reason": "Import/export symbol issue vs `_is_hf_initialized` constructor incompatibility." }, { - "left": "issue:42491", - "right": "issue:44368", + "left": "issue:43881", + "right": "issue:43883", "accept": false, - "reason": "Different Qwen LoRA-related warnings/failures with different underlying causes." + "reason": "Different model classes and different attribute/load failures." }, { - "left": "issue:43881", - "right": "issue:43883", + "left": "issue:42491", + "right": "issue:44368", "accept": false, - "reason": "Separate model-loading bugs in different model classes." + "reason": "One is a LoRA compatibility problem across branches; the other is a warning about tie_word_embeddings." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Remote_code breakage is unrelated to Gemma4 forward device placement." + "reason": "Broad remote_code regressions are not the same as a Gemma4 device-map forward mismatch." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Tokenizer mapping bug in add-new-model-like is a different code path from remote_code loading regressions." + "reason": "Tokenizer mapping edge case vs remote_code regression; different concrete bugs." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "Autocast dtype mismatch and device mismatch are different execution-path failures." + "reason": "Different runtime errors: autocast dtype mismatch vs device placement mismatch." }, { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Both are tokenizer performance regressions, but they affect different conversion APIs and are not the same bug." + "reason": "Both are tokenizer performance regressions, but they affect different conversion APIs and do not look like one bug." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Mistral tokenizer regex crash vs flash-attn fallback checks; distinct failures." + "reason": "Tokenizer regex crash vs flash-attn fallback logic; unrelated." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Remote_code/model loading regressions vs tokenizer codec handling and warning behavior." + "reason": "Remote_code regressions and Kimi tokenizer codec issues are different failures." }, { "left": "issue:43742", "right": "issue:45399", "accept": false, - "reason": "KeyError loading a specific model vs flash-attn fallback blocking; unrelated." + "reason": "Model load key error vs flash-attn fallback gating; not the same issue." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "lm_head tying bug vs non-persistent buffer corruption; different subsystems." + "reason": "Different code paths: tied weights on Mistral3 vs non-persistent buffer corruption in from_pretrained()." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same docs regression: v5 removed pipeline tasks, and both issues complain that docs still reference them." + "reason": "Same docs regression: v5 removed pipeline tasks, but the docs still reference them." }, { "left": "issue:43257", "right": "issue:43866", "accept": false, - "reason": "Qwen3 MoE loading/deepspeed issue vs a corrupted Ovis2 checkpoint." + "reason": "Different problems: Qwen3 MoE loading/conversion vs a corrupted Ovis2 checkpoint." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same underlying _is_hf_initialized argument incompatibility in newer transformers/init-empty-weights/bitsandbytes loading paths." + "reason": "Same `_is_hf_initialized` compatibility regression when loading with `init_empty_weights`." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "Different Qwen MoE/fp8 vs bitsandbytes incompatibility bugs." + "reason": "Different bugs: Qwen3 MoE FP8 failure vs bitsandbytes constructor incompatibility." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Whisper failing to run and an fbank feature-extraction issue are not clearly the same defect." + "reason": "The Whisper issue is too broad to \u092a\u0941\u0937\u094d\u091f\u093f this as the same `_torch_extract_fbank_features()` bug from the title alone." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Different DeepSpeed training problems: weight conversion vs gradient averaging." - }, - { - "left": "issue:43940", - "right": "issue:45313", - "accept": false, - "reason": "Both are ZeRO-3 loading failures, but the models and symptoms differ enough that this is not a clear duplicate." + "reason": "Different DeepSpeed-related issues: weight conversion on load vs gradient averaging with GAS." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Completely different runtime errors: loss kwargs mismatch vs tensor mask shape mismatch." + "reason": "Completely unrelated training API TypeError vs tensor-mask shape IndexError." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": true, + "reason": "Same underlying Qwen3-family ZeRO-3 weight-loading failure; both describe missing weights under DeepSpeed ZeRO-3." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Gradient checkpointing ineffective in PEFT LoRA is not the same as a missing supports_gradient_checkpointing flag on a trust_remote_code model." + "reason": "Related theme, but not the same concrete bug: one is PEFT LoRA checkpointing effectiveness, the other is a missing flag on a trust_remote_code variant." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it treats the cluster as mixed rather than collapsing it into one bug, and the two accepted soft pairs are well-supported by the issue titles and descriptions. The remaining pair rejections are appropriately cautious and do not overstate duplicate evidence." + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The selected canonical anchor (43872) is supported by the near-duplicate 44291, and the soft-edge accepts that were marked true are plausible from the titles alone. No obvious overstatement or mergeability issue." }, "evaluator_used": true, "retried": false, @@ -1626,8 +1638,8 @@ ], "soft_pairs": [ "issue:43854|issue:44863", - "issue:44928|issue:45588", "issue:41553|issue:44991", + "issue:44928|issue:45588", "issue:41093|issue:44805", "issue:44479|issue:44560", "issue:44530|issue:44863", @@ -1659,14 +1671,14 @@ "issue:43317|issue:44756", "issue:38175|issue:39692" ], - "prepared_review_unit_hash": "a02aadc11414bc46ad72cbd2354f163739e501877a4242327692b6d14c0315b1", + "prepared_review_unit_hash": "9a42506daa19e7c38fe81d37b6448c8364c2d33893d497ac6a25ae30acea000c", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, @@ -1681,210 +1693,210 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a mixed cluster of Transformers 5.x regressions around loading/init plus a few model-specific multimodal bugs. Only a small number of soft edges look like true duplicates; most are just same subsystem or same model family.", - "confidence": 0.68, - "canonical_issue_reason": "43957 is the broadest, most representative issue in the set: a cross-model loading regression tied to Transformers 5+ and meta-device initialization, which matches the dominant theme of the cluster.", + "summary": "Mostly a heterogeneous issue cluster, but there are a few likely duplicate subpairs: custom model init in Transformers v5, SigLIP2 bad outputs, mask-shape IndexError, and a Qwen video-input regression.", + "confidence": 0.67, + "canonical_issue_reason": "Issue 43646 is the broadest, clearest representative of the custom-model initialization regression; 43645 is the notebook-specific variant of the same underlying bug.", "canonical_pr_reason": null, - "best_issue_reason": "43957 is the best single anchor for the cluster because it has a clear, reproducible failure mode and generalizes across multiple model-loading regressions.", + "best_issue_reason": "43646 is the best single issue to anchor the cluster because it is the most general formulation of one of the strongest duplicate subclusters.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Both are model-loading failures, but they concern different architectures and different failure modes; not the same bug." + "reason": "Different model-loading failures for unrelated model families; same broad symptom, but not the same bug." }, { - "left": "issue:44928", - "right": "issue:45588", + "left": "issue:41553", + "right": "issue:44991", "accept": false, - "reason": "Different Qwen3.5 attention-path problems: gradient/SDPA collapse versus a flash-attention AttributeError on sink-less models." + "reason": "Both are tokenizer-load problems, but they affect different models and likely different code paths." }, { - "left": "issue:41553", - "right": "issue:44991", + "left": "issue:44928", + "right": "issue:45588", "accept": false, - "reason": "Both are tokenizer-loading complaints, but for unrelated models and different symptoms." + "reason": "Different failures: RLHF numerical explosion vs flash-attention AttributeError on sink-less models." }, { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "Near-identical IndexError about mask/tensor shape mismatch; same failure signature." + "reason": "Same underlying mask/index shape-mismatch IndexError; 44805 looks like a later generic report of the same bug." }, { "left": "issue:44479", "right": "issue:44560", "accept": true, - "reason": "Both describe the same Qwen video-input regression in transformers 5.3.0; 44560 is a model-specific StopIteration symptom." + "reason": "Both describe the same Qwen video-input regression in transformers 5.3.0, with 44560 as a concrete variant." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Qwen3.5 PagedAttentionCache crash and NemotronH checkpoint loading are unrelated." + "reason": "Unrelated failures: Qwen3.5 PagedAttentionCache crash vs NemotronH checkpoint loading." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Tokenizer error vs a separate model-loading failure for a different model." + "reason": "Separate tokenizer-loading issues for different models; no clear shared root cause." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Different subsystems entirely: DeepSpeed training weight shape error versus image-processor void-mask handling." + "reason": "Completely different subsystems and errors: DeepSpeed training vs image processor segmentation maps." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Different bugs: Qwen2.5-VL rope indexing versus Qwen3.5-MoE rotary embedding initialization/memory issues." + "reason": "Different bugs in different model components: Qwen2.5-VL rope indexing vs Qwen3.5-MoE meta materialization." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "GGUF/gpt-oss support and Phi-tiny-MoE autocast dtype mismatch are unrelated." + "reason": "Unrelated: GGUF gpt-oss support vs Phi MoE autocast dtype mismatch." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Same Gemma4 family, but one is FSDP2 training state rebuild and the other is device-map inference mismatch." + "reason": "Both are Gemma4 issues, but one is an FSDP2 KeyError and the other is a device-map forward mismatch." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "Different Qwen3-MoE issues: architecture support versus sliding-window behavior." + "reason": "Different problems in different model families: GGUF architecture support vs Qwen3-MoE sliding window behavior." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "AudioFlamingo3 batched embedding leak and Whisper empty-transcription handling are unrelated." + "reason": "Different audio/Whisper bugs with no shared code-path indicated." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Different bugs: non-persistent buffer corruption versus Kosmos2.5 long-OCR indexing." + "reason": "Different regressions: non-persistent buffer corruption vs long OCR index error." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Video regression and an MPS histogram dtype limitation are unrelated." + "reason": "Video-input regression vs an MPS histogram integer support error; not the same bug." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Import error for a missing class and meta-device loading failures are different problems." + "reason": "Both are load-time failures, but one is a missing import and the other is meta-device loading behavior." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "torch.compile+SDPA failure on Bamba and Qwen3.5 flash-attention position-id misinterpretation are different code paths." + "reason": "Different attention/kernel failures: torch.compile+SDPA vs Flash Attention misreading position IDs." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Same Transformers 5.0 custom-model initialization regression; the notebook case is just a narrower presentation." + "reason": "Same custom-model initialization regression; 43645 is just the Jupyter-notebook manifestation of 43646." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "FineGrainedFP8/Qwen3MoE failure is a different loading/configuration issue from meta-device model loading." + "reason": "Different issues: FineGrainedFP8Config failure vs meta-device model loading." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "Different failure classes: GGUF architecture support versus missing Float8 storage type." + "reason": "Unrelated: GGUF model architecture support vs missing Float8 storage type." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Video StopIteration in Qwen3-vl-embedding and TRL SFT embedding unpacking are unrelated." + "reason": "Both mention Qwen3.5/video-adjacent behavior, but one is video loading and the other is SFT embedding unpacking." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel.py execution issue and GGUF gpt-oss support are different topics." + "reason": "Different topics entirely: 3d_parallel.py runtime vs GGUF gpt-oss support." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "GenerationMixin import error and Qwen3MoE FP8 failure are unrelated." + "reason": "Different import-path regression vs FP8 configuration failure." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "Tokenizer-mapping edge case and Gemma4 device mismatch are different bugs." - }, - { - "left": "issue:43596", - "right": "issue:44661", - "accept": false, - "reason": "DeepSpeed ZeRO-3/BERT index error and add-new-model-like tokenizer mapping failure are unrelated." + "reason": "Different config/mapping logic: tokenizer mapping edge case vs Gemma4 device mismatch." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Qwen video StopIteration and a generic mask-shape IndexError are unrelated." + "reason": "Different errors and subsystems: video StopIteration vs mask/index shape mismatch." }, { "left": "issue:43632", - "right": "issue:45834", + "right": "issue:45030", "accept": false, - "reason": "_is_hf_initialized regression and Kosmos2.5 OCR indexing are different issues." + "reason": "Different config/init regressions with no strong evidence of a shared root cause." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Same SigLIP2 model producing wrong/empty outputs across different entry points; likely the same preprocessing/model mismatch." + "reason": "Same SigLIP2 model-output bug; the later issue is a broader description of the same incorrect behavior." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both mention stricter config validation, but they involve different models and are not clearly the same bug." + "reason": "Both involve stricter config validation, but they affect different model configs and failures." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "ZeRO-3/BERT index error and tokenizer-mapping failure are unrelated." + "reason": "Different bugs: DeepSpeed ZeRO3 empty-index error vs tokenizer mapping failure." }, { - "left": "issue:43317", - "right": "issue:44756", + "left": "issue:43596", + "right": "issue:44661", "accept": false, - "reason": "Device-map/offload loading failure and mmap OOM avoidance on Strix Halo are different concerns." + "reason": "Different bugs with different triggers and code paths." }, { - "left": "issue:38175", - "right": "issue:39692", + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Unrelated: tokenizer mapping edge case vs generic mask/index shape mismatch." + }, + { + "left": "issue:43317", + "right": "issue:44756", "accept": false, - "reason": "Same SigLIP2 family, but 39692 is a documentation-example issue with additional quantization failure; not the same underlying bug." + "reason": "Both are loading/memory related, but one is offload failure and the other is Strix Halo mmap OOM." } ] }, "evaluator_result": { "accept": true, - "feedback": "The anchor choice and cluster summary are grounded in the packet, and the soft-edge verdicts are generally conservative. The accepted pairs all share a plausible common failure signature or clearly related model/regression context, and the rejected pairs avoid overclaiming duplicates across different models or subsystems." + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The accepted soft pairs are plausible duplicate subclusters from the titles alone, and the rejected pairs avoid overclaiming despite superficial symptom overlap. The canonical/best-issue choice around 43646 vs 43645 is also well supported." }, "evaluator_used": true, "retried": false, @@ -1975,241 +1987,247 @@ "issue:44960|issue:46050", "issue:45362|issue:45406", "issue:42491|issue:45292", - "issue:44164|issue:44291", "issue:44910|issue:45588", + "issue:44164|issue:44291", "issue:45276|issue:45335", "issue:43404|issue:43454", "issue:41720|issue:44155", "issue:43425|issue:43957", "issue:39692|issue:44960", - "issue:42947|issue:45704", "issue:44402|issue:44488", + "issue:42947|issue:45704", "issue:36010|issue:39404", "issue:43873|issue:44387" ], - "prepared_review_unit_hash": "93dae5bf66a1e1d6f57d1ac3f14f4078789a37ed2c12f085c749f3089226bcd0", + "prepared_review_unit_hash": "202ba87d9a872a938c251f60ee1f7701189946c941224bb516d9682d09e4c845", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 32, - "serialized_chars": 36708, - "estimated_input_tokens": 9177, - "estimated_eval_tokens": 18610 + "serialized_chars": 36710, + "estimated_input_tokens": 9178, + "estimated_eval_tokens": 18612 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous, but it supports two real duplicate subclusters rather than just one: 43644\u219444534 for the non-persistent buffer junk regression, and 45276\u219445335 for resize_token_embeddings not propagating to downstream embeddings/output heads. The rest are at most theme-level similarities, not the same concrete bug.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43644 is still the best canonical anchor because it has the clearest direct-duplicate evidence and a well-scoped Transformers 5.0 regression; the resize_token_embeddings pair is a separate duplicate subcluster, not a reason to replace it.", + "summary": "This cluster is mostly a set of distinct Transformers regression reports, with only a couple of clear duplicate/near-duplicate pairs around v5 model initialization and embedding-resize behavior.", + "confidence": 0.69, + "canonical_issue_reason": "Issue 43644 is the broadest and clearest regression report in the set: a Transformers 5.0.0 initialization bug affecting non-persistent buffers, which matches the general v5 regression theme better than the more model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the strongest issue to keep as the cluster representative: specific, clearly scoped, and directly duplicated by issue:44534, while the other accepted pair is a separate but narrower duplicate subcluster.", + "best_issue_reason": "Issue 43644 is the best single issue to represent the cluster because it is concrete, widely scoped, and closest to the repeated v5 loading/initialization failures seen across the nearby reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Same underlying bug: Transformers v5 filling non-persistent buffers with junk / non-deterministic buffer init." + "reason": "Same underlying bug: both report Transformers v5 filling non-persistent buffers with junk, and the titles are effectively the same regression." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Both involve loading/quantization regressions, but they affect different models and code paths (Qwen3 MoE LoRA vs BitNet packed weights)." + "reason": "Different failures: one is Qwen3 MoE LoRA incompatibility across branches, the other is BitNet packed-weight loading with accelerate." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Different failures: Qwen3 auto device mapping CUDA assert vs Qwen3 MoE FineGrainedFP8Config load failure." + "reason": "Both are Qwen3-family, but the concrete failures differ: cudaErrorAssert on device mapping vs FineGrainedFP8Config loading." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Unrelated models and symptoms; GLM5 issue is not the same bug as Qwen3.5 flash-attention generation failure." + "reason": "Different models and different failure modes; one is GLM5, the other is Qwen3.5 flash-attention generation." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Both are Qwen3.5-related, but one is ZeRO-3 loading of language_model weights and the other is incorrect visual encoder keys on save_pretrained." + "reason": "Both involve Qwen3.5, but one is ZeRO-3 weight loading and the other is incorrect visual encoder keys on save_pretrained." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are model-loading complaints for different BERT models, but there is no clear evidence they share the same root cause." + "reason": "Both are model-loading failures, but they target different models and do not clearly point to the same underlying bug." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Completely different problems: FP8/Qwen3 MoE vs mmap/OOM on Strix Halo." + "reason": "Unrelated issues: Qwen3Moe FP8 loading failure vs a mmap/OOM problem on Strix Halo." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Different model families and different bugs; LoRA compatibility regression is not the same as resize_token_embeddings not updating Gemma4 embedding layers." + "reason": "Different areas: LoRA model compatibility on branch changes vs resize_token_embeddings not propagating to embeddings." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint resume step math and SGD argument handling are unrelated." + "reason": "Trainer checkpoint-resume step math and unused SGD args are unrelated training bugs." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different models and failures: unit-test load issue for GLM-4.7-Flash vs shape mismatch for Qwen3-VL." + "reason": "Different models and different load-time failures: GLM-4.7 unit test loading vs Qwen3-VL weight shape mismatch." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Buffer initialization regression vs notebook custom-model initialization breakage; related to v5, but not the same concrete bug." + "reason": "Both are Transformers 5 regressions, but one is junk-filled buffers and the other is custom-model/Jupyter initialization." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model loading regressions, but they concern different models and likely different causes." + "reason": "Both mention tokenizer/loading regressions, but they concern different models and likely different root causes." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Same SigLIP2 family, but one is a docs/example mismatch and quantization issue while the other is bad outputs in AutoModel/pipeline." + "reason": "One is a documentation example problem, the other is an inference-quality regression for SigLIP2; not the same bug." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different subsystems and models: Qwen3 MoE loading/offload vs GLM-4.7-Flash test loading." + "reason": "Different Qwen3/GLM loading problems with different symptoms and code paths." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different tokenizer paths and different model families." + "reason": "Both are tokenizer regressions, but they affect different tokenizer families and model sets; too broad to treat as one bug." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Qwen3 MoE training memory usage vs quantization offloading behavior are not the same bug." + "reason": "Qwen3 MoE training memory use and quantization/offloading behavior are distinct issues." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both mention memory/quantization, but one is training efficiency and the other is int4 CUDA reserved-memory/OOM regression." + "reason": "Different memory regressions: MoE training memory usage vs int4 CUDA reserved-memory/OOM." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "Different model families and different failures; LoRA compatibility vs Qwen3-VL weight shape mismatch." + "reason": "Unrelated: Qwen3 MoE LoRA branch compatibility vs Qwen3-VL weight shape mismatch." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "Unrelated issues: GLM5 general issue vs sinq quantized save/load failure." + "reason": "Different model families and different lifecycle bugs; no clear shared code-path." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different models and code paths: Qwen3.5 chat crash vs Gemma4Processor missing _tokenizer in transformers serve." + "reason": "Qwen3.5 chat crash and Gemma4 processor attribute error are unrelated failures." }, { - "left": "issue:44164", - "right": "issue:44291", + "left": "issue:42491", + "right": "issue:45292", "accept": false, - "reason": "Both are model lifecycle/loading problems, but one is extra_state handling and the other is init_empty_weights/_is_hf_initialized compatibility." + "reason": "LoRA/branch compatibility is unrelated to resize_token_embeddings not updating output embeddings." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Both are flash-attention-related, but one is a packed-sequence/position_ids crash and the other is an s_aux=None AttributeError." + "reason": "Both involve flash attention, but the concrete crashes are different: packed position_ids vs s_aux=None attribute error." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different loading/save bugs: extra_state serialization vs init_empty_weights/_is_hf_initialized argument handling." }, { "left": "issue:45276", "right": "issue:45335", "accept": true, - "reason": "Same concrete regression: resize_token_embeddings fails to update downstream embedding/output modules; the model wrappers differ, but the bug pattern is the same." + "reason": "Same underlying resize_token_embeddings regression: embeddings/output heads are not updated for related Gemma/T5Gemma model variants." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both concern lm_head tying, but they are different model families with distinct symptoms and likely distinct fixes." + "reason": "Both mention untied lm_head weights, but they are different model families and the evidence is not strong enough to treat them as the same concrete bug." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Completely unrelated: Qwen3 device-mapping CUDA assert vs AudioFlamingo3 token/embedding leakage across tracks." + "reason": "Different subsystems: Qwen3 device-mapping crash vs AudioFlamingo3 batched inference token/embedding leakage." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch version compatibility and meta-device loading are different regressions." + "reason": "Torch version compatibility and meta-device model-loading breakage are distinct problems." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "No substantive overlap beyond both being model-specific issues." + "reason": "Different models and different kinds of failure; one is a doc example, the other is a general GLM5 issue." }, { - "left": "issue:42947", - "right": "issue:45704", + "left": "issue:44402", + "right": "issue:44488", "accept": false, - "reason": "Different areas entirely: gradient checkpointing with PEFT LoRA vs T5 apex RMSNorm memory leak." + "reason": "Different model-loading complaints; one is tokenizer vocab mismatch, the other is a broader load failure for a different model." }, { - "left": "issue:44402", - "right": "issue:44488", + "left": "issue:42947", + "right": "issue:45704", "accept": false, - "reason": "Both involve BERT model loading/tokenizer problems, but the reported failures are different and not clearly the same root cause." + "reason": "Gradient checkpointing with PEFT LoRA is unrelated to a T5/apex RMSNorm memory leak." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "Import path breakage for GenerationMixin and Whisper pipeline return_language are unrelated failures." + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language failure are unrelated." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Both are quantization/memory complaints, but one is offloading not working and the other is an int4 reserved-memory/OOM regression." + "reason": "Both are memory-related, but one is quantization offloading and the other is increased CUDA reserved memory under int4 quantization." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The analysis is mostly conservative, and the 43644\u219444534 duplicate looks well supported. However, the summary is a bit too categorical in saying there is \"only one clear duplicate pair\": 45276\u219445335 is a very close resize_token_embeddings regression pair with highly similar symptoms, so the packet does not support dismissing the cluster as having just one duplicate pair with that level of certainty." + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays conservative. The two accepted soft pairs are well-supported by near-identical titles and closely matching regression descriptions, while the rejects avoid overclaiming shared root cause across mostly model-specific failures." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -2294,8 +2312,8 @@ "issue:45684|issue:45698", "issue:45588|issue:45663", "issue:43646|issue:43708", - "issue:43645|issue:44164", "issue:45812|issue:46315", + "issue:43645|issue:44164", "issue:45292|issue:45335", "issue:43828|issue:43957", "issue:44402|issue:44451", @@ -2312,20 +2330,20 @@ "issue:39404|issue:44912", "issue:44265|issue:44898" ], - "prepared_review_unit_hash": "701d457aa311fdf66267f43588fbaa60cfc9c086129f902d1d257b6f4958071f", + "prepared_review_unit_hash": "ad8c47ae41574a4ff2aeb1d99d67b01c1450c4b01e032829cb05197f4bf2ca53", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 38047, + "serialized_chars": 38046, "estimated_input_tokens": 9512, "estimated_eval_tokens": 19280 }, @@ -2334,210 +2352,222 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Most items are unrelated issue reports across different models and regressions. The only clear duplicate-like family is the resize_token_embeddings bug affecting output/decoder embeddings across Gemma4, generic output_embeddings, and T5Gemma variants.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 45292 is the broadest, most generic formulation of the resize_token_embeddings regression, and it cleanly covers the more specific Gemma4 and T5Gemma manifestations.", + "summary": "The set is mostly a grab bag of unrelated Transformers bugs. The only clear duplicate cluster is the resize_token_embeddings regression trio: issues 45276, 45292, and 45335 all describe the same output-embedding propagation bug, with 45292 as the broadest representative.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 45292 is the best canonical issue because it states the core bug generically (`resize_token_embeddings` not affecting `output_embeddings`) and covers the more specific model variants in 45276 and 45335.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45292 is the best cluster representative because it states the core bug succinctly without being over-specialized to one model family.", + "best_issue_reason": "Issue 45292 is the strongest global issue anchor: concise, generic, and directly matched by the two accepted duplicate edges.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Different failures: Gemma4 FSDP shared_kv_states KeyError vs multimodal Qwen config unwrapping in from_config." + "reason": "Different subsystems and failure modes: Gemma4 FSDP attention state vs Qwen multimodal config unwrapping." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized argument handling is unrelated to Strix Halo mmap OOM." + "reason": "One is an init_empty_weights argument regression; the other is mmap/OOM on Strix Halo. Not the same bug." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Different code paths and symptoms: Gemma4 attention state rebuild vs Qwen3.5 MoE rotary embedding CPU-offload incompatibility." + "reason": "Gemma4 FSDP KeyError and Qwen3.5 MoE CPU-offload rotary embedding incompatibility are unrelated." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "AudioFlamingo3 batched inference token leak is unrelated to Gemma4 PLE device mismatch." + "reason": "AudioFlamingo3 batched inference token leak is unrelated to Gemma4 device placement under auto device map." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Int4 CUDA reserved-memory growth is a quantization memory regression, not the mmap/OOM-on-Strix-Halo bug." + "reason": "Int4 quantization memory growth and Strix Halo mmap OOM are different problems." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are model-loading/quantization issues, but they affect different models and different failure modes (NVFP4 run error vs MXFP4 fallback to bf16)." + "reason": "Both involve quantized model loading, but the models and concrete failures differ; not enough evidence of one bug." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug is unrelated to Gemma4 FSDP attention-state handling." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to Gemma4 FSDP attention state rebuilding." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Meta-device loading regression is a separate loading path from disabling mmap on Strix Halo." + "reason": "Meta-device loading failures are a different class of issue than mmap/OOM on Strix Halo." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Both mention gradient checkpointing, but one is about ineffective LoRA checkpointing and the other about eval-mode prohibition." + "reason": "Both mention gradient checkpointing, but one is ineffectiveness with PEFT LoRA and the other is an eval-mode restriction." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same underlying resize_token_embeddings regression; 45276 is a Gemma4-specific manifestation of the broader output_embeddings issue in 45292." + "reason": "Same underlying bug: `resize_token_embeddings` does not propagate to output embeddings; 45276 is a model-specific instance." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "Model loading test failure for GLM-4.7-Flash is unrelated to the Qwen3.5 save_pretrained regression." + "reason": "GLM-4.7-Flash loading test failure is unrelated to Qwen3.5 save_pretrained checkpoint regression." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "save/from_pretrained extra_state handling is a serialization bug, not an mmap/OOM memory issue." + "reason": "Custom `extra_state` save/load handling is unrelated to Strix Halo mmap/OOM." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction is unrelated to the Qwen3.5 RLHF NaN/SDPA fallback problem." + "reason": "Eval-mode checkpointing restriction and RLHF NaN explosion from 3D position ids are different bugs." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Video-input regression for Qwen VL models is unrelated to strict Granite speech config loading." + "reason": "Video-input regression for Qwen VL models is not the same as strict-config loading for granite_speech." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "One is about read-only file permissions during save_pretrained; the other is loading the wrong custom module after save_pretrained." + "reason": "One concerns read-only permissions during save_pretrained; the other is wrong custom-module resolution after save_pretrained." }, { "left": "issue:45588", "right": "issue:45663", "accept": false, - "reason": "Flash-attention sink-less s_aux=None crash is a different attention-path bug from the Gemma4 FSDP KeyError." + "reason": "Flash-attention s_aux=None crash is unrelated to Gemma4 shared_kv_states rebuilding." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom model initialization breakage in Transformers 5.0 is unrelated to resume_from_checkpoint max_steps calculation." + "reason": "Custom model initialization in notebooks and resume_from_checkpoint max_steps calculation are unrelated." }, { - "left": "issue:43645", - "right": "issue:44164", + "left": "issue:45812", + "right": "issue:46315", "accept": false, - "reason": "Notebook-defined custom model initialization is not the same bug as save/from_pretrained extra_state handling." + "reason": "Both are tokenizer-related, but one is incorrect token IDs and the other is a performance regression." }, { - "left": "issue:45812", - "right": "issue:46315", + "left": "issue:43645", + "right": "issue:44164", "accept": false, - "reason": "One is a correctness regression in token IDs, the other is a performance regression in convert_tokens_to_ids." + "reason": "Jupyter custom-model init breakage is unrelated to save/from_pretrained extra_state handling." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Same resize_token_embeddings family of bug; 45335 is the T5Gemma decoder.embed_tokens manifestation of the broader output_embeddings regression." + "reason": "Same resize_token_embeddings propagation bug; 45335 is a t5gemma-specific manifestation." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch on Phi-tiny-MoE is unrelated to meta-device loading failures." + "reason": "Autocast dtype mismatch on Phi-tiny-MoE is a different failure from meta-device loading issues." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Tokenizer vocab-size mismatch is not the same underlying issue as failing to load ScandiBERT." + "reason": "Tokenizer vocab-size mismatch and ScandiBERT loading failure are not clearly the same defect." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch 2.10 incompatibility is too broad and not the same concrete dtype-mismatch failure as Phi-tiny-MoE autocast." + "reason": "Torch 2.10 incompatibility is a version support issue, not the same as an autocast dtype mismatch on a specific model." }, { "left": "issue:43257", "right": "issue:44863", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate+deepspeed is a different checkpoint-loading bug from NemotronH checkpoint loading." + "reason": "Qwen3 MOE deepspeed conversion and NemotronH checkpoint loading are unrelated model-loading bugs." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap OOM regression." + "reason": "Float8 storage lookup failure is unrelated to disabling mmap on Strix Halo." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper pipeline return_language regression is unrelated to a tokenizer-version behavior change." + "reason": "Whisper pipeline return_language regression is unrelated to tokenizer behavior changes." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 is a different runtime failure from Gemma4 PLE device mismatch." + "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 PLE device mismatch are different code paths." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both involve Transformers 5.0 and custom models, but one is about base_model_prefix loading and the other about notebook-defined model initialization." + "reason": "Both involve custom models, but one is `base_model_prefix` loading and the other is notebook initialization; not the same bug." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch version incompatibility is unrelated to Gemma4 PLE device mismatch." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch 2.10 incompatibility and Mistral-3 dtype mismatch are different issues." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention group-type crash is unrelated to the non-template-node compilation TypeError." + "reason": "PagedAttentionCache group-type crash and non-template-node compilation error are unrelated." }, { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Isolated reproduction of issue 38071 does not clearly match the Qwen3.5 PagedAttentionCache linear_attention crash." + "reason": "Isolated reproduction of another issue is not the same as the Qwen3.5 PagedAttentionCache crash." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper return_language pipeline regression is unrelated to MXFP4 quantization falling back to bf16." + "reason": "Whisper return_language regression is unrelated to MXFP4 quantization fallback." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export/torch_compilable_check failure is a different export path from Perceiver non-default-resolution image classification." + "reason": "torch.export failure with torch_compilable_check is unrelated to Perceiver interpolation failure." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative overall. The only accepted soft-edge family, resize_token_embeddings across 45276/45292/45335, is strongly supported by the titles and similarity scores. The other pair rejections are appropriately cautious and do not overclaim duplication from the available evidence." + "feedback": "Summary is grounded in the packet, and the soft-edge verdicts are conservative. The only accepted cluster is the resize_token_embeddings trio (45276/45292/45335), which is well supported by the issue titles and similarity signal." }, "evaluator_used": true, "retried": false, @@ -2616,14 +2646,14 @@ "issue:43425|issue:44918", "issue:43299|issue:43931", "issue:43994|issue:44079", - "issue:44164|issue:45759", "issue:36032|issue:43452", + "issue:44164|issue:45759", "issue:44671|issue:44991", "issue:43596|issue:44683", "issue:44991|issue:45081", "issue:43856|issue:44756", - "issue:39404|issue:44671", "issue:44164|issue:44479", + "issue:39404|issue:44671", "issue:44737|issue:45859", "issue:43582|issue:43632", "issue:44488|issue:45081", @@ -2640,20 +2670,20 @@ "issue:44811|issue:45584", "issue:43541|issue:45084" ], - "prepared_review_unit_hash": "d11ed3da89adc4c0e34132488fb1daa4ce178baab33ea77df31f044be30ec821", + "prepared_review_unit_hash": "4ab15ccd02b1d39fca4b2684553f4c76096db813caa4fffc5b49baf3f2125f91", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37000, + "serialized_chars": 36997, "estimated_input_tokens": 9250, "estimated_eval_tokens": 18756 }, @@ -2662,198 +2692,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose collection of mostly unrelated Transformers regressions, many centered on v5 loading/tokenizer/runtime issues but not the same underlying bug. I would reject every soft edge pair; they are too model-specific or hit different code paths. If one issue must represent the cluster, the broadest concrete regression is the v5 model-loading failure around `base_model_prefix` handling.", - "confidence": 0.83, - "canonical_issue_reason": "issue:43611 is the clearest broad, concrete regression in the set: it describes a Transformers 5.0.0 model-loading breakage that can plausibly represent the common v5 loading theme better than the more specialized reports.", + "summary": "These items are mostly separate Transformers regressions across different models, tokenizers, and runtime paths; the soft pairs share theme words but not the same underlying bug. I\u2019d keep the broad v5 model-loading regression as the representative issue.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43611 is the clearest broad, concrete regression report in the set: Transformers 5.0.0 breaks model loading for configs using `base_model_prefix`, making it a good representative of the loading-regression subgroup.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the most representative and actionable issue here because it is a clear, reproducible loading regression with a general API impact, whereas many others are narrower model- or feature-specific failures.", + "best_issue_reason": "issue:43611 is the most suitable overall representative because it is specific, actionable, and broadly applicable to the v5 loading-breakage theme without being overly model-specific.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Different bugs: video-input regression in Qwen/VL models vs a generic mask-shape IndexError." + "reason": "Different failures: a v5 video-input regression for specific Qwen VL models vs a generic mask/index shape mismatch." }, { "left": "issue:43906", "right": "issue:46326", "accept": false, - "reason": "One is an isolated reproduction of another external issue; the other is a mistral_common tokenizer `continue_final_message` bug. No shared code path." + "reason": "Only one side references another issue; the titles do not establish the same concrete tokenizer bug." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "Quantized save/load corruption is unrelated to tokenizer message-finalization behavior." + "reason": "Quantized save/load corruption and a mistral_common continue_final_message bug are distinct code paths." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Float8 storage lookup failure and template-compilation errors are unrelated runtime/compiler issues." + "reason": "Float8 storage lookup failure vs template-node compilation error are unrelated runtime/compiler problems." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but they cover different API/task deprecations and are not the same change." + "reason": "Both are documentation-related, but they concern different pipeline APIs and different removal/mismatch issues." }, { "left": "issue:43425", "right": "issue:44918", "accept": false, - "reason": "Torch version incompatibility is unrelated to the Qwen3.5 embedding unpacking failure in TRL SFT training." + "reason": "Torch version incompatibility and Qwen3.5 input-embedding unpacking with TRL are separate issues." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Both involve Qwen3-VL loading, but one is a dev-version/model-family regression and the other is a specific weight-shape mismatch; not the same bug." + "reason": "One is a v5 loading regression for Qwen3VL MoE models; the other is a weight-shape mismatch on a specific Qwen3-VL checkpoint." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "A bad model/pipeline output quality report does not match a `ModelOutput` key-assignment bug." + "reason": "Siglip2 nonsensical outputs and `ModelOutput` key assignment with prior `None` are different bugs." }, { - "left": "issue:44164", - "right": "issue:45759", + "left": "issue:36032", + "right": "issue:43452", "accept": false, - "reason": "`extra_state` handling in save/from_pretrained is a different path from `text_config` unwrapping in `from_config`." + "reason": "T5 tokenizer special-token method conflict and gguf_file loading breakage are unrelated loading failures." }, { - "left": "issue:36032", - "right": "issue:43452", + "left": "issue:44164", + "right": "issue:45759", "accept": false, - "reason": "Tokenizer method-name conflict and gguf loading breakage are separate loading failures." + "reason": "`extra_state` save/load handling and `from_config` not unwrapping `text_config` are separate serialization/config issues." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "Incorrect CamemBERT MLM predictions and tokenizer loading failure for a different model are not the same issue." + "reason": "CamemBERT masked-LM correctness and tokenizer loading for EMBEDDIA/est-roberta are different problems." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "Deepspeed zero3/BertModel index error is unrelated to compiled flex_attention compatibility with newer torch." + "reason": "DeepSpeed ZeRO-3 init index error and compiled flex_attention failure on torch>=2.9 are unrelated." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-loading issues, but one is Mistral regex patching and the other is a different model's tokenizer load failure; different causes and fixes." + "reason": "Both involve tokenizers, but they target different models and different root causes." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "General training memory inefficiency vs Strix Halo mmap OOM are different resource problems." + "reason": "Qwen3 MoE training memory inefficiency and Strix Halo mmap OOM avoidance are not the same defect." }, { - "left": "issue:39404", - "right": "issue:44671", + "left": "issue:44164", + "right": "issue:44479", "accept": false, - "reason": "Whisper pipeline language-return behavior is unrelated to CamemBERT masked-LM prediction quality." + "reason": "Serialization of `extra_state` is unrelated to the Qwen VL video-input regression." }, { - "left": "issue:44164", - "right": "issue:44479", + "left": "issue:39404", + "right": "issue:44671", "accept": false, - "reason": "Save/load `extra_state` handling and a video-input regression are different subsystems." + "reason": "Whisper pipeline `return_language` regression and CamemBERT MLM prediction errors are different model-specific bugs." }, { "left": "issue:44737", "right": "issue:45859", "accept": false, - "reason": "Both mention device placement, but one is XLNet CPU arithmetic and the other is Qwen3.5 MoE CPU-offload compatibility; not the same code path." + "reason": "XLNet CPU placement bug and Qwen3_5 MoE CPU-offload incompatibility affect different code paths." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon warmup type error is unrelated to the `_is_hf_initialized` regression." + "reason": "Apple Silicon allocator warmup TypeError and `_is_hf_initialized` flag breakage are separate issues." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer loading failures with different models and failure mechanisms." + "reason": "A model-loading failure for `cjvt/sleng-bert` is not the same as the Mistral regex patch crash." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "`add-new-model-like` mapping logic and flex_attention compilation on torch>=2.9 are unrelated." + "reason": "An add-new-model-like mapping bug and a flex_attention compile failure are unrelated." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Both are model loading failures, but they affect different models and appear to stem from different loader paths." + "reason": "Both are loading failures, but for different models and different failure mechanisms." }, { "left": "issue:32090", "right": "issue:39404", "accept": false, - "reason": "Trainer broadcast of `NoneType` and Whisper return_language failure are unrelated bugs." + "reason": "Trainer broadcast TypeError and Whisper `return_language` pipeline regression are unrelated." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are v5 regressions, but one is a model-loading breakage around `base_model_prefix` and the other is a non-persistent buffer initialization bug." + "reason": "Both are v5 regressions, but one is about `base_model_prefix` loading and the other about non-persistent buffers being initialized with junk." }, { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "Perplexity degradation on one model has nothing to do with tokenizer post-processor updates at runtime." + "reason": "High perplexity on GPT-OSS and tokenizer backend post-processor updates are different concerns." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Image-preprocessor dtype mismatch is unrelated to template-node compilation errors." + "reason": "Mistral-3 image dtype mismatch and template-node compilation errors are unrelated." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "Tokenizer special-token conflict and FineGrainedFP8Config failure are different classes of bug." + "reason": "T5 tokenizer special-token conflict and Qwen3Moe FineGrainedFP8Config failure are separate model/tokenizer issues." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "AutoConfig field loss with `trust_remote_code=True` is unrelated to PagedAttentionCache crashing on `linear_attention` groups." + "reason": "Missing fields from `trust_remote_code=True` config loading and PagedAttentionCache linear_attention crashes are different bugs." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Trainer token-count tracking and multi-image iterable-dataset finetuning are different issues." + "reason": "TrainerState token counting and Qwen2.5-VL multi-image iterable-dataset training failure are unrelated." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "Qwen2.5-VL iterable-dataset finetuning failure is unrelated to mistral_common tokenizer message handling." + "reason": "Multi-image Qwen2.5-VL training failure and mistral_common `continue_final_message` behavior are not the same issue." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Both Whisper-related, but one is `batch_decode` option handling and the other is empty-transcription generation after align_special_tokens; not the same bug." - }, - { - "left": "issue:43541", - "right": "issue:45084", - "accept": false, - "reason": "Mixtral torch-dynamo tracing failure is unrelated to non-template node compilation." + "reason": "Both are Whisper-related, but one is batch_decode skipping special tokens and the other is empty-transcription generation after align_special_tokens." } ] }, @@ -2954,256 +2978,256 @@ "issue:43377|issue:43819", "issue:44451|issue:45081", "issue:44265|issue:44610", - "issue:44991|issue:46315", "issue:44987|issue:45081", + "issue:44991|issue:46315", "issue:42915|issue:44291", "issue:45216|issue:45357", "issue:44220|issue:44683", "issue:36032|issue:43742", "issue:44062|issue:44987" ], - "prepared_review_unit_hash": "d9f8dadf162dbec14259df27040474039dda4e2c2b43dba6270a9778855869df", + "prepared_review_unit_hash": "63da7724a872f51ec47ac8cd9052944ec3ae048df636b4544ac21ed5b348b496", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 38038, - "estimated_input_tokens": 9510, - "estimated_eval_tokens": 19276 + "serialized_chars": 38036, + "estimated_input_tokens": 9509, + "estimated_eval_tokens": 19274 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated v5 regressions across loading, init, audio, and generation paths. Only one clear duplicate family shows up strongly: the GGUF/from_pretrained loading failure pair.", - "confidence": 0.58, - "canonical_issue_reason": "issue:43452 is the clearest, broadest loading-path regression in the set (`gguf_file` + `from_pretrained`) and has the strongest duplicate-like match to issue:43482.", + "summary": "Mostly heterogeneous model-loading/regression reports. The clearest duplicate-like pair in this packet is the nearly identical non-persistent-buffer regression: issue:43644 and issue:44534. The 43644/43646 pairing is too title-only to merge confidently.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43644 is the best anchor because it matches issue:44534 almost verbatim and sits in the clearest duplicate-like v5 buffer-regression subcluster; it is the most concrete representative available here.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43452 is the best single representative of the only clearly matching duplicate family in this cluster.", + "best_issue_reason": "issue:43644 is the strongest standalone issue to keep because it is specific, actionable, and closest to the near-identical 44534 report; it is a better cluster anchor than the broader custom-init titles.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Different model-load regressions; same broad theme, but not the same underlying bug." + "reason": "Both are load failures, but they target different models and no shared concrete root cause is evident." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Backend install metadata vs backend registry error; distinct issues." + "reason": "Backend-related, but one is about pip extras and the other about a missing backend mapping; different bugs." }, { "left": "issue:44530", "right": "issue:46304", "accept": false, - "reason": "Unrelated subsystems: Qwen3.5 paged-attention cache vs distributed PyTorch build support." + "reason": "Different subsystems and symptoms: cache/group-type handling versus distributed build support." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Related v5 init regressions, but one is buffer corruption and the other is generic custom-model initialization." + "reason": "The titles suggest related v5 breakage, but with only titles available there is not enough evidence to say they are the same underlying defect." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "Both audio-related, but one is a feature-extractor arg regression and the other is a low-level fbank helper issue." + "reason": "Both are audio-related, but they point to different code paths and failure modes." }, { "left": "issue:43452", "right": "issue:43482", - "accept": true, - "reason": "Both point to the same GGUF/from_pretrained loading regression in v5; Qwen2.5-GGUF is a concrete instance of the broader bug." + "accept": false, + "reason": "Both involve loading, but one is a general API break and the other is a specific Qwen2.5-GGUF regression." }, { "left": "issue:44464", "right": "issue:45987", "accept": false, - "reason": "Chunked generation/compiled forward inconsistency is not the same bug as a cache length return-type contract issue." + "reason": "Chunked generation/compiled forward and seq-length return-type issues are distinct bugs." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Different model-loading failures with different symptoms and code paths." + "reason": "Missing tied-weights metadata in Molmo and a model-specific load failure for sleng-bert are unrelated." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "Tokenizer special-token conflict vs notebook custom-model init regression are not the same bug." + "reason": "Tokenizer method-name conflict and notebook custom-model initialization breakage are different regressions." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Dequantized offload loading failure is unrelated to GGUF loading." + "reason": "GPU/CPU offload loading and Qwen2.5-GGUF loading are different failure paths." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Template compilation error vs invalid method call in GraniteMoEHybrid; different defects." + "reason": "Template compilation error and GraniteMoEHybrid invalid method call are unrelated." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Different Qwen model bugs: hidden-states output vs rope-index scaling." + "reason": "Different Qwen families and different bugs: hidden-state handling versus rope-index scaling." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Same general area, but different models and different load failures." + "reason": "Both are version-specific load failures, but on different models with no evidence of the same root cause." }, { "left": "issue:43746", "right": "issue:46326", "accept": false, - "reason": "PEFT local-checkpoint loading bug is unrelated to mistral_common continue_final_message handling." + "reason": "PEFT/local checkpoint loading and continue_final_message handling are separate issues." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio chat-template sampling-rate issue vs MIMI batching/padding-mask mismatch." + "reason": "Audio chat-template sampling-rate defaulting and MIMI padding-mask batching are different bugs." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Notebook custom-model init regression is not the same concrete bug as non-persistent buffer junk." + "reason": "Notebook custom-model init and non-persistent-buffer junk are both v5 regressions, but not clearly the same concrete bug." }, { "left": "issue:43646", "right": "issue:44534", "accept": false, - "reason": "Generic custom-model init and buffer junk are related, but not clearly the same underlying bug." + "reason": "Broad custom-model initialization breakage and non-persistent-buffer corruption are related only at a high level, not enough to merge." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Weight-tying bug vs serialization inconsistency; same theme, different failure mode." + "reason": "lm_head tying in Mistral3 and tied-weight serialization differences are distinct failure modes." }, { "left": "issue:43089", "right": "issue:45987", "accept": false, - "reason": "Generation performance overhead and cache length type contract are unrelated." + "reason": "Generation performance overhead and StaticCache seq-length type mismatch are unrelated." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Loading noise issue vs Mistral regex patch crash are different problems." + "reason": "Logging/noise during loading and a Mistral regex patch crash are not the same issue." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Different loading failures for different models and code paths." + "reason": "Missing tied-weights metadata and tokenizer loading failure are unrelated." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "FP8 config failure and template compilation failure are unrelated." + "reason": "FineGrainedFP8Config failure and template compilation errors are separate bugs." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Llava Next segfault and Qwen3OmniMoe video-processing ValueError are not the same defect." + "reason": "Segfault on Llava Next and video-input ValueError in Qwen3OmniMoe are different multimodal bugs." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Siglip2 inference quality issue vs OmDet-Turbo processor size mismatch." + "reason": "Nonsensical SigLIP2 outputs and OmDet-Turbo image-size mismatch are distinct model/processor issues." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "MIMI padding-mask batching bug vs DAC from_latents/STE mismatch are different code paths." + "reason": "Missing padding-mask support and DAC STE mismatch are unrelated audio-model defects." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "Different model load failure vs tokenizer regex patch crash." + "reason": "General model-load failure and Mistral tokenizer regex patch crash are not the same bug." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "torch.export check failure and processor size mismatch are unrelated." + "reason": "torch.export failure and processor image-size mismatch are unrelated." }, { - "left": "issue:44991", - "right": "issue:46315", + "left": "issue:44987", + "right": "issue:45081", "accept": false, - "reason": "Tokenizer loading failure and convert_tokens_to_ids slowdown are different regressions." + "reason": "Different models and different failure mechanisms; no shared code-path evidence." }, { - "left": "issue:44987", - "right": "issue:45081", + "left": "issue:44991", + "right": "issue:46315", "accept": false, - "reason": "Model loading failure and Mistral tokenizer regex patch crash are not the same bug." + "reason": "Tokenizer load failure and convert_tokens_to_ids performance regression are different issues." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "FineGrainedFP8Config failure vs init_empty_weights argument error are distinct regressions." + "reason": "FineGrainedFP8Config failure and init_empty_weights / unexpected _is_hf_initialized argument are separate initialization bugs." }, { "left": "issue:45216", "right": "issue:45357", "accept": false, - "reason": "Same model family and API, but one is a broad save_pretrained regression and the other is a narrower visual-encoder-key follow-up; not clearly the same bug." + "reason": "Both are Qwen3.5 save_pretrained regressions, but one is incorrect checkpoint content and the other is incorrect visual encoder keys." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Audio fbank helper issue vs compiled flex_attention failure are unrelated." + "reason": "Audio fbank extraction and compiled flex_attention on torch>=2.9 are unrelated." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "T5 tokenizer add_special_tokens conflict and MobileLLM key error are different loading bugs." + "reason": "Tokenizer special-token conflict and a MobileLLM loading key error are different bugs." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "AddedToken special-arg duplication vs a model-specific tokenizer loading failure are not the same underlying issue." + "reason": "AddedToken special-argument duplication and a model loading failure for physical-intelligence/fast are unrelated." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft-edge pair, issue:43452 and issue:43482, is a plausible GGUF/from_pretrained duplicate family based on the titles alone, while the other pairs are appropriately rejected as distinct or not clearly the same bug. The overall characterization of the cluster as mostly unrelated regressions is supported." + "accept": false, + "feedback": "The write-up is mostly grounded, but it is not conservative enough on the soft-edge accept for issue:43644 vs issue:43646. From the packet we only have titles, and \u2018fills non-persistent buffers with junk\u2019 versus \u2018breaks custom model initialization\u2019 is suggestive but not enough to assert the same underlying defect. The summary is also a bit too strong in calling this the only obvious duplicate subcluster, since issue:43644 and issue:44534 have nearly identical titles and look like the clearest duplicate pair in the packet. Overall: reasonable cluster impression, but one soft-edge verdict overstates the evidence." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -3270,23 +3294,23 @@ "issue:46326" ], "soft_pairs": [ - "issue:43632|issue:44756", "issue:46209|issue:46326", + "issue:43632|issue:44756", "issue:44479|issue:45325", "issue:42915|issue:44589", "issue:43404|issue:45127", "issue:36032|issue:41720", "issue:43531|issue:44877", - "issue:40444|issue:45325", "issue:43611|issue:44534", + "issue:40444|issue:45325", "issue:43296|issue:43531", "issue:43425|issue:44403", "issue:40990|issue:44811", + "issue:43792|issue:44589", "issue:43632|issue:44534", "issue:44079|issue:45663", - "issue:43792|issue:44589", - "issue:33666|issue:40444", "issue:44403|issue:44991", + "issue:33666|issue:40444", "issue:43421|issue:44568", "issue:44683|issue:45020", "issue:43381|issue:45923", @@ -3301,221 +3325,224 @@ "issue:42175|issue:45542", "issue:44079|issue:44534" ], - "prepared_review_unit_hash": "b4e0b4cc3f73d767b8875cb5880d0060f2de2adfdb64008cd4f87f6678b329b6", + "prepared_review_unit_hash": "f84c92c54d53776663ef07a06329c200ed41415d3eda4780cdace061f61f8c09", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36698, - "estimated_input_tokens": 9175, - "estimated_eval_tokens": 18606 + "serialized_chars": 36693, + "estimated_input_tokens": 9174, + "estimated_eval_tokens": 18604 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are mostly unrelated issue reports that only share broad subsystem or version-regression themes. None of the soft candidate pairs look like the same underlying bug, so no pair should be merged. The strongest standalone issue is the broad Granite tokenizer regression.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45812 is the best representative issue: it describes a concrete, silent regression affecting all Granite models, has multiple inbound references, and points to a clear tokenizer-loading path.", + "summary": "This cluster is mostly heterogeneous. The only strong duplicate-like pair is the AutoTokenizer/tokenizer.json regression; the rest are different model-, backend-, or feature-specific bugs with only loose thematic overlap.", + "confidence": 0.77, + "canonical_issue_reason": "issue:45812 is the clearest high-impact, broad tokenizer regression in the set: it affects all Granite models, has inbound references, and aligns with the strongest duplicate-like evidence around AutoTokenizer loading behavior.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is also the best global issue candidate because it is the most broadly impactful and well-scoped user-facing regression in the set.", + "best_issue_reason": "issue:45812 is the best overall issue to keep because it is specific enough to be actionable, broad enough to represent a real cluster theme, and appears to cover a silent v4\u2192v5 tokenization regression with wide impact.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43632", - "right": "issue:44756", + "left": "issue:46209", + "right": "issue:46326", "accept": false, - "reason": "Different bugs: HF initialization flag handling vs mmap/OOM behavior on Strix Halo." + "reason": "Unrelated topics: FP8 inference on B200 vs tokenizer continue_final_message behavior." }, { - "left": "issue:46209", - "right": "issue:46326", + "left": "issue:43632", + "right": "issue:44756", "accept": false, - "reason": "FP8 inference on B200 and tokenizer continue_final_message are unrelated code paths." + "reason": "Different failure modes and code paths: _is_hf_initialized / v5 loading flag vs mmap OOM on Strix Halo." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both involve Qwen VL/video-temporal logic, but one is a broad video-input regression and the other is a still-image rope/position-id scaling bug." + "reason": "Both are Qwen VL regressions, but the titles point to different bugs: generic video-input regression vs still-image temporal position_id scaling." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "Both mention FP8, but one is a model/config failure and the other is a missing storage object type error." + "reason": "Both mention FP8, but one is a Qwen3Moe FineGrainedFP8Config failure and the other is a missing Float8 storage type error; not the same bug." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Mistral3 lm_head tying and LoRA vocabulary merge collapse are different defects." + "reason": "Both involve tied embeddings/lm_head, but one is a model-definition tie bug and the other is a LoRA-merging collapse issue." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer attribute conflict vs Qwen3 auto device-mapping cuda assert are unrelated." + "reason": "Tokenizer attribute conflict vs device-map CUDA assert; no shared underlying bug." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE and strict config loading for granite_speech are separate issues." + "reason": "Qwen3-MoE sliding_window behavior is unrelated to granite_speech config strictness." }, { - "left": "issue:40444", - "right": "issue:45325", + "left": "issue:43611", + "right": "issue:44534", "accept": false, - "reason": "Both touch Qwen2.5-VL, but training with multiple images per prompt is not the same as still-image temporal position-id scaling." + "reason": "Both are transformers v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; different code paths." }, { - "left": "issue:43611", - "right": "issue:44534", + "left": "issue:40444", + "right": "issue:45325", "accept": false, - "reason": "Base-model-prefix loading and non-persistent buffer initialization are different transformer v5 regressions." + "reason": "Qwen2.5-VL multi-image dataset finetuning failure is not the same as the rope-index temporal position_ids bug." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL/vLLM loading failure and Qwen3-MoE sliding_window are unrelated." + "reason": "PaddleOCR-VL loading in vLLM and Qwen3-MoE sliding_window are unrelated." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch 2.10 compatibility and noisy loading logs are not the same bug." + "reason": "Torch compatibility packaging issue vs noisy transformer loading messages; different problems." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on GPT-OSS and Whisper batch_decode special-token handling are unrelated." + "reason": "High perplexity on gpt-oss-20b is unrelated to Whisper processor.batch_decode skipping special tokens." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper runtime failure and missing Float8 storage are unrelated." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "_is_hf_initialized flag handling is unrelated to non-persistent buffer junk values." + "reason": "Both are v5 regressions, but the titles indicate distinct bugs: _is_hf_initialized handling vs junk-filled non-persistent buffers." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment with None and Gemma4 FSDP2 shared_kv_states rebuilding are different layers of the stack." + "reason": "ModelOutput key assignment and Gemma-4 FSDP2 shared_kv_states rebuilding are different defects." }, { - "left": "issue:43792", - "right": "issue:44589", + "left": "issue:44403", + "right": "issue:44991", "accept": false, - "reason": "Whisper runtime failure and Float8 storage lookup are unrelated failures." + "reason": "Generic loading noise vs a specific tokenizer loading failure for EMBEDDIA/est-roberta." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both are Qwen VL training issues, but multi-GPU training and iterable-dataset multi-image failure are not the same concrete bug." - }, - { - "left": "issue:44403", - "right": "issue:44991", - "accept": false, - "reason": "Loading noise and AutoTokenizer ignoring tokenizer.json are different tokenizer/loading issues." + "reason": "Both are Qwen VL training issues, but multi-GPU training and multi-image iterable-dataset failures are not the same bug." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Runtime post-processor updates and mdeberta add_special_tokens behavior are separate tokenizer bugs." + "reason": "Runtime special-token post-processor updates and add_special_tokens BOS/EOS behavior are related to tokenization, but not the same concrete failure." }, { "left": "issue:44683", "right": "issue:45020", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and remote_code model loading regressions are unrelated." + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to remote_code model loading regressions." }, { "left": "issue:43381", "right": "issue:45923", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction and a missing supports_gradient_checkpointing flag are different problems." + "reason": "Gradient checkpointing in eval mode is different from a missing supports_gradient_checkpointing flag." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image finetuning failure." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image finetuning." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Unexpected _is_hf_initialized argument and missing Float8 storage object are different v5 failures." + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a different initialization-path bug than missing Float8 storage." }, { "left": "issue:44829", "right": "issue:45479", "accept": false, - "reason": "FlashAttention-3 training degeneration and single-label-classification num_labels=1 zero-loss are different training bugs." + "reason": "Both concern degenerate classification training, but one is flash_attention_3-specific and the other is the num_labels=1 loss formulation." }, { "left": "issue:44462", "right": "issue:45812", - "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and Granite token-ID corruption are related only at a high level; the concrete failure modes differ." + "accept": true, + "reason": "Both are AutoTokenizer/tokenizer.json regressions that affect token IDs; they plausibly share the same tokenizer-loading code path." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer max-length default are unrelated." + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer max_length behavior are unrelated tokenizer issues." }, { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Dequantized model offload loading and Qwen3 MoE memory use are distinct memory-related issues." + "reason": "device_map=auto offload loading failure is unrelated to Qwen3 MoE training memory inefficiency." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Qwen2.5-GGUF loading and Llama4Config missing pad_token_id are unrelated model-specific bugs." + "reason": "GGUF loading failure and Llama4Config missing pad_token_id are different model-loading problems." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Video-LLaVA temporal-attention/weight-sharing bug and Qwen3.5 save_pretrained key regression are different visual-model issues." + "reason": "Video-LLaVA tower weight/temporal-attention bug is unrelated to Qwen3.5 save_pretrained key mapping." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Both involve TensorFlow packaging, but one is missing TF in the torch extra and the other is tensorboard-only causing backend errors; not the same bug." + "reason": "Both concern TensorFlow packaging, but one is missing TF in the torch extra while the other is an undefined tf backend when only tensorboard is installed; different dependency issues." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "ModelOutput key assignment and buffer initialization corruption are unrelated." + "reason": "ModelOutput key handling and non-persistent buffer initialization are unrelated defects." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The canonical issue choice is supported by the packet title/inbound refs, and the soft-edge decisions mostly stick to visible title-level evidence without overclaiming. The 44462\u219445812 acceptance is cautious enough to be plausible, while the rest of the rejects are appropriately conservative." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -3591,8 +3618,8 @@ "issue:44265|issue:45072", "issue:43749|issue:44079", "issue:43611|issue:43646", - "issue:43632|issue:45072", "issue:45325|issue:45381", + "issue:43632|issue:45072", "issue:42175|issue:43296", "issue:43792|issue:43957", "issue:43994|issue:44898", @@ -3606,28 +3633,28 @@ "issue:43606|issue:44265", "issue:43262|issue:43746", "issue:43296|issue:45800", - "issue:40990|issue:45584", "issue:34634|issue:35707", + "issue:40990|issue:45584", "issue:35707|issue:41720", "issue:44568|issue:45920", "issue:44186|issue:44265", "issue:41720|issue:45823", "issue:43606|issue:43819" ], - "prepared_review_unit_hash": "e26d7dc6dc94c598595e7862388d9728ec60dfa07d238150fbd73a2d3c724093", + "prepared_review_unit_hash": "8adeedbd6b245f0594d976a6dd595647daeae6dbe6af32304dc24bee7b5295f6", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37100, + "serialized_chars": 37099, "estimated_input_tokens": 9275, "estimated_eval_tokens": 18806 }, @@ -3636,198 +3663,198 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a loose similarity cluster, not a true duplicate set: the items span many unrelated model-loading, tokenizer, multimodal, and device-mapping regressions. The strongest common thread is a newer transformers v5 tokenizer/regression subgroup, but most soft pairs are not the same underlying bug.", - "confidence": 0.74, - "canonical_issue_reason": "issue:45920 is the broadest and most representative report in the set, covering a transformers v5 tokenizer regression across multiple models rather than a narrow model-specific symptom.", + "summary": "This cluster is mostly heterogeneous. A few pairs share broad themes like Transformers v5 model-loading regressions or Qwen2.5-VL position-id bugs, but none are clearly the same concrete bug/change, so I would not collapse any of the soft edges.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 43611 is the closest cluster center because several nearby items are Transformer v5 model-init/loading regressions, but the overall set is too diverse for a true duplicate canonical.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 best captures the cluster\u2019s main recurring theme: a generalized tokenizer regression affecting several models, with broader applicability than the other issue reports.", + "best_issue_reason": "43611 is the cleanest representative of the only coherent subtheme in the set: v5 model-loading / initialization breakage.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "Different subsystems and failures: tokenizer loading/error messaging vs Apple Silicon warmup TypeError." + "reason": "Unrelated failure modes: tokenizer loading error message vs Apple Silicon warmup TypeError." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Both involve loading, but one is a Qwen3 device-mapping assert and the other is a Mistral regex patch AttributeError." + "reason": "Different bugs in different code paths: auto device mapping CUDA assert vs tokenizer regex patch crash." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior and init_empty_weights/_is_hf_initialized loading error are distinct bugs." + "reason": "Both mention v5 regressions, but one is sliding-window behavior and the other is init_empty_weights/_is_hf_initialized loading." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "Both are batching/masking-related, but they affect different models and different code paths." + "reason": "Different model families and symptoms: batched-vs-single padding mask issue vs Gemma3 sliding-window NaNs." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Qwen2.5-VL multi-image finetuning failure and video position-id regression are different multimodal bugs." + "reason": "Both are Qwen multimodal regressions, but one is training with multiple images per prompt and the other is video position IDs." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Non-persistent buffer serialization and ModelOutput key assignment are unrelated." + "reason": "Serialization/tied-weights issue vs ModelOutput key assignment bug; not the same code-path." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export failure via torch_compilable_check is unrelated to bfloat16 dtype mismatches." + "reason": "Export failure due to torch_compilable_check is unrelated to bf16 dtype mismatch inference." }, { "left": "issue:43749", "right": "issue:44079", "accept": false, - "reason": "FSDP CPU RAM efficient loading and ModelOutput key handling do not share the same bug." + "reason": "FSDP CPU RAM loading breakage is unrelated to ModelOutput key assignment." }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are v5 loading regressions, but one is base_model_prefix handling and the other is custom model initialization." + "reason": "Close theme, but still different bugs: base_model_prefix loading vs custom model initialization." }, { - "left": "issue:43632", - "right": "issue:45072", + "left": "issue:45325", + "right": "issue:45381", "accept": false, - "reason": "_is_hf_initialized breakage is separate from dtype mismatch issues in inference." + "reason": "Same model family, but one is still-image temporal scaling and the other is video vision_position_ids; too specific to merge blindly." }, { - "left": "issue:45325", - "right": "issue:45381", + "left": "issue:43632", + "right": "issue:45072", "accept": false, - "reason": "Both touch qwen2.5-vl position IDs, but one is still-image temporal scaling and the other is video vision_position_ids; not clearly the same bug." + "reason": "_is_hf_initialized flag breakage is unrelated to bf16 dtype mismatch in inference." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Packaging/backend dependency issue and vLLM model loading failure are unrelated." + "reason": "Different packaging/loading issues: missing TensorFlow backend in extras vs PaddleOCR-VL load failure in vLLM." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "Whisper runtime failure and meta-device loading breakage are different problems." + "reason": "Whisper model runtime failure is not the same as meta-device model loading breakage." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "SigLIP2 bad outputs and Perceiver resize failures are separate vision-model issues." + "reason": "Both are vision-model quality issues, but they involve different models and likely different preprocessing paths." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Both involve Qwen2.5-VL, but one is a rope-index scaling bug and the other is an OCR long-input index error." + "reason": "Both are multimodal, but one is Qwen2.5-VL rope/position IDs and the other is Kosmos2.5 OCR index bounds." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Tying lm_head weights and serialization inconsistency are related in theme, but not the same concrete failure." + "reason": "Tied-weight loading/generation bug vs inconsistent serialization of lm_head.weight; related area, not same defect." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen2.5-GGUF loading failure are distinct model-format problems." + "reason": "Different failures: FineGrainedFP8Config on Qwen3Moe vs GGUF loading on Qwen2.5." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "Checkpoint resume key mismatch is unrelated to qwen2.5-vl rope index scaling." + "reason": "Checkpoint key mismatch is unrelated to Qwen2.5-VL position-id computation." }, { "left": "issue:44479", "right": "issue:45823", "accept": false, - "reason": "Both mention multimodal/device regressions, but the affected models and failure modes differ." + "reason": "Different model regressions: Qwen video input path vs Gemma4 PLE device mismatch." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "Unexpected init argument and strict config loading are different load-time failures." + "reason": "Unexpected _is_hf_initialized arg is a loading/init bug; strict granite_speech config is a separate config-validation issue." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "Quantization/offloading behavior and tokenization changes are not the same issue." + "reason": "Quantization/offloading issue vs tokenizer behavior change across versions; no shared concrete bug." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU offload device mismatch for Bark and torch.export failures for other models are unrelated." + "reason": "CPU offload device mismatch for Bark is unrelated to torch.export/exportability with torch_compilable_check." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio processor sample-rate defaulting and GraniteSpeech PEFT checkpoint loading are distinct bugs." + "reason": "Audio chat-template sampling-rate default is unrelated to PEFT checkpoint loading." }, { "left": "issue:43296", "right": "issue:45800", "accept": false, - "reason": "PaddleOCR-VL load failure in vLLM and torch/transformers incompatibility are different compatibility issues." + "reason": "Version compatibility reports, but for different stacks and failure modes." }, { - "left": "issue:40990", - "right": "issue:45584", + "left": "issue:34634", + "right": "issue:35707", "accept": false, - "reason": "High perplexity on gpt-oss and Whisper empty-transcription generation failure are unrelated." + "reason": "BarkProcessor voice_preset bug and progressive generation with inputs_embeds/past_key_values are unrelated." }, { - "left": "issue:34634", - "right": "issue:35707", + "left": "issue:40990", + "right": "issue:45584", "accept": false, - "reason": "Bark voice_preset and progressive generation with inputs_embeds/past_key_values are separate bugs." + "reason": "Completely different models and symptoms: high perplexity vs empty-transcription generation failure." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "Progressive generation cache handling and Qwen3 auto device-map asserts do not match." + "reason": "Progressive generation bug is unrelated to Qwen3 auto device mapping CUDA assert." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "Both are tokenizer regressions, but one is missing BOS/EOS on a specific tokenizer while the other is broad wrong-token-ID behavior across models." + "reason": "One tokenizer-specific BOS/EOS regression vs a broader AutoTokenizer token-id regression; not the same concrete bug." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "LayoutLMv2Tokenizer padding/NER crashes are unrelated to torch.export/exportability." + "reason": "Tokenizer padding/truncation crash is unrelated to torch.export failure on compilable-check models." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Qwen3 cuda assert and Gemma4 PLE device mismatch are different device-placement failures." + "reason": "Both involve device placement, but different models and different failure mechanisms." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Bark CPU offload device mismatch and DAC.from_latents/STE mismatch are different audio-model bugs." + "reason": "CPU offload device mismatch and DAC forward/from_latents mismatch are unrelated bugs." } ] }, @@ -3906,8 +3933,8 @@ "issue:46305|issue:46315", "issue:43881|issue:44488", "issue:43377|issue:44610", - "issue:43454|issue:45072", "issue:43257|issue:45923", + "issue:43454|issue:45072", "issue:43572|issue:45759", "issue:43317|issue:43873", "issue:43377|issue:44898", @@ -3919,8 +3946,8 @@ "issue:43749|issue:44534", "issue:44466|issue:45357", "issue:41720|issue:44977", - "issue:42617|issue:42915", "issue:44991|issue:46291", + "issue:42617|issue:42915", "issue:43526|issue:43883", "issue:40990|issue:45701", "issue:44387|issue:45704", @@ -3931,229 +3958,229 @@ "issue:44008|issue:45823", "issue:43404|issue:45357", "issue:36331|issue:42915", - "issue:43450|issue:45701", - "issue:44877|issue:44991" + "issue:44877|issue:44991", + "issue:43450|issue:45701" ], - "prepared_review_unit_hash": "5aa10f7de0ac0f45c843fe94101e0791bc8a30e91aad6f8c1728481ccfd55523", + "prepared_review_unit_hash": "3e0158722ad37f927444479efbbae2caebc7b30413d25887e515b15460e8191f", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37525, - "estimated_input_tokens": 9382, - "estimated_eval_tokens": 19020 + "serialized_chars": 37521, + "estimated_input_tokens": 9381, + "estimated_eval_tokens": 19018 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is heterogeneous: the soft pairs share broad Transformers/model-loading themes, but they do not look like the same underlying bug or change. I would not deduplicate any of the proposed soft edges. I\u2019d anchor the cluster on the broad offloading/quantization regression (#43873).", - "confidence": 0.84, - "canonical_issue_reason": "#43873 is the broadest and most central regression in this set: it covers offloading/quantization behavior, has strong discussion/inbound-reference activity, and is a plausible umbrella for nearby loading/offload reports better than the narrower model-specific bugs.", + "summary": "This cluster is highly heterogeneous: most items are unrelated model-specific regressions, tokenizer/loading issues, quantization/offload problems, or training quirks. The soft pairs share only superficial wording and should not be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: the items span unrelated bugs across training, quantization, tokenization, loading, and model-specific behavior, with only superficial title overlap.", "canonical_pr_reason": null, - "best_issue_reason": "#43873 is the best standalone issue to represent the cluster because it is broad, active, and closer to a cross-cutting runtime regression than the other narrowly scoped model- or tokenizer-specific reports.", + "best_issue_reason": "No single issue is a strong global representative; if forced, the broadest regression-style report is the closest anchor, but it still does not match most items in the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "AyaVision lm_head tying is a different bug from Bamba-9B-v2 torch.compile/SDPA failure." + "reason": "Different code paths and models: lm_head tying vs torch.compile/SDPA behavior." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 doc example errors are unrelated to Qwen2.5-VL iterable-dataset multi-image finetuning failures." + "reason": "Documentation example errors are unrelated to Qwen2.5-VL finetuning with multi-image iterable datasets." }, { "left": "issue:46305", "right": "issue:46315", "accept": false, - "reason": "MoonshineStreaming off-by-one masking and tokenizer conversion slowdown are unrelated regressions." + "reason": "Different subsystems: Moonshine sliding-window masking vs slow-tokenizer conversion performance." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "GLM-4V-9B loading failure and cjvt/sleng-bert tokenizer loading are separate model-loading issues." + "reason": "Different failures: model loading for glm-4v-9b vs tokenizer loading for sleng-bert." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "MIMI batched-vs-single padding-mask bug is not the same as OmDet-Turbo processor/model image-size mismatch." + "reason": "Different bugs: MIMI batching/padding-mask inconsistency vs OmDet-Turbo processor/model resolution mismatch." }, { - "left": "issue:43454", - "right": "issue:45072", + "left": "issue:43257", + "right": "issue:45923", "accept": false, - "reason": "AyaVision weight tying and dtype mismatches in SwitchTransformers/TimmWrapperModel are different code paths." + "reason": "Different models and mechanisms: Qwen3 MoE weight conversion vs missing gradient-checkpointing flag." }, { - "left": "issue:43257", - "right": "issue:45923", + "left": "issue:43454", + "right": "issue:45072", "accept": false, - "reason": "Qwen3 MOE weight conversion with accelerate/deepspeed is unrelated to a missing gradient-checkpointing flag." + "reason": "Unrelated issues: tied lm_head weights vs bfloat16 dtype mismatches in inference." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "StableLmConfig pad_token_idx regression and AutoModel text_config unwrapping are different config-loading bugs." + "reason": "Different configuration problems: missing pad_token_idx vs unwrapping text_config in composite multimodal configs." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Both touch offloading/quantization, but one is a dequantized model loading failure and the other is a broader quantized offloading regression." + "reason": "Both mention offloading/quantization, but the concrete failure modes are different enough to avoid deduping." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "MIMI padding-mask output mismatch is unrelated to Perceiver non-default-resolution interpolation failure." + "reason": "Different models and symptoms: MIMI batch masking vs Perceiver non-default resolution handling." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "High perplexity on gpt-oss-20b is not the same bug as PEFT LoRA gradient checkpointing being ineffective." + "reason": "Perplexity regression is unrelated to LoRA gradient-checkpointing ineffectiveness." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next segfault on load and sinq save/load breakage are different model-loading failures." + "reason": "Different issues: Llava Next loading segfault vs sinq quantized save/load breakage." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "AyaVision weight tying is unrelated to BitNet packed-weight unpacking during accelerate loading." + "reason": "Different code paths: lm_head tying vs BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:45325", "right": "issue:46315", "accept": false, - "reason": "Qwen2.5-VL rope-index scaling and tokenizer conversion slowdown are unrelated." + "reason": "Different bugs: Qwen2.5-VL rope/position_ids scaling vs tokenizer conversion speed regression." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "Qwen3 auto device-map cuda assert and Llava Next segfault are distinct loading/runtime crashes." + "reason": "Different model families and failure types: cudaErrorAssert on Qwen3 vs segfault loading Llava Next." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk are different regression reports." + "reason": "Broken FSDP CPU RAM-efficient loading is unrelated to non-persistent buffers being filled with junk." }, { "left": "issue:44466", "right": "issue:45357", "accept": false, - "reason": "lm_head serialization inconsistency and visual-encoder key serialization are separate save_pretrained issues." + "reason": "Both involve saving, but one is tied lm_head serialization and the other is incorrect visual encoder keys." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Qwen3 auto device-map crash and Qwen3.5 flash-attention generation bug are not the same failure." + "reason": "Different Qwen3 failures: device-mapping cudaErrorAssert vs flash-attention generation issues." }, { - "left": "issue:42617", - "right": "issue:42915", + "left": "issue:44991", + "right": "issue:46291", "accept": false, - "reason": "3d_parallel.py startup failure and Qwen3Moe FineGrainedFP8Config failure are different issues." + "reason": "Tokenizer loading regression is unrelated to kernels v0.15 requiring version/revision." }, { - "left": "issue:44991", - "right": "issue:46291", + "left": "issue:42617", + "right": "issue:42915", "accept": false, - "reason": "Tokenizer loading regression and kernels version/revision import failure are unrelated." + "reason": "Different models and errors: 3d_parallel.py runtime failure vs FineGrainedFP8Config failure in Qwen3Moe." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BEiT label reduction bug and Molmo all_tied_weights_keys attribute error do not share the same code path." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Molmo tied-weights attribute error." }, { "left": "issue:40990", "right": "issue:45701", "accept": false, - "reason": "Perplexity on a specific model is unrelated to a generic tokenization-version change." + "reason": "High perplexity and tokenizer behavior changes are only loosely related; not the same underlying bug." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "int4 quantization OOM and T5 apex RMSNorm memory leak are different memory regressions." + "reason": "Different memory issues with different causes: int4 reserved-memory growth vs apex FusedRMSNorm leak." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "GLM loading failure and tokenizer loading failure are separate problems." + "reason": "Different models and failure modes: glm-4v-9b loading vs EMBEDDIA tokenizer loading." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "CustomTrainer compute_loss signature mismatch is unrelated to 3d_parallel.py failing to run." + "reason": "Trainer compute_loss API mismatch is unrelated to 3d_parallel.py execution failure." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Blip2 dtype propagation and Trainer resume/max_steps recalculation are different trainer/runtime bugs." + "reason": "BLIP2 dtype propagation and resume_from_checkpoint step calculation are unrelated bugs." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "LayoutLMv2Tokenizer crashes and bf16 dtype mismatches are unrelated." + "reason": "Tokenizer padding/truncation crash is unrelated to dtype mismatches in inference." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma 3n variable-name collision and Gemma4 PLE device mismatch are different Gemma-family bugs." + "reason": "Different Gemma issues: variable name collision in forward vs device_map auto device mismatch." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Mistral3 lm_head tying and Qwen3.5 visual encoder key serialization are unrelated model-serialization issues." + "reason": "Different model families and different save/load problems; not the same tied-weights bug." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "CustomTrainer kwarg signature error and Qwen3Moe FP8 failure are different bugs." + "reason": "Unexpected compute_loss keyword argument is unrelated to Qwen3Moe FineGrainedFP8Config failure." }, { - "left": "issue:43450", - "right": "issue:45701", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "Video processor batching shape bug and tokenization-version change are unrelated." + "reason": "Strict config parsing for granite_speech is unrelated to tokenizer loading for EMBEDDIA/est-roberta." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:43450", + "right": "issue:45701", "accept": false, - "reason": "Strict granite_speech config loading and tokenizer loading regression are separate config/tokenizer issues." + "reason": "Video processor batch-shape bug is unrelated to tokenization changes across versions." } ] }, @@ -4257,20 +4284,20 @@ "issue:33357|issue:35707", "issue:30064|issue:34634" ], - "prepared_review_unit_hash": "3eab3acfc8fc2b19a246bbb2aaa302da87bdeb3b206d69235871acc3a54b6f49", + "prepared_review_unit_hash": "f400c1c1a6ca70c2441e7b258dcc37a022558599ddb12c4920ceaabd5901bd0b", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36346, + "serialized_chars": 36348, "estimated_input_tokens": 9087, "estimated_eval_tokens": 18430 }, @@ -4279,186 +4306,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely heterogeneous: the suggested soft pairs share broad subsystem themes, but they do not look like the same underlying bug/change. I would reject all soft edges; the closest matches are still different concrete failure modes, models, or code paths.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 45701 is the broadest umbrella-style report here because it describes a version-dependent tokenization regression and has the most cross-references, but it still does not subsume the other reports in this cluster.", + "summary": "These items do not form a duplicate cluster: the soft pairs are all cross-subsystem or cross-model false positives with different failure modes. No soft edge should be merged.", + "confidence": 0.94, + "canonical_issue_reason": "Issue 45812 is the strongest representative: it describes a concrete, silent tokenization regression affecting all Granite models with clear user impact and good diagnostic value.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45701 is the best representative issue for the cluster as a whole, mainly because it is broad and central to the tokenization-regression theme; however, the cluster is too mixed to treat it as a true duplicate anchor for most other items.", + "best_issue_reason": "Issue 45812 is the best global issue candidate because it is specific, reproducible, and broadly impactful compared with the more niche or ambiguous reports here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Different failure modes: quantization/offloading behavior vs mmap-related OOM on Strix Halo." + "reason": "Quantization/offload behavior vs mmap OOM workaround; different failure modes and likely different fixes." }, { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "Unrelated bugs: torch.compile/SDPA failure in Bamba vs unpacking packed weights during accelerate loading in BitNet." + "reason": "torch.compile + SDPA on Bamba is unrelated to accelerate loading and unpacking packed weights." }, { "left": "issue:42175", "right": "issue:46291", "accept": false, - "reason": "Packaging/backend install issue vs kernels dependency/version requirement; not the same code path." + "reason": "Missing TensorFlow extra in pip install is a packaging issue, not the kernels version/revision import breakage." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Qwen3 auto device-mapping CUDA assert is unrelated to not being able to run 3d_parallel.py." + "reason": "Qwen3 auto device mapping CUDA assert and failing 3d_parallel.py are unrelated problems." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both hit transformers serve, but one is a 'str' .to() crash and the other is a missing _tokenizer attribute; different paths." + "reason": "Continuous-batching multimodal crash differs from Gemma4Processor missing _tokenizer; separate code paths." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Different model-specific issues: torch.compile with SDPA vs CPU offload device mismatch." + "reason": "Bamba torch.compile/SDPA bug is not the same as Bark-small CPU offload device mismatch." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both involve image sizing, but they are different models with different preprocessing/model-expectation mismatches." + "reason": "Both involve image sizing, but they hit different models and different logic (processor output size vs interpolation on Perceiver)." }, { "left": "issue:44186", - "right": "issue:44361", + "right": "issue:44610", "accept": false, - "reason": "Different tokenizer classes and task failures; only broad 'tokenizer crash' similarity." + "reason": "Tokenizer NER padding/truncation crash is unrelated to an image processor producing the wrong resolution." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache-path collision and read-only permission propagation are separate filesystem bugs." + "reason": "Cache-path collision for trust_remote_code is unrelated to permission propagation in save_pretrained." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Specific processor.tokenizer/images regression vs a broader tokenization-change report; not the same concrete bug." + "reason": "A regression around processor.tokenizer/image passing is not the same concrete bug as general version-dependent tokenization changes." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "FineGrainedFP8Config load failure is unrelated to dequantized model offload with device_map=auto." + "reason": "FineGrainedFP8 Qwen3Moe failure and dequantized model offload loading failure are different quantization/offload bugs." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "ImportError for GenerationMixin and a script execution problem are unrelated." + "reason": "GenerationMixin import failure is unrelated to a script runtime failure in 3d_parallel.py." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch compatibility and kernels version/revision requirements are different dependency issues." + "reason": "Torch 2.10 incompatibility is a dependency/version issue, not the kernels import requirement change." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Different distributed/model-loading problems: CUDA assert on auto device mapping vs TP not working for GPT-OSS MoE." + "reason": "Qwen3 auto device mapping CUDA assert and GPT-OSS MoE-only-TP support are different model/runtime issues." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked-LM regression and Qwen3_5 cache-state reset are unrelated bugs." + "reason": "CamemBERT masked LM regression and Qwen3.5 recurrent-state reset bug are unrelated model behaviors." }, { "left": "issue:43525", "right": "issue:46291", "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to the kernels dependency import failure." + "reason": "Missing pad_token_id in Llama4Config is unrelated to the kernels package import constraint." }, { "left": "issue:45081", "right": "issue:45812", "accept": false, - "reason": "Different tokenizer regressions: Mistral regex patch crash vs Granite token-ID mismatch." + "reason": "Mistral regex patch/backend_tokenizer crash is a tokenizer backend attribute issue, not a Granite token-ID regression." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and StableLm pad_token_idx missing are separate config/model issues." + "reason": "Qwen3-MoE sliding_window behavior and StableLmConfig pad_token_idx missing are different config/model bugs." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Gemma3TextConfig API mismatch and GPTNeoX rotary_pct reload regression affect different config fields." + "reason": "Gemma3TextConfig API discrepancy is unrelated to GPTNeoX rotary_pct not persisting on reload." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Qwen3OmniMoe video-processing failure and Qwen3VL weight_only load error are different code paths." + "reason": "Video-input processing error in Qwen3OmniMoe is not the same as Qwen3VL weight_only loading failure." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "FP8 loading failure is not the same as a segmentation fault while loading Llava Next." + "reason": "FineGrainedFP8 loading failure and Llava Next segfault are distinct model-loading bugs." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Different tokenizer/task failures with no evidence of a shared underlying defect." + "reason": "LayoutLMv2Tokenizer NER/padding crash and MLukeTokenizer task AttributeError are different tokenizer issues." }, { "left": "issue:43421", "right": "issue:45812", "accept": false, - "reason": "Runtime post-processor update problem and Granite tokenization regression are distinct bugs." + "reason": "Runtime post-processor update omission is a different tokenizer backend bug than Granite token-ID misgeneration." }, { "left": "issue:36331", "right": "issue:41720", "accept": false, - "reason": "CustomTrainer compute_loss signature breakage is unrelated to Qwen3 device-mapping CUDA asserts." + "reason": "CustomTrainer compute_loss signature mismatch is unrelated to the Qwen3 device-mapping CUDA assert." }, { "left": "issue:43526", "right": "issue:43881", "accept": false, - "reason": "BeitImageProcessorFast label reduction and glm-4v-9b loading failure are unrelated." + "reason": "BeitImageProcessorFast label reduction bug has no overlap with glm-4v-9b loading failure." }, { "left": "issue:29159", "right": "issue:36032", "accept": false, - "reason": "Slow/fast tokenizer inconsistency is not the same bug as T5Tokenizer's add_special_tokens conflict." + "reason": "Slow-vs-fast tokenizer inconsistency and T5Tokenizer add_special_tokens load conflict are different tokenizer bugs." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric gathering label truncation and SAM3 text/box behavior are different issues." + "reason": "Metric gathering last-batch label dropping is unrelated to SAM3 text/bbox weird behavior." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "MacOS bus error in CLIP loading and progressive generation with embeds/past_key_values are unrelated." + "reason": "MacOS bus error in a CLIP model and progressive generation with inputs_embeds/past_key_values are unrelated." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Void segmentation-map processing and BarkProcessor voice_preset failure are different processors and failure modes." + "reason": "Void segmentation map processing and BarkProcessor voice_preset failure are different processor bugs." } ] }, @@ -4533,17 +4560,17 @@ ], "soft_pairs": [ "issue:41720|issue:43746", - "issue:43873|issue:45704", "issue:43550|issue:43653", + "issue:43873|issue:45704", "issue:43606|issue:43720", "issue:45406|issue:45464", "issue:44530|issue:46291", "issue:43421|issue:45701", "issue:34634|issue:41720", "issue:43531|issue:45759", + "issue:44492|issue:44509", "issue:43606|issue:43653", "issue:42490|issue:43329", - "issue:44492|issue:44509", "issue:34567|issue:35707", "issue:44625|issue:45479", "issue:44514|issue:44545", @@ -4551,31 +4578,31 @@ "issue:45759|issue:45859", "issue:44077|issue:44625", "issue:43421|issue:43927", - "issue:44898|issue:45072", "issue:43531|issue:44589", + "issue:44898|issue:45072", "issue:45588|issue:45759", "issue:44368|issue:45357", - "issue:44987|issue:45042", "issue:44062|issue:44451", + "issue:44987|issue:45042", "issue:43782|issue:45381", - "issue:39692|issue:44610", "issue:44918|issue:44936", + "issue:39692|issue:44610", "issue:43746|issue:46050" ], - "prepared_review_unit_hash": "1e2d560b40a7a2497f7634c1569a3c358a5dbff0bb4d909dd431d04622c17984", + "prepared_review_unit_hash": "3ccd9e5587441cd95d0bf962d5b8b62ee9166f6055b032cd491e0c90bc23f58b", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36672, + "serialized_chars": 36669, "estimated_input_tokens": 9168, "estimated_eval_tokens": 18592 }, @@ -4584,192 +4611,198 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a collection of unrelated closed issues across different models/subsystems. Only one soft pair is a clear duplicate: the two Qwen2_5_VLProcessor.apply_chat_template batched-input padding=False crashes.", - "confidence": 0.97, - "canonical_issue_reason": "issue:44514 is the cleanest canonical for the only true duplicate pair: it was opened first and describes the batched-input/padding=False crash in Qwen2_5_VLProcessor.apply_chat_template clearly.", + "summary": "This cluster is mostly heterogeneous; the only clear duplicate pair is the Qwen2_5_VLProcessor.apply_chat_template crash on batched input with padding=False (44514/44545).", + "confidence": 0.93, + "canonical_issue_reason": "issue:44514 is the best canonical for the duplicate pair because it was opened first and describes the same concrete crash as 44545 with the same repro conditions.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44514 is the best representative issue overall because it is the earliest and most complete report for the one confirmed duplicate cluster.", + "best_issue_reason": "issue:44514 is the strongest representative issue in the cluster: specific, reproducible, and the clearest duplicate target for 44545.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Different failures and code paths: Qwen3 auto device mapping CUDA assert vs GraniteSpeech PEFT local checkpoint loading." + "reason": "Different failures: Qwen3 auto device-mapping CUDA assert vs GraniteSpeech PEFT local checkpoint loading." }, { - "left": "issue:43873", - "right": "issue:45704", + "left": "issue:43550", + "right": "issue:43653", "accept": false, - "reason": "Both involve model internals, but one is quantization/offload behavior and the other is a T5/apex RMSNorm memory leak; not the same bug." + "reason": "Unrelated bugs in different areas: torch.compile/SDPA on Bamba vs BigBirdTokenizer special-token decode behavior." }, { - "left": "issue:43550", - "right": "issue:43653", + "left": "issue:43873", + "right": "issue:45704", "accept": false, - "reason": "Unrelated subsystems: Bamba SDPA/torch.compile failure vs BigBirdTokenizer special-token decode issue." + "reason": "Different code paths: quantization/offloading behavior vs T5 apex FusedRMSNorm memory leak." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "Both are accelerate/loading-related, but they affect different models with different root causes and symptoms." + "reason": "Different models and causes: CPU offload device mismatch vs BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Different processors and failure modes: Gemma4Processor missing _tokenizer vs Qwen3.5 streaming chat/completions failure." + "reason": "Different processors and failure modes: missing _tokenizer in Gemma4Processor vs Qwen3.5 streaming chat/completions failure." }, { "left": "issue:44530", "right": "issue:46291", "accept": false, - "reason": "PagedAttentionCache on Qwen3.5 is unrelated to the kernels import/version requirement issue." + "reason": "Different subsystems: PagedAttentionCache group-type crash vs import-time kernel version/revision validation." }, { "left": "issue:43421", "right": "issue:45701", "accept": false, - "reason": "One is a specific runtime special-token/post-processor sync bug; the other is a broad tokenization regression report." + "reason": "Both mention tokenization, but one is runtime special-token/post-processor syncing and the other is version-dependent tokenization changes." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping are unrelated." + "reason": "No shared underlying bug: BarkProcessor voice_preset vs Qwen3 auto device-mapping CUDA assert." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Different Qwen3-related bugs: sliding_window handling in MoE models vs config unwrapping for multimodal from_config." + "reason": "Different Qwen3.5 issues: sliding_window behavior vs from_config not unwrapping text_config." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A typo in cache strategies is unrelated to removed pipeline-task docs." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "Device-mismatch CPU-offload bug vs tokenizer mask-token registration bug; not the same underlying issue." + "reason": "Different bugs in different components: CPU offload mismatch vs tokenizer special-token registration/decode output." }, { "left": "issue:42490", "right": "issue:43329", "accept": false, - "reason": "Local-path trailing-slash warning is unrelated to the multimodal token counting video-branch undefined-variable bug." - }, - { - "left": "issue:44492", - "right": "issue:44509", - "accept": false, - "reason": "A documentation typo is unrelated to removed pipeline-task docs." + "reason": "Local-path trailing-slash import warning is unrelated to multimodal token counting video-branch undefined names." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "TrainerState token counters not updating is unrelated to generation with inputs_embeds/past_key_values." + "reason": "Trainer token counter not updating is unrelated to progressive generation with inputs_embeds and past_key_values." }, { "left": "issue:44625", "right": "issue:45479", "accept": false, - "reason": "Both mention classification config, but one is Qwen3.5 config propagation and the other is a cross-model zero-loss regression with num_labels=1." + "reason": "Different classification bugs: num_labels propagation vs degenerate zero loss for num_labels=1." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Same exact crash in Qwen2_5_VLProcessor.apply_chat_template on batched input when padding=False; same code path and symptom." + "reason": "Same exact bug: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model-loading regressions, but they concern different models and are not clearly the same root cause." + "reason": "Different tokenizer/load failures for different models and different root causes." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Unwrapping text_config for multimodal configs is unrelated to Qwen3_5MoeTextRotaryEmbedding CPU-offload compatibility." + "reason": "Different Qwen3.5 model-loading/runtime issues: multimodal config unwrapping vs CPU-offload incompatibility in rotary embedding." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "PatchTSMixer post_init policy and Qwen3.5 num_labels propagation are unrelated." + "reason": "PatchTSMixer post-init policy is unrelated to Qwen3.5 num_labels propagation." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Special-token/post-processor runtime update bug is different from DiaConfig losing custom token IDs after save/load." + "reason": "Runtime special-token/post-processor updates are unrelated to DiaConfig losing custom token IDs on save/load." }, { - "left": "issue:44898", - "right": "issue:45072", + "left": "issue:43531", + "right": "issue:44589", "accept": false, - "reason": "Perceiver image-size interpolation failure and bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel are different bugs." + "reason": "Sliding-window behavior for Qwen3-MoE is unrelated to Float8 storage lookup failures." }, { - "left": "issue:43531", - "right": "issue:44589", + "left": "issue:44898", + "right": "issue:45072", "accept": false, - "reason": "Qwen3-MoE sliding_window issue is unrelated to the Float8 storage TypeError." + "reason": "Different image-model dtype/resolution issues: Perceiver non-default resolution vs SwitchTransformers/TimmWrapperModel bfloat16 mismatches." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "flash_attention s_aux=None crash is unrelated to config unwrapping for Qwen multimodal models." + "reason": "flash_attention s_aux=None crash is unrelated to Qwen multimodal config handling." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "A warning about tie_word_embeddings and a regression in saving visual encoder keys are different issues." + "reason": "Different Qwen3.5 LoRA/save_pretrained warnings vs incorrect visual encoder key persistence." }, { - "left": "issue:44987", - "right": "issue:45042", + "left": "issue:44062", + "right": "issue:44451", "accept": false, - "reason": "Loading a specific model and PIL backend requiring torchvision are unrelated." + "reason": "AddedToken keyword-argument error is unrelated to loading a specific model checkpoint." }, { - "left": "issue:44062", - "right": "issue:44451", + "left": "issue:44987", + "right": "issue:45042", "accept": false, - "reason": "AddedToken special-argument conflict and ScandiBERT loading failure are different tokenizer issues with no clear shared root cause." + "reason": "Loading physical-intelligence/fast is unrelated to PIL backend processors requiring torchvision." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Qwen3VL weight_only load failure and Qwen2.5-VL video vision_position_ids bug are different multimodal problems." + "reason": "Different VL bugs: weight_only load error vs wrong vision_position_ids for video input." }, { - "left": "issue:39692", - "right": "issue:44610", + "left": "issue:44918", + "right": "issue:44936", "accept": false, - "reason": "Both involve processor/example mismatches, but they affect different models and different failure modes." + "reason": "Unpacking Qwen3.5 input embeddings with TRL is unrelated to trainer.evaluate() failing after train()." }, { - "left": "issue:44918", - "right": "issue:44936", + "left": "issue:39692", + "right": "issue:44610", "accept": false, - "reason": "Different training/evaluation failures with different APIs and symptoms." + "reason": "Docs/example errors are not the same bug as OmDet-Turbo processor/model resolution mismatch." }, { "left": "issue:43746", "right": "issue:46050", "accept": false, - "reason": "PEFT local-checkpoint loading and sinq quantized save/load breakage are unrelated model-loading problems." + "reason": "PEFT checkpoint loading for GraniteSpeech is unrelated to sinq quantized model save/load corruption." + }, + { + "left": "issue:44936", + "right": "issue:44987", + "accept": false, + "reason": "Training/evaluation state issue is unrelated to loading physical-intelligence/fast." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the lone accepted duplicate pair (issue:44514 / issue:44545) is well supported by the identical title and matching failure description. The other soft-edge rejections are appropriately conservative given the different model names, symptoms, and subsystems in the issue titles." + "feedback": "The assessment is grounded in the packet: 44514/44545 are the same issue by title and chronology, and the rest of the soft-edge rejections stay conservative by citing distinct models, code paths, or failure modes. I don\u2019t see any overstatement that would force a reject." }, "evaluator_used": true, "retried": false, @@ -4841,14 +4874,14 @@ ], "soft_pairs": [ "issue:45704|issue:45923", - "issue:45663|issue:45923", - "issue:43421|issue:43653", "issue:43381|issue:43844", + "issue:43421|issue:43653", "issue:42886|issue:44843", + "issue:45663|issue:45923", "issue:45704|issue:45859", "issue:41720|issue:45704", - "issue:43994|issue:44671", "issue:43531|issue:43645", + "issue:43994|issue:44671", "issue:43421|issue:44186", "issue:43746|issue:43873", "issue:45081|issue:45356", @@ -4865,234 +4898,237 @@ "issue:43550|issue:45072", "issue:43295|issue:45834", "issue:43618|issue:43697", - "issue:43550|issue:43819", "issue:44062|issue:44488", + "issue:43550|issue:43819", "issue:45183|issue:45479", "issue:43388|issue:46032", - "issue:44186|issue:44898", - "issue:43322|issue:45685" + "issue:43322|issue:45685", + "issue:44186|issue:44898" ], - "prepared_review_unit_hash": "f83766eb9d788e8ab4615c8273f0210a13d28093d00c98ab2dba1e46ff785c93", + "prepared_review_unit_hash": "d8228334165485221e6783549ea0cb20274c99432afd814cad7bcf195a65a8e1", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37482, - "estimated_input_tokens": 9371, - "estimated_eval_tokens": 18998 + "serialized_chars": 37488, + "estimated_input_tokens": 9372, + "estimated_eval_tokens": 19000 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated tokenizer, quantization/offload, model-loading, and model-specific regression reports. The soft-similarity links are mostly topical rather than true duplicates, so none should be merged as the same underlying bug/change.", - "confidence": 0.82, - "canonical_issue_reason": "issue:43873 is the strongest cluster anchor by activity and breadth (open, many discussion/inbound refs) and is the most generally framed problem in the set.", + "summary": "The cluster is mostly a set of unrelated issue reports. Only one soft pair looks like the same underlying bug: offline tokenizer loading broken by `_patch_mistral_regex` / `model_info` behavior. The rest are at best broad subsystem overlaps, not true duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "issue:44843 is the strongest canonical issue because it names the concrete root cause (`model_info()` called unconditionally in `_patch_mistral_regex`) and directly explains the offline failure; issue:42886 is the broader symptom report.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best representative issue here because it has the highest apparent demand signal and the broadest scope among the listed issues.", + "best_issue_reason": "issue:44843 is the best representative of the cluster\u2019s only clear duplicate-like bug: it is specific, root-cause oriented, and covers the offline tokenizer regression more precisely than the symptom-level report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45704", "right": "issue:45923", "accept": false, - "reason": "Different failures: T5/apex RMSNorm memory leak vs missing supports_gradient_checkpointing on a trust_remote_code variant." + "reason": "Different problems: T5 RMSNorm memory leak vs missing `supports_gradient_checkpointing` flag in a trust-remote-code model." }, { - "left": "issue:45663", - "right": "issue:45923", + "left": "issue:43381", + "right": "issue:43844", "accept": false, - "reason": "Both are Gemma/Nemotron-related model bugs only at a broad theme level; the concrete code paths and symptoms are unrelated." + "reason": "Both involve training/runtime stability, but one is eval-mode gradient checkpointing and the other is ZeRO-3 gradient explosion; not the same bug." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Both involve tokenizer special-token handling, but one is runtime post-processor updates while the other is BigBird mask-token registration and decode output." - }, - { - "left": "issue:43381", - "right": "issue:43844", - "accept": false, - "reason": "Different training/runtime behaviors: eval-mode gradient checkpointing guard vs gradient inflation under ZeRO-3." + "reason": "Both concern tokenizer special-token behavior, but one is runtime post-processor updates and the other is BigBird mask-token registration / decode output." }, { "left": "issue:42886", "right": "issue:44843", + "accept": true, + "reason": "Same offline-tokenizer bug: both describe failure under `HF_HUB_OFFLINE=1`, and 44843 gives the specific root cause in `_patch_mistral_regex`/`model_info()`." + }, + { + "left": "issue:45663", + "right": "issue:45923", "accept": false, - "reason": "Both mention offline/tokenizer loading, but one is cache loading failure and the other is a specific mistral regex patch making a network call." + "reason": "Gemma4 FSDP2 shared-state KeyError is unrelated to the Nemotron gradient-checkpointing flag issue." }, { "left": "issue:45704", "right": "issue:45859", "accept": false, - "reason": "Apex RMSNorm memory leak and Qwen3_5Moe CPU-offload incompatibility are separate defects." + "reason": "Both mention performance/offload-ish areas, but one is an apex RMSNorm memory leak and the other is Qwen3 MoE CPU-offload incompatibility." }, { "left": "issue:41720", "right": "issue:45704", "accept": false, - "reason": "Different subsystems and symptoms: Qwen3 auto device-map CUDA assert vs T5/apex norm leak." + "reason": "CUDA assert in Qwen3 auto device mapping is unrelated to T5/apex RMSNorm memory leak." }, { - "left": "issue:43994", - "right": "issue:44671", + "left": "issue:43531", + "right": "issue:43645", "accept": false, - "reason": "Both are model/tokenizer regressions, but one is SigLIP2 inference misbehavior and the other is add-new-model-like mapping logic." + "reason": "Qwen3-MoE sliding-window behavior and Jupyter custom-model initialization are unrelated." }, { - "left": "issue:43531", - "right": "issue:43645", + "left": "issue:43994", + "right": "issue:44671", "accept": false, - "reason": "Sliding-window handling for Qwen3-MoE is unrelated to notebook custom-model initialization in Transformers 5.0.0." + "reason": "Different model failures: SigLIP2 AutoModel/pipeline output issue vs CamemBERT masked-LM prediction regression." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Tokenizer special-token post-processing and LayoutLMv2 NER/padding crashes are different code paths." + "reason": "Both are tokenizer bugs, but they affect different code paths and symptoms; no evidence they are the same underlying defect." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "PEFT local-checkpoint loading for GraniteSpeech is unrelated to quantization/offloading behavior." + "reason": "Local checkpoint loading with PEFT adapters is not the same as quantization/offloading not working as expected." }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Both concern mistral-regex/tokenizer behavior, but one is a crash on backend_tokenizer access and the other is codec handling plus a misleading warning." + "reason": "Both touch `fix_mistral_regex`, but one is a crash on loading Mistral tokenizer while the other is a Kimi-K2.5 codec/warning regression; not clearly one bug." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "SigLIP2 documentation example errors and quantization/offloading are not the same bug." + "reason": "SigLIP2 docs example errors are unrelated to quantization/offloading behavior." }, { "left": "issue:43540", "right": "issue:46294", "accept": false, - "reason": "Different multimodal video-processing failures for different Qwen models; not the same concrete defect." + "reason": "Both are video-input issues in Qwen models, but they fail in different layers (`process`/model input vs chat-template token-feature mismatch)." }, { "left": "issue:42915", "right": "issue:43746", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure and GraniteSpeech PEFT checkpoint loading involve different mechanisms and errors." + "reason": "Qwen3Moe FP8 config failure is unrelated to GraniteSpeech PEFT checkpoint loading." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Kimi-K2.5 tokenizer regression and list_repo_files not forwarding revision are unrelated." + "reason": "Tokenizer regression and CI `list_repo_files` revision forwarding are unrelated." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "DeepSpeed ZeRO-3/BertModel initialization error versus an audio feature extraction issue." + "reason": "Deepspeed ZeRO-3/BertModel index error is unrelated to `_torch_extract_fbank_features()`." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Image segmentation-map processing and GenerationMixin import failure are unrelated." + "reason": "Image processor segmentation-map handling is unrelated to the `GenerationMixin` import error." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "BitNet packed-weight loading and DAC.from_latents/STE mismatch are different model-specific bugs." + "reason": "BitNet packed-weight loading and DAC `from_latents` vs forward mismatch are different model-specific bugs." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Tokenizer special-token registration and BitNet accelerate loading are unrelated defects." + "reason": "BigBird special-token decode bug is unrelated to BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "Audio fbank extraction and add-new-model-like/tokenizer mapping logic do not share the same failure mode." + "reason": "Audio feature extraction issue and `add-new-model-like` failure in tokenizer mapping are unrelated." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Video input validation in Qwen3OmniMoe and a Gemma3n variable-name collision are unrelated." + "reason": "Video-input failure in Qwen3OmniMoe is unrelated to Gemma 3n audio-mask variable collision." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "Both are dtype/compile-related in broad terms, but Bamba SDPA torch.compile failure and SwitchTransformers/TimmWrapper bfloat16 mismatch are different bugs." + "reason": "Different paths: torch.compile/SDPA on Bamba vs bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "Custom-model processor/tokenizer regression and Kosmos2.5 OCR long-input index error are different code paths." + "reason": "Processor/tokenizer regression is unrelated to Kosmos2.5 long-OCR index error." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "Missing CLIPOutput attentions assignment and RTDetrV2 output divergence in v5 are not the same issue." + "reason": "CLIP attentions being unassigned is unrelated to RTDetrV2 output differences in v5." }, { - "left": "issue:43550", - "right": "issue:43819", + "left": "issue:44062", + "right": "issue:44488", "accept": false, - "reason": "Both affect model behavior, but torch.compile/SDPA vs DAC latent conversion are unrelated." + "reason": "`AddedToken` keyword collision and failure to load `cjvt/sleng-bert` are different tokenizer-loading issues." }, { - "left": "issue:44062", - "right": "issue:44488", + "left": "issue:43550", + "right": "issue:43819", "accept": false, - "reason": "AddedToken keyword collision and inability to load a specific model are different tokenizer/model-loading problems." + "reason": "Bamba torch.compile/SDPA failure is unrelated to DAC `from_latents` mismatch." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "XOR validation messaging across models and single-label classification zero-loss are separate issues." + "reason": "Input validation error messaging and degenerate zero loss from `num_labels=1` are different bugs." }, { "left": "issue:43388", "right": "issue:46032", "accept": false, - "reason": "Metric gathering/truncation in the last batch is unrelated to Mamba2Mixer use_cache correctness." + "reason": "`gather_for_metrics` last-batch label dropping is unrelated to Mamba2Mixer `use_cache` correctness." }, { - "left": "issue:44186", - "right": "issue:44898", + "left": "issue:43322", + "right": "issue:45685", "accept": false, - "reason": "LayoutLMv2 tokenizer batching crash and Perceiver interpolation failure are different model-specific bugs." + "reason": "Llava Next segmentation fault and MPS histogram-on-int error are unrelated." }, { - "left": "issue:43322", - "right": "issue:45685", + "left": "issue:44186", + "right": "issue:44898", "accept": false, - "reason": "Llava Next segmentation fault and MoE MPS histogram Int support are unrelated." + "reason": "LayoutLMv2 tokenizer NER/padding crash and Perceiver non-default-resolution failure are unrelated." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall and appropriately conservative. The one accepted soft pair (42886/44843) is supported by the packet, and the remaining pairs are rejected for clearly different bugs rather than loose subsystem similarity. The canonical/best-issue choice is also reasonable." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5160,216 +5196,216 @@ ], "soft_pairs": [ "issue:43335|issue:43756", - "issue:43531|issue:43643", "issue:43716|issue:44008", + "issue:43531|issue:43643", "issue:43653|issue:43819", - "issue:43950|issue:45987", "issue:44843|issue:45081", + "issue:43950|issue:45987", "issue:43643|issue:44877", "issue:30064|issue:36331", "issue:44448|issue:44671", "issue:43540|issue:45381", - "issue:43421|issue:44811", "issue:41720|issue:44991", - "issue:43782|issue:44977", + "issue:43421|issue:44811", "issue:43577|issue:43818", + "issue:43782|issue:44977", "issue:39692|issue:45834", "issue:44821|issue:45084", "issue:45584|issue:45834", - "issue:44186|issue:44442", "issue:43644|issue:43749", + "issue:44186|issue:44442", "issue:43606|issue:45072", - "issue:43577|issue:43582", "issue:44479|issue:45834", "issue:34634|issue:36010", + "issue:43577|issue:43582", "issue:43825|issue:43994", "issue:44805|issue:44936", "issue:44485|issue:44509" ], - "prepared_review_unit_hash": "517e864268991e21a241f80c8c7ae57e3744ee2625a4c48423b4987d8c69904e", + "prepared_review_unit_hash": "93bba59b05dbd9c2b7f7b158c1669fd4d3b6858010fc18c06c746b1580faae51", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 35696, - "estimated_input_tokens": 8924, - "estimated_eval_tokens": 18104 + "serialized_chars": 35697, + "estimated_input_tokens": 8925, + "estimated_eval_tokens": 18106 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: the pairwise similarities are driven by shared subsystems or API names, not the same concrete bug. I would not merge any of the soft-edge pairs, and there are no PRs to evaluate.", - "confidence": 0.91, - "canonical_issue_reason": null, + "summary": "This cluster is mostly heterogeneous. A few pairs share a subsystem or helper function, but none look like the same underlying bug/change closely enough to merge. The strongest recurring theme is non-persistent buffer/loading regressions, with issue:43950 as the clearest representative.", + "confidence": 0.69, + "canonical_issue_reason": "issue:43950 is the most specific and broadly useful statement of a core regression in the cluster: non-persistent buffer corruption during loading. It reads like the best representative of that loading-path bug family.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43950 is the strongest standalone issue in the set: it describes a clear, broad 5.x regression in `from_pretrained()` buffer handling with direct impact and a well-scoped failure mode.", + "best_issue_reason": "issue:43950 is the best issue to anchor this cluster because it is concrete, regression-oriented, and broadly applicable to a core model-loading path rather than a narrow model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "Different models and different bugs: SwitchTransformers sparse-layer config vs Smollm3 RoPE layer count." - }, - { - "left": "issue:43531", - "right": "issue:43643", - "accept": false, - "reason": "Unrelated failures: Qwen3-MoE sliding window behavior vs AutoConfig missing fields with `trust_remote_code=True`." + "reason": "Both are model/configuration bugs, but they affect different models and different mechanics (sparse-layer setup vs RoPE-layer dropping). Not the same underlying issue." }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Different code paths and symptoms: image preprocessor dtype mismatch vs Gemma3n audio variable name collision." + "reason": "Both are multimodal model bugs, but the failures are unrelated: dtype mismatch in image preprocessing vs a Gemma3n variable/attribute collision." }, { - "left": "issue:43653", - "right": "issue:43819", + "left": "issue:43531", + "right": "issue:43643", "accept": false, - "reason": "Tokenizer special-token registration bug and DAC latent/STE mismatch are unrelated." + "reason": "One is a Qwen3-MoE sliding-window behavior issue; the other is a config-loading bug with trust_remote_code. Same area only at a very high level." }, { - "left": "issue:43950", - "right": "issue:45987", + "left": "issue:43653", + "right": "issue:43819", "accept": false, - "reason": "Both involve state/cache semantics, but one is a loading regression and the other is a `get_seq_length()` return-type contract issue." + "reason": "Tokenizer special-token registration and DAC latent/forward consistency are unrelated bugs with no shared code path implied." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Same helper function name, but distinct concrete bugs: offline model-info call vs missing `backend_tokenizer` attribute." + "reason": "Both hit _patch_mistral_regex, but one is an offline model_info call and the other is a tokenizer-API AttributeError. Same helper, different root causes." + }, + { + "left": "issue:43950", + "right": "issue:45987", + "accept": false, + "reason": "Both mention loading/state handling regressions, but one is buffer corruption on load and the other is a StaticCache return-type contract bug." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both concern config loading, but one is missing fields from remote config and the other is strict validation rejecting `granite_speech`." + "reason": "These are distinct config-loading issues: missing fields from AutoConfig with trust_remote_code versus strict config rejection for granite_speech." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Image segmentation map processing and trainer loss signature mismatch are unrelated." + "reason": "An image-segmentation processing failure and a Trainer compute_loss signature mismatch are unrelated." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Both are v5 output regressions, but they affect different models and different inference paths." + "reason": "Both are output regressions in v5, but they concern different models and different failure modes; no shared underlying bug is evident." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Different Qwen multimodal video bugs with different failure modes and model families." - }, - { - "left": "issue:43421", - "right": "issue:44811", - "accept": false, - "reason": "Special-token/post-processor update issue is unrelated to Whisper batch decode skipping special tokens." + "reason": "Both are video-input regressions in Qwen-family models, but they affect different model lines and likely different preprocessing/code paths." }, { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Qwen3 auto device-map CUDA assert and tokenizer loading for est-roberta are unrelated." + "reason": "Auto device mapping CUDA assertions and tokenizer-loading failures are unrelated." }, { - "left": "issue:43782", - "right": "issue:44977", + "left": "issue:43421", + "right": "issue:44811", "accept": false, - "reason": "Qwen3VL `weight_only=True` load failure is separate from Qwen3.5 flash-attention generation issues." + "reason": "Special-token/post-processor update behavior and Whisper batch_decode skipping special tokens are different tokenizer-layer problems." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "BLIP2 dtype propagation and Video-LLaVA temporal-attention/weight-sharing are different bugs." + "reason": "A dtype-loading issue in BLIP2 is not the same bug as a Video-LLaVA temporal-attention/weight-sharing problem." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Both involve Qwen models, but one is a from_pretrained weight_only error and the other is flash-attention generation behavior. Different code paths." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 doc-example errors and Kosmos2.5 long-OCR index error are not the same underlying problem." + "reason": "A SigLIP2 docs/example problem and a Kosmos2.5 OCR index error are not the same underlying issue." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "AutoImageProcessor URL loading and template compilation errors are unrelated." + "reason": "AutoImageProcessor URL loading and Jinja/template compilation errors are completely separate." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty-transcription generation and Kosmos2.5 long OCR input errors are different failure cases." - }, - { - "left": "issue:44186", - "right": "issue:44442", - "accept": false, - "reason": "LayoutLMv2 tokenizer padding/NER crash and FastSpeech2Conformer tokenizer loading failure are unrelated." + "reason": "Whisper empty-transcription generation and Kosmos2.5 long-OCR indexing are unrelated." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Non-persistent buffer corruption in `from_pretrained()` is unrelated to FSDP CPU RAM-efficient loading." + "reason": "Both concern loading regressions, but one is non-persistent buffer corruption and the other is broken FSDP CPU RAM efficient loading. Too different to merge." }, { - "left": "issue:43606", - "right": "issue:45072", + "left": "issue:44186", + "right": "issue:44442", "accept": false, - "reason": "Bark CPU-offload device mismatch and dtype mismatches in SwitchTransformers/TimmWrapperModel are distinct issues." + "reason": "LayoutLMv2 NER/padding crashes and AutoTokenizer failing to load a specific tokenizer are different tokenizer issues." }, { - "left": "issue:43577", - "right": "issue:43582", + "left": "issue:43606", + "right": "issue:45072", "accept": false, - "reason": "BLIP2 dtype propagation and Apple Silicon warmup TypeError are unrelated." + "reason": "Device-mismatch under CPU offload for Bark is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel inference." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "Video-input regression across Qwen models and Kosmos2.5 OCR index error are different bugs." + "reason": "A broad Qwen video-input regression and a Kosmos2.5 OCR index error do not look like the same bug." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "BarkProcessor voice preset failure and GenerationMixin import error are unrelated." + "reason": "A BarkProcessor voice_preset issue and a GenerationMixin import failure are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype loading and Apple Silicon caching_allocator_warmup TypeError are unrelated." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "Pipeline task deprecation messaging and SigLIP2 output quality issues do not share the same bug." + "reason": "Pipeline deprecation/error-message behavior and SigLIP2 nonsensical outputs are distinct issues." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "Mask shape IndexError and trainer evaluate-after-train failure are different code paths." + "reason": "A mask-shape IndexError and a trainer.evaluate() failure after training are not the same bug." }, { "left": "issue:44485", "right": "issue:44509", "accept": false, - "reason": "GLM-5 RoPE implementation discussion and docs still mentioning removed pipeline tasks are not the same change." + "reason": "GLM-5 RoPE implementation discussion and stale docs mentioning removed pipeline tasks are unrelated." } ] }, @@ -5445,214 +5481,208 @@ "issue:44561|issue:45020", "issue:45987|issue:46192", "issue:28218|issue:29159", - "issue:44610|issue:45072", "issue:44671|issue:45701", + "issue:44610|issue:45072", "issue:43720|issue:45072", "issue:43782|issue:45325", "issue:43825|issue:43827", "issue:43749|issue:44291", - "issue:43582|issue:44857", "issue:43906|issue:46209", + "issue:43582|issue:44857", "issue:45081|issue:45406", "issue:42175|issue:45081", "issue:45372|issue:46326", "issue:43262|issue:43819", "issue:45003|issue:46285", "issue:34634|issue:36331", - "issue:43653|issue:44912", "issue:44060|issue:44368", + "issue:43653|issue:44912", "issue:43526|issue:43697", "issue:43742|issue:45020", - "issue:43526|issue:44554", - "issue:45084|issue:45538", "issue:43650|issue:43906", + "issue:45084|issue:45538", + "issue:43526|issue:44554", "issue:43329|issue:45600", "issue:44589|issue:45042" ], - "prepared_review_unit_hash": "53336f22552f76ca132f6bec9c5b9255ac7334f152974966ac3770b211bda172", + "prepared_review_unit_hash": "26bb9c2a906d317100e840151c983dd40ddfe023533dcac5f5a13556dedb835c", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 26, - "serialized_chars": 35137, - "estimated_input_tokens": 8785, - "estimated_eval_tokens": 17826 + "serialized_chars": 35154, + "estimated_input_tokens": 8789, + "estimated_eval_tokens": 17834 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous; the soft-similarity pairs share keywords but usually describe different bugs, regressions, or model-specific failures rather than the same underlying change.", - "confidence": 0.88, - "canonical_issue_reason": "Issue 45020 is the broadest and most representative regression report in the set, centered on recent-version breakage for remote_code models.", + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenizer, model-loading, processor, cache, and CI issues. None of the soft pairs look like safe duplicates of the same underlying bug/change.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue is safe here; the items span unrelated failures across many subsystems, so picking one representative would be misleading.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45020 is the best umbrella issue here because it is user-facing, high-signal, and broad enough to represent the closest matching compatibility regressions in the cluster.", + "best_issue_reason": "No issue stands out as a global best representative because the cluster is not centered on one bug family.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44561", "right": "issue:45020", "accept": false, - "reason": "Both mention remote_code/v5 breakage, but 44561 is a specific missing-helper removal while 45020 is a broader compatibility report; not the same concrete bug." + "reason": "Both involve remote_code breakage, but one is specifically about removing is_torch_fx_available while the other is a broad recent-version regression; different root causes and fixes." }, { "left": "issue:45987", "right": "issue:46192", "accept": false, - "reason": "Different code paths: StaticCache return type vs. text_kwargs ignoring return_tensors." + "reason": "These are unrelated API/behavior bugs: StaticCache sequence length vs text_kwargs/return_tensors handling." }, { "left": "issue:28218", "right": "issue:29159", "accept": false, - "reason": "Both are tokenizer-related, but one is an added-token spacing bug and the other is a slow-vs-fast tokenizer inconsistency; too vague to be the same defect." + "reason": "Both touch tokenizer behavior, but one is about added-token spacing and the other is a broad slow/fast tokenizer inconsistency; too little evidence they are the same bug." }, { - "left": "issue:44610", - "right": "issue:45072", + "left": "issue:44671", + "right": "issue:45701", "accept": false, - "reason": "Different failures and models: OmDet-Turbo image size mismatch vs. SwitchTransformers/TimmWrapper dtype issues." + "reason": "CamemBERT masked-LM prediction regressions and a general tokenization change report are not the same concrete issue." }, { - "left": "issue:44671", - "right": "issue:45701", + "left": "issue:44610", + "right": "issue:45072", "accept": false, - "reason": "CamemBERT masked-LM predictions and generic tokenization changes are not the same bug." + "reason": "Completely different components and failure modes: image processor sizing vs bfloat16 dtype mismatch." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weight loading during accelerate is unrelated to bfloat16 dtype mismatches." + "reason": "Packed-weight loading in BitNet and dtype mismatches in inference are separate code paths with different fixes." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Different Qwen-VL regressions: weight_only loading error vs. rope index scaling bug." + "reason": "Both are vision-language related, but one is a from_pretrained weight_only load failure and the other is a rope-index position_id bug." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "One is a runtime error message about pipeline support, the other is documentation still referencing pipeline; related fallout, but not the same artifact or bug." + "reason": "Both relate to v5 pipeline removals, but one is an error-message wording problem and the other is stale docs; not the same bug/change." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "Both involve loading, but FSDP CPU RAM-efficient loading and init_empty_weights/_is_hf_initialized are separate mechanisms and failures." + "reason": "Both concern loading flows, but FSDP CPU RAM efficient loading and init_empty_weights argument handling are distinct issues." }, { - "left": "issue:43582", - "right": "issue:44857", + "left": "issue:43906", + "right": "issue:46209", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and CUDA float16 loss crash are unrelated." + "reason": "An isolated reproduction of an older issue is unrelated to FP8 inference on B200." }, { - "left": "issue:43906", - "right": "issue:46209", + "left": "issue:43582", + "right": "issue:44857", "accept": false, - "reason": "An isolated reproduction of an older issue is not the same as an FP8 inference failure on B200." + "reason": "Apple Silicon allocator warmup and CUDA float16 AMP crashes are unrelated hardware/backend bugs." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Mistral regex patch crash and Gemma4Processor missing _tokenizer are distinct tokenizer/processor load failures." + "reason": "Both are AttributeErrors, but they arise in different processors and different loading paths; not the same underlying defect." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Packaging backend dependency issue vs. tokenizer regex patch crash; no shared underlying bug." + "reason": "TensorFlow extra-install packaging and Mistral tokenizer regex patching are unrelated." }, { "left": "issue:45372", "right": "issue:46326", "accept": false, - "reason": "Both touch mistral_common-adjacent loading, but one is an import/dependency break and the other is continue_final_message behavior." + "reason": "Both mention mistral_common, but one is an import-time dependency break and the other is a chat-template behavior bug; different failures and fixes." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Audio chat-template sampling-rate default and DAC from_latents/STE mismatch are unrelated." + "reason": "Audio chat-template sampling-rate defaults and DAC latent reconstruction mismatch are unrelated." }, { "left": "issue:45003", "right": "issue:46285", "accept": false, - "reason": "Unsafe sys.modules access and CLIP text-model structure inconsistency are separate regressions." + "reason": "Unsafe sys.modules access and CLIP text-model structural inconsistency are different model utility issues." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "BarkProcessor voice_preset and CustomTrainer compute_loss are entirely different subsystems." + "reason": "BarkProcessor voice preset and CustomTrainer compute_loss signature mismatch are unrelated." }, { - "left": "issue:43653", - "right": "issue:44912", + "left": "issue:44060", + "right": "issue:44368", "accept": false, - "reason": "Tokenizer special-token registration and MXFP4 quantization fallback are unrelated." + "reason": "Both mention tied-weights warnings, but the affected models, triggers, and remedies differ; not enough to merge as one bug." }, { - "left": "issue:44060", - "right": "issue:44368", + "left": "issue:43653", + "right": "issue:44912", "accept": false, - "reason": "Both involve Qwen3 warnings, but one is a mistaken tied-weights mapping and the other is a LoRA warning about tie_word_embeddings; not the same defect." + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization loading are unrelated." }, { "left": "issue:43526", "right": "issue:43697", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels and RTDetrV2 output differences are unrelated." + "reason": "BeitImageProcessorFast label reduction and RTDetrV2 output drift are different model/processor regressions." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "MobileLLM key error while loading is not the same as broad remote_code breakage." + "reason": "A key error on one specific model and broad remote_code breakage are not the same concrete issue." }, { - "left": "issue:43526", - "right": "issue:44554", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "Different model families and failure modes: label reduction bug vs. MPS attention correctness issue." + "reason": "No meaningful overlap; one is a vague request, the other a reproduction of a separate issue." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template compilation error and CLIPTokenizer model_max_length behavior are unrelated." + "reason": "Template compilation TypeError and CLIPTokenizer max-length handling are unrelated tokenizer/model config bugs." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:43526", + "right": "issue:44554", "accept": false, - "reason": "\u201cADD THE DATA\u201d is not the same as an isolated reproduction issue; no shared bug." + "reason": "These concern different domains: image label reduction vs MPS attention correctness." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "Undefined video-branch variables in multimodal token counting and removed auto-mapping config references are different issues." - }, - { - "left": "issue:44589", - "right": "issue:45042", - "accept": false, - "reason": "Float8 storage lookup failure and PIL-backend torchvision dependency regression are unrelated." + "reason": "A multimodal-token counting bug and stale auto_mappings references are unrelated CI/runtime problems." } ] }, @@ -5728,19 +5758,19 @@ "soft_pairs": [ "issue:43746|issue:44811", "issue:43452|issue:44991", - "issue:44387|issue:44617", "issue:43638|issue:44936", + "issue:44387|issue:44617", "issue:43550|issue:44912", "issue:43819|issue:43927", "issue:43257|issue:43329", - "issue:36010|issue:43645", "issue:43854|issue:44977", - "issue:39692|issue:44912", + "issue:36010|issue:43645", "issue:43720|issue:44610", + "issue:39692|issue:44912", "issue:38175|issue:44610", "issue:36010|issue:36331", - "issue:43454|issue:43577", "issue:43531|issue:45042", + "issue:43454|issue:43577", "issue:45459|issue:45803", "issue:43257|issue:43696", "issue:42490|issue:43257", @@ -5756,14 +5786,14 @@ "issue:29159|issue:34567", "issue:42915|issue:43742" ], - "prepared_review_unit_hash": "644db3f35c4d6d77d6abfdb84ebf39f9f021319e7e2316b3c561294536743a67", + "prepared_review_unit_hash": "7ef9706be5904a0ee202ae0542ae0d7b6bdbab1aba0bbc0a5fc1c0d4280dc3b4", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, @@ -5778,186 +5808,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft pair looks like a true duplicate. Most are adjacent subsystem regressions, but they affect different code paths, models, or failure modes. I picked one high-signal issue as the best representative overall.", - "confidence": 0.36, - "canonical_issue_reason": "issue:44387 is a clear, broadly scoped regression report with strong discussion activity and an explicit, reproducible failure mode under int4 quantization in Transformers 5.x.", + "summary": "This set is heterogeneous: the soft pairs share broad subsystem vocabulary but not the same underlying bug or fixable PR. I would not merge any of the candidate pairs as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44387 is the strongest standalone issue here: well-scoped, broadly relevant, and likely to be easy to triage or search against.", + "best_issue_reason": "issue:43257 is the strongest standalone bug report here: it has a concrete Qwen3 MoE loading failure, a clear reproduction context, and the most actionable description among the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "Different domains: PEFT/local checkpoint loading vs Whisper batch decoding and skip_special_tokens handling." + "reason": "Different bugs and code paths: PEFT/local checkpoint loading vs Whisper batch decoding token handling." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model-loading failures, but for different mechanisms and targets: gguf_file loading vs a specific tokenizer regression for EMBEDDIA/est-roberta." + "reason": "Both involve loading, but one is gguf_file path handling for Auto* classes and the other is a tokenizer regression for a specific model." }, { - "left": "issue:44387", - "right": "issue:44617", + "left": "issue:43638", + "right": "issue:44936", "accept": false, - "reason": "Both mention OOM, but one is an int4 quantization memory regression and the other is a model-specific video OOM; not the same bug." + "reason": "Different trainer/runtime failures: DeepSpeed ZeRO-3 index error vs trainer.evaluate() breaking after train()." }, { - "left": "issue:43638", - "right": "issue:44936", + "left": "issue:44387", + "right": "issue:44617", "accept": false, - "reason": "Different trainer failures and different triggers: DeepSpeed ZeRO-3 index error vs evaluate-after-train behavior." + "reason": "Both are OOMs, but from different models and likely different memory paths; not the same bug." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "Unrelated issues: torch.compile/SDPA failure in Bamba vs MXFP4 loading fallback in git-oss-20b." + "reason": "Different features: torch.compile+SDPA on Bamba vs MXFP4 quantization fallback on gpt-oss." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "Different bug classes: DAC forward mismatch from missing STE vs DiaConfig losing custom token IDs after save/load." + "reason": "Unrelated model logic: DAC latent reconstruction mismatch vs DiaConfig token ID persistence." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is unrelated to the multimodal token-counting video branch bug." - }, - { - "left": "issue:36010", - "right": "issue:43645", - "accept": false, - "reason": "Both concern import/model initialization regressions, but they affect different APIs and failure paths." + "reason": "Different failures: Qwen3 MoE weight conversion under accelerate/deepspeed vs undefined helpers in a video token counting branch." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "Both are model-loading/generation issues, but one is a GLM-4.7-Flash test loading problem and the other is a flash-attention generation regression for Qwen3.5." + "reason": "Different models and symptoms: GLM-4.7-Flash test loading issue vs Qwen3.5 flash-attention generation bug." }, { - "left": "issue:39692", - "right": "issue:44912", + "left": "issue:36010", + "right": "issue:43645", "accept": false, - "reason": "Documentation/example errors are not the same as a quantization-loading fallback bug." + "reason": "Both mention loading/imports, but one is a missing GenerationMixin import and the other is a notebook/custom-model initialization regression." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to an image processor/model input size mismatch." + "reason": "Distinct subsystems: BitNet packed-weight loading vs OmDet-Turbo processor image-size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Docs example errors for SigLIP2 are not the same as the gpt-oss MXFP4 quantization fallback bug." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Different models and symptoms: zero probabilities in SigLIP2 vs OmDet-Turbo image size mismatch." + "reason": "Different models and symptoms: zero probabilities in SigLIP2 vs incorrect processor output size in OmDet-Turbo." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "ImportError for GenerationMixin and a custom trainer loss-signature TypeError are unrelated." + "reason": "Importing GenerationMixin is unrelated to the compute_loss signature change with num_items_in_batch." }, { - "left": "issue:43454", - "right": "issue:43577", + "left": "issue:43531", + "right": "issue:45042", "accept": false, - "reason": "Weight tying causing garbage generation is a different defect than dtype propagation staying float32 in BLIP2." + "reason": "Qwen3-MoE sliding-window behavior is unrelated to PIL backend image processors requiring torchvision." }, { - "left": "issue:43531", - "right": "issue:45042", + "left": "issue:43454", + "right": "issue:43577", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to PIL backend processors requiring torchvision." + "reason": "Both are model-loading/architecture issues, but one is lm_head weight tying and the other is dtype propagation for Blip2/QFormer." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both involve exception handling, but they are different code paths and different failure surfaces." + "reason": "Both involve exception handling, but they are different functions and different error-masking bugs." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Qwen3 MoE loading/weight conversion is unrelated to GPT-oss-20b CUDA OOM." + "reason": "Qwen3 MoE weight conversion is unrelated to a plain CUDA OOM on gpt-oss-20b." }, { "left": "issue:42490", "right": "issue:43257", "accept": false, - "reason": "Trailing-slash local model path warning is unrelated to Qwen3 MoE weight conversion under accelerate+deepspeed." + "reason": "Path-slash import warning and Qwen3 MoE deepspeed loading are different problems." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio chat-template sampling-rate defaults are unrelated to Whisper batch_decode special-token handling." + "reason": "Both are audio-related, but one is chat-template sampling-rate defaulting and the other is batch_decode skip_special_tokens handling." }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Missing fields from trust_remote_code AutoConfig and missing text_config unwrapping in AutoModelForCausalLM are different config-loading bugs." + "reason": "Different config-loading bugs: missing fields with trust_remote_code vs not unwrapping text_config for composite multimodal configs." }, { "left": "issue:43329", "right": "issue:44479", "accept": false, - "reason": "Both are video-related, but one is a specific undefined-variable bug in multimodal token counting while the other is a broader video-input regression across Qwen models." + "reason": "Both are video-related, but one is undefined helper usage in multimodal token counting and the other is a broader video input regression." }, { "left": "issue:43329", "right": "issue:43450", "accept": false, - "reason": "A token-counting video-branch bug is not the same as batched video processor shape handling." + "reason": "Undefined video token helper references are not the same as batched video processor shape errors." }, { "left": "issue:43606", "right": "issue:45859", "accept": false, - "reason": "Both mention CPU offload, but they affect different models and code paths." + "reason": "Both mention CPU offload, but they are different models and different device-mismatch failures." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "TrainerState token-count tracking is unrelated to GPT-oss-20b perplexity on WikiText-2." + "reason": "TrainerState token counting is unrelated to GPT-oss perplexity behavior." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "Both are loading/memory issues, but one is FSDP CPU RAM efficient loading and the other is a Strix Halo mmap/OOM workaround." + "reason": "Different memory-related issues: FSDP CPU RAM efficient loading vs mmap on Strix Halo." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Qwen3 MoE loading with accelerate+deepspeed and Qwen3-VL weight-shape mismatch are different load failures." + "reason": "Both are Qwen3-family loading issues, but the failure modes are different: deepspeed conversion vs VL weight-shape mismatch." }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Tokenizer special-token insertion regression and Kimi-K2.5 codec/warning regression are distinct tokenizer bugs." + "reason": "Tokenizer regressions, but on different models and different root causes (special-token insertion vs codec handling/warning behavior)." }, { "left": "issue:29159", "right": "issue:34567", "accept": false, - "reason": "Tokenizer consistency issues are unrelated to TrainerState input-token accounting." + "reason": "Tokenizer slow/fast inconsistency is unrelated to TrainerState.num_input_tokens_seen not updating." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "FineGrainedFP8Config failure for Qwen3Moe and a key error loading MobileLLM-125M are different model-loading bugs." + "reason": "Different model/config loading failures with no shared underlying code path." } ] }, @@ -6034,9 +6064,9 @@ "issue:44291|issue:45042", "issue:42915|issue:44821", "issue:43452|issue:43582", - "issue:43927|issue:44912", "issue:44849|issue:45701", "issue:44987|issue:45538", + "issue:43927|issue:44912", "issue:43697|issue:45357", "issue:42915|issue:45538", "issue:44568|issue:45488", @@ -6051,231 +6081,243 @@ "issue:45137|issue:45405", "issue:43540|issue:44560", "issue:24643|issue:29159", - "issue:43577|issue:45072", "issue:43742|issue:44291", "issue:44987|issue:45907", + "issue:43577|issue:45072", "issue:28282|issue:29127", "issue:42915|issue:43749", "issue:43257|issue:43756", - "issue:43582|issue:43782", "issue:38175|issue:44898", "issue:42175|issue:44488", + "issue:43582|issue:43782", "issue:44737|issue:45588" ], - "prepared_review_unit_hash": "dda1b2c4c060e2552790ef72f7a3b240b13a719280b76428e8422c01c4bf7a2a", + "prepared_review_unit_hash": "fea73ee08d838d48913f2cc00a4a2629278fe8e9b632f6a7e4c58f18f649ffd6", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36742, - "estimated_input_tokens": 9186, - "estimated_eval_tokens": 18628 + "serialized_chars": 36747, + "estimated_input_tokens": 9187, + "estimated_eval_tokens": 18630 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated issue reports spanning tokenization, vision/video, loading, dtype, and distributed-training bugs. Only one soft pair looks like the same underlying bug: the Qwen video-input regression/position-id problem.", - "confidence": 0.87, - "canonical_issue_reason": "No single cluster-wide canonical issue exists because the set is heterogeneous; if forced, issue 44479 is the best representative of the only clear subcluster since it describes the broad Qwen video regression that also covers 45381.", + "summary": "Most pairs are false positives: they share a broad subsystem (tokenizers, loading, video, dtype, quantization) but not the same concrete bug. The only strong duplicate-looking pair is the Qwen video-input regression pair.", + "confidence": 0.9, + "canonical_issue_reason": "issue:44479 is the broadest and most central report in the only plausible duplicate pair, and it directly covers the specific qwen2.5-vl video-position-id symptom described in issue:45381.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the strongest standalone representative: it is a broad, user-facing regression report, and it overlaps the only near-duplicate issue in the set.", + "best_issue_reason": "issue:44479 is the best representative issue for this set because it captures the shared Qwen video-input regression theme more broadly than the more specific follow-up report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44291", "right": "issue:45042", "accept": false, - "reason": "Different failures: empty-weights loading arg mismatch vs torchvision dependency in PIL image processors." + "reason": "Different bugs: init_empty_weights/_is_hf_initialized loading error vs image-processor torchvision dependency regression." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Both are loading-related, but one is FineGrainedFP8/Qwen3Moe and the other is AutoImageProcessor URL handling." + "reason": "Different failures: FineGrainedFP8Config model loading vs AutoImageProcessor URL loading." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "GGUF tokenizer/model loading bug vs Apple Silicon caching_allocator_warmup TypeError; unrelated code paths." - }, - { - "left": "issue:43927", - "right": "issue:44912", - "accept": false, - "reason": "DiaConfig save/load token-ID bug and MXFP4 quantization fallback are separate model/config issues." + "reason": "Unrelated code paths: gguf_file tokenizer/model loading vs Apple Silicon warmup TypeError." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "Both mention tokenization behavior, but one is a Qwen hidden-states bug and the other is a generic version-dependent tokenization change." + "reason": "Different scope: Qwen3.5 hidden-states bug vs a broad tokenization behavior change report." }, { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Model loading failure for physical-intelligence/fast and CLIPTokenizer model_max_length are unrelated." + "reason": "Different issues: model loading for physical-intelligence/fast vs CLIPTokenizer model_max_length handling." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "Different bugs: DiaConfig save/load token IDs vs MXFP4 quantization fallback." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "RTDetrV2 output regression and Qwen3.5 save_pretrained key mismatch are different model-specific bugs." + "reason": "Different model behaviors: RTDetrV2 output mismatch vs Qwen3.5 save_pretrained key regression." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "Different bugs: Qwen3Moe FP8 loading vs CLIPTokenizer max_length misconfiguration." + "reason": "Different bugs and models: Qwen3Moe FP8 config failure vs CLIPTokenizer max length issue." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Both are tokenizer regressions, but one is add_special_tokens BOS/EOS handling and the other is LlamaTokenizer overriding tokenizer.json pre-tokenizer." + "reason": "Both are tokenizer regressions, but one is add_special_tokens behavior and the other is LlamaTokenizer pre-tokenizer override; not the same bug." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Different Qwen runtime failures: auto device-mapping CUDA assert vs video-input ValueError." + "reason": "Different failures: auto device mapping cuda assert vs video-input ValueError in Qwen3OmniMoe." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "CLIP MacOS bus error and TrainerState token counter not updating are unrelated." + "reason": "Completely different symptoms: Mac bus error on CLIP vs TrainerState token-count update bug." }, { "left": "issue:44479", "right": "issue:45381", "accept": true, - "reason": "Both describe the same Qwen video-input regression in 5.3.0, centered on wrong video position-id handling." + "reason": "Same underlying Qwen video-input regression; 45381 is a specific qwen2.5-vl symptom within the broader regression reported in 44479." }, { - "left": "issue:42175", - "right": "issue:43824", + "left": "issue:45137", + "right": "issue:45405", "accept": false, - "reason": "Backend packaging omission for TensorFlow is unrelated to the Qwen2.5-VL import error." + "reason": "Different areas: DeepSpeed ZeRO3 deque error vs dependency-version bump for PEFT." }, { - "left": "issue:43317", - "right": "issue:45538", + "left": "issue:43742", + "right": "issue:44291", "accept": false, - "reason": "Device-map/offload dequantization failure and tokenizer max-length bug are unrelated." + "reason": "Both are loading-related, but one is a key error for MobileLLM and the other is an unexpected constructor argument in init_empty_weights." }, { - "left": "issue:43540", - "right": "issue:44479", + "left": "issue:44987", + "right": "issue:45907", "accept": false, - "reason": "Both involve video inputs, but the model families and failure modes differ; not the same concrete bug." + "reason": "Unrelated: a model-loading failure vs a repo-files helper not forwarding revision." }, { - "left": "issue:44062", - "right": "issue:44991", + "left": "issue:43577", + "right": "issue:45072", "accept": false, - "reason": "AddedToken special-argument error and tokenizer loading failure for est-roberta are different issues." + "reason": "Both concern dtype, but they affect different models and code paths; not one concrete bug." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "PyTorch-missing ImportError and Llama 3.2 Vision model-loading regression are unrelated." + "reason": "Different problems: missing PyTorch import detection vs a specific Llama Vision model-loading regression." }, { - "left": "issue:45137", + "left": "issue:43749", "right": "issue:45405", "accept": false, - "reason": "DeepSpeed ZeRO3 deque underflow is unrelated to the PEFT minimum-version bump." + "reason": "Unrelated: FSDP CPU RAM loading breakage vs minimum PEFT version metadata issue." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both are video-related, but one is Qwen3OmniMoe input processing and the other is Qwen3-vl-embedding StopIteration." + "reason": "Both mention video inputs, but the models and failure modes differ; not enough evidence of one shared bug." }, { "left": "issue:24643", "right": "issue:29159", "accept": false, - "reason": "DeepSpeed weight-dimension error and tokenizer slow/fast inconsistency are unrelated." + "reason": "Different domains: DeepSpeed training runtime error vs tokenizer inconsistency." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Different loading failures with different root causes; no shared code-path bug shown." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Same repo surface area, but distinct bugs and fixes." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both involve dtype, but the affected models and the underlying mismatch differ." + "reason": "Distinct dtype issues on different model families; not mergeable as one PR." }, { - "left": "issue:43742", - "right": "issue:44291", + "left": "issue:42175", + "right": "issue:43824", "accept": false, - "reason": "MobileLLM key error and init_empty_weights unexpected argument are distinct loading bugs." + "reason": "Different packaging/import issues: missing tensorflow backend vs missing Qwen2_5_VL class export." }, { - "left": "issue:44987", - "right": "issue:45907", + "left": "issue:44291", + "right": "issue:45042", "accept": false, - "reason": "Model load failure and list_repo_files revision forwarding are different hub-related problems." + "reason": "Different loading paths and different symptoms; no same underlying bug." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "PyTorch import failure and LayoutLMv3 box-content error are unrelated." + "reason": "Missing PyTorch import vs a clarity/improved error-message request for LayoutLMv3." }, { - "left": "issue:42915", - "right": "issue:43749", + "left": "issue:41720", + "right": "issue:44560", "accept": false, - "reason": "Qwen3Moe FP8 loading and FSDP CPU RAM efficient loading are different distributed/quantization issues." + "reason": "Different Qwen video/device failures with different stacks and symptoms." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Qwen3 MoE weight conversion with accelerate+deepspeed and Smollm3 RoPE-layer mismatch are unrelated." - }, - { - "left": "issue:43582", - "right": "issue:43782", - "accept": false, - "reason": "Apple Silicon TypeError in warmup and Qwen3VL weight_only loading error are different failures." + "reason": "MOE weight conversion with accelerate/deepspeed is unrelated to Smollm3 RoPE layer dropping." }, { "left": "issue:38175", "right": "issue:44898", "accept": false, - "reason": "Different vision-model bugs: zero probabilities in SigLIP2 vs Perceiver non-default-resolution failure." + "reason": "Both are vision-model issues, but one is zero probabilities in SigLIP2 and the other is non-default-resolution Perceiver classification failure." }, { "left": "issue:42175", "right": "issue:44488", "accept": false, - "reason": "Packaging/backend dependency issue and a model loading failure for cjvt/sleng-bert are unrelated." + "reason": "Packaging/backend availability issue vs a model-loading failure; unrelated." + }, + { + "left": "issue:43582", + "right": "issue:43782", + "accept": false, + "reason": "Different bugs: Apple Silicon warmup TypeError vs Qwen3VL weight_only loading error." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "XLNet CPU-device inefficiency and flash_attention crash on sink-less models are separate code paths." + "reason": "Different code paths: XLNet CPU tensor creation vs flash-attention sink handling." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft edge, issue:44479 with issue:45381, is well-supported by the titles and both point to the same Qwen video-input / position-id regression. The other verdicts are phrased cautiously and do not overclaim overlap. The representative-issue rationale is also consistent with the mostly heterogeneous set." + "feedback": "Grounded in the packet: the only accepted pair is the clearly related Qwen video-input regression/symptom pair, and the other verdicts stay conservative by rejecting only when titles point to different model families or code paths. No obvious evidence overstatement." }, "evaluator_used": true, "retried": false, @@ -6306,7 +6348,6 @@ "issue:43493", "issue:43526", "issue:43531", - "issue:43577", "issue:43643", "issue:43697", "issue:43720", @@ -6343,7 +6384,8 @@ "issue:45920", "issue:45987", "issue:46050", - "issue:46121" + "issue:46121", + "issue:46202" ], "soft_pairs": [ "issue:43493|issue:43994", @@ -6352,8 +6394,8 @@ "issue:43526|issue:45479", "issue:44291|issue:44821", "issue:43756|issue:45325", - "issue:43906|issue:45381", "issue:44062|issue:45081", + "issue:43906|issue:45381", "issue:44913|issue:46121", "issue:44792|issue:44964", "issue:45042|issue:45081", @@ -6370,202 +6412,202 @@ "issue:41762|issue:44936", "issue:44493|issue:46050", "issue:44514|issue:45381", - "issue:43452|issue:43577", + "issue:43906|issue:46202", "issue:44466|issue:45005" ], - "prepared_review_unit_hash": "c043df886d02f1eddecaf96b0c5ae11cb92fbd783aea6c52c9066d103de42503", + "prepared_review_unit_hash": "a685fa590df622e5b75e2bba4a4a9c0d04bec551cf34a655ec936342c8184131", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 36005, - "estimated_input_tokens": 9002, - "estimated_eval_tokens": 18260 + "serialized_chars": 35939, + "estimated_input_tokens": 8985, + "estimated_eval_tokens": 18226 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a mix of unrelated Transformers issues. The only well-grounded duplicate pair in the packet is the v5 tied-weights/lm_head serialization regression. The SigLIP2 pair is only title-level similar and lacks enough body evidence to call it a duplicate, so it should remain a tentative similarity rather than an accepted merge candidate.", - "confidence": 0.64, - "canonical_issue_reason": "issue:44466 is the best canonical issue because it describes a concrete, broad v5 tied-weights/lm_head serialization regression and can subsume the more specific translation-model report.", + "summary": "The cluster is mostly a grab-bag of transformer v5/model-loading regressions, but only two soft pairs look like true duplicates: SigLIP2 output mismatch and tied-weight serialization/regression issues. Most other pairs share a subsystem name only and should stay separate.", + "confidence": 0.77, + "canonical_issue_reason": "issue:44466 is the clearest representative of the cluster\u2019s strongest concrete theme: v5 tied-weight / `lm_head.weight` serialization behavior with a reproducible regression and broad model impact.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44466 is the strongest global issue representative: it has the clearest underlying defect, broadest scope among the plausible duplicates, and the best evidence for a real duplicate relationship.", + "best_issue_reason": "issue:44466 is the best single issue to anchor this cluster because it is specific, regression-oriented, and matches one of the few genuinely duplicate-looking pairs.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43493", "right": "issue:43994", - "accept": false, - "reason": "The titles are similar, but the packet does not show enough issue-body evidence that they share the same root cause; this is better treated as tentative similarity, not a duplicate." + "accept": true, + "reason": "Both describe SigLIP2 producing wrong outputs versus the expected implementation; same underlying model-behavior bug." }, { "left": "issue:43404", "right": "issue:45987", "accept": false, - "reason": "Different subsystems and symptoms: Mistral weight tying versus a TrainerState token counter not updating." + "reason": "One is a Mistral3 tied-weight bug; the other is a TrainerState counter not updating. Different code paths and symptoms." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "Trainer state bookkeeping bug versus a custom trainer signature incompatibility; not the same defect." + "reason": "Trainer token-count tracking vs a custom `compute_loss` signature break. Related training area, but not the same bug." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BEiT image label reduction bug is unrelated to the zero-loss single-label classification issue." + "reason": "BEiT image-label reduction bug versus sequence-classification zero-loss behavior. Different components and failure modes." }, { "left": "issue:44291", "right": "issue:44821", "accept": false, - "reason": "Different loading paths and failures: init_empty_weights argument handling versus loading AutoImageProcessor from URL." + "reason": "`init_empty_weights` loading error versus loading `AutoImageProcessor` from a URL. Not the same underlying issue." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Both mention RoPE, but they affect different models and different computations; not the same bug." + "reason": "Both involve RoPE/position handling, but they affect different models and different logic paths." }, { - "left": "issue:43906", - "right": "issue:45381", + "left": "issue:44062", + "right": "issue:45081", "accept": false, - "reason": "Too little evidence that the isolated reproduction of #38071 is the same as the Qwen2.5-VL video position-id bug." + "reason": "Tokenizer `AddedToken` duplication error versus a Mistral regex patch crash. Different tokenizer bugs." }, { - "left": "issue:44062", - "right": "issue:45081", + "left": "issue:43906", + "right": "issue:45381", "accept": false, - "reason": "Tokenizer AddedToken special-argument error and Mistral regex patch crash are distinct tokenizer-loading issues." + "reason": "The first is only an isolated reproduction of another issue; the second is a Qwen2.5-VL video position-id bug. Not clearly the same defect." }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "Both are rope/config related, but one is silent config reload loss and the other is a TypeError in rope validation conversion." + "reason": "Config reload defaulting and rope-validation list handling are separate configuration bugs." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Different multimodal models and different failure modes; not the same underlying issue." + "reason": "Janus image-generation test failure versus Phi-4 multimodal load failure. Different model-specific regressions." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "Image processor dependency handling is unrelated to Mistral tokenizer regex patching." + "reason": "Image-processor torchvision dependency regression versus tokenizer regex patch crash. Different subsystems and fixes." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both involve loading failures, but the titles do not indicate the same root cause or code path." + "reason": "GGUF loading breakage versus a specific model that also fails to load. Same broad loading domain, but not the same concrete bug." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "KeyError on MobileLLM loading and missing Float8 storage object are different loading failures." + "reason": "MobileLLM load key error versus missing Float8 storage type. Distinct runtime failures." }, { "left": "issue:45357", "right": "issue:45920", "accept": false, - "reason": "Qwen3.5 checkpoint serialization and AutoTokenizer token-ID regression are different problems." + "reason": "Incorrect save_pretrained visual encoder keys versus wrong token IDs from AutoTokenizer. Different serialization/tokenization bugs." }, { "left": "issue:43531", "right": "issue:44821", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to loading an AutoImageProcessor from a URL." + "reason": "Qwen3-MoE sliding-window behavior versus AutoImageProcessor URL loading. No meaningful bug match." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "PEFT version bump and auto_mappings CI breakage are unrelated maintenance issues." + "reason": "Unreleased PEFT version bump versus broken auto_mappings references. CI/package-version issue versus code-reference cleanup." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "General remote-code AutoConfig field loss and a specific Qwen config field being dropped by @strict are not clearly the same defect." + "reason": "Generic remote-code config field loss versus a specific Qwen3.5MoEVision config field being dropped. Similar area, but not the same bug." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Qwen3.5 num_labels propagation and strict loading blocking granite_speech are different config-loading issues." + "reason": "Qwen3.5 `num_labels` propagation bug versus granite_speech strict-config loading failure. Different config issues." }, { "left": "issue:24643", "right": "issue:29127", "accept": false, - "reason": "DeepSpeed training weight-shape error and LayoutLMv3 box-validation message quality are unrelated." + "reason": "DeepSpeed training runtime error versus LayoutLMv3 error-message clarity issue. Not the same bug." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "RTDetrV2 output mismatch and Qwen3.5 save_pretrained regression are different model-specific regressions." + "reason": "RTDetrV2 output regression versus Qwen3.5 save_pretrained checkpoint regression. Separate models and code paths." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "BitNet accelerate-loading/unpacked-weight bug is unrelated to Perceiver non-default-resolution failure." + "reason": "BitNet packed-weight loading bug versus Perceiver interpolation failure. Same release window, different defects." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "Gemma3 DeepSpeed ZeRO-3 loading crash and trainer.evaluate() after train() are different workflows and failures." + "reason": "Gemma3 ZeRO-3 load index error versus trainer.evaluate() failing after training. Different failure surfaces." }, { "left": "issue:44493", "right": "issue:46050", "accept": false, - "reason": "Unexpected position-id keys during loading and sinq quantization save/load breakage do not look like the same issue." + "reason": "Unexpected position-id keys in many models versus sinq quantization save/load breakage. Only loosely related by serialization." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen2.5-VL, but one is batched chat-template padding behavior and the other is video position-id computation." + "reason": "Chat-template batching crash versus video position-id bug. Both Qwen2.5-VL, but not the same code-path problem." }, { - "left": "issue:43452", - "right": "issue:43577", + "left": "issue:43906", + "right": "issue:46202", "accept": false, - "reason": "gguf_file loading breakage and Blip2 dtype preservation are unrelated issues." + "reason": "An isolated reproduction note versus a Gemma pipeline roundtrip issue. No clear duplicate relationship." }, { "left": "issue:44466", "right": "issue:45005", "accept": true, - "reason": "Both describe the same tied-weights/lm_head serialization regression in v5, with 45005 being a broader manifestation on translation models." + "reason": "Both are about tied weights / `lm_head.weight` serialization regressions in v5, and could plausibly be fixed by one change." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The tied-weights duplicate (44466 \u2194 45005) is well grounded, but the accepted SigLIP2 pair (43493 \u2194 43994) is too strong for the evidence in the packet. We only have title-level similarity and no issue-body evidence showing the same root cause, so it should be treated as a tentative similarity rather than an accepted duplicate. Overall the summary is mostly conservative, but this one soft-edge accept overstates the available evidence." + "accept": true, + "feedback": "The summary stays grounded in the packet, and the soft-edge verdicts are mostly conservative. The two accepted pairs are defensible from the titles alone (SigLIP2 output discrepancy; tied-weight / lm_head serialization regression), while the rest are appropriately rejected as different bug surfaces. No mergeability issue applies here because these are issue pairs, not PR pairs." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -6603,14 +6645,15 @@ "issue:43742", "issue:43749", "issue:43819", - "issue:43906", "issue:43927", "issue:44361", "issue:44442", "issue:44464", + "issue:44479", "issue:44484", "issue:44514", "issue:44568", + "issue:44625", "issue:44704", "issue:44743", "issue:44811", @@ -6629,30 +6672,29 @@ "issue:45823", "issue:46002", "issue:46170", - "issue:46178", - "issue:46202" + "issue:46178" ], "soft_pairs": [ - "issue:43906|issue:46202", + "issue:43452|issue:43577", "issue:43688|issue:43749", "issue:43452|issue:44843", "issue:45372|issue:46170", "issue:30064|issue:33666", "issue:28282|issue:32090", - "issue:43819|issue:44811", "issue:44936|issue:45823", + "issue:43819|issue:44811", "issue:28282|issue:33357", "issue:43295|issue:45356", - "issue:43653|issue:44361", "issue:28282|issue:35141", + "issue:43653|issue:44361", "issue:44484|issue:44704", "issue:44361|issue:44442", "issue:38175|issue:43295", "issue:45542|issue:46178", "issue:44464|issue:44743", - "issue:44514|issue:45290", - "issue:43577|issue:43697", "issue:43653|issue:44568", + "issue:43577|issue:43697", + "issue:44514|issue:45290", "issue:29159|issue:33357", "issue:44843|issue:44991", "issue:43646|issue:43749", @@ -6663,222 +6705,229 @@ "issue:45491|issue:46002", "issue:43688|issue:45701", "issue:44464|issue:44977", - "issue:43550|issue:43927" + "issue:43550|issue:43927", + "issue:44479|issue:44625" ], - "prepared_review_unit_hash": "eff42a47b59f77fcdfc6a2dfb1e144ba900f416b3ee0720ab8ba5fa511e23645", + "prepared_review_unit_hash": "4ccfd8d4285794da75d7e74380410aff3cc0d8d412f53c4dcfa75eb943e41e8b", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 37083, - "estimated_input_tokens": 9271, - "estimated_eval_tokens": 18798 + "soft_pair_count": 32, + "serialized_chars": 37414, + "estimated_input_tokens": 9354, + "estimated_eval_tokens": 18964 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose mix of Transformers issues, mostly around tokenization/loading regressions but with several unrelated model/trainer/runtime bugs. I do not see any soft pair that confidently describes the same underlying bug or change, so the edges should be rejected.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45701 is the broadest tokenizer-regression report in the set and has the most downstream references, making it the best umbrella representative for this heterogeneous cluster.", + "summary": "The set is mostly a grab bag of unrelated Transformers bugs. There is a loose tokenizer/processor regression theme, but none of the proposed soft pairs look like the same concrete defect, so all soft edges are rejected.", + "confidence": 0.7, + "canonical_issue_reason": "issue:45701 is the broadest and most central tokenizer-regression report in the set, and it already attracts other related reports via references.", "canonical_pr_reason": null, - "best_issue_reason": "45701 is the most general and discussion-heavy issue here, so it is the best single issue to keep as the representative triage target.", + "best_issue_reason": "issue:45701 is the best representative issue for the cluster\u2019s loose tokenizer-change theme because it is broad, still open, and closer to an umbrella report than the narrower model-specific bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43906", - "right": "issue:46202", + "left": "issue:43452", + "right": "issue:43577", "accept": false, - "reason": "Both involve tokenization/roundtrip behavior, but one is an isolated reproduction of an older issue and the other is a Gemma pipeline-specific failure; not clearly the same root bug." + "reason": "Different subsystems: tokenizer/model loading with gguf_file vs Blip2 dtype propagation." }, { "left": "issue:43688", "right": "issue:43749", "accept": false, - "reason": "Different problems: aux-loss normalization in MoE models versus broken FSDP CPU RAM-efficient loading." + "reason": "Different bugs: auxiliary loss normalization vs FSDP CPU RAM efficient loading." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "Both affect loading paths, but one is gguf-based AutoTokenizer/AutoModel loading and the other is an offline model_info call in a Mistral regex patch." + "reason": "Both involve loading, but one is gguf_file support and the other is offline model_info handling in tokenizer patching." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "Different failure modes and platforms: Gemma processor import error versus Windows/ROCm torch.distributed dependency issues." + "reason": "Unrelated failures: Gemma processor import error vs Windows ROCm torch.distributed dependency issue." }, { "left": "issue:30064", "right": "issue:33666", "accept": false, - "reason": "Image processor segmentation-map handling and Qwen2-VL multi-GPU training are unrelated." + "reason": "Image processor void-segmentation failure is unrelated to Qwen2-VL multi-GPU training." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "PyTorch-missing import error and Trainer NoneType broadcast error are distinct failures." - }, - { - "left": "issue:43819", - "right": "issue:44811", - "accept": false, - "reason": "DAC latent conversion mismatch and Whisper batch_decode skip_special_tokens handling are different code paths and bugs." + "reason": "PyTorch missing import error and trainer NoneType broadcast error are different code paths." }, { "left": "issue:44936", "right": "issue:45823", "accept": false, - "reason": "Trainer post-train evaluate failure and Gemma4 PLE device mismatch are separate runtime issues." + "reason": "Trainer evaluate-after-train issue is unrelated to Gemma4 device-map mismatch." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC latent conversion mismatch is unrelated to Whisper batch_decode special-token handling." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "Missing PyTorch import error and MacOS bus error in CLIP are not the same bug." + "reason": "Missing PyTorch import and MacOS CLIP bus error are distinct problems." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer/processor regressions, but they target different model families and different breakages." + "reason": "Both are tokenizer regressions, but one is processor.tokenizer/image passing and the other is Kimi codec handling; not the same bug." }, { - "left": "issue:43653", - "right": "issue:44361", + "left": "issue:28282", + "right": "issue:35141", "accept": false, - "reason": "BigBird mask-token decoding and MLuke task AttributeError are unrelated tokenizer issues." + "reason": "ImportError for PyTorch and embedding reinitialization after resize are unrelated." }, { - "left": "issue:28282", - "right": "issue:35141", + "left": "issue:43653", + "right": "issue:44361", "accept": false, - "reason": "Import-time PyTorch absence and embedding reinit after resize are unrelated." + "reason": "Different tokenizer bugs affecting different models and failure modes." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "Save_pretrained shard-size behavior and AutoProcessor kwargs propagation are different API concerns." + "reason": "save_pretrained shard size question is unrelated to AutoProcessor kwargs forwarding." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "MLukeTokenizer AttributeError and AutoTokenizer failing on FastSpeech2ConformerTokenizer are separate tokenizer-loading problems." + "reason": "MLukeTokenizer AttributeError and FastSpeech2ConformerTokenizer loading failure are not the same defect." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Siglip2 zero probabilities and processor/tokenizer regression are different symptoms and code paths." + "reason": "SigLIP2 zero probabilities are unrelated to the processor.tokenizer regression." }, { "left": "issue:45542", "right": "issue:46178", "accept": false, - "reason": "Both are backend-related, but tensorboard-without-TensorFlow and tensorflow_text backend mapping are distinct issues." + "reason": "Both touch TF backend detection, but one is missing TensorFlow with tensorboard installed and the other is an unmapped tensorflow_text backend." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Chunked generation with compiled forward and recurrent-state reset under cache are different generation bugs." + "reason": "Compiled chunked generation inconsistency and recurrent-state reset under cache are separate generation bugs." }, { - "left": "issue:44514", - "right": "issue:45290", + "left": "issue:43653", + "right": "issue:44568", "accept": false, - "reason": "Both hit apply_chat_template, but one is batching/padding=False and the other is assistant tool-call content handling; not clearly one bug." + "reason": "Both are special-token issues, but they affect different tokenizers and different behaviors." }, { "left": "issue:43577", "right": "issue:43697", "accept": false, - "reason": "Mixed-precision dtype loading for Blip2 and RTDetrV2 output drift are unrelated." + "reason": "Blip2 dtype loading and RTDetrV2 output divergence are unrelated." }, { - "left": "issue:43653", - "right": "issue:44568", + "left": "issue:44514", + "right": "issue:45290", "accept": false, - "reason": "Both involve tokenizer special tokens, but BigBird mask registration and mdeberta BOS/EOS omission are different failures." + "reason": "Both are apply_chat_template crashes, but the triggers differ: batched padding=False vs tool-call messages with no content." }, { "left": "issue:29159", "right": "issue:33357", "accept": false, - "reason": "Slow/fast tokenizer inconsistency and MacOS bus error with CLIP are not the same underlying bug." + "reason": "Slow/fast tokenizer inconsistency is unrelated to the CLIP MacOS bus error." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Offline model_info call in AutoTokenizer and tokenizer loading failure for EMBEDDIA/est-roberta are different breakages." + "reason": "Both involve tokenizer loading, but one is HF_HUB_OFFLINE/model_info patching and the other is a specific EMBEDDIA tokenizer regression." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Custom model initialization breakage and FSDP loading breakage are separate regression reports." + "reason": "Custom model initialization breakage and FSDP loading regression are different issues." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "Aux-loss normalization in MoE models and T5 RMSNorm memory leak are unrelated." + "reason": "Aux-loss normalization and T5 apex RMSNorm memory leak are unrelated." }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error are different model-specific bugs." + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error are different bugs." }, { "left": "issue:43688", "right": "issue:45440", "accept": false, - "reason": "OLMoE/GPT Oss aux-loss normalization and DeepseekV3MoE divergence are different MoE issues with different causes." + "reason": "Aux-loss normalization is unrelated to DeepseekV3MoE implementation divergence." }, { "left": "issue:43653", "right": "issue:43927", "accept": false, - "reason": "BigBird special-token registration and DiaConfig custom token IDs after save/load are different tokenizer/config bugs." + "reason": "BigBird special-token registration and DiaConfig custom token IDs are distinct tokenizer/config bugs." }, { "left": "issue:45491", "right": "issue:46002", "accept": false, - "reason": "Both mention NaNs, but Gemma3 sliding-window all-padding windows and classification-head init/checkpoint NaNs are not the same failure." + "reason": "Gemma3 NaNs from sliding-window attention are unrelated to classification head NaNs on checkout/init." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "OLMoE/GPT Oss aux-loss normalization is unrelated to the broad tokenization regression theme of 45701." + "reason": "Aux-loss normalization does not match the broad tokenization-change report." }, { "left": "issue:44464", "right": "issue:44977", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency and Qwen3.5 flash-attention generation issues are distinct model/runtime bugs." + "reason": "Compiled chunked generation inconsistency and flash-attention generation failure are different code paths." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba and DiaConfig token-ID loss are unrelated." + "reason": "Bamba torch.compile/SDPA failure is unrelated to DiaConfig token-ID loss." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Video-input regression in Qwen VL models is unrelated to Qwen3.5 num_labels propagation." } ] }, @@ -6934,7 +6983,6 @@ "issue:44451", "issue:44464", "issue:44466", - "issue:44479", "issue:44554", "issue:44561", "issue:44610", @@ -6942,7 +6990,9 @@ "issue:44821", "issue:44849", "issue:44913", + "issue:44964", "issue:44977", + "issue:45405", "issue:45588", "issue:45823", "issue:46002", @@ -6951,25 +7001,24 @@ "issue:46209" ], "soft_pairs": [ - "issue:44479|issue:44625", "issue:41720|issue:44464", - "issue:43550|issue:44361", "issue:43643|issue:44561", + "issue:43550|issue:44361", "issue:44977|issue:46209", - "issue:43577|issue:45588", "issue:32090|issue:34567", + "issue:43577|issue:45588", "issue:43577|issue:44554", "issue:32090|issue:33666", "issue:44466|issue:46050", "issue:43504|issue:43720", - "issue:43504|issue:43550", "issue:33666|issue:36010", + "issue:43504|issue:43550", "issue:43504|issue:44610", "issue:43742|issue:44821", "issue:45823|issue:46050", "issue:43262|issue:43295", - "issue:43720|issue:43927", "issue:43452|issue:44451", + "issue:43720|issue:43927", "issue:43606|issue:43927", "issue:43994|issue:46192", "issue:44849|issue:46209", @@ -6981,222 +7030,223 @@ "issue:43756|issue:44913", "issue:43493|issue:43825", "issue:28218|issue:33357", - "issue:39692|issue:43493" + "issue:39692|issue:43493", + "issue:44964|issue:45405" ], - "prepared_review_unit_hash": "640cad28ab32a26c6dc82c8918247d304242d6c84cd568dbfa99632722e3fe90", + "prepared_review_unit_hash": "74135cd37031932f82489f28c6e35596862d5fd9d715a053e5997c298e4dbd92", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36437, - "estimated_input_tokens": 9110, - "estimated_eval_tokens": 18476 + "serialized_chars": 36990, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The items are a loose mix of unrelated Transformers bugs/regressions; none of the soft-similarity pairs look like the same underlying defect strongly enough to merge. I\u2019d use a generic loading/regression report as the cluster representative, but not as a duplicate match.", - "confidence": 0.33, - "canonical_issue_reason": "Issue 43504 is the best representative of the cluster\u2019s broadest recurring theme: model-loading/preset breakage from a concrete regression, and it sits near several other soft-similarity loading-related reports.", + "summary": "This cluster is mostly a loose mix of unrelated Transformers bug reports around loading, config, training, and model-specific regressions. None of the soft pairs look like safe duplicates, so I would not merge any of them. If forced to pick one representative issue, the broadest cluster anchor is the v5 `trust_remote_code` regression.", + "confidence": 0.83, + "canonical_issue_reason": "`issue:44561` is the broadest, most central symptom in the set: a core v5 regression that breaks `trust_remote_code` models. It is more representative of the cluster than the narrower model-specific or single-path failures.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43504 is the clearest general representative among these mostly unrelated bug reports; it is concrete, regression-oriented, and closer to the cluster\u2019s repeated load/config failure pattern than the more niche entries.", + "best_issue_reason": "`issue:44561` is the best stand-in for this loose cluster because it names a concrete upstream regression with wider impact and clearer actionability than the other, more model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44479", - "right": "issue:44625", - "accept": false, - "reason": "Both are Qwen3.5-related, but one is a video-input regression and the other is num_labels propagation; different code paths and failure modes." - }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "One is a Qwen3 device-map cuda assert, the other is chunked generation mismatch under compiled forward; not the same bug." + "reason": "Different failures: device-mapping CUDA assert vs compiled-forward chunked generation inconsistency. Same general area, not the same bug." }, { - "left": "issue:43550", - "right": "issue:44361", + "left": "issue:43643", + "right": "issue:44561", "accept": false, - "reason": "Bamba SDPA compile failure and MLukeTokenizer task AttributeError are unrelated subsystems and symptoms." + "reason": "Both mention `trust_remote_code`, but one is missing config fields from `AutoConfig` while the other is a v5 `is_torch_fx_available` removal regression. Different code paths and fixes." }, { - "left": "issue:43643", - "right": "issue:44561", + "left": "issue:43550", + "right": "issue:44361", "accept": false, - "reason": "Both mention trust_remote_code, but one is missing fields from AutoConfig and the other is a v5 regression from removing is_torch_fx_available; different defects." + "reason": "Unrelated: SDPA/torch.compile failure in Bamba vs tokenizer AttributeError in MLuke tasks." }, { "left": "issue:44977", "right": "issue:46209", "accept": false, - "reason": "Flash-attention generation bug for Qwen3.5 is unrelated to FP8 inference failure on B200 hardware." + "reason": "Different hardware/model issues: Qwen3.5 flash-attention generation bug vs FP8 inference failure on B200." }, { - "left": "issue:43577", - "right": "issue:45588", + "left": "issue:32090", + "right": "issue:34567", "accept": false, - "reason": "Blip2 dtype handling and flash_attention.py crashing on s_aux=None are separate code paths and distinct bugs." + "reason": "Trainer broadcast TypeError and TrainerState token counter not updating are separate training bugs." }, { - "left": "issue:32090", - "right": "issue:34567", + "left": "issue:43577", + "right": "issue:45588", "accept": false, - "reason": "Trainer NoneType broadcast error and num_input_tokens_seen not updating are different Trainer issues with no shared failure mechanism." + "reason": "Blip2 dtype mismatch on load is not the same as flash-attention crashing on `s_aux=None`." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Blip2 dtype persistence and an MPS attention correctness issue are not the same underlying problem." + "reason": "Different subsystems: model dtype propagation vs an MPS attention correctness issue." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Both touch training, but one is a Trainer broadcast TypeError and the other is a Qwen2-VL multi-GPU training question; not duplicates." + "reason": "Generic multi-GPU/trainer error versus Qwen2-VL multi-GPU training request; not the same underlying defect." }, { "left": "issue:44466", "right": "issue:46050", "accept": false, - "reason": "Serialization of tied weights depending on device and quantized-model save/load breakage are different save/load bugs." + "reason": "Both are save/load related, but one is tied-weight serialization and the other is sinq quantized model persistence. Not one concrete bug." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Legacy-field loading failure in BeitForSemanticSegmentation and BitNet packed-weight unpacking during accelerate loading are separate loading regressions." + "reason": "Legacy field loading failure in BEiT is unrelated to BitNet packed-weight unpacking during accelerate loading." }, { - "left": "issue:43504", - "right": "issue:43550", + "left": "issue:33666", + "right": "issue:36010", "accept": false, - "reason": "Beit preset loading failure and Bamba SDPA torch.compile failure affect different models and different execution paths." + "reason": "Multi-GPU training for a vision-language model is unrelated to a `GenerationMixin` import error." }, { - "left": "issue:33666", - "right": "issue:36010", + "left": "issue:43504", + "right": "issue:43550", "accept": false, - "reason": "Multi-GPU training for Qwen2-VL and GenerationMixin import failure are unrelated." + "reason": "Model-loading legacy field bug is not the same as a compile/SDPA runtime failure." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Model preset loading failure and processor image-size mismatch for OmDet-Turbo are different bugs in different components." + "reason": "Different model-specific loading/preprocessing failures; no shared underlying code path." }, { "left": "issue:43742", "right": "issue:44821", "accept": false, - "reason": "MobileLLM load key error and AutoImageProcessor-from-URL failure are different loading paths and error classes." + "reason": "KeyError loading MobileLLM is unrelated to loading an AutoImageProcessor from URL." }, { "left": "issue:45823", "right": "issue:46050", "accept": false, - "reason": "Gemma4 device-map forward mismatch and sinq quantization save/load corruption are unrelated defects." + "reason": "Device-mapping forward mismatch and quantized save/load corruption are separate problems." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio processor chat-template sampling-rate default and processor.tokenizer/image-passing regression are different processor bugs." + "reason": "Audio `apply_chat_template` sampling-rate defaulting and processor/tokenizer regression are different issues." }, { - "left": "issue:43720", - "right": "issue:43927", + "left": "issue:43452", + "right": "issue:44451", "accept": false, - "reason": "BitNet packed-weight loading and DiaConfig losing custom token IDs after save/load are both load-related but not the same bug." + "reason": "Both are load failures, but for different artifacts and failure modes (`gguf_file` path vs loading ScandiBERT)." }, { - "left": "issue:43452", - "right": "issue:44451", + "left": "issue:43720", + "right": "issue:43927", "accept": false, - "reason": "gguf_file breaking AutoTokenizer/AutoModel loading and failure to load a specific model are too broad to treat as the same concrete defect." + "reason": "Packed-weight loading in BitNet and DiaConfig losing token IDs after save/load are distinct persistence bugs." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch for bark-small and DiaConfig field loss after save/load are unrelated." + "reason": "CPU offload device mismatch in bark-small is unrelated to DiaConfig save/load token loss." }, { "left": "issue:43994", "right": "issue:46192", "accept": false, - "reason": "SigLIP2 nonsensical outputs and return_tensors being ignored with text_kwargs are different API bugs." + "reason": "SigLIP2 bad outputs and `return_tensors` being ignored are unrelated API bugs." }, { "left": "issue:44849", "right": "issue:46209", "accept": false, - "reason": "Qwen3.5 hidden-states bug and FP8 inference failure on B200 are distinct model/runtime issues." + "reason": "Qwen3.5 hidden-state output bug is not the same as FP8 inference failure on B200." }, { "left": "issue:33666", "right": "issue:36331", "accept": false, - "reason": "Qwen2-VL multi-GPU training and CustomTrainer.compute_loss signature mismatch are separate Trainer-related problems." + "reason": "Qwen2-VL multi-GPU training and `compute_loss` signature mismatch are different training issues." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Embedding dtype drifting during eval and auxiliary-loss normalization in OLMoE/GPT Oss are different training behaviors." + "reason": "One is dtype drift between train/eval; the other is auxiliary-loss normalization. Same broad area, not one bug." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Missing pad_token_id vs missing pad_token_idx are similar at a naming level, but they reference different configs and likely different fixes." + "reason": "Both are missing-pad-token config regressions, but for different models and different fields (`pad_token_id` vs `pad_token_idx`)." }, { "left": "issue:44625", "right": "issue:46002", "accept": false, - "reason": "Qwen3.5 num_labels propagation and NaNs in classification heads after init/checkout are unrelated issues." + "reason": "`num_labels` propagation in Qwen3.5 is unrelated to NaNs in classification heads." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "BitNet loading with accelerate and MLukeTokenizer task AttributeError are not the same code path." + "reason": "BitNet loading/unpacking and MLukeTokenizer task AttributeError are unrelated." }, { "left": "issue:43756", "right": "issue:44913", "accept": false, - "reason": "Smollm3 RoPE-layer discrepancy and GPTNeoXConfig rotary_pct reload defaulting are both rotary-related, but one is an implementation mismatch and the other is config persistence." + "reason": "Smollm3 RoPE-layer count bug and GPTNeoX rotary_pct reload issue both touch RoPE, but they are not the same defect." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 HF-vs-JAX discrepancy and pipeline translation-task error message regression are unrelated." + "reason": "SigLIP2 implementation discrepancy is unrelated to a pipeline error message about translation tasks." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer added-token spacing and a MacOS bus error with a CLIP model are entirely different bugs." + "reason": "Tokenizer added-space behavior and MacOS bus error in CLIP are unrelated." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "SigLIP2 doc example errors and SigLIP2 implementation mismatch overlap on model name only; one is documentation/example breakage, the other is core numerical behavior." + "reason": "Docs example mistakes are not the same as an HF-vs-JAX implementation discrepancy." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Model loading failure and an unreleased PEFT version bump are unrelated." } ] }, @@ -7240,6 +7290,7 @@ "issue:43756", "issue:43824", "issue:43844", + "issue:43873", "issue:43927", "issue:43994", "issue:44112", @@ -7254,7 +7305,6 @@ "issue:44792", "issue:44843", "issue:44849", - "issue:44964", "issue:44977", "issue:45083", "issue:45200", @@ -7266,23 +7316,23 @@ "issue:45759", "issue:45812", "issue:45823", + "issue:46050", "issue:46285" ], "soft_pairs": [ - "issue:44964|issue:45405", "issue:44849|issue:45357", "issue:44568|issue:45812", "issue:43540|issue:44977", "issue:45200|issue:45229", "issue:43493|issue:43844", - "issue:44488|issue:44843", "issue:43278|issue:43844", + "issue:44488|issue:44843", "issue:39692|issue:43295", "issue:44493|issue:46285", "issue:43756|issue:45083", "issue:43646|issue:44466", - "issue:43334|issue:43572", "issue:43257|issue:43299", + "issue:43334|issue:43572", "issue:33357|issue:33666", "issue:43334|issue:45759", "issue:45229|issue:45823", @@ -7291,216 +7341,217 @@ "issue:43824|issue:44589", "issue:43504|issue:43606", "issue:43278|issue:43701", - "issue:44464|issue:45584", "issue:43572|issue:44291", + "issue:44464|issue:45584", "issue:43575|issue:45405", - "issue:44792|issue:45405", "issue:43504|issue:43653", + "issue:44792|issue:45405", "issue:43606|issue:44112", - "issue:43994|issue:44493" + "issue:43994|issue:44493", + "issue:43873|issue:46050" ], - "prepared_review_unit_hash": "e28b7974dbe2bd147f1406fa7ce5b70f40313f6192d24ce8df9a674847511067", + "prepared_review_unit_hash": "bffba67cf4554cc523ffee0b853e2bf13c8d284b20b651fae0018bf8456374bd", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36003, - "estimated_input_tokens": 9001, - "estimated_eval_tokens": 18258 + "serialized_chars": 36546, + "estimated_input_tokens": 9137, + "estimated_eval_tokens": 18530 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the issues cover unrelated bugs across different models, tokenizers, configs, training/inference paths, and CI regressions. None of the soft pairs look like the same underlying bug/change, so no duplicates are accepted.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "This cluster is heterogeneous: the paired items share broad model-family or regression themes, but none appear to be the same concrete bug/change. I would not merge any soft edge pair.", + "confidence": 0.91, + "canonical_issue_reason": "No true duplicate center stands out; issue 43873 is the broadest currently open item with the most discussion, so it is the best representative if one must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 43873 is the most general and most active issue in the set, making it the safest single representative for the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44964", - "right": "issue:45405", - "accept": false, - "reason": "Model loading failure for Phi-4-multimodal-instruct vs an unreleased PEFT version bump; unrelated causes." - }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Qwen3.5 hidden-state output bug vs incorrect visual encoder keys on save_pretrained; different code paths." + "reason": "Both concern Qwen3.5, but one is about output_hidden_states and the other about save_pretrained visual encoder keys; different code paths and failures." }, { "left": "issue:44568", "right": "issue:45812", "accept": false, - "reason": "mDeBERTa special-token handling vs Granite token ID regression; same broad tokenizer area but different failures." + "reason": "Tokenizer-related, but one is a special-tokens/BOS-EOS regression for mdeberta and the other is Granite token ID mapping; not the same defect." }, { "left": "issue:43540", "right": "issue:44977", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError vs Qwen3.5 flash-attention generation issue; not the same bug." + "reason": "Both mention Qwen3-family multimodal behavior, but one is a video-processing ValueError and the other is a flash-attention generation issue." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Text-only mm_token_type_ids defaulting vs multi-GPU CUDA OOM; distinct Gemma4 issues." + "reason": "One is about missing mm_token_type_ids defaults for text-only fine-tuning, the other is a multi-GPU inference CUDA OOM; unrelated failures." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 implementation mismatch vs DeepSpeed ZeRO-3 gradient growth; different symptoms and root causes." + "reason": "SigLIP2 is the only shared theme; one is an implementation discrepancy, the other is a DeepSpeed ZeRO-3 gradient increase bug." }, { - "left": "issue:44488", - "right": "issue:44843", + "left": "issue:43278", + "right": "issue:43844", "accept": false, - "reason": "Loading a specific model fails vs unconditional model_info() call breaking offline mode; separate tokenizer/load issues." + "reason": "Embedding dtype drift in eval is unrelated to the ZeRO-3 gradient anomaly, despite both involving training/evaluation stability." }, { - "left": "issue:43278", - "right": "issue:43844", + "left": "issue:44488", + "right": "issue:44843", "accept": false, - "reason": "Embedding dtype drift in evaluation vs gradient explosion under ZeRO-3; unrelated behavior." + "reason": "One is a model load failure for a specific checkpoint, the other is an offline-mode failure caused by an unconditional hub call." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "SigLIP2 doc example errors vs processor.tokenizer / image-to-tokenizer regression; different problems." + "reason": "Both are regression-style model issues, but one is a SigLIP2 doc example/model-processor mismatch and the other is a custom processor.tokenizer API break." }, { "left": "issue:44493", "right": "issue:46285", "accept": false, - "reason": "Unexpected position-id key warnings vs CLIP text-model structure change; not the same code-path bug." + "reason": "Both mention unexpected keys/structure, but one is a position-id key warning across models and the other is a CLIPTextModel class-structure inconsistency." }, { "left": "issue:43756", "right": "issue:45083", "accept": false, - "reason": "Smollm3 RoPE layer mismatch vs qwen3_omni_moe feature-length helper behavior; unrelated." + "reason": "Smollm3 RoPE-layer loss and qwen3_omni_moe feature-length helper behavior are different implementation bugs." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization breakage vs lm_head.weight serialization inconsistency; different failures." + "reason": "One breaks custom model initialization in v5, the other is inconsistent lm_head serialization; different code paths and symptoms." }, { - "left": "issue:43334", - "right": "issue:43572", + "left": "issue:43257", + "right": "issue:43299", "accept": false, - "reason": "Missing pad_token_id in Qwen3-VL vs missing pad_token_idx in StableLmConfig; similar theme but not the same bug." + "reason": "Both involve Qwen3 MoE loading, but one is accelerate+deepspeed conversion and the other is a transformers dev loading regression for Qwen3VL MoE models." }, { - "left": "issue:43257", - "right": "issue:43299", + "left": "issue:43334", + "right": "issue:43572", "accept": false, - "reason": "Qwen3 MoE weight conversion under accelerate/deepspeed vs Qwen3VL MoE loading in dev; related family, different concrete issues." + "reason": "Both are missing-config-attribute bugs, but for different configs and different attribute names; not the same underlying issue." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "MacOS CLIP bus error vs Qwen2-VL multi-GPU training; unrelated models and failures." + "reason": "Completely different areas: MacOS bus error loading CLIP vs Qwen2-VL multi-GPU training." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "pad_token_id missing in Qwen3-VL vs AutoModelForCausalLM not unwrapping composite multimodal text_config; different configuration bugs." + "reason": "Both concern multimodal Qwen configs, but one is a missing pad_token_id attribute and the other is unwrapping text_config in AutoModelForCausalLM.from_config." }, { "left": "issue:45229", "right": "issue:45823", "accept": false, - "reason": "Gemma4 multi-GPU OOM vs device mismatch with device_map='auto'; both on Gemma4 but different bugs." + "reason": "Both are Gemma4 multi-GPU/device issues, but one is OOM and the other is a device mismatch during forward." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Float8 storage lookup error vs flash_attn import KeyError on Python 3.13; unrelated compatibility issues." + "reason": "Unrelated import/runtime failures: missing Float8 storage vs flash_attn KeyError on Python 3.13." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "DiaConfig save/load token ID loss vs MLukeTokenizer task AttributeError; different model/tokenizer problems." + "reason": "Both are tokenizer/config issues, but one is save/load token ID loss and generation IndexError, while the other is an AttributeError in MLukeTokenizer tasks." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Qwen2.5-VL import failure vs float8 storage TypeError; no shared underlying bug." + "reason": "Different failures: missing Qwen2_5_VL import vs Float8 storage lookup/type error." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Beit legacy-field loading issue vs bark-small CPU offload device mismatch; unrelated." + "reason": "Legacy-field loading bug in BeitForSemanticSegmentation is unrelated to bark-small CPU offload device mismatch." }, { "left": "issue:43278", "right": "issue:43701", "accept": false, - "reason": "Embedding dtype change on evaluate vs resume_from_checkpoint key mismatch; different regressions." - }, - { - "left": "issue:44464", - "right": "issue:45584", - "accept": false, - "reason": "Compiled-forward chunked generation inconsistency vs Whisper empty transcription after align_special_tokens; unrelated models and paths." + "reason": "Embedding dtype drift during eval and resume_from_checkpoint key mismatch are unrelated problems." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "StableLmConfig pad_token_idx issue vs unexpected _is_hf_initialized argument with init_empty_weights; different compatibility bugs." + "reason": "StableLmConfig pad_token_idx regression and init_empty_weights unexpected argument error are different initialization bugs." }, { - "left": "issue:43575", - "right": "issue:45405", + "left": "issue:44464", + "right": "issue:45584", "accept": false, - "reason": "Tensor-parallel OOM on Qwen2-57B vs PEFT version availability; unrelated." + "reason": "Chunked generation inconsistencies with compiled forward and Whisper empty-transcription failure are unrelated generation bugs." }, { - "left": "issue:44792", + "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "janus test failure vs PEFT release-version issue; unrelated." + "reason": "Qwen2 large-model TP OOM and an unsupported PEFT version bump have no shared defect." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "Beit legacy-field load failure vs BigBirdTokenizer mask-token special-token bug; different components." + "reason": "A pretrained-model legacy-field load failure is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and a PEFT version requirement issue are unrelated." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "CPU offload device mismatch in bark-small vs stale GraniteSpeech CI test; same general area but not the same bug." + "reason": "One is a real CPU-offload device mismatch in bark-small; the other is a stale CI test in GraniteSpeech." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 wrong outputs with AutoModel/pipeline vs position-id key warnings; unrelated." + "reason": "SigLIP2 inference corruption and a generic position-id key warning are not the same bug." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Both mention quantization/offloading, but one is offloading not behaving as expected and the other is sinq save/load breakage; different code paths." } ] }, @@ -7533,6 +7584,7 @@ "issue:34567", "issue:37428", "issue:38175", + "issue:39692", "issue:42490", "issue:43065", "issue:43232", @@ -7544,7 +7596,6 @@ "issue:43696", "issue:43701", "issue:43844", - "issue:43873", "issue:43957", "issue:44186", "issue:44423", @@ -7574,7 +7625,6 @@ "issue:46369" ], "soft_pairs": [ - "issue:43873|issue:46050", "issue:43065|issue:46369", "issue:45588|issue:46050", "issue:44423|issue:44734", @@ -7600,198 +7650,199 @@ "issue:45381|issue:46209", "issue:43295|issue:43450", "issue:44554|issue:46365", - "issue:43295|issue:44610" + "issue:43295|issue:44610", + "issue:39692|issue:46192" ], - "prepared_review_unit_hash": "b97f131d992b25019fe157cb83e1ae95e056f03a91fbcc4ef4fa47b31e3625a9", + "prepared_review_unit_hash": "bff1ebee622686eafe14607aa67e20765bc2df790c3d9b3be43b92eac36e392d", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35792, - "estimated_input_tokens": 8948, - "estimated_eval_tokens": 18152 + "serialized_chars": 35839, + "estimated_input_tokens": 8960, + "estimated_eval_tokens": 18176 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster looks heterogeneous: the soft pairs span unrelated tokenizer, serving, model-loading, training, and backend issues. None of the candidate pairs appears to be the same underlying bug/change, so no soft duplicates are accepted.", - "confidence": 0.93, - "canonical_issue_reason": "issue:44521 is the strongest anchor in this set: it is open, well-scoped, has multiple inbound references, and its multimodal apply_chat_template bug is clearly described.", + "summary": "This set looks like a heterogeneous grab bag of unrelated issues; the soft-similarity pairs do not appear to be duplicate reports of the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44521 is the best standalone issue here because it has the clearest reproduction path and the most evidence of active external impact.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43873", - "right": "issue:46050", - "accept": false, - "reason": "Both mention quantization/offloading, but one is an offloading behavior report and the other is a save/load failure for a quantized model; different code paths and symptoms." - }, { "left": "issue:43065", "right": "issue:46369", "accept": false, - "reason": "Same model family (SAM3) but different problems: a dummy Conv2d implementation vs a memory overflow during inference." + "reason": "Different Sam3 problems: one is a dummy Conv2d placeholder in Sam3PixelDecoder, the other is a memory overflow when using sam3-litetext." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "One is a flash-attention integration crash on s_aux=None; the other is a quantized-model serialization regression. Not the same bug." + "reason": "Unrelated failure modes: flash_attention.py crashes on s_aux=None, while sinq quantization breaks after save/load." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serving crashes, but one is a multimodal model AttributeError and the other is a KV-cache continuation indexing bug; different failure modes." + "reason": "Both hit serve crashes, but one is multimodal continuous-batching with a string/.to error and the other is KV-cache continuation tensor indexing." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Flash Attention 2 test failures and MLflow reporting on failed jobs are unrelated subsystems and symptoms." + "reason": "Completely different subsystems: Flash Attention 2 test failures vs MLflow reporting not updating after a failed training job." }, { "left": "issue:42490", "right": "issue:43696", "accept": false, - "reason": "A trailing-slash import warning and a CUDA OOM on GPT-oss-20b are unrelated." + "reason": "No overlap beyond being runtime issues: local-path import warning vs GPT-oss-20b CUDA OOM." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Gemma4Processor missing _tokenizer and a rope-parameter TypeError are separate issues with no shared code path indicated." + "reason": "Different code paths: Gemma4Processor missing _tokenizer in serve, versus JSON-loaded list type handling in rope validation." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading regressions are different training/loading bugs." + "reason": "Different problems and models: ZeRO-3 gradient growth vs Phi-4 multimodal loading failure." }, { "left": "issue:29159", "right": "issue:32090", "accept": false, - "reason": "Tokenizer consistency problems and a Trainer broadcast TypeError do not describe the same bug." + "reason": "Tokenizer slow/fast inconsistency is unrelated to the Trainer NoneType GPU broadcast error." }, { "left": "issue:28218", "right": "issue:34567", "accept": false, - "reason": "An added-token spacing issue is unrelated to TrainerState's num_input_tokens_seen not updating." + "reason": "Both concern tokenizer/training behavior, but one is added-token spacing and the other is TrainerState token counting not updating." }, { "left": "issue:43582", "right": "issue:44977", "accept": false, - "reason": "Apple Silicon warmup TypeError and Qwen3.5 flash-attention generation issues are different backend/model bugs." + "reason": "Unrelated: Apple Silicon caching_allocator_warmup TypeError versus Qwen3.5 flash-attention generation bug." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "MPS attention correctness with mismatched head dims is unrelated to an AuxRequest PyTorch version check." + "reason": "Different layers: MPS attention correctness with mismatched head dims vs a PyTorch version check for flex_attention imports." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Template-node compilation and flash_attn import-key lookup are different import/compile failures." + "reason": "Different root causes: template-node compilation error versus import_utils KeyError for flash_attn on Python 3.13." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript export failing on GeneratorExp is unrelated to SigLIP2 producing unexpected zero probabilities." + "reason": "No meaningful match: torch.jit.script GeneratorExp export issue versus SigLIP2 zero-probability outputs." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Video-input processing in Qwen3OmniMoe and Gemma4 mm_token_type_ids defaults are separate multimodal issues." + "reason": "Both are multimodal-related, but one is video input processing in Qwen3OmniMoe and the other is Gemma 4 token-type IDs for text-only fine-tuning." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "Both involve apply_chat_template, but one is incorrect assistant masks for multimodal inputs while the other is a crash on tool-call assistant messages with no content." + "reason": "Same API family but different bugs: all-zero assistant masks for multimodal inputs versus crash on tool-call assistant messages with empty content." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Gemma4 mm_token_type_ids defaults and a device_map='auto' PLE device mismatch are different bugs." + "reason": "Different Gemma4 issues: defaulting mm_token_type_ids to zeros versus device-mismatch behavior with device_map=\"auto\"." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "bfloat16 dtype mismatches in CI and Wav2Vec2 save/tokenization failures are not the same underlying issue." + "reason": "Different model classes and symptoms: dtype mismatches in inference versus Wav2Vec2 save_pretrained/tokenization failure." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "Missing flash_attention_utils symbol and meta-device model loading breakage are separate loading/import regressions." + "reason": "Both involve model loading/import paths, but one is a missing flash-attention symbol import and the other is meta-device loading breaking some models." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "A SigLIP2 implementation discrepancy and a resume_from_checkpoint key mismatch are unrelated." + "reason": "Different concerns: SigLIP2 HF-vs-JAX discrepancy versus checkpoint resume key mismatch." }, { "left": "issue:44186", "right": "issue:46192", "accept": false, - "reason": "LayoutLMv2Tokenizer crashes on NER/batched padding and return_tensors being ignored are different tokenizer API bugs." + "reason": "Tokenizer-related but not the same bug: LayoutLMv2Tokenizer crashes on NER/batched padding, while return_tensors is ignored when text_kwargs is passed." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs after sync_gpus and Qwen3.5 output_hidden_states behavior are unrelated generation issues." + "reason": "Both are generation-related, but they target different code paths: sync_gpus kwargs update versus output_hidden_states handling in Qwen3.5." }, { "left": "issue:45823", "right": "issue:46192", "accept": false, - "reason": "A Gemma4 device placement mismatch and silently ignored return_tensors are unrelated." + "reason": "Different areas: Gemma4 device placement with auto mapping versus return_tensors precedence behavior." }, { "left": "issue:45381", "right": "issue:46209", "accept": false, - "reason": "Qwen2.5-VL video position IDs and FP8 inference on B200 are unrelated issues." + "reason": "Completely different topics: Qwen2.5-VL video vision_position_ids versus FP8 inference failure on B200." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor.tokenizer access regression and batched video processor shape errors are different multimodal bugs." + "reason": "Both are processor issues, but one is processor.tokenizer access/images-to-tokenizer regression and the other is batched video processor shape." }, { "left": "issue:44554", "right": "issue:46365", "accept": false, - "reason": "MPS attention correctness and importing Trainer failing unless datasets is uninstalled are unrelated." + "reason": "Unrelated: MPS attention correctness bug versus an import conflict around Trainer and datasets." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Processor/tokenizer regression and OmDet-Turbo processor output size mismatch are different problems." + "reason": "Different processor regressions: custom code accessing processor.tokenizer/passing images to tokenizer versus OmDet-Turbo processor output size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 documentation/model-processor mismatch and quantization failure is unrelated to return_tensors being ignored with text_kwargs." } ] }, @@ -7819,10 +7870,10 @@ "issue:35141", "issue:37428", "issue:38175", - "issue:39692", "issue:41720", "issue:42490", "issue:42915", + "issue:43262", "issue:43278", "issue:43295", "issue:43334", @@ -7865,153 +7916,154 @@ "issue:46326" ], "soft_pairs": [ - "issue:39692|issue:46192", - "issue:44554|issue:45356", "issue:43572|issue:44987", + "issue:44554|issue:45356", "issue:43295|issue:43873", "issue:43572|issue:43957", "issue:43950|issue:46285", "issue:45520|issue:46178", "issue:44466|issue:45127", - "issue:37428|issue:43824", "issue:43550|issue:43844", + "issue:37428|issue:43824", "issue:44898|issue:45412", "issue:45200|issue:45325", "issue:43450|issue:43575", "issue:43334|issue:43525", "issue:44792|issue:45563", - "issue:42490|issue:43493", "issue:41720|issue:46209", + "issue:42490|issue:43493", + "issue:43388|issue:43526", "issue:42915|issue:43824", "issue:43824|issue:43975", - "issue:43388|issue:43526", "issue:28218|issue:35141", "issue:43278|issue:43856", "issue:38175|issue:43493", "issue:43526|issue:46192", - "issue:43653|issue:44112", "issue:45290|issue:46326", + "issue:43653|issue:44112", "issue:43975|issue:45520", - "issue:44442|issue:45198" + "issue:44442|issue:45198", + "issue:43262|issue:43450", + "issue:38175|issue:45412" ], - "prepared_review_unit_hash": "09d431bedc8cf73b2f504e84c9abfb8e170e4c79c18f75a6423353136113d812", + "prepared_review_unit_hash": "d7fdbf54992daa34ddf72dab138198064c2029b6b6cd54cd8bc63897a51c70c9", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36484, - "estimated_input_tokens": 9121, - "estimated_eval_tokens": 18498 + "soft_pair_count": 29, + "serialized_chars": 36814, + "estimated_input_tokens": 9204, + "estimated_eval_tokens": 18664 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the issues cover unrelated tokenizer, model-loading, quantization, vision, and CI regressions. I don\u2019t see any pair that shares the same underlying bug, so no soft merge should be accepted.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue: the set is too diverse, spanning unrelated bugs like tokenizer spacing, config attribute errors, quantization/offload failures, vision processor shape issues, and CI regressions.", + "summary": "This set is mostly a grab bag of unrelated Transformers bug reports that only share broad surface similarity (tokenizers, model loading, configs, processors, regressions). I do not see any soft pair that clearly describes the same underlying bug or change, so no duplicate edges should be accepted.", + "confidence": 0.29, + "canonical_issue_reason": "issue:43957 is the broadest representative of the cluster\u2019s dominant theme: a general Transformers 5.x model-loading regression involving `torch.device(\"meta\")`, which is more central and reusable as a cluster anchor than the many narrower, model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative for the cluster because there is no shared underlying defect across the items.", + "best_issue_reason": "issue:43957 is the most generally useful issue to keep as the representative problem statement because it covers a cross-model loading failure rather than a single model or one-off symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:39692", - "right": "issue:46192", + "left": "issue:43572", + "right": "issue:44987", "accept": false, - "reason": "Different problems: SigLIP2 documentation/example errors vs a processor/text_kwargs handling regression." + "reason": "Both are loading-related, but one is a missing config field and the other is a different failure when loading a specific model; not the same bug." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "Unrelated model/runtime bugs: MPS attention correctness vs Kimi-K2.5 tokenizer codec handling." - }, - { - "left": "issue:43572", - "right": "issue:44987", - "accept": false, - "reason": "Both involve loading/configuration, but one is a missing pad_token_idx attribute and the other is a transformers>=5.1 model load failure; not the same bug." + "reason": "Different subsystems entirely: MPS attention correctness vs tokenizer regression in Kimi-K2.5." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Custom processor/tokenizer regression is unrelated to quantization offloading behavior." + "reason": "One is a processor/tokenizer regression, the other is quantization offloading behavior; no shared code-path bug." }, { "left": "issue:43572", "right": "issue:43957", "accept": false, - "reason": "Different loading failures: missing pad_token_idx vs meta-device loading breaks some models." + "reason": "Both involve loading, but one is a StableLmConfig attribute issue and the other is a meta-device load regression affecting other models." }, { "left": "issue:43950", "right": "issue:46285", "accept": false, - "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to CLIP class structure changes." + "reason": "Non-persistent buffer corruption on load is unrelated to CLIPTextModel structural inconsistency." }, { "left": "issue:45520", "right": "issue:46178", "accept": false, - "reason": "Different import/backends issues: flash_attn on Python 3.13 vs tensorflow_text missing from BACKENDS_MAPPING." + "reason": "Different import/backend lookup failures in unrelated modules." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "Both concern serialization/tied weights, but one is device-dependent lm_head.weight saving and the other is LoRA merge collapse after vocab extension; not the same code-path." + "reason": "Both mention tied weights, but one is serialization/device-dependent and the other is LoRA merge collapse after vocab extension; not the same concrete bug." }, { - "left": "issue:37428", - "right": "issue:43824", + "left": "issue:43550", + "right": "issue:43844", "accept": false, - "reason": "Different import errors for different symbols/models and likely different root causes." + "reason": "A torch.compile+SDPA model failure and a DeepSpeed gradient issue are not the same problem." }, { - "left": "issue:43550", - "right": "issue:43844", + "left": "issue:37428", + "right": "issue:43824", "accept": false, - "reason": "torch.compile+SDPA failure is unrelated to gradient blow-up with DeepSpeed ZeRO-3." + "reason": "Both are import errors, but for different missing symbols and different modules." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Perceiver positional encoding failure and RT-DETR memory release issue are different bugs." + "reason": "Image resolution handling vs model memory release are unrelated." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Gemma mm_token_type_ids defaults vs Qwen2.5-VL rope position scaling are unrelated multimodal issues." + "reason": "Different multimodal bugs: missing mm token type defaults vs incorrect RoPE scaling." }, { "left": "issue:43450", "right": "issue:43575", "accept": false, - "reason": "Video processor batching shape bug is unrelated to tensor-parallel OOM when loading Qwen2-57B-A14B-Instruct." + "reason": "Video processor shape handling and tensor-parallel OOM are unrelated." }, { "left": "issue:43334", "right": "issue:43525", "accept": false, - "reason": "Both are pad_token_id-related load failures, but for different models/config classes (Qwen3-VL vs Llama4) and likely different fixes." + "reason": "Same symptom class (missing `pad_token_id`), but different model families and likely separate config implementations." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "A failed janus image-generation test is not the same as a stale generate() warning about num_return_sequences." + "reason": "A failed image-generation test and a paged-generate warning are not the same bug." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/backend failures for different model classes." }, { "left": "issue:42490", @@ -8020,76 +8072,82 @@ "reason": "Local-path import warning and SigLIP2 implementation discrepancy are unrelated." }, { - "left": "issue:41720", - "right": "issue:46209", + "left": "issue:43388", + "right": "issue:43526", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert and FP8 inference on B200 are different hardware/runtime issues." + "reason": "Both involve labels, but one is metrics gathering and the other is image processor label reduction." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure is unrelated to Qwen2_5_VL import errors." + "reason": "FineGrainedFP8Config failure and missing model import are unrelated." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "ImportError for a missing class is not the same as DeepSeek-Coder detokenization regression." - }, - { - "left": "issue:43388", - "right": "issue:43526", - "accept": false, - "reason": "Both are label-processing bugs, but one is gather_for_metrics dropping tuple elements and the other is BeitImageProcessorFast reduce_labels returning only one label; different code paths." + "reason": "Importing a class from transformers and incorrect detokenization are different bugs." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer extra-space behavior and embedding reinitialization after resize are unrelated." + "reason": "Tokenizer spacing after added tokens and embedding reinitialization on resize are different code paths." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Embedding dtype drift in evaluate is unrelated to Qwen3 MoE training memory usage." + "reason": "Training/eval dtype change is unrelated to Qwen3 MoE memory usage." }, { "left": "issue:38175", "right": "issue:43493", "accept": false, - "reason": "Unexpected zero probabilities in SigLIP2 and JAX/HF implementation discrepancy are related by model family but not the same concrete bug." + "reason": "Both mention SigLIP2, but one is zero probabilities and the other is a broader HF-vs-JAX discrepancy; not enough to treat as the same bug." }, { "left": "issue:43526", "right": "issue:46192", "accept": false, - "reason": "Different bugs in different layers: BeitImageProcessorFast label reduction vs return_tensors being ignored with text_kwargs." + "reason": "Image label reduction and `return_tensors` being ignored are unrelated processor bugs." }, { - "left": "issue:43653", - "right": "issue:44112", + "left": "issue:45290", + "right": "issue:46326", "accept": false, - "reason": "Both are CI-related, but one is BigBirdTokenizer special-token registration and the other is a GraniteSpeech stale device override test." + "reason": "Both are chat-template related, but one is a tool-call crash and the other is `continue_final_message`; different failure modes." }, { - "left": "issue:45290", - "right": "issue:46326", + "left": "issue:43653", + "right": "issue:44112", "accept": false, - "reason": "apply_chat_template crashing on tool-call assistant messages is a different tokenizer path from continue_final_message behavior." + "reason": "Tokenizer special-token registration and a GraniteSpeech CI test issue are unrelated." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Tokenizer detokenization regression is unrelated to flash_attn import-key handling on Python 3.13." + "reason": "Tokenizer detokenization regression and Python 3.13 flash-attn import failure are unrelated." }, { "left": "issue:44442", "right": "issue:45198", "accept": false, - "reason": "FastSpeech2ConformerTokenizer loading failure and Wav2Vec2 save_pretrained/tokenization failure are different tokenizer bugs." + "reason": "Different tokenizers/models and different failure modes; not the same underlying issue." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and batched video-shape handling are unrelated processor bugs." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 probability behavior and RT-DETR memory release are unrelated." } ] }, @@ -8117,17 +8175,18 @@ "issue:29942", "issue:32090", "issue:37428", - "issue:38175", - "issue:43262", + "issue:39692", + "issue:43278", "issue:43335", "issue:43425", - "issue:43450", "issue:43504", + "issue:43526", "issue:43550", "issue:43606", "issue:43611", "issue:43653", "issue:43819", + "issue:43825", "issue:43827", "issue:43901", "issue:43931", @@ -8149,7 +8208,6 @@ "issue:45245", "issue:45290", "issue:45381", - "issue:45412", "issue:45520", "issue:45584", "issue:45588", @@ -8162,239 +8220,245 @@ "issue:46285" ], "soft_pairs": [ - "issue:43262|issue:43450", - "issue:38175|issue:45412", "issue:45759|issue:46285", "issue:45520|issue:46170", "issue:44514|issue:44521", "issue:44610|issue:44857", "issue:44561|issue:46234", - "issue:43819|issue:44112", "issue:43827|issue:43901", "issue:44206|issue:45584", + "issue:43819|issue:44112", "issue:43819|issue:45083", "issue:44857|issue:45593", "issue:43335|issue:43550", - "issue:37428|issue:43425", "issue:43606|issue:45083", + "issue:37428|issue:43425", "issue:29942|issue:32090", "issue:44448|issue:44743", "issue:45127|issue:45245", "issue:45200|issue:45588", "issue:43611|issue:43931", "issue:43335|issue:43653", - "issue:43819|issue:44493", "issue:44855|issue:45083", "issue:43550|issue:44112", + "issue:43819|issue:44493", "issue:44206|issue:45823", - "issue:37428|issue:44351", "issue:45290|issue:45381", + "issue:37428|issue:44351", "issue:43335|issue:43504", + "issue:29127|issue:32090", "issue:45588|issue:45859", - "issue:29127|issue:32090" + "issue:43278|issue:43825", + "issue:39692|issue:43526" ], - "prepared_review_unit_hash": "088934eeaa7e57ebd11286e6ddc051490cc6018c3b3da4258b87fc5680cbab38", + "prepared_review_unit_hash": "66fba942fccfd485e39d0179ab849448eb9ec397ad710c60c72b22656f059250", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36367, - "estimated_input_tokens": 9092, - "estimated_eval_tokens": 18440 + "serialized_chars": 36401, + "estimated_input_tokens": 9101, + "estimated_eval_tokens": 18458 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated issues across chat templates, import utilities, vision/audio processors, model loading, and CI/runtime regressions. I don\u2019t see any soft edge that clearly represents the same underlying bug/change, so all candidate links are rejected.", - "confidence": 0.24, - "canonical_issue_reason": "No strong canonical issue: the cluster is too mixed to represent one duplicate bug. If a representative must be chosen, issue 44521 is the broadest open item in the chat_template/multimodal subtheme and has the most incoming references among the listed issues.", + "summary": "Heterogeneous cluster of mostly unrelated issues spanning model loading, chat templates, import compatibility, CI, and processor bugs; none of the soft pairs look like true duplicates or mergeable PR-style overlaps.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue: the cluster mixes unrelated regressions across chat templates, imports, configs, and CI, so no item clearly subsumes the rest.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44521 is the best representative only by breadth and activity: it is open, has multiple inbound references, and centers on a widely used multimodal chat_template path.", + "best_issue_reason": "No strong best issue for dedupe anchoring; the soft pairs are mostly superficial theme matches rather than the same underlying bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43262", - "right": "issue:43450", + "left": "issue:45759", + "right": "issue:46285", "accept": false, - "reason": "Different processors and different failure modes: audio chat-template sampling-rate default vs batched video shape mismatch." + "reason": "Different problems: composite multimodal config unwrapping vs CLIP text-model structure inconsistency." }, { - "left": "issue:38175", - "right": "issue:45412", + "left": "issue:45520", + "right": "issue:46170", "accept": false, - "reason": "Unrelated problems: SigLIP2 zero probabilities vs RT-DETR memory not being released." - }, - { - "left": "issue:45759", - "right": "issue:46285", - "accept": false, - "reason": "Different areas and symptoms: composite Qwen multimodal config unwrapping vs CLIP text-model structure inconsistency." - }, - { - "left": "issue:45520", - "right": "issue:46170", - "accept": false, - "reason": "Both are import/runtime issues, but they concern different missing dependencies and platforms, not the same bug." + "reason": "Both are import/runtime compatibility issues, but on different platforms and different missing dependencies." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Both hit apply_chat_template, but one is a batching crash and the other is assistant_mask generation; too different to be the same bug." + "reason": "Same area, but one is a batching/padding crash and the other is incorrect assistant masks for multimodal inputs." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "Different models and code paths: processor output size mismatch vs AMP crash in LwDetrImageLoss." + "reason": "Processor input-size mismatch vs AMP float16 loss crash; different code paths and failure modes." }, { "left": "issue:44561", "right": "issue:46234", "accept": false, - "reason": "Both involve import_utils availability helpers, but the removed/missing symbols and affected models are different." - }, - { - "left": "issue:43819", - "right": "issue:44112", - "accept": false, - "reason": "Different subsystems: DAC latent reconstruction mismatch vs a GraniteSpeech CI device-override test." + "reason": "Both involve import_utils, but they concern different missing helpers and different breakages." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs issues, but they cover different pipeline APIs and behaviors." + "reason": "Both are docs-related, but one is about pipeline removals in v5 and the other about return_all_scores behavior." }, { "left": "issue:44206", "right": "issue:45584", "accept": false, - "reason": "Different regressions: feature extractor argument handling vs Whisper empty-transcription generation after token alignment." + "reason": "Different bugs: unsupported feature-extractor arg vs Whisper empty transcription after special-token alignment." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent mismatch vs a stale CI device override test; not the same underlying defect." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "Different helper/model behavior: DAC latent mismatch vs qwen3_omni_moe feature-length calculation." + "reason": "Unrelated code paths: DAC from_latents mismatch vs qwen3_omni_moe helper length behavior." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "Different failures in different models: AMP crash vs auxiliary losses omitted when denoising is off." + "reason": "Loss AMP crash vs auxiliary-loss logic when denoising is off; different symptoms and fixes." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "Different bugs: SwitchTransformers sparse-layer config vs Bamba torch.compile/SDPA failure." + "reason": "Config bug creating sparse layers at zero vs torch.compile/SDPA failure; distinct issues." }, { - "left": "issue:37428", - "right": "issue:43425", + "left": "issue:43606", + "right": "issue:45083", "accept": false, - "reason": "One is a missing flash-attention helper import, the other is broad Torch 2.10 incompatibility." + "reason": "CPU offload device mismatch vs helper output-length behavior; no shared underlying bug." }, { - "left": "issue:43606", - "right": "issue:45083", + "left": "issue:37428", + "right": "issue:43425", "accept": false, - "reason": "Different models and issues: Bark CPU-offload device mismatch vs qwen3_omni_moe feature-length helper behavior." + "reason": "Missing internal import vs Torch 2.10 incompatibility; one is an API reference issue, the other a version break." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 test failures are unrelated to Trainer _gpu_broadcast_one NoneType errors." + "reason": "Flash Attention 2 test failures and NoneType broadcast errors are unrelated failures." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Both are model behavior regressions, but they affect different architectures and code paths." + "reason": "Different model behavior bugs in different modules; not one fixable code-path problem." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "Unrelated: LoRA merge collapse with tied embeddings vs category-count runtime limit." + "reason": "LoRA merge collapse and category-count limit errors are unrelated." }, { "left": "issue:45200", "right": "issue:45588", "accept": false, - "reason": "Different functionality: token-type defaults for Gemma4 fine-tuning vs flash_attention sink-less crash." + "reason": "mm_token_type_ids defaulting issue vs flash_attention s_aux=None crash; different components." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Both are model-loading problems, but one is a base_model_prefix compatibility break and the other is a specific Qwen3-VL shape mismatch." + "reason": "Base-model-prefix loading regression vs Qwen3-VL weight-shape mismatch; separate model-loading problems." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Different models and root causes: sparse-layer config bug vs tokenizer special-token registration/decode issue." - }, - { - "left": "issue:43819", - "right": "issue:44493", - "accept": false, - "reason": "Different errors: DAC latent mismatch vs unexpected position-id keys across models." + "reason": "Sparse-layer config bug vs tokenizer special-token registration bug; different subsystems." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 JIT parsing/import issue is unrelated to qwen3_omni_moe helper behavior." + "reason": "Python 3.13 import parse error vs helper length behavior; no duplication." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "Different regressions: Bamba torch.compile/SDPA vs GraniteSpeech stale CI device override." + "reason": "torch.compile SDPA failure and a stale CI device-override test are not the same bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different issues: DAC latent mismatch vs unexpected position-id keys in model loading." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "Both are device-related, but one is an audio feature-extractor regression and the other is a Gemma4 PLE device_map mismatch." + "reason": "Unsupported feature-extractor arg vs Gemma4 device-map mismatch; unrelated." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call crash vs wrong vision_position_ids for video input; distinct bugs." }, { "left": "issue:37428", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different code paths." + "reason": "Missing flash-support import vs missing HybridCache import; different symbols and breakages." }, { - "left": "issue:45290", - "right": "issue:45381", + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Sparse-layer creation bug vs BEiT legacy-field loading bug; unrelated." + }, + { + "left": "issue:29127", + "right": "issue:32090", "accept": false, - "reason": "Different multimodal problems: chat_template tool-call crash vs wrong vision_position_ids in Qwen2.5-VL video input." + "reason": "LayoutLMv3 error-message clarity issue vs Trainer NoneType broadcast crash; not duplicates." }, { "left": "issue:45588", "right": "issue:45859", "accept": false, - "reason": "Different fixes needed: flash_attention s_aux=None handling vs Qwen3_5Moe rotary embedding CPU-offload compatibility." + "reason": "Both mention offload/attention-adjacent code, but one is an AttributeError on s_aux and the other is CPU-offload incompatibility in a rotary embedding." }, { - "left": "issue:29127", - "right": "issue:32090", + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype regression vs pipeline task-support messaging; no common bug." + }, + { + "left": "issue:39692", + "right": "issue:43526", "accept": false, - "reason": "Completely different issues: clearer LayoutLMV3 error messaging vs Trainer broadcast TypeError." + "reason": "Docs example errors vs BeitImageProcessorFast reduce_labels bug; different artifacts and fixes." } ] }, @@ -8426,6 +8490,7 @@ "issue:33357", "issue:34689", "issue:35141", + "issue:37428", "issue:38175", "issue:39692", "issue:40444", @@ -8441,7 +8506,6 @@ "issue:43701", "issue:43720", "issue:43746", - "issue:43825", "issue:43854", "issue:43873", "issue:43927", @@ -8453,6 +8517,7 @@ "issue:44779", "issue:44857", "issue:44977", + "issue:44987", "issue:45072", "issue:45083", "issue:45229", @@ -8467,10 +8532,8 @@ "issue:46234" ], "soft_pairs": [ - "issue:43278|issue:43825", - "issue:45459|issue:46234", - "issue:39692|issue:43526", "issue:45310|issue:45478", + "issue:45459|issue:46234", "issue:43335|issue:45083", "issue:39692|issue:44493", "issue:43720|issue:44112", @@ -8478,261 +8541,256 @@ "issue:43335|issue:44112", "issue:43335|issue:45593", "issue:42490|issue:43854", - "issue:43606|issue:44857", "issue:34689|issue:35141", + "issue:43606|issue:44857", "issue:44360|issue:44561", "issue:45920|issue:46192", "issue:38175|issue:44493", - "issue:44857|issue:45072", "issue:44977|issue:45229", - "issue:28218|issue:32090", + "issue:44857|issue:45072", "issue:44448|issue:44977", + "issue:28218|issue:32090", "issue:43746|issue:46209", "issue:40444|issue:43526", "issue:43701|issue:45083", - "issue:29942|issue:35141", "issue:32090|issue:33357", + "issue:29942|issue:35141", "issue:43550|issue:45083", "issue:43873|issue:45229", - "issue:42490|issue:43575", "issue:43927|issue:45698", - "issue:10000|issue:15354", + "issue:42490|issue:43575", "issue:44561|issue:44779", + "issue:10000|issue:15354", "issue:43278|issue:43329", - "issue:41720|issue:45229" + "issue:41720|issue:45229", + "issue:37428|issue:44987" ], - "prepared_review_unit_hash": "e7bc074d6bf585333f51936d6344e662d6d77124027aafbb32e190f2331faa95", + "prepared_review_unit_hash": "5fdbf31e0516f5ec74fdd918e21b708c2d8e0b133ea8f9a24f61c778dda471ef", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 33, - "serialized_chars": 37043, - "estimated_input_tokens": 9261, - "estimated_eval_tokens": 18778 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37339, + "estimated_input_tokens": 9335, + "estimated_eval_tokens": 18926 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a grab-bag of unrelated issues across different Transformers subsystems. I found one clear duplicate pair: the two Qwen3.5 Moe from_pretrained reports. All other soft matches look like broad thematic similarity rather than the same bug.", - "confidence": 0.95, - "canonical_issue_reason": "Issue 45310 is the better canonical for the Qwen3.5 Moe from_pretrained failure: it is the earlier report with more discussion and the same concrete failure mode as 45478.", + "summary": "The only clear duplicate pair is the Qwen3.5 MoE `from_pretrained` regression. All other soft pairs describe different models, subsystems, or failure modes and should stay separate.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45310 is the earlier, more detailed report of the same `transformers>=5.4.0` Qwen3.5 MoE `from_pretrained` failure; issue:45478 reads like a later duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45310 is the strongest issue to keep overall because it is the clearest exact-duplicate target among the cluster and has the most complete original report.", + "best_issue_reason": "issue:45310 is the best representative issue here because it is the clearest and most established report of the only true duplicate pair in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43278", - "right": "issue:43825", - "accept": false, - "reason": "Different bugs: one is a training/evaluate dtype mismatch, the other is a pipeline error-message regression about translation tasks." + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same exact Qwen3.5 MoE `from_pretrained` regression on `transformers>=5.4.0`; later report duplicates the earlier one." }, { "left": "issue:45459", "right": "issue:46234", "accept": false, - "reason": "Both involve import/availability handling, but they are unrelated code paths: tokenizer protobuf error masking vs a missing utility symbol." - }, - { - "left": "issue:39692", - "right": "issue:43526", - "accept": false, - "reason": "Different models and failures: SigLIP2 documentation example errors vs BEiTImageProcessorFast reduce_labels behavior." - }, - { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "Same underlying bug: both report transformers>=5.4.0 causing Qwen3.5 Moe from_pretrained loading failures, with matching titles and symptoms." + "reason": "Unrelated bugs: tokenizer error handling with missing protobuf vs a missing import_utils symbol for Falcon Mamba." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Both are model/config helper bugs, but one is SwitchTransformers sparse-layer creation and the other is qwen3_omni_moe feature-length calculation." + "reason": "Different components and failures: SwitchTransformers sparse-layer config vs qwen3_omni_moe feature-length helper behavior." }, { "left": "issue:39692", "right": "issue:44493", "accept": false, - "reason": "The first is a SigLIP2 docs/model-processing issue; the second is a position-id key warning across many models." + "reason": "SigLIP2 doc/quantization issues are not the same as the broad unexpected-position-id regression." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "Both touch CI/accelerate loading, but one is about unpacking packed weights and the other is a stale GraniteSpeech device override test." + "reason": "Different problems: packed-weight loading in BitNet vs a stale CI device-override test in GraniteSpeech." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Both are v4/v5 behavior regressions, but they affect different models and different code paths: Pegasus generation vs DeepSeek tokenization." + "reason": "Different regressions: Pegasus v4/v5 output mismatch vs DeepSeek tokenizer changes in v5." }, { "left": "issue:43335", "right": "issue:44112", "accept": false, - "reason": "SwitchTransformers sparse layer construction is unrelated to GraniteSpeech CI device override failures." + "reason": "No shared underlying bug; config-layer creation and GraniteSpeech CI test failure are distinct." }, { "left": "issue:43335", "right": "issue:45593", "accept": false, - "reason": "Different issues: sparse encoder layer creation vs D-FINE auxiliary loss behavior when denoising is disabled." + "reason": "SwitchTransformers config bug and D-FINE auxiliary-loss behavior are unrelated." }, { "left": "issue:42490", "right": "issue:43854", "accept": false, - "reason": "Both concern loading/import behavior, but one is a local-path slash warning and the other is a model-loading failure for GLM-4.7-Flash." + "reason": "Local-path trailing-slash warning is unrelated to GLM-4.7-Flash loading failures." }, { - "left": "issue:43606", - "right": "issue:44857", + "left": "issue:34689", + "right": "issue:35141", "accept": false, - "reason": "Different failures in different settings: CPU offload device mismatch for bark-small vs float16 AMP crash in LwDetrImageLoss." + "reason": "Model-loading regression for Llama 3.2 Vision is unrelated to embedding reinitialization after resize." }, { - "left": "issue:34689", - "right": "issue:35141", + "left": "issue:43606", + "right": "issue:44857", "accept": false, - "reason": "Distinct model-loading problems: Llama 3.2 Vision load breakage vs embedding reinitialization after resizing token embeddings." + "reason": "CPU offload device mismatch and float16 AMP crash are different code paths and symptoms." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "The DSA indexer discussion is unrelated to the v5 removal of is_torch_fx_available breaking trust_remote_code models." + "reason": "DSA indexer activation question is unrelated to the removal of `is_torch_fx_available` breaking trust-remote-code models." }, { "left": "issue:45920", "right": "issue:46192", "accept": false, - "reason": "AutoTokenizer wrong token IDs across models is a tokenizer regression; the other is return_tensors being ignored when text_kwargs is passed." + "reason": "Wrong token IDs in AutoTokenizer is a different bug from `return_tensors` being ignored with explicit `text_kwargs`." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "One is unexpected zero probabilities in SigLIP2; the other is unexpected key warnings with position ids." - }, - { - "left": "issue:44857", - "right": "issue:45072", - "accept": false, - "reason": "Both are numeric-precision issues, but they are different models and code paths: LwDetrImageLoss AMP crash vs dtype mismatches in inference." + "reason": "Zero probabilities in SigLIP2 are not the same as unexpected position-id keys across models." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Different model families and symptoms: Qwen3.5 flash-attention generation vs Gemma4 multi-GPU CUDA OOM." + "reason": "Flash-attention generation issue for Qwen3.5 is unrelated to Gemma4 multi-GPU CUDA OOM." }, { - "left": "issue:28218", - "right": "issue:32090", + "left": "issue:44857", + "right": "issue:45072", "accept": false, - "reason": "Tokenizer-added-space issue is not the same as Trainer/_gpu_broadcast_one failing on NoneType." + "reason": "Different dtype/runtime issues in different models; not the same underlying bug." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Pegasus output drift between v4 and v5 is unrelated to Qwen3.5 flash-attention generation failures." + "reason": "Pegasus v4/v5 output drift and Qwen3.5 flash-attention generation failures are unrelated." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer spacing bug and Trainer `_gpu_broadcast_one` NoneType error are unrelated." }, { "left": "issue:43746", "right": "issue:46209", "accept": false, - "reason": "PEFT-adapter local checkpoint loading and FP8 inference on B200 are separate problems with no shared concrete failure mode." + "reason": "PEFT local-checkpoint loading issue is unrelated to FP8 inference failure on B200." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "Multimodal iterable dataset finetuning failure is unrelated to BEiTImageProcessorFast reduce_labels returning only one label." + "reason": "Qwen2.5-VL iterable-dataset multi-image failure is unrelated to BEiT label reduction behavior." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "resume_from_checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + "reason": "Checkpoint key mismatch and qwen3_omni_moe helper behavior are different issues." }, { - "left": "issue:29942", - "right": "issue:35141", + "left": "issue:32090", + "right": "issue:33357", "accept": false, - "reason": "Flash Attention 2 test failures and embedding reinitialization after resize are different defects." + "reason": "Trainer broadcast type error and MacOS bus error with CLIP are unrelated." }, { - "left": "issue:32090", - "right": "issue:33357", + "left": "issue:29942", + "right": "issue:35141", "accept": false, - "reason": "Trainer NoneType broadcast error is unrelated to the MacOS bus error with community CLIP loading." + "reason": "Flash Attention 2 test failures are unrelated to embedding dtype reinitialization on resize." }, { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba-9B-v2 does not match qwen3_omni_moe helper behavior." + "reason": "Bamba torch.compile SDPA failure is unrelated to qwen3_omni_moe output-length helper behavior." }, { "left": "issue:43873", "right": "issue:45229", "accept": false, - "reason": "Quantization offloading problems are not the same bug as Gemma4 multi-GPU OOM." - }, - { - "left": "issue:42490", - "right": "issue:43575", - "accept": false, - "reason": "Trailing-slash local model warning is unrelated to Qwen2-57B-A14B-Instruct tensor-parallel OOM." + "reason": "Quantization offloading issue and Gemma4 OOM are different failure modes." }, { "left": "issue:43927", "right": "issue:45698", "accept": false, - "reason": "DiaConfig custom token ID persistence and wrong custom-module loading after save_pretrained are different save/load bugs." + "reason": "Custom token ID loss on save/load is different from loading the wrong custom module after save_pretrained." }, { - "left": "issue:10000", - "right": "issue:15354", + "left": "issue:42490", + "right": "issue:43575", "accept": false, - "reason": "These are unrelated historical issues: German DistilBertModel error vs GeneratorExp unsupported by torch.jit.script." + "reason": "Trailing-slash path warning is unrelated to tp-induced OOM when loading Qwen2-57B-A14B-Instruct." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "Removal of is_torch_fx_available breaking trust_remote_code models is unrelated to DeepSeek tokenizer regression." + "reason": "Removed utility breaking trust_remote_code models is unrelated to DeepSeek tokenizer regression." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel issue and GeneratorExp torch.jit.script export failure are unrelated." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "Embedding dtype drift between train/eval is unrelated to the multimodal token-count helper video branch bug." + "reason": "Embedding dtype mismatch in evaluation is unrelated to undefined variables in multimodal token counting." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Qwen3 auto device mapping cudaErrorAssert and Gemma4 multi-GPU OOM are different model/runtime failures." + "reason": "Qwen3 device-mapping assert on A800 is a different bug from Gemma4 multi-GPU OOM." + }, + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "Missing flash-attention import symbol and loading physical-intelligence/fast failure are not the same issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The evaluation is grounded in the packet and stays conservative overall. The only accepted pair, issue:45310 vs issue:45478, is well supported by identical titles plus closely matching Qwen3.5 Moe from_pretrained failure reports, and choosing the earlier issue as canonical is reasonable. The remaining soft matches are appropriately rejected as same-theme but different bug/path cases. No mergeability issues apply here since these are issue pairs, not PR pairs." + "feedback": "The summary is grounded in the packet: issue:45310 and issue:45478 have identical titles and later/earlier timestamps, so treating them as the duplicate pair is reasonable. The other pairings are described conservatively as unrelated based on clearly different titles and failure modes. No overstatement stands out." }, "evaluator_used": true, "retried": false, @@ -8755,11 +8813,12 @@ "nodes": [ "issue:25251", "issue:29127", + "issue:29942", + "issue:30064", "issue:32090", "issue:33290", "issue:33357", "issue:34567", - "issue:37428", "issue:39692", "issue:42490", "issue:42915", @@ -8768,6 +8827,7 @@ "issue:43278", "issue:43381", "issue:43412", + "issue:43575", "issue:43611", "issue:43653", "issue:43873", @@ -8784,7 +8844,6 @@ "issue:44857", "issue:44871", "issue:44977", - "issue:44987", "issue:45198", "issue:45245", "issue:45372", @@ -8802,7 +8861,6 @@ "issue:46285" ], "soft_pairs": [ - "issue:37428|issue:44987", "issue:44977|issue:45563", "issue:44857|issue:45588", "issue:44568|issue:45491", @@ -8819,8 +8877,8 @@ "issue:45245|issue:45563", "issue:43653|issue:44743", "issue:43611|issue:46121", - "issue:44206|issue:44479", "issue:32090|issue:33290", + "issue:44206|issue:44479", "issue:45459|issue:45712", "issue:29127|issue:34567", "issue:33357|issue:39692", @@ -8830,210 +8888,225 @@ "issue:44871|issue:44977", "issue:43065|issue:43994", "issue:44361|issue:45198", - "issue:42915|issue:42994" + "issue:42915|issue:42994", + "issue:45372|issue:45459", + "issue:29942|issue:30064", + "issue:43278|issue:43575" ], - "prepared_review_unit_hash": "8909718c591854269f34ca2c53e82d208a8e9095ab7287be00efac0ec8cee389", + "prepared_review_unit_hash": "d14fdb644b9ed7763a8d7a0065a48535d320624ba6cb87908c23461d1a0876db", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 36114, - "estimated_input_tokens": 9029, - "estimated_eval_tokens": 18314 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37175, + "estimated_input_tokens": 9294, + "estimated_eval_tokens": 18844 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a loose semantic cluster rather than true duplicates: the issues cover unrelated models, tokenizers, quantization, serving, and eval/training regressions. I found no soft pair that clearly describes the same underlying bug/change.", - "confidence": 0.91, - "canonical_issue_reason": "Issue 43873 is the broadest and most active item in the set (open, 12 discussions, 5 inbound refs) and best fits as the representative issue for the cluster, even though the surrounding issues are largely unrelated.", + "summary": "These are mostly distinct bug reports across unrelated components; the soft similarity pairs do not describe the same underlying issue, so none should be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single true canonical issue emerges because the set is not a duplicate cluster; if a representative is needed, issue:45372 is the clearest, most actionable report.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the best global issue candidate because it is the most broadly scoped and most externally referenced item here.", + "best_issue_reason": "issue:45372 is the strongest standalone representative: it has a concrete failure, version scope, and enough context to reproduce, but it is not a duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:37428", - "right": "issue:44987", - "accept": false, - "reason": "Different failures: a flash-attention import error vs a model loading failure for physical-intelligence/fast." - }, { "left": "issue:44977", "right": "issue:45563", "accept": false, - "reason": "Qwen3.5 flash-attention generation regression and a stale num_return_sequences warning are unrelated." + "reason": "Qwen3.5 flash-attn generation regression vs a stale warning in paged generate; different symptoms and code paths." }, { "left": "issue:44857", "right": "issue:45588", "accept": false, - "reason": "Different code paths: LwDetrImageLoss AMP/CUDA crash vs flash_attention.py crashing on s_aux=None." + "reason": "LwDetrImageLoss AMP/CUDA crash vs flash_attention.py failing on s_aux=None; unrelated failures." }, { "left": "issue:44568", "right": "issue:45491", "accept": false, - "reason": "Tokenizer special-token insertion bug and Gemma3 NaN embeddings are different bugs." + "reason": "Tokenizer special-token omission vs Gemma3 NaN embeddings on mixed-length GPU batches; distinct bugs." }, { "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Gemma 4 processor import/dependency failure is not the same as the flash-attention s_aux=None crash." + "reason": "Gemma4 processor import failure vs flash_attention AttributeError; different failure points." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Both involve eval-mode behavior, but one is a dtype change and the other is gradient checkpointing support." + "reason": "Embedding dtype drift in eval vs gradient-checkpointing disallowed in eval mode; related to eval but not the same bug." }, { "left": "issue:43412", "right": "issue:45588", "accept": false, - "reason": "Executorch export failures for Gemma3n are unrelated to the flash_attention.py AttributeError." + "reason": "ExecuTorch export guard/erfinv issue vs flash_attention None handling; different paths and fixes." }, { "left": "issue:43611", "right": "issue:46285", "accept": false, - "reason": "Both touch model-loading structure, but one is a base_model_prefix regression and the other is a CLIP class layout inconsistency." + "reason": "base_model_prefix loading regression vs CLIP text model structure inconsistency; separate model-loading issues." }, { "left": "issue:45588", "right": "issue:45823", "accept": false, - "reason": "Flash-attention s_aux handling and Gemma4 device-map mismatch are different forward-path bugs." + "reason": "flash_attention None crash vs Gemma4 PLE device mismatch; not the same underlying problem." }, { "left": "issue:45372", "right": "issue:45406", "accept": false, - "reason": "They both mention Gemma4, but one is a missing import from mistral_common and the other is a missing _tokenizer attribute in serve." + "reason": "Both involve Gemma4, but one is processor import loading and the other is transformers serve missing _tokenizer; distinct code paths." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "A generation output regression and a tied-weights serialization inconsistency are separate issues." + "reason": "Different outputs for Pegasus vs inconsistent lm_head serialization; both v5 regressions but not one bug." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Python-version compatibility and DeepseekV3MoE divergence are unrelated." + "reason": "Python version compatibility regression vs DeepseekV3MoE divergence; unrelated." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Wav2Vec2 save/tokenization failure is not the same as Fuyu's bare except swallowing errors." + "reason": "Wav2Vec2 save/tokenization failure vs a bare-except bug in Fuyu batching; different model/util code." }, { "left": "issue:25251", "right": "issue:32090", "accept": false, - "reason": "Pipeline top_k nesting behavior and Trainer _gpu_broadcast_one NoneType error are unrelated." + "reason": "Pipeline top_k nesting bug vs Trainer _gpu_broadcast_one NoneType error; unrelated subsystems." }, { "left": "issue:45245", "right": "issue:45563", "accept": false, - "reason": "Category-count limit error and stale warning emission are not the same bug." + "reason": "Category-count runtime limit vs stale generate warning; different generate-related issues." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "BigBirdTokenizer special-token registration and Qwen recurrent-state reset are different code paths." + "reason": "BigBirdTokenizer special-token registration vs Qwen3.5 recurrent state reset; unrelated." }, { "left": "issue:43611", "right": "issue:46121", "accept": false, - "reason": "Loading break with base_model_prefix and rope-validation TypeError are unrelated." + "reason": "base_model_prefix loading regression vs rope-validation TypeError on JSON-loaded lists; different mechanisms." }, { - "left": "issue:44206", - "right": "issue:44479", + "left": "issue:32090", + "right": "issue:33290", "accept": false, - "reason": "A LasrFeatureExtractor argument crash and a Qwen video-input regression are different model/input bugs." + "reason": "Trainer NoneType broadcast error vs deepspeed Adafactor OOM; unrelated training issues." }, { - "left": "issue:32090", - "right": "issue:33290", + "left": "issue:44206", + "right": "issue:44479", "accept": false, - "reason": "Trainer broadcast TypeError and deepspeed Adafactor OOM are not the same underlying issue." + "reason": "Feature-extractor center-arg crash vs video input regression for Qwen models; different components." }, { "left": "issue:45459", "right": "issue:45712", "accept": false, - "reason": "Tokenizer exception masking and leftover dummy classes are separate maintenance bugs." + "reason": "Tokenizer decode-error masking vs leftover dummy classes/check_repo cleanup; not the same bug." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 error-message clarity and TrainerState token counting are unrelated." + "reason": "LayoutLMv3 error-message clarity vs TrainerState num_input_tokens_seen not updating; unrelated." }, { "left": "issue:33357", "right": "issue:39692", "accept": false, - "reason": "MacOS CLIP bus error and SigLIP2 documentation mistakes are different problems." + "reason": "MacOS CLIP bus error vs SigLIP2 docs example issues; different artifacts and failure modes." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "Different tokenizers/models and different failure modes." + "reason": "LayoutLMv2Tokenizer NER/padding crash vs Wav2Vec2 save/tokenization failure; different tokenizer families." }, { "left": "issue:42490", "right": "issue:43873", "accept": false, - "reason": "Local-path slash warning and quantization/offloading malfunction are not the same bug." + "reason": "Local-path import warning vs quantization offloading behavior; not the same underlying issue." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "Sam3PixelDecoder dummy layer and RT-DETR memory leak are unrelated." + "reason": "Dummy Conv2d placeholder cleanup vs RT-DETR memory release on delete; different bugs." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Gemma3 EOS-token config mismatch is unrelated to Qwen3.5 flash-attention generation failure." + "reason": "Gemma3 eos_token_id mismatch vs Qwen3.5 flash-attention generation regression; unrelated model issues." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "Dummy Conv2d in SAM3 and SigLIP2 nonsense outputs are different issues." + "reason": "Sam3PixelDecoder dummy layer vs SigLIP2 nonsensical outputs; different code paths." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure do not share a concrete root cause." + "reason": "MLukeTokenizer AttributeError vs Wav2Vec2 save/tokenization failure; unrelated." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "Both involve quantization, but one is a config/runtime failure for Qwen3Moe and the other is a quantized-model saving bug." + "reason": "Qwen3Moe FP8 config failure vs quantized model saving failure; both quantization-related but not the same defect." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma4 import regression vs protobuf decode-error masking; distinct tokenizer/import concerns." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures vs void segmentation-map processing failure; different issues." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Eval dtype change vs tp-load OOM on Qwen2-57B; unrelated." } ] }, @@ -9059,7 +9132,6 @@ "nodes": [ "issue:29127", "issue:29942", - "issue:30064", "issue:33357", "issue:35141", "issue:36010", @@ -9067,10 +9139,10 @@ "issue:42915", "issue:43066", "issue:43262", - "issue:43278", "issue:43334", "issue:43381", "issue:43421", + "issue:43452", "issue:43526", "issue:43531", "issue:43575", @@ -9090,269 +9162,263 @@ "issue:44462", "issue:44464", "issue:44625", + "issue:44743", + "issue:44811", "issue:44857", "issue:44977", "issue:45072", "issue:45198", "issue:45245", - "issue:45372", "issue:45412", "issue:45440", - "issue:45459", "issue:45491", "issue:45584", + "issue:45588", "issue:45593", "issue:45834", "issue:45859", "issue:46369" ], "soft_pairs": [ - "issue:29942|issue:30064", - "issue:45372|issue:45459", - "issue:43278|issue:43575", "issue:43066|issue:44462", "issue:45412|issue:46369", "issue:43994|issue:45412", "issue:44265|issue:45198", - "issue:43381|issue:43701", "issue:43756|issue:45440", + "issue:43381|issue:43701", "issue:43421|issue:45491", "issue:43653|issue:44442", "issue:29127|issue:33357", "issue:43824|issue:44279", "issue:45440|issue:45593", - "issue:29942|issue:36010", "issue:43994|issue:46369", + "issue:29942|issue:36010", "issue:43526|issue:45834", "issue:29127|issue:35141", "issue:43526|issue:44186", "issue:44625|issue:45584", "issue:43761|issue:45859", "issue:43575|issue:43701", - "issue:44442|issue:45072", "issue:42915|issue:43931", + "issue:44442|issue:45072", "issue:42490|issue:43381", "issue:43575|issue:44387", - "issue:43334|issue:43531", "issue:44977|issue:45245", + "issue:43334|issue:43531", "issue:43526|issue:44857", "issue:44464|issue:45491", - "issue:43262|issue:44206" + "issue:43262|issue:44206", + "issue:44743|issue:44811", + "issue:43452|issue:45588" ], - "prepared_review_unit_hash": "8635b74dfce5ec0df88243042b3ce66050c1a1dd3b4e0a21fc9c70c667ea089b", + "prepared_review_unit_hash": "282496423c60409b5b809ca68fc67a1709cd26b5276936cb84b7d8be2fd8dd97", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36944, - "estimated_input_tokens": 9236, - "estimated_eval_tokens": 18728 + "soft_pair_count": 30, + "serialized_chars": 36644, + "estimated_input_tokens": 9161, + "estimated_eval_tokens": 18578 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Loose issue cluster of mostly unrelated Transformers regressions and model-specific bugs. None of the soft edges look like true duplicates; the closest subtheme is tokenizer/loading behavior, with issue 44462 as the best representative.", - "confidence": 0.71, - "canonical_issue_reason": "No strong duplicate cluster emerges. issue:44462 is the cleanest, most broadly applicable loader regression in the set (AutoTokenizer ignoring tokenizer.json), so it is the best issue representative.", + "summary": "The cluster is heterogeneous: most items are distinct model-, tokenizer-, or runtime-specific bugs rather than one shared underlying defect. I found no soft pair that should be merged. The broadest representative issue is the AutoTokenizer repository-loading regression.", + "confidence": 0.66, + "canonical_issue_reason": "issue:44462 is the most general, reusable regression in the set: it affects AutoTokenizer loading from repository files and has downstream references, making it the best anchor for a tokenizer/loading subcluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44462 is user-visible, reproducible, and broader than the model-specific edge cases here, making it the most suitable standalone issue to keep as the representative of the cluster's closest theme.", + "best_issue_reason": "issue:44462 is the clearest cross-model loading bug among the candidates; it is broader and more central than the model-specific failures in the rest of the set.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:29942", - "right": "issue:30064", - "accept": false, - "reason": "Unrelated failures: Flash Attention tests vs image processor handling of void segmentation maps." - }, - { - "left": "issue:45372", - "right": "issue:45459", - "accept": false, - "reason": "Different tokenizer/processor loading problems with different causes: missing ReasoningEffort import vs protobuf error masking." - }, - { - "left": "issue:43278", - "right": "issue:43575", - "accept": false, - "reason": "Different bug classes: dtype change in eval vs tensor-parallel load OOM." - }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Both touch tokenizer loading, but one is decoder-type regression and the other is AutoTokenizer source selection; not the same bug." + "reason": "Both involve tokenization behavior, but one is a decoder-type regression in v5 and the other is repository tokenizer loading; different code paths and failures." }, { "left": "issue:45412", "right": "issue:46369", "accept": false, - "reason": "Both mention memory, but one is cleanup on deletion and the other is overflow during model use; different code paths." + "reason": "Both mention memory, but one is a deletion/GC leak in RT-DETR and the other is an overflow while running sam3-litetext; not the same bug." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "Different model/runtime issues: nonsensical outputs vs memory release on deletion." + "reason": "Unrelated issues: SigLIP inference quality vs RT-DETR memory cleanup." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export failure with torch_compilable_check is unrelated to save_pretrained/tokenization failure." + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failures." }, { - "left": "issue:43381", - "right": "issue:43701", + "left": "issue:43756", + "right": "issue:45440", "accept": false, - "reason": "Gradient checkpointing in eval mode is unrelated to resume_from_checkpoint key mismatch." + "reason": "Both are model-specific implementation mismatches, but they affect different architectures and different internal mechanisms." }, { - "left": "issue:43756", - "right": "issue:45440", + "left": "issue:43381", + "right": "issue:43701", "accept": false, - "reason": "Different model implementations and different divergence symptoms; not mergeable as one fix." + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate training-time problems." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "Special-token post-processor update bug is unrelated to NaN embeddings in mixed-length GPU batching." + "reason": "Runtime special-token/post-processor updates are unrelated to Gemma3 NaNs from sliding-window all-padding batches." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "BigBird special-token registration bug is not the same as AutoTokenizer ignoring tokenizer.json." + "reason": "Both are tokenizer-related, but one is BigBird special-token registration and the other is AutoTokenizer failing to load a specific tokenizer class." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 error-message clarity issue and CLIP MacOS bus error are unrelated." + "reason": "LayoutLMv3 error messaging and a MacOS CLIP bus error are unrelated." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "ImportError for a missing model class is not the same as a generic dependency issue with transformers." + "reason": "A missing Qwen2_5_VL import is not the same as a generic dependency issue." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "DeepseekV3MoE implementation divergence is unrelated to D-FINE auxiliary-loss behavior." + "reason": "DeepSeekV3 implementation divergence and D-FINE auxiliary-loss behavior are different model bugs." }, { - "left": "issue:29942", - "right": "issue:36010", + "left": "issue:43994", + "right": "issue:46369", "accept": false, - "reason": "Failing Flash Attention tests and GenerationMixin import failure are unrelated." + "reason": "Different models and different failure modes: bad outputs vs memory overflow." }, { - "left": "issue:43994", - "right": "issue:46369", + "left": "issue:29942", + "right": "issue:36010", "accept": false, - "reason": "Different symptoms and models: SigLIP2 bad outputs vs sam3-litetext memory overflow." + "reason": "Failing Flash Attention 2 tests are not the same as a GenerationMixin import error." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "BEiT label reduction bug is unrelated to Kosmos2.5 OCR index errors." + "reason": "Beit label reduction and Kosmos2.5 OCR index errors are distinct image-processing issues." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "Error-message clarity for LayoutLMv3 is unrelated to embedding reinitialization in post_init." + "reason": "An error-message clarity bug is unrelated to token-embedding reinitialization." }, { "left": "issue:43526", "right": "issue:44186", "accept": false, - "reason": "ImageProcessor label reduction and LayoutLMv2 tokenizer crashes are different bugs." + "reason": "Both touch vision preprocessing, but one is label reduction in BeitImageProcessorFast and the other is LayoutLMv2 tokenization/padding crashes." }, { "left": "issue:44625", "right": "issue:45584", "accept": false, - "reason": "num_labels propagation in Qwen3.5 is unrelated to Whisper empty-transcription generation failures." + "reason": "Qwen3.5 num_labels propagation and Whisper empty-transcription generation are unrelated." }, { "left": "issue:43761", "right": "issue:45859", "accept": false, - "reason": "CLIPVision hidden_states regression is unrelated to Qwen3_5Moe CPU-offload compatibility." + "reason": "Hidden-states forwarding in CLIPVisionModel is unrelated to Qwen3_5Moe CPU-offload incompatibility." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "Tensor-parallel OOM and resume-from-checkpoint key mismatch do not share a concrete code path." + "reason": "A Qwen2 model-load OOM is not the same as a checkpoint key mismatch." }, { - "left": "issue:44442", - "right": "issue:45072", + "left": "issue:42915", + "right": "issue:43931", "accept": false, - "reason": "Tokenizer loading failure and bfloat16 dtype mismatch are unrelated." + "reason": "FineGrainedFP8Config failure and Qwen3-VL weight-shape mismatch are separate Qwen model issues." }, { - "left": "issue:42915", - "right": "issue:43931", + "left": "issue:44442", + "right": "issue:45072", "accept": false, - "reason": "FineGrainedFP8Config failure for Qwen3Moe and Qwen3-VL shape mismatch are different model-loading bugs." + "reason": "Tokenizer loading and bfloat16 dtype mismatches are unrelated." }, { "left": "issue:42490", "right": "issue:43381", "accept": false, - "reason": "Trailing-slash import warning is unrelated to gradient checkpointing eval-mode behavior." + "reason": "A local-path import warning is unrelated to gradient checkpointing behavior." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "Both involve OOM, but one is tp loading and the other is int4 reserved-memory growth; too different to merge." + "reason": "Both are OOM-related, but one is a specific model load with tensor parallelism and the other is a global int4 quantization memory regression." }, { - "left": "issue:43334", - "right": "issue:43531", + "left": "issue:44977", + "right": "issue:45245", "accept": false, - "reason": "Missing pad_token_id in Qwen3-VL is unrelated to Qwen3-MoE sliding_window behavior." + "reason": "Qwen3.5 flash-attention generation and category-count limits are unrelated." }, { - "left": "issue:44977", - "right": "issue:45245", + "left": "issue:43334", + "right": "issue:43531", "accept": false, - "reason": "Flash-attention generation bug is unrelated to the categories > 2^24 runtime limit." + "reason": "Missing pad_token_id for Qwen3-VL and sliding-window behavior in Qwen3-MoE are different bugs." }, { "left": "issue:43526", "right": "issue:44857", "accept": false, - "reason": "Label-array reduction bug and float16 AMP crash are unrelated." + "reason": "Beit label reduction and LwDetrImageLoss AMP crashes do not share the same code path." }, { "left": "issue:44464", "right": "issue:45491", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is unrelated to mixed-length NaN embeddings." + "reason": "Compiled-forward generation inconsistency and Gemma3 NaNs from padding/windowing are different failure modes." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Default 16kHz sample-rate handling is unrelated to the unsupported center argument crash." + "reason": "Audio chat-template sampling-rate defaults and a feature-extractor center-arg crash are distinct audio bugs." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen recurrent-state reset and Whisper batch_decode skip_special_tokens handling are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:45588", + "accept": false, + "reason": "GGUF loading failures and flash-attention sinkless-model crashes involve different subsystems and symptoms." } ] }, @@ -9389,10 +9455,10 @@ "issue:43316", "issue:43381", "issue:43388", - "issue:43452", "issue:43493", "issue:43525", "issue:43611", + "issue:43643", "issue:43646", "issue:43720", "issue:43723", @@ -9408,40 +9474,38 @@ "issue:44442", "issue:44479", "issue:44743", - "issue:44811", "issue:44871", "issue:44898", + "issue:44913", "issue:44987", "issue:45003", "issue:45216", "issue:45290", "issue:45440", "issue:45479", - "issue:45588", "issue:45698", + "issue:45759", "issue:45803", "issue:45907", "issue:46178", "issue:46234" ], "soft_pairs": [ - "issue:44743|issue:44811", - "issue:43452|issue:45588", "issue:43646|issue:44077", "issue:33290|issue:34689", - "issue:33290|issue:34567", "issue:44062|issue:45907", + "issue:33290|issue:34567", "issue:33290|issue:35141", "issue:43381|issue:43873", - "issue:44442|issue:44898", "issue:45003|issue:45440", - "issue:43761|issue:44079", + "issue:44442|issue:44898", "issue:43723|issue:45479", + "issue:43761|issue:44079", "issue:42490|issue:44743", "issue:43316|issue:43525", + "issue:43232|issue:43388", "issue:34689|issue:43611", "issue:42175|issue:46234", - "issue:43232|issue:43388", "issue:43720|issue:44442", "issue:43761|issue:44188", "issue:43931|issue:45216", @@ -9454,216 +9518,218 @@ "issue:44442|issue:45803", "issue:42907|issue:45698", "issue:43873|issue:44871", - "issue:43749|issue:43931" + "issue:43749|issue:43931", + "issue:43643|issue:44913", + "issue:43316|issue:45759" ], - "prepared_review_unit_hash": "3c56b08f1e14adeed3d53a9bedd52ed194b7f7757fc6a6bd6f71b3629a787d6b", + "prepared_review_unit_hash": "10638528ff09d95bb91ebdc8fb040ee9eab7a627cd796d174195f30cce3e01db", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36305, - "estimated_input_tokens": 9077, - "estimated_eval_tokens": 18410 + "serialized_chars": 36329, + "estimated_input_tokens": 9083, + "estimated_eval_tokens": 18422 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is noisy and mostly mixes unrelated Transformers regressions across loading, generation, tokenization, and model-output behavior. None of the soft pairs look like the same underlying bug/change closely enough to merge or dedupe.", - "confidence": 0.89, - "canonical_issue_reason": "issue:45698 is the best representative anchor: it is an active, broad from_pretrained/save_pretrained custom-module loading regression with clear user impact and good centrality among the loading-related issues.", + "summary": "The cluster is heterogeneous: most items are unrelated Transformers regressions across loading, config, generation, tokenization, and docs. I did not find any soft pair that looks like the same underlying bug/change, so all soft edges should be rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the best broad representative of the cluster because it concerns a general Transformers 5.0 custom initialization regression, but the items are not true duplicates of one another.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45698 is the strongest standalone issue in the set because it is open, specific, actionable, and closest to the common loading/saving regression theme that appears repeatedly in the cluster.", + "best_issue_reason": "issue:43646 is the most representative issue by breadth and centrality, but only as a loose theme anchor rather than a duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44743", - "right": "issue:44811", - "accept": false, - "reason": "Different code paths: recurrent-state reset in modular Qwen vs Whisper processor batch_decode skip_special_tokens handling." - }, - { - "left": "issue:43452", - "right": "issue:45588", - "accept": false, - "reason": "One is tokenizer/model loading via gguf, the other is a flash_attention crash on s_aux=None; not the same bug." - }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "Broad v5 custom model init regression vs patchtsmixer post_init policy change; related area but not one concrete defect." + "reason": "Both mention initialization/post_init behavior, but they describe different models and different bugs." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "OOM with Adafactor/DeepSpeed is unrelated to Llama 3.2 Vision model loading breakage." + "reason": "OOM with Adafactor/DeepSpeed is unrelated to Llama 3.2 model loading." }, { - "left": "issue:33290", - "right": "issue:34567", + "left": "issue:44062", + "right": "issue:45907", "accept": false, - "reason": "Optimizer memory issue and TrainerState token counter bug are unrelated." + "reason": "Tokenizers AddedToken keyword collision is unrelated to list_repo_files revision forwarding." }, { - "left": "issue:44062", - "right": "issue:45907", + "left": "issue:33290", + "right": "issue:34567", "accept": false, - "reason": "AddedToken special-argument duplication vs list_repo_files not forwarding revision are separate APIs and failures." + "reason": "Optimizer memory usage and TrainerState token counting are unrelated." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "OOM during training is not the same as output embedding reinitialization after resize_token_embeddings." + "reason": "OOM in training is unrelated to embedding reinitialization during post_init." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient checkpointing eval-mode behavior and quantization offloading are different subsystems and failure modes." - }, - { - "left": "issue:44442", - "right": "issue:44898", - "accept": false, - "reason": "Tokenizer loading failure vs Perceiver image classification with interpolate_pos_encoding are not the same bug." + "reason": "Gradient checkpointing in eval mode and quantization offloading are different failure modes." }, { "left": "issue:45003", "right": "issue:45440", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils is unrelated to DeepseekV3MoE behavioral divergence." + "reason": "Unsafe sys.modules access and DeepSeekV3 divergence are different code paths." }, { - "left": "issue:43761", - "right": "issue:44079", + "left": "issue:44442", + "right": "issue:44898", "accept": false, - "reason": "CLIPVision hidden_states regression and ModelOutput key assignment are distinct issues." + "reason": "Tokenizer loading failure and Perceiver image classification resizing failure are unrelated." }, { "left": "issue:43723", "right": "issue:45479", "accept": false, - "reason": "Tokenizer loading in v5 and zero-loss sequence classification are unrelated bug classes." + "reason": "AutoTokenizer loading in v5 and sequence-classification zero-loss behavior are unrelated bugs." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression and ModelOutput key assignment bug are not the same issue." }, { "left": "issue:42490", "right": "issue:44743", "accept": false, - "reason": "Local-path slash warning is unrelated to recurrent-state reset with cache." + "reason": "Local-path slash warning and Qwen recurrent-state reset are unrelated." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Llama4Config pad_token_id missing are different config-model mismatches." + "reason": "Both are config API issues, but for different models and different missing fields." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs handling and gather_for_metrics label truncation are unrelated." }, { "left": "issue:34689", "right": "issue:43611", "accept": false, - "reason": "Both involve model loading, but they are different concrete regressions: Llama 3.2 Vision loading vs base_model_prefix handling." + "reason": "Llama 3.2 Vision loading failure and base_model_prefix loading regression are different bugs." }, { "left": "issue:42175", "right": "issue:46234", "accept": false, - "reason": "Missing tensorflow backend in extras install is not the same as a missing import_utils helper symbol." - }, - { - "left": "issue:43232", - "right": "issue:43388", - "accept": false, - "reason": "Generation kwargs after sync_gpus and last-batch label truncation are unrelated." + "reason": "Missing tensorflow backend in install extras and missing import_utils helper are unrelated packaging issues." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "BitNet packed-weight unpacking during accelerate loading is unrelated to AutoTokenizer loading FastSpeech2ConformerTokenizer." + "reason": "BitNet packed-weight loading and tokenizer loading failure are unrelated." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "CLIPVision hidden_states regression is unrelated to attention-kernel divergence under torch.compile." + "reason": "Vision model hidden_states regression and attention-kernel divergence under torch.compile are unrelated." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and Qwen3.5 save_pretrained regression are separate model-loading/saving bugs." + "reason": "Qwen3-VL weight-shape mismatch on load and Qwen3.5 save_pretrained regression are different bugs." }, { "left": "issue:34689", "right": "issue:36010", "accept": false, - "reason": "Llama model loading breakage is not the same as the GenerationMixin import error." + "reason": "Model loading breakage and GenerationMixin import error are unrelated." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "Optional post_init allowance in patchtsmixer and video-input regression in Qwen VL models are unrelated." + "reason": "PatchTSMixer post_init policy and Qwen video-input regression are unrelated." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation discrepancy and docs still referencing pipeline() are not the same defect." + "reason": "SigLIP2 implementation discrepancy and docs still referencing pipeline() are unrelated." }, { "left": "issue:33290", "right": "issue:36331", "accept": false, - "reason": "Adafactor/DeepSpeed OOM and CustomTrainer.compute_loss signature mismatch are unrelated training bugs." + "reason": "Adafactor/DeepSpeed OOM and custom Trainer compute_loss signature mismatch are unrelated." }, { "left": "issue:44987", "right": "issue:46178", "accept": false, - "reason": "Loading physical-intelligence/fast on v5.1.0 and tensorflow_text backend mapping are different installation/loading problems." + "reason": "Physical-intelligence/fast loading failure and tensorflow_text backend mapping error are unrelated." }, { "left": "issue:44479", "right": "issue:45290", "accept": false, - "reason": "Video-input regression for Qwen VL models and chat-template crashes on tool-call assistant messages are different code paths." + "reason": "Video-input regression and chat-template crash on tool-call messages are unrelated." }, { "left": "issue:44442", "right": "issue:45803", "accept": false, - "reason": "Tokenizer loading failure and FuyuBatchFeature bare-except error are unrelated." + "reason": "Tokenizer loading failure and Fuyu bare-except error are different bugs." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals and wrong custom-module loading after save_pretrained are not the same bug." + "reason": "Dequantized save failure and wrong custom-module loading after save_pretrained are both save/load themed but not the same underlying bug." }, { "left": "issue:43873", "right": "issue:44871", "accept": false, - "reason": "Quantization offloading behavior and Gemma eos_token_id inconsistency are distinct issues." + "reason": "Quantization/offloading behavior and Gemma-3 eos_token_id inconsistency are unrelated." }, { "left": "issue:43749", "right": "issue:43931", "accept": false, - "reason": "FSDP CPU RAM efficient loading and Qwen3-VL shape mismatch are different loading failures." + "reason": "FSDP CPU RAM efficient loading and Qwen3-VL weight mismatch are different load-path failures." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "trust_remote_code missing fields and rotary_pct reverting on reload are both config-related but distinct bugs." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both involve config handling, but one is a Gemma3TextConfig API discrepancy and the other is unwrapping text_config for multimodal Qwen models." } ] }, @@ -9699,12 +9765,13 @@ "issue:43388", "issue:43404", "issue:43441", + "issue:43452", "issue:43475", "issue:43540", "issue:43550", "issue:43575", "issue:43606", - "issue:43643", + "issue:43646", "issue:43650", "issue:43688", "issue:43701", @@ -9728,7 +9795,6 @@ "issue:44913", "issue:45137", "issue:45584", - "issue:45759", "issue:45803", "issue:46050", "issue:46170", @@ -9737,18 +9803,16 @@ "issue:46369" ], "soft_pairs": [ - "issue:43643|issue:44913", - "issue:43316|issue:45759", "issue:44190|issue:44496", "issue:44855|issue:45137", "issue:43650|issue:46202", "issue:45803|issue:46365", "issue:25251|issue:28282", "issue:43295|issue:43825", - "issue:42617|issue:43575", "issue:43550|issue:43701", - "issue:43065|issue:44617", + "issue:42617|issue:43575", "issue:43606|issue:43701", + "issue:43065|issue:44617", "issue:43575|issue:43742", "issue:16998|issue:22355", "issue:43540|issue:44514", @@ -9764,198 +9828,206 @@ "issue:44871|issue:46050", "issue:43388|issue:43688", "issue:43761|issue:44479", - "issue:41950|issue:43475" + "issue:41950|issue:43475", + "issue:43646|issue:44913", + "issue:43452|issue:44496" ], - "prepared_review_unit_hash": "81725582b4d17bbe39a2aef7c9546fad89a63623ed9ab7b6b33b32acc9c7cda2", + "prepared_review_unit_hash": "ea791a0b5b9cc70dfad39efdde92bacd30eab91ed9dc61aa3221a60fec95c75d", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 35892, - "estimated_input_tokens": 8973, - "estimated_eval_tokens": 18202 + "serialized_chars": 35819, + "estimated_input_tokens": 8955, + "estimated_eval_tokens": 18166 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is dominated by unrelated transformer bugs and regressions with only superficial topical overlap. None of the soft pairs look like the same concrete bug or change, so all pair candidates are rejected.", - "confidence": 0.41, - "canonical_issue_reason": "issue:41950 is the strongest standalone bug report here: it is open, has active discussion and inbound references, and describes a concrete, reproducible pipeline regression. It is the best canonical anchor among these items.", + "summary": "This cluster is a broad mix of unrelated Transformers issues; the soft-similarity links are mostly false positives across different models, environments, and failure modes.", + "confidence": 0.94, + "canonical_issue_reason": "No single issue is a good canonical representative: the items span unrelated bugs in pipelines, configs, model loading, runtime errors, and model-specific regressions.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41950 is the most suitable global issue anchor because it is specific, current, and has the most external signal among the listed issues.", + "best_issue_reason": "No issue stands out as a strong global representative for deduplication; the cluster is too heterogeneous to map to one underlying bug.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43643", - "right": "issue:44913", - "accept": false, - "reason": "Both are config round-trip problems, but one is about trust_remote_code loading and missing fields while the other is a GPTNeoX rotary_pct reload regression; different code paths and failures." - }, - { - "left": "issue:43316", - "right": "issue:45759", - "accept": false, - "reason": "Both concern multimodal/text_config API handling, but one is a Gemma3TextConfig inconsistency and the other is AutoModelForCausalLM.from_config failing to unwrap text_config for Qwen 3.5/3.6; not the same bug." - }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Completely different failures: local dataset loading in an example script versus an unrecognized model/config error." + "reason": "Different problems: local dataset loading in a training script vs an unrecognized model/config error when loading a checkpoint." }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Unrelated issues: Python 3.13 torch.jit parsing/IndentationError versus DeepSpeed ZeRO3 deque underflow." + "reason": "Unrelated failures: Python 3.13 import/annotation parsing vs a DeepSpeed ZeRO3 deque underflow." }, { "left": "issue:43650", "right": "issue:46202", "accept": false, - "reason": "No meaningful overlap: a placeholder issue title versus a Gemma pipeline roundtrip bug." + "reason": "No overlap in bug shape; one is a low-information post, the other is a Gemma pipeline roundtrip issue." }, { "left": "issue:45803", "right": "issue:46365", "accept": false, - "reason": "Different subsystems and symptoms: exception handling in Fuyu batch conversion versus Trainer import failure caused by datasets." + "reason": "Different code paths: Fuyu tensor conversion swallowing exceptions vs Trainer import being affected by installed datasets." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Different areas entirely: pipeline top_k output shape versus missing PyTorch dependency for AutoModel." + "reason": "Pipeline output-shape bug vs missing PyTorch dependency on AutoModel import; not the same underlying issue." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Both are regressions in v5-era processor/pipeline behavior, but they affect different APIs and different failures; not one concrete bug." + "reason": "One is a processor/tokenizer regression, the other is an error-message/docstring issue about translation support." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and resume checkpoint key mismatch are unrelated." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "Unrelated: a script execution failure versus loading a large Qwen model with tensor-parallel OOM." + "reason": "A script execution problem is not the same as an OOM when loading a large Qwen model with tensor parallelism." }, { - "left": "issue:43550", + "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "Different bugs: torch.compile/SDPA failure in Bamba versus checkpoint key mismatch on resume." + "reason": "Device-mismatch CPU offload bug vs checkpoint key mismatch; different failure modes and code paths." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "Same model family name, but one is a dummy Conv2d concern in Sam3PixelDecoder and the other is a CUDA OOM in Sam3Video; not the same bug." - }, - { - "left": "issue:43606", - "right": "issue:43701", - "accept": false, - "reason": "No shared underlying issue: CPU offload device mismatch versus resume_from_checkpoint key mismatch." + "reason": "Dummy Conv2d implementation issue vs SAM3 video OOM; not a duplicate." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "A tensor-parallel OOM and a key error while loading MobileLLM are distinct failures." + "reason": "Large-model OOM vs MobileLLM key loading error; unrelated symptoms and root causes." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "Completely unrelated: model_max_length behavior versus missing transformers.onnx module." + "reason": "Model_max_length behavior question and missing transformers.onnx module are unrelated." }, { "left": "issue:43540", "right": "issue:44514", "accept": false, - "reason": "Both are Qwen multimodal processing bugs, but one is a video-input ValueError in Qwen3OmniMoe and the other is batched apply_chat_template padding behavior in Qwen2_5_VL; different code paths." + "reason": "Qwen3OmniMoe video-processing ValueError differs from a batched chat-template padding crash." }, { "left": "issue:43316", "right": "issue:43716", "accept": false, - "reason": "Different model-specific problems: Gemma3TextConfig API inconsistency versus Mistral-3 image preprocessor dtype mismatch." + "reason": "API/config discrepancy in Gemma3TextConfig is unrelated to an image-preprocessor dtype mismatch." }, { "left": "issue:44871", "right": "issue:45584", "accept": false, - "reason": "Different models and defects: Gemma eos_token_id mismatch versus Whisper empty-transcription generation after align_special_tokens." + "reason": "Gemma eos_token_id inconsistency and Whisper empty-transcription handling are separate model-specific bugs." }, { "left": "issue:44610", "right": "issue:46369", "accept": false, - "reason": "Unrelated issues: OmDet-Turbo image-size mismatch versus SAM3 LiteText memory overflow." + "reason": "Processor size mismatch for OmDet-Turbo vs memory overflow in sam3-litetext; different root causes." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Both involve processor/model config regressions, but the concrete bugs are different and not mergeable as one fix." + "reason": "Processor/tokenizer regression and Gemma eos_token_id config mismatch do not share a code path." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Gemma3TextConfig API mismatch is not the same as strict config rejection for granite_speech." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "Different scope: ESM2 model breakage versus a janus test failure." + "reason": "ESM2 breakage and a janus image-generation test failure are unrelated." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "One is about optional post_init being disallowed for patchtsmixer; the other is about non-persistent buffers being filled with junk. Different issues." + "reason": "Optional post_init allowance vs junk-filled non-persistent buffers are distinct v5 issues." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "Related model family, but one is lm_head weight tying and the other is FlashAttention failure; distinct concrete bugs." + "reason": "Weight tying bug in Mistral3 is unrelated to Ministral FlashAttention failures." }, { "left": "issue:45803", "right": "issue:46170", "accept": false, - "reason": "Different concerns: Fuyu exception swallowing versus Windows/ROCm import dependency issues." + "reason": "Bare except in Fuyu tensor conversion and Windows/ROCm torch.distributed import issues are unrelated." }, { "left": "issue:44871", "right": "issue:46050", "accept": false, - "reason": "No direct relation: Gemma token ID inconsistency versus quantized model save/load breakage." + "reason": "Gemma eos_token_id config mismatch is not the same as sinq quantized save/load breakage." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Both touch training/metrics behavior, but one is gather_for_metrics label truncation and the other is auxiliary-loss normalization; not the same bug." + "reason": "Metric gathering truncation bug and auxiliary-loss normalization bug affect different training internals." }, { "left": "issue:43761", "right": "issue:44479", "accept": false, - "reason": "Different regressions: CLIPVisionModel hidden_states handling versus Qwen video input regression." + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are different model families and code paths." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Different SAM/video-related failures: pipeline processor selection versus missing fpn_position_embeddings on a vision encoder output." + "reason": "Video-classification processor selecting image processors is unrelated to SAM3 output missing fpn_position_embeddings." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage in v5 and GPTNeoX rotary_pct not persisting are separate config/init issues." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "gguf_file handling in from_pretrained is unrelated to an unrecognized model_type in config.json." } ] }, @@ -9984,23 +10056,22 @@ "issue:41720", "issue:42490", "issue:43065", - "issue:43452", "issue:43504", "issue:43525", "issue:43531", "issue:43572", - "issue:43646", + "issue:43644", "issue:43673", "issue:43761", "issue:43827", "issue:43931", + "issue:43976", "issue:43994", "issue:44060", "issue:44190", "issue:44242", "issue:44464", "issue:44479", - "issue:44496", "issue:44521", "issue:44534", "issue:44610", @@ -10010,11 +10081,13 @@ "issue:44792", "issue:44855", "issue:44913", + "issue:44933", "issue:44964", "issue:44977", "issue:45127", "issue:45200", "issue:45245", + "issue:45362", "issue:45381", "issue:45406", "issue:45542", @@ -10028,231 +10101,238 @@ "issue:46369" ], "soft_pairs": [ - "issue:43646|issue:44913", - "issue:43452|issue:44496", - "issue:45127|issue:45200", "issue:43673|issue:44743", + "issue:45127|issue:45200", "issue:32090|issue:35141", "issue:43572|issue:46178", - "issue:44617|issue:45584", "issue:43504|issue:45698", + "issue:44617|issue:45584", "issue:44734|issue:45406", "issue:45542|issue:46234", - "issue:43525|issue:43531", "issue:45563|issue:45584", + "issue:43525|issue:43531", "issue:44792|issue:44977", "issue:43761|issue:44242", - "issue:45381|issue:46369", "issue:43931|issue:44060", + "issue:45381|issue:46369", "issue:44534|issue:44913", - "issue:44792|issue:45600", "issue:44190|issue:44964", + "issue:44792|issue:45600", "issue:44464|issue:45563", "issue:35141|issue:42490", - "issue:43827|issue:43994", "issue:44479|issue:44734", + "issue:43827|issue:43994", "issue:44855|issue:45588", - "issue:44617|issue:46369", "issue:41720|issue:45245", + "issue:44617|issue:46369", + "issue:43065|issue:44610", "issue:44521|issue:45381", - "issue:43065|issue:44610" + "issue:44933|issue:45362", + "issue:43644|issue:44913", + "issue:43976|issue:44855" ], - "prepared_review_unit_hash": "af901f971abb35a457eb89802116dd18f3eb26c6bd1d969d65fb903756acedb6", + "prepared_review_unit_hash": "8b926d133c9d10ec5aafaa61768c9fc8c82f0d31b5ca7c54704548d8b8fb23e5", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 28, - "serialized_chars": 35645, - "estimated_input_tokens": 8912, - "estimated_eval_tokens": 18080 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36375, + "estimated_input_tokens": 9094, + "estimated_eval_tokens": 18444 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose semantic grouping of Transformers issues around v5 regressions, loading/configuration edge cases, and model-specific bugs. I don\u2019t see any true duplicate pairs among the soft edges; the titles point to different code paths, models, or failure modes.", - "confidence": 0.37, - "canonical_issue_reason": "issue:43646 is the broadest, most umbrella-like report in the set: \"Transformers 5.0.0 breaks custom model initialization.\" It best captures the cluster\u2019s general v5 regression theme, even though the cluster itself is not a tight duplicate set.", + "summary": "This cluster is mostly unrelated issue reports spanning different models, backends, and failure modes. The only obvious duplicate family is the Transformers v5 non-persistent-buffer regression, with 44534 being the cleanest representative.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44534 is the clearest, most generic statement of the recurring v5 non-persistent-buffer bug and is a better canonical representative than the more version-specific wording in 43644.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43646 is the best single representative issue because it is high-level and spans the widest portion of the shared theme (v5 initialization/loading regressions) without being tied to one narrow model or API surface.", + "best_issue_reason": "issue:44534 is the best single issue to keep because it most directly captures the underlying regression that appears to have a near-duplicate sibling (43644) and is phrased broadly enough to anchor that family.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43646", - "right": "issue:44913", - "accept": false, - "reason": "Broad v5 initialization regression vs a GPTNeoXConfig reload defaulting bug; different failure modes and code paths." - }, - { - "left": "issue:43452", - "right": "issue:44496", + "left": "issue:43673", + "right": "issue:44743", "accept": false, - "reason": "GGUF loading path breakage in tokenizer/model loaders vs generic unrecognized model_type/config loading failure." + "reason": "Both mention cache/state behavior, but one is a missing GenerationMixin cache during chunked_prefill while the other is recurrent states resetting in modular_qwen3_5; different bugs and code paths." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "LoRA merge with extended vocab collapse vs Gemma 4 token-type defaulting for text-only fine-tuning; unrelated bugs." - }, - { - "left": "issue:43673", - "right": "issue:44743", - "accept": false, - "reason": "Missing GenerationMixin cache in chunked_prefill vs recurrent states resetting in modular_qwen3_5 with seq_len>1; different internals." + "reason": "LoRA merge/tie_word_embeddings corruption and Gemma 4 mm_token_type_ids defaulting are unrelated model-loading/training issues." }, { "left": "issue:32090", "right": "issue:35141", "accept": false, - "reason": "Trainer GPU broadcast TypeError vs embedding reinitialization after resize; not the same underlying defect." + "reason": "Trainer NoneType GPU broadcast failure is unrelated to output embeddings being reinitialized after resizing token embeddings." }, { "left": "issue:43572", "right": "issue:46178", "accept": false, - "reason": "StableLmConfig pad_token_idx regression vs backend registry missing tensorflow_text; unrelated configuration/import issues." + "reason": "Missing pad_token_idx in StableLmConfig and an undefined tensorflow_text backend entry are unrelated configuration/import problems." }, { - "left": "issue:44617", - "right": "issue:45584", + "left": "issue:43504", + "right": "issue:45698", "accept": false, - "reason": "Sam3Video memory/OOM issue vs Whisper empty transcription failure after align_special_tokens; different models and symptoms." + "reason": "A legacy field breaking Beit preset loading is not the same as from_pretrained selecting the wrong custom module after save_pretrained." }, { - "left": "issue:43504", - "right": "issue:45698", + "left": "issue:44617", + "right": "issue:45584", "accept": false, - "reason": "Legacy field breaking Beit semantic segmentation model preset loading vs custom module selection after save/load; different loaders." + "reason": "Sam3Video CUDA OOM and Whisper empty-transcription generation failure are different models and different failure modes." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "KV-cache continuation tensor indexing crash in serve vs Gemma4Processor missing _tokenizer in serve; different crashes." + "reason": "Wrong KV-cache tensor indexing in transformers serve is unrelated to Gemma4Processor missing _tokenizer." }, { "left": "issue:45542", "right": "issue:46234", "accept": false, - "reason": "TensorFlow backend detection issue from tensorboard-only install vs missing is_falcon_mamba_ssm_available helper; unrelated import-utils bugs." + "reason": "TensorBoard/TensorFlow backend setup error and missing is_falcon_mamba_ssm_available helper are unrelated import/system issues." }, { - "left": "issue:43525", - "right": "issue:43531", + "left": "issue:45563", + "right": "issue:45584", "accept": false, - "reason": "Llama4Config missing pad_token_id vs Qwen3-MoE sliding_window behavior; not the same bug." + "reason": "A stale paged-generate warning about num_return_sequences is unrelated to Whisper failing on empty transcription after align_special_tokens." }, { - "left": "issue:45563", - "right": "issue:45584", + "left": "issue:43525", + "right": "issue:43531", "accept": false, - "reason": "Paged generate warning cleanup vs Whisper alignment failure on empty transcription; different generation paths." + "reason": "Llama4Config missing pad_token_id and Qwen3-MoE sliding_window behavior are different config bugs." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Janus image-generation test failure vs Qwen3.5 flash-attention generation bug; different models/code paths." + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are different model-specific problems." }, { "left": "issue:43761", "right": "issue:44242", "accept": false, - "reason": "CLIPVisionModel hidden_states regression vs load-balancing loss omission when router logits are off; unrelated behavior." + "reason": "CLIPVisionModel returning hidden_states=None and missing load-balancing loss are not the same bug." }, { - "left": "issue:45381", - "right": "issue:46369", + "left": "issue:43931", + "right": "issue:44060", "accept": false, - "reason": "Qwen2.5-VL video position IDs bug vs Sam3-LiteText memory overflow; not the same underlying issue." + "reason": "Qwen3-VL weight shape mismatch and a wrong tied-weights warning in Qwen3-Next are distinct issues." }, { - "left": "issue:43931", - "right": "issue:44060", + "left": "issue:45381", + "right": "issue:46369", "accept": false, - "reason": "Qwen3-VL weight shape mismatch on load vs incorrect tied-weights warning in Qwen3-Next; different model-loading problems." + "reason": "Qwen2.5-VL video vision_position_ids regression and Sam3-LiteText memory overflow are unrelated." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffers filled with junk vs GPTNeoX rotary_pct reload defaulting; separate v5 regressions." + "reason": "Non-persistent buffers filled with junk and GPTNeoX rotary_pct reverting on reload are different config/state bugs." }, { - "left": "issue:44792", - "right": "issue:45600", + "left": "issue:44190", + "right": "issue:44964", "accept": false, - "reason": "Janus test failure vs stale auto_mappings references to removed Sam3LiteText configs; unrelated issues." + "reason": "Local dataset loading in no-trainer image classification and Phi-4 multimodal model loading errors are unrelated." }, { - "left": "issue:44190", - "right": "issue:44964", + "left": "issue:44792", + "right": "issue:45600", "accept": false, - "reason": "Local dataset loading in a training script vs loading a specific multimodal model with latest transformers; different scopes." + "reason": "A Janus image-generation test failure is unrelated to auto_mappings referencing removed Sam3LiteText configs." }, { "left": "issue:44464", "right": "issue:45563", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency vs stale num_return_sequences warning; not the same bug." + "reason": "Chunked generation inconsistencies with compiled forward and a stale num_return_sequences warning do not describe the same problem." }, { "left": "issue:35141", "right": "issue:42490", "accept": false, - "reason": "Embedding reinit after resize vs trailing-slash local model import warning; distinct loading/initialization concerns." + "reason": "Resizing token embeddings reinitializing the output embedding is unrelated to a local-path trailing-slash import warning." }, { - "left": "issue:43827", - "right": "issue:43994", + "left": "issue:44479", + "right": "issue:44734", "accept": false, - "reason": "Docs still referencing removed pipeline() vs SigLIP2 model producing nonsensical outputs with AutoModel/pipeline; documentation vs model behavior." + "reason": "A v5.3 video-input regression in Qwen VL models and a transformers serve KV-cache indexing crash are different code paths." }, { - "left": "issue:44479", - "right": "issue:44734", + "left": "issue:43827", + "right": "issue:43994", "accept": false, - "reason": "Qwen video-input regression across several models vs serve KV-cache continuation indexing crash; separate regressions." + "reason": "Outdated docs referencing pipeline() and SigLIP2 producing nonsensical outputs with AutoModel/pipeline are not the same bug." }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Python 3.13 import/parsing failure in DebertaV2Model vs flash_attention sink-less-model AttributeError; different subsystems." + "reason": "A Python 3.13 IndentationError in DebertaV2Model is unrelated to a flash_attention crash on s_aux=None." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device mapping cuda asserts and the 2^24 category limit error are entirely different failures." }, { "left": "issue:44617", "right": "issue:46369", "accept": false, - "reason": "Sam3Video CUDA OOM vs Sam3-LiteText memory overflow; same family but different models and failure characteristics, not clearly one bug." + "reason": "Both concern Sam3-family memory usage, but CUDA OOM in Sam3Video and memory overflow in sam3-litetext are different models and not the same concrete bug." }, { - "left": "issue:41720", - "right": "issue:45245", + "left": "issue:43065", + "right": "issue:44610", "accept": false, - "reason": "Auto device mapping cudaErrorAssert on A800 vs category-count limit RuntimeError; unrelated." + "reason": "A dummy Conv2d in Sam3PixelDecoder and a processor/model input-size mismatch are different Sam3 issues." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "All-zero assistant masks in multimodal chat templates vs incorrect video vision_position_ids; both multimodal, but different defects." + "reason": "All-zero assistant masks for multimodal chat templates and wrong vision_position_ids for Qwen2.5-VL video inputs are different multimodal bugs." }, { - "left": "issue:43065", - "right": "issue:44610", + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "A nonexistent image_utils import and Qwen3.5 chat crashes are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "The non-persistent-buffer junk regression and GPTNeoX rotary_pct reload behavior are different bugs despite both touching model configuration/state." + }, + { + "left": "issue:43976", + "right": "issue:44855", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder vs OmDet-Turbo processor producing the wrong image size; different problems." + "reason": "A Python-version compatibility regression in Transformers 5.1.0 is unrelated to the Python 3.13 DebertaV2Model parsing failure." } ] }, @@ -10285,11 +10365,11 @@ "issue:43295", "issue:43335", "issue:43404", + "issue:43479", "issue:43540", "issue:43550", - "issue:43644", + "issue:43643", "issue:43650", - "issue:43976", "issue:44016", "issue:44077", "issue:44112", @@ -10299,13 +10379,12 @@ "issue:44246", "issue:44279", "issue:44462", + "issue:44496", "issue:44683", "issue:44734", "issue:44737", "issue:44855", "issue:44877", - "issue:44913", - "issue:44933", "issue:44938", "issue:45003", "issue:45127", @@ -10313,7 +10392,7 @@ "issue:45245", "issue:45276", "issue:45357", - "issue:45362", + "issue:45375", "issue:45446", "issue:45542", "issue:45584", @@ -10326,19 +10405,16 @@ "issue:46234" ], "soft_pairs": [ - "issue:44933|issue:45362", - "issue:43644|issue:44913", - "issue:43976|issue:44855", "issue:43550|issue:44190", - "issue:45632|issue:46234", "issue:44077|issue:44877", + "issue:45632|issue:46234", "issue:43335|issue:44855", "issue:45127|issue:45357", "issue:43540|issue:44734", "issue:44462|issue:45907", "issue:44737|issue:45803", - "issue:44112|issue:45600", "issue:43335|issue:44190", + "issue:44112|issue:45600", "issue:43116|issue:44190", "issue:46082|issue:46121", "issue:44938|issue:45542", @@ -10348,209 +10424,205 @@ "issue:36246|issue:41628", "issue:43404|issue:45216", "issue:43650|issue:44220", - "issue:44246|issue:45003", "issue:42175|issue:44279", "issue:44186|issue:44734", + "issue:44246|issue:45003", "issue:41950|issue:44016", "issue:36246|issue:38617", - "issue:44683|issue:45446" + "issue:44683|issue:45446", + "issue:43479|issue:45375", + "issue:43643|issue:44496" ], - "prepared_review_unit_hash": "284c11e41b56ba24e96147214f1433afe6dad03398c37a92a3f5964babf41233", + "prepared_review_unit_hash": "044ed1389f9f929d31f918eae4652653f934d9ce2fa362acac79f0ef1c0f9986", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36089, - "estimated_input_tokens": 9023, - "estimated_eval_tokens": 18302 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35440, + "estimated_input_tokens": 8860, + "estimated_eval_tokens": 17976 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a heterogeneous mix of unrelated Transformers bugs, with a few loose subthemes around import/export regressions, config loading, and model-specific runtime failures. None of the proposed soft pairs look like the same underlying bug or change, so I would not merge any of them.", - "confidence": 0.74, - "canonical_issue_reason": "issue:41628 is the clearest broad representative of the import/export regression subset: it is a user-facing import failure in the top-level Transformers API, with a concise title and likely high impact.", + "summary": "This cluster is heterogeneous: most items are unrelated Transformers bugs that only share broad themes like imports, configs, or multimodal models. None of the soft pairs look like the same concrete bug/change, so all are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated bugs across imports, configs, examples, CI, and model-specific regressions.", "canonical_pr_reason": null, - "best_issue_reason": "issue:41628 is the best issue to anchor this cluster because it is explicit, well-scoped, and more representative than the narrower model-specific import errors nearby.", + "best_issue_reason": "issue:41628 is the best representative issue because it is a concrete, user-facing top-level import failure with a clear repro, but it is still not a good duplicate sink for most of the other items.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44933", - "right": "issue:45362", - "accept": false, - "reason": "Both are model-related failures, but one is an image_utils import error and the other is a Qwen3.5 chat crash; different code paths and fixes." - }, - { - "left": "issue:43644", - "right": "issue:44913", - "accept": false, - "reason": "One is about corrupted non-persistent buffers in 5.0.0, the other is GPTNeoXConfig losing rotary_pct on reload; distinct serialization/config bugs." - }, - { - "left": "issue:43976", - "right": "issue:44855", - "accept": false, - "reason": "These are separate Python-version compatibility problems with different symptoms and root causes." - }, { "left": "issue:43550", "right": "issue:44190", "accept": false, - "reason": "Torch.compile/SDPA on Bamba-9B-v2 and local dataset loading in an image-classification example are unrelated." + "reason": "Different failures and code paths: torch.compile with SDPA in Bamba vs local dataset loading in an image-classification example." }, { - "left": "issue:45632", - "right": "issue:46234", + "left": "issue:44077", + "right": "issue:44877", "accept": false, - "reason": "A cache-path collision for trust_remote_code is unrelated to a missing import-utils helper function." + "reason": "Both are config-related, but one is about an optional post_init field and the other about strict loading of granite_speech config; not the same bug." }, { - "left": "issue:44077", - "right": "issue:44877", + "left": "issue:45632", + "right": "issue:46234", "accept": false, - "reason": "Both involve config behavior, but one is patchtsmixer post_init validation and the other is granite_speech config strictness; not the same bug." + "reason": "Different APIs and symptoms: cache-path collision for trust_remote_code local models vs a missing utility export in import_utils." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "SwitchTransformers sparse-layer construction and a DebertaV2Model import parse error are unrelated." + "reason": "Unrelated issues: SwitchTransformers sparse layer construction vs a Python 3.13 import/indentation parsing failure in DebertaV2Model." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "These are both Qwen-related, but one is LoRA/vocab weight collapse and the other is incorrect visual encoder keys on save_pretrained; different defects." + "reason": "Same model family, but different save/merge problems and likely different code paths; not one concrete underlying bug." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "Qwen3OmniMoe video-input validation and transformers serve KV-cache continuation indexing are different code paths." + "reason": "Both involve video/serve-ish contexts, but one is a processing error in Qwen3OmniMoe and the other is KV-cache continuation tensor indexing in transformers serve." }, { "left": "issue:44462", "right": "issue:45907", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and list_repo_files not forwarding revision are unrelated repo-access behaviors." + "reason": "Different functionality: AutoTokenizer ignoring tokenizer.json vs list_repo_files failing to forward revision." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "XLNet device placement and FuyuBatchFeature exception handling are separate bugs." + "reason": "XLNet device placement bug vs Fuyu exception handling bug; no shared underlying defect." }, { - "left": "issue:44112", - "right": "issue:45600", + "left": "issue:43335", + "right": "issue:44190", "accept": false, - "reason": "Both affect CI, but they come from different stale references in different subsystems." + "reason": "Different layers of the stack: model config construction vs example-script dataset loading." }, { - "left": "issue:43335", - "right": "issue:44190", + "left": "issue:44112", + "right": "issue:45600", "accept": false, - "reason": "SwitchTransformers config construction and a local dataset loading script issue do not share an underlying cause." + "reason": "One is a stale CI device-override test, the other is auto_mappings referencing removed configs; same broad CI area only." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "A multi-label classification example bug and a local dataset loading bug are separate example-script failures." + "reason": "Classification example output bug vs image-classification example input-loading bug; not the same behavior or code path." }, { "left": "issue:46082", "right": "issue:46121", "accept": false, - "reason": "LlamaConfig head_dim validation and rope-params dict conversion with JSON-loaded lists are different config-validation issues." + "reason": "Unrelated config validation problems: Llama head_dim rejection vs rope-validation list handling." }, { "left": "issue:44938", "right": "issue:45542", "accept": false, - "reason": "Python 3.14 import failure and missing TensorFlow backend due to pip extras are unrelated compatibility/dependency issues." + "reason": "Python version load failure vs backend-installation mismatch; different root causes." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Gemma4 embedding resizing and Whisper empty-transcription generation failure are different model behaviors." + "reason": "Both are multimodal, but one is embedding resize propagation in gemma4 and the other is Whisper empty-transcription generation after align_special_tokens." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "A regression in processor/tokenizer access is unrelated to Gemma4 resize_token_embeddings not propagating." + "reason": "Regression in processor/tokenizer image passing vs gemma4 embedding resize behavior; distinct regressions." }, { "left": "issue:45245", "right": "issue:45584", "accept": false, - "reason": "Category-cardinality runtime error and Whisper empty-transcription handling are unrelated." + "reason": "Category cardinality limit error is unrelated to Whisper empty-transcription generation." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they affect different symbols and modules, so they likely need separate fixes." + "reason": "Both are import errors, but for different symbols and modules; not the same missing-export bug." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Mistral3 lm_head tying and Qwen3.5 checkpoint save_pretrained correctness are different model-specific regressions." + "reason": "Model-specific lm_head tying bug vs Qwen3.5 save_pretrained checkpoint regression; different mechanisms and symptoms." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "The placeholder 'ADD THE DATA' issue is not the same as a feature-extraction bug." - }, - { - "left": "issue:44246", - "right": "issue:45003", - "accept": false, - "reason": "Import-time slowness and unsafe sys.modules access are related to importing, but not the same concrete bug." + "reason": "Placeholder-style issue title vs fbank feature extraction bug; no meaningful duplicate evidence." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "Missing TensorFlow in a torch extra install and a generic dependency issue are not the same underlying problem." + "reason": "TensorFlow backend installation issue vs generic dependency issue; too vague and not the same defect." }, { "left": "issue:44186", "right": "issue:44734", "accept": false, - "reason": "LayoutLMv2 tokenizer NER/padding failure and transformers serve KV-cache indexing are unrelated." + "reason": "Tokenizer crash on NER padding/truncation is unrelated to KV-cache continuation indexing in serve." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import latency complaint vs unsafe sys.modules access; different problem class and code path." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Video-classification pipeline image-processor lookup and a notebook syntax error are unrelated." + "reason": "Video-classification pipeline bug vs notebook syntax error; no shared underlying change." }, { "left": "issue:36246", "right": "issue:38617", "accept": false, - "reason": "Two different missing imports from different modules; same broad symptom, but not the same bug." + "reason": "Both are import failures, but one is a missing model class export and the other is a missing configuration utility symbol." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and an incorrect PyTorch version check for AuxRequest are separate flex_attention issues." + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to a version-check bug for AuxRequest import." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Both involve multimodal config fields, but different models and different missing defaults/strict-dropped fields." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "trust_remote_code field-loss issue vs unrecognized model/config issue; not the same load path failure." } ] }, @@ -10581,11 +10653,9 @@ "issue:43299", "issue:43381", "issue:43441", - "issue:43479", "issue:43526", "issue:43531", "issue:43577", - "issue:43643", "issue:43650", "issue:43673", "issue:43824", @@ -10597,6 +10667,7 @@ "issue:44075", "issue:44112", "issue:44220", + "issue:44242", "issue:44279", "issue:44336", "issue:44368", @@ -10608,10 +10679,11 @@ "issue:44987", "issue:45042", "issue:45103", + "issue:45200", "issue:45216", "issue:45335", + "issue:45357", "issue:45372", - "issue:45375", "issue:45584", "issue:45588", "issue:45593", @@ -10624,12 +10696,10 @@ "issue:46369" ], "soft_pairs": [ - "issue:43479|issue:45375", - "issue:43643|issue:44496", "issue:38175|issue:43065", - "issue:43854|issue:44964", "issue:44964|issue:45600", "issue:44016|issue:44220", + "issue:43854|issue:44964", "issue:44112|issue:44964", "issue:46123|issue:46369", "issue:44561|issue:45632", @@ -10639,8 +10709,8 @@ "issue:42907|issue:43441", "issue:43526|issue:43901", "issue:43295|issue:45335", - "issue:45372|issue:46234", "issue:44496|issue:45042", + "issue:45372|issue:46234", "issue:43441|issue:45588", "issue:44016|issue:44062", "issue:43299|issue:45216", @@ -10654,222 +10724,224 @@ "issue:45335|issue:45584", "issue:43650|issue:46209", "issue:45103|issue:45372", - "issue:43577|issue:45216" + "issue:43577|issue:45216", + "issue:44242|issue:45200", + "issue:43299|issue:45357" ], - "prepared_review_unit_hash": "64e8522151837e8148b73b2dc2457f1ab5fbf2354463d3a2cddb959bc1fabbba", + "prepared_review_unit_hash": "72fc8e582fd784f2f9630ba1a607c33ad560e3ca7cd1cb434cebecf91f648a2d", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37074, - "estimated_input_tokens": 9269, - "estimated_eval_tokens": 18794 + "serialized_chars": 37030, + "estimated_input_tokens": 9258, + "estimated_eval_tokens": 18772 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the items span unrelated model-loading, config, generation, attention, notebook, CI, and docs issues. None of the soft pairs look like the same underlying bug/change, so there is no credible duplicate/canonical merge candidate here.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "The cluster is highly heterogeneous: most items are distinct bugs spanning model loading, generation, processors, docs, CI, and utilities. I do not see any soft pair that clearly represents the same underlying change or defect, so no candidate merges are justified.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43824 is the most representative single artifact here: it has the highest discussion/inbound-reference signal and is a concrete reproducible regression, while the rest of the cluster is too unrelated to treat as one duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:43824 is the best stand-in for this cluster because it has the strongest activity signal and a clearly scoped user-facing failure; there is no meaningful PR candidate in the provided items.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43479", - "right": "issue:45375", - "accept": false, - "reason": "Both are config-default bugs, but they affect different model families and different fields/strict-handling behavior." - }, - { - "left": "issue:43643", - "right": "issue:44496", - "accept": false, - "reason": "One is about trust_remote_code dropping config fields; the other is about an unrecognized model_type in config.json." - }, { "left": "issue:38175", "right": "issue:43065", "accept": false, - "reason": "Different models and different symptoms: zero probabilities in SigLIP2 vs a dummy Conv2d in Sam3PixelDecoder." - }, - { - "left": "issue:43854", - "right": "issue:44964", - "accept": false, - "reason": "Both are multimodal model loading failures, but they involve different models and likely different code paths." + "reason": "Different models and different failure modes: Siglip probability output vs a dummy Conv2d in Sam3PixelDecoder." }, { "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "One is a specific model load failure; the other is stale auto_mappings references breaking CI." + "reason": "One is a Phi-4 multimodal loading regression; the other is an auto_mappings/removed-config CI break. Different bug sources." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "A notebook syntax error is unrelated to an audio feature extraction bug." + "reason": "Notebook syntax error and audio feature extraction issue are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different code paths; not the same bug." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "CI device-override test flakiness is not the same bug as a model loading failure." + "reason": "CI test flake in GraniteSpeech is unrelated to Phi-4 multimodal loading." }, { "left": "issue:46123", "right": "issue:46369", "accept": false, - "reason": "One is a batching/partial-output bug; the other is a memory overflow in sam3-litetext." + "reason": "MaskGenerationPipeline batching bug and sam3-litetext memory overflow are unrelated." }, { "left": "issue:44561", "right": "issue:45632", "accept": false, - "reason": "Both mention trust_remote_code, but one is a missing removed API import and the other is a local cache-path collision." + "reason": "One is a missing helper breaking trust_remote_code models; the other is a cache-path collision bug. Different root causes." }, { "left": "issue:44734", "right": "issue:45632", "accept": false, - "reason": "KV-cache tensor indexing in serve is unrelated to trust_remote_code cache-path collisions." + "reason": "KV-cache tensor indexing in transformers serve is unrelated to local-model cache path collisions." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "Python-version support regression is different from a specific model load failure in transformers>=5.1.0." + "reason": "Python version compatibility issue and a model-loading failure for physical-intelligence/fast are unrelated." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior is unrelated to broken auto_mappings references for Sam3LiteText." + "reason": "Qwen3-MoE sliding_window behavior and Sam3LiteText auto_mappings breakage do not share a code path." }, { "left": "issue:42907", "right": "issue:43441", "accept": false, - "reason": "Same model family, but one is about saving dequantized weights while the other is a FlashAttention load failure." + "reason": "Both mention Ministral/Devstral or FlashAttention, but one is save/dequantization and the other is a runtime attention failure; not the same defect." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "A processor label-reduction bug is unrelated to a documentation mismatch in TextClassificationPipeline." + "reason": "BeitImageProcessorFast label reduction bug and pipeline docs mismatch are unrelated." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "These are different model APIs: processor/tokenizer regression vs token-embedding resizing not reaching decoder.embed_tokens." + "reason": "Processor.tokenizer regression and resize_token_embeddings not updating decoder embeddings are different subsystems." }, { - "left": "issue:45372", - "right": "issue:46234", + "left": "issue:44496", + "right": "issue:45042", "accept": false, - "reason": "Different missing imports in different subsystems; they do not share a concrete code-path failure." + "reason": "Unrecognized model type in config and PIL image processor torchvision dependency regression are unrelated." }, { - "left": "issue:44496", - "right": "issue:45042", + "left": "issue:45372", + "right": "issue:46234", "accept": false, - "reason": "Model config parsing failure is unrelated to image processor torchvision dependency handling." + "reason": "Gemma 4 processor import failure and missing falcon mamba availability helper are different missing-symbol issues." }, { "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "Both involve FlashAttention, but they affect different code paths and failure modes, so they are not mergeable as one fix." + "reason": "Both involve FlashAttention, but one is model-specific and the other is a generic s_aux=None crash; not enough to treat as one bug." }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error and AddedToken constructor argument conflict are unrelated bugs." + "reason": "Notebook syntax error and AddedToken special-argument conflict are unrelated." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Both mention Qwen, but one breaks loading and the other breaks save_pretrained checkpoint output." + "reason": "Qwen3VL MoE loading breakage and Qwen3.5 save_pretrained checkpoint corruption are separate regressions." }, { "left": "issue:43673", "right": "issue:44336", "accept": false, - "reason": "Generation cache missing during chunked_prefill is unrelated to ANSI escape codes in loading_report." + "reason": "GenerationMixin cache missing in chunked_prefill and ANSI code emission in loading_report are unrelated." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "A docs/API wording issue is not the same as an unrecognized-model config loading error." + "reason": "Pipeline documentation issue and model config loading failure do not describe the same bug." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary-loss behavior are different features and failure modes." + "reason": "Gradient checkpointing eval-mode restriction and D-FINE auxiliary-loss omission are unrelated training behaviors." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols from different modules." + "reason": "Different missing imports from different modules; same broad symptom, but not the same underlying defect." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "General dependency/version compatibility is too broad; these are not the same concrete defect." + "reason": "Python support regression and generic dependency issue are not the same code-path problem." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Both are model/config loading issues, but one is unrecognized model_type while the other is failing to unwrap text_config for composite multimodal configs." + "reason": "Unrecognized model type vs composite Qwen config unwrapping are different model-loading failures." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Unused SGD args and a tie_word_embeddings warning are unrelated training/configuration concerns." + "reason": "Unused SGD args and a tie_word_embeddings warning are unrelated." }, { "left": "issue:45335", "right": "issue:45584", "accept": false, - "reason": "Token-embedding resize propagation is unrelated to Whisper empty-transcription generation after align_special_tokens." + "reason": "Embedding resize bug and Whisper empty-transcription generation failure are different areas." }, { "left": "issue:43650", "right": "issue:46209", "accept": false, - "reason": "A vague placeholder issue is not the same as FP8 inference failure on B200." + "reason": "Placeholder issue and FP8/B200 inference failure are unrelated." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "Auto-docstring annotation handling and a missing ReasoningEffort import are different failures in different code paths." + "reason": "auto_docstring annotation crash and Gemma 4 processor import error are unrelated." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "BLIP2 dtype propagation and Qwen3.5 save_pretrained regression are distinct model-specific bugs." + "reason": "BLIP2 dtype propagation bug and Qwen3.5 save_pretrained regression are different issues." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss omission and Gemma 4 text-only mm_token_type_ids defaulting are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 regressions, but one is model loading and the other is incorrect save_pretrained visual encoder keys; not one bug." } ] }, @@ -10893,6 +10965,7 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ + "issue:33290", "issue:38175", "issue:39692", "issue:41950", @@ -10910,7 +10983,6 @@ "issue:43901", "issue:43976", "issue:44016", - "issue:44242", "issue:44279", "issue:44291", "issue:44297", @@ -10925,16 +10997,15 @@ "issue:44779", "issue:45020", "issue:45081", - "issue:45200", "issue:45276", "issue:45278", "issue:45325", - "issue:45357", "issue:45397", "issue:45405", "issue:45542", "issue:45593", "issue:45600", + "issue:45656", "issue:45712", "issue:45812", "issue:45823", @@ -10943,183 +11014,164 @@ "issue:46132" ], "soft_pairs": [ - "issue:44242|issue:45200", "issue:38175|issue:44617", - "issue:43299|issue:45357", "issue:44521|issue:45325", "issue:44279|issue:45405", - "issue:44779|issue:45812", "issue:44351|issue:45278", "issue:39692|issue:43901", + "issue:44779|issue:45812", "issue:41950|issue:45020", "issue:43295|issue:44779", "issue:43479|issue:43525", "issue:41950|issue:45823", "issue:43299|issue:43854", - "issue:44683|issue:45542", "issue:43976|issue:45405", + "issue:44683|issue:45542", "issue:44462|issue:45081", "issue:43650|issue:46132", "issue:44297|issue:44462", "issue:43450|issue:44279", - "issue:44466|issue:45276", "issue:45081|issue:45907", + "issue:44466|issue:45276", "issue:44016|issue:44291", "issue:43441|issue:45834", "issue:42994|issue:43317", "issue:43653|issue:45593", "issue:45600|issue:45712", "issue:44623|issue:44749", - "issue:44623|issue:45397" + "issue:44623|issue:45397", + "issue:33290|issue:45656" ], - "prepared_review_unit_hash": "76a5aa1984dd7411d13b3da741103a7197ddd974e8b770c87eccf83f497796ce", + "prepared_review_unit_hash": "0af117dd4f98706cf13871d92a938456a32389d607eea813bb708c345704eed9", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36038, - "estimated_input_tokens": 9010, - "estimated_eval_tokens": 18276 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35097, + "estimated_input_tokens": 8775, + "estimated_eval_tokens": 17806 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: it mixes tokenizer regressions, multimodal processor bugs, config/serialization issues, quantization/loading failures, docs problems, and CI housekeeping. None of the soft-edge pairs look like the same underlying bug/change, so no duplicates are accepted.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue stands out; the items are thematically broad but not the same defect, so a representative issue would be misleading.", - "canonical_pr_reason": "No PRs are present in this cluster.", - "best_issue_reason": "No single issue is a strong global representative because the cluster spans many unrelated failure modes rather than one root cause.", - "best_pr_reason": "No PRs are present in this cluster.", + "summary": "All soft pairs are false positives: they share only broad topical overlap, while each item points to a different bug, model family, or subsystem. No duplicate merge is justified.", + "confidence": 0.97, + "canonical_issue_reason": "No exact canonical duplicate stands out in this heterogeneous set; if a representative issue is needed, issue 44779 is the clearest concrete regression report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44779 is a focused, reproducible tokenizer regression with clear impact and a well-defined failure mode.", + "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44242", - "right": "issue:45200", - "accept": false, - "reason": "Different model families and different bugs: missing load-balancing loss vs. a multimodal config default for text-only fine-tuning." - }, { "left": "issue:38175", "right": "issue:44617", "accept": false, - "reason": "Unrelated problems: siglip2 zero probabilities versus a Sam3Video CUDA OOM." - }, - { - "left": "issue:43299", - "right": "issue:45357", - "accept": false, - "reason": "Both involve Qwen models, but one is a loading regression and the other is incorrect save_pretrained key serialization." + "reason": "Different model and failure mode: SigLIP2 zero probabilities vs Sam3Video CUDA OOM." }, { "left": "issue:44521", "right": "issue:45325", "accept": false, - "reason": "Different multimodal code paths: assistant mask generation versus rope/position-id scaling." + "reason": "Both are multimodal, but one is assistant mask generation and the other is temporal position ID scaling; different code paths." }, { "left": "issue:44279", "right": "issue:45405", "accept": false, - "reason": "A generic dependency issue is not the same as a specific MIN_PEFT_VERSION bump problem." - }, - { - "left": "issue:44779", - "right": "issue:45812", - "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model tokenizers and likely different code paths." + "reason": "Dependency/versioning problems in different places: a transformers dependency issue vs a PEFT minimum-version bump." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "One is a missing HybridCache import; the other is a broad import-error report after upgrade." + "reason": "45278 is a broad import-breakage report; 44351 is a specific HybridCache import error. Too general to call the same bug." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "Different documentation problems affecting different pipelines and behaviors." + "reason": "Both mention docs, but they are unrelated docs issues: SigLIP2 example errors vs TextClassificationPipeline parameter docs." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths; not enough to merge as one bug." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "A pipeline processor selection bug is not the same as remote_code model breakage." + "reason": "Video-classification processor lookup and remote_code model breakage are unrelated subsystems." }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Different regressions: processor/tokenizer image handling versus wrong token IDs for Deepseek." + "reason": "Both involve tokenizers, but one is processor.tokenizer / multimodal input handling and the other is Deepseek tokenization correctness." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Different config classes and different missing/default attribute bugs." + "reason": "Different config bugs: default multimodal sub-config initialization vs missing pad_token_id on Llama4Config." }, { "left": "issue:41950", "right": "issue:45823", "accept": false, - "reason": "Same broad multimodal area, but different failures and code paths; not mergeable as one fix." + "reason": "Video pipeline processor selection and Gemma4 device-mapping mismatch are unrelated." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Different Qwen loading bug versus a GLM-4.7 test-loading issue." + "reason": "Distinct model-loading failures for different models and different symptoms." }, { - "left": "issue:44683", - "right": "issue:45542", + "left": "issue:43976", + "right": "issue:45405", "accept": false, - "reason": "Different subsystems: compiled flex_attention on new torch versus tensorboard/TensorFlow backend detection." + "reason": "Python-version compatibility vs an unreleased PEFT minimum version are not the same issue." }, { - "left": "issue:43976", - "right": "issue:45405", + "left": "issue:44683", + "right": "issue:45542", "accept": false, - "reason": "Python compatibility failure is unrelated to a PEFT version pin bump." + "reason": "Torch flex_attention compilation failure and TensorBoard/TensorFlow backend error are unrelated." }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "Tokenizer loading from repo JSON versus a Mistral regex patch crash are distinct defects." + "reason": "Both are tokenizer-related, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash; different paths." }, { "left": "issue:43650", "right": "issue:46132", "accept": false, - "reason": "Non-technical placeholder issue versus an AttentionInterface.register behavior change." + "reason": "Unrelated: a vague data request vs a behavior change in AttentionInterface.register." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Serialization metadata mismatch is not the same as AutoTokenizer ignoring tokenizer.json on load." + "reason": "Tokenizer save metadata mismatch and AutoTokenizer ignoring tokenizer.json are different serialization/loading bugs." }, { "left": "issue:43450", "right": "issue:44279", "accept": false, - "reason": "A batched video-processor shape bug is unrelated to a dependency issue." - }, - { - "left": "issue:44466", - "right": "issue:45276", - "accept": false, - "reason": "Different Gemma4/v5 issues: tied-weight serialization versus resize_token_embeddings not updating related embeddings." + "reason": "Video processor batch-shape bug vs dependency issue; no shared underlying defect." }, { "left": "issue:45081", @@ -11127,47 +11179,59 @@ "accept": false, "reason": "Tokenizer regex crash and list_repo_files revision forwarding are unrelated." }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both involve model weights/embeddings, but one is serialization inconsistency and the other is resize_token_embeddings behavior." + }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error and init_empty_weights TypeError are unrelated." + "reason": "Notebook syntax error and init_empty_weights argument handling are different failure modes." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "Different models and failures: FlashAttention on Ministral-3 versus OCR long-input indexing in Kosmos2.5." + "reason": "Different models and errors: FlashAttention failure for Ministral-3 vs OCR length index error in Kosmos2.5." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Both touch quantization/offload, but one is saving a quantized model and the other is loading a dequantized model with device_map=auto." + "reason": "Quantized saving failure and dequantized loading/offload failure are related only broadly to quantization, not the same bug." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "Tokenizer special-token registration and D-FINE auxiliary-loss handling are unrelated." + "reason": "Tokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Both are repo maintenance/CI problems, but they concern different stale class lists and different failure modes." + "reason": "Both are repo-maintenance/CI style issues, but one is stale auto-mapping references and the other is leftover dummy classes." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained missing files is unrelated to a performance regression in data filtering after upgrade." + "reason": "Processor save_pretrained missing files and a slowdown regression after upgrade are unrelated." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Missing processor save files and gemma-4 zero3 from_pretrained are different bugs." + "reason": "Different failures: processor serialization vs Gemma4 zero3 from_pretrained." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention deepspeed/optimizer, but one is OOM with Adafactor and the other is double optimizer stepping; not the same code-path problem." } ] }, @@ -11191,10 +11255,10 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:33290", "issue:41628", "issue:42175", "issue:42491", + "issue:42886", "issue:43262", "issue:43299", "issue:43441", @@ -11236,235 +11300,235 @@ "issue:45561", "issue:45563", "issue:45593", - "issue:45656", - "issue:45823" + "issue:45823", + "issue:46132" ], "soft_pairs": [ - "issue:33290|issue:45656", - "issue:44361|issue:45593", "issue:44261|issue:44263", + "issue:44361|issue:45593", "issue:44671|issue:44779", "issue:44016|issue:45823", + "issue:42175|issue:43976", "issue:43441|issue:44683", "issue:43901|issue:43994", - "issue:42175|issue:43976", "issue:43723|issue:44162", "issue:43262|issue:44521", "issue:43673|issue:43761", "issue:43824|issue:43976", - "issue:43525|issue:43881", "issue:44297|issue:45397", + "issue:43525|issue:43881", "issue:43901|issue:44246", "issue:43525|issue:43643", "issue:43784|issue:45542", + "issue:44188|issue:44336", "issue:41628|issue:43976", "issue:44589|issue:45362", - "issue:44188|issue:44336", "issue:45561|issue:45563", "issue:44336|issue:44373", - "issue:44246|issue:44496", "issue:42491|issue:43299", + "issue:44246|issue:44496", "issue:44295|issue:45397", "issue:44242|issue:45593", "issue:43531|issue:45362", - "issue:43526|issue:45003" + "issue:43526|issue:45003", + "issue:42886|issue:46132" ], - "prepared_review_unit_hash": "87d8b5c11eecadf612331ef5dccf1b1c5b6aab38da8964ab3ce35db5633d3fad", + "prepared_review_unit_hash": "d3dec6a0acf8b6ed76691a572f359738311d507ec4691f3fd4264d6e8f9ed7a6", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35219, - "estimated_input_tokens": 8805, - "estimated_eval_tokens": 17866 + "serialized_chars": 35251, + "estimated_input_tokens": 8813, + "estimated_eval_tokens": 17882 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft pair looks like the same underlying bug/change. Most pairs share only a subsystem keyword (deepspeed, Qwen, apply_chat_template, config loading, etc.) but differ in concrete failure mode, model, or code path.", - "confidence": 0.93, - "canonical_issue_reason": "None of the candidate pairs is a convincing duplicate merge target, so there is no canonical issue to collapse onto.", + "summary": "This looks like a loose bag of Transformers bug reports with only superficial lexical similarity. The soft pairs mostly share subsystem terms (tokenizers, Qwen, docs, configs, etc.) but describe different failures, so I do not see a real duplicate cluster.", + "confidence": 0.92, + "canonical_issue_reason": "No single issue is a good canonical anchor: the items span unrelated regressions in tokenization, model loading, generation, audio processors, attention kernels, docs, and utilities. The pairwise similarities are mostly keyword-level rather than same-bug duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "There is no single issue that cleanly represents the whole set; the items are heterogeneous across unrelated bugs, regressions, docs, and packaging problems.", + "best_issue_reason": "If forced to pick one representative issue, issue:43643 is the broadest regression-style report, but it still only represents a narrow config-loading problem and does not actually unify the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:33290", - "right": "issue:45656", + "left": "issue:44261", + "right": "issue:44263", "accept": false, - "reason": "Both mention deepspeed/optimizers, but one is Adafactor OOM and the other is double optimizer stepping; different bugs." + "reason": "Different bugs: one is missing config.rms_norm_eps in MLA q_a_layernorm, the other is torch.split return values in GlmMoeDsaIndexer." }, { "left": "issue:44361", "right": "issue:45593", "accept": false, - "reason": "Unrelated failures: MLukeTokenizer AttributeError vs D-FINE auxiliary loss behavior." - }, - { - "left": "issue:44261", - "right": "issue:44263", - "accept": false, - "reason": "Different models and code paths: MLA layernorm epsilon precision issue vs torch.split return handling in GlmMoeDsaIndexer." + "reason": "Unrelated: MLukeTokenizer AttributeError on tasks versus a stale warning in paged generate()." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "Both are model regressions, but CamemBERT masked LM and DeepSeek tokenizer issues are separate." + "reason": "Different components and symptoms: CamemBERT masked LM predictions vs Deepseek tokenizer output regression." }, { "left": "issue:44016", "right": "issue:45823", "accept": false, - "reason": "Notebook syntax error and Gemma4 device-mismatch forward crash are unrelated." + "reason": "Notebook syntax error and Gemma4 PLE device mismatch are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Package dependency/backend install issue versus a Python-version compatibility problem." }, { "left": "issue:43441", "right": "issue:44683", "accept": false, - "reason": "Different attention bugs on different models/torch versions; not the same concrete code path." + "reason": "Both involve attention/FlashAttention-ish failures, but they are different code paths and different failure modes." }, { "left": "issue:43901", "right": "issue:43994", "accept": false, - "reason": "Docs mismatch for a pipeline parameter vs incorrect SigLIP2 model outputs; distinct issues." - }, - { - "left": "issue:42175", - "right": "issue:43976", - "accept": false, - "reason": "Packaging dependency omission vs Python version compatibility failure; not duplicates." + "reason": "Docs mismatch for return_all_scores is not the same as incorrect model/pipeline outputs for SigLIP2." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "Tokenizer loading regression and ESM2 model breakage are unrelated." + "reason": "Tokenizer loading in v5 is separate from ESM2 model breakage." }, { "left": "issue:43262", "right": "issue:44521", "accept": false, - "reason": "Same API area, but audio sample-rate default bug and multimodal assistant_mask bug are different behaviors." + "reason": "Both mention chat template/audio or multimodal processing, but the bugs are distinct and not the same underlying defect." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "Generation cache missing in v5 chunked_prefill and CLIPVision hidden_states regression are separate code paths." + "reason": "Cache missing during chunked_prefill is unrelated to CLIPVisionModel hidden_states being None." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "Import error for Qwen2.5-VL model class vs Python 3.9+ incompatibility; not the same bug." + "reason": "Missing Qwen2.5-VL import and Python version incompatibility are unrelated." }, { - "left": "issue:43525", - "right": "issue:43881", + "left": "issue:44297", + "right": "issue:45397", "accept": false, - "reason": "Llama4Config missing pad_token_id and glm-4v-9b loading failure are different model/config issues." + "reason": "Tokenizer save_pretrained class mismatch is unrelated to gemma-4 zero3 from_pretrained loading." }, { - "left": "issue:44297", - "right": "issue:45397", + "left": "issue:43525", + "right": "issue:43881", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch vs Gemma4 zero3 from_pretrained failure; unrelated." + "reason": "Different models and different AttributeError/loader failures; no shared code-path evidence." }, { "left": "issue:43901", "right": "issue:44246", "accept": false, - "reason": "Docs issue vs intermittent slow import; no shared underlying defect." + "reason": "Docs inconsistency versus import-time slowness are unrelated." }, { "left": "issue:43525", "right": "issue:43643", "accept": false, - "reason": "Both are config-loading related, but missing pad_token_id and missing fields from trust_remote_code are not the same bug." + "reason": "Both are model/config loading errors, but for different models and different missing fields; not the same bug." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Sentence-transformers import NameError and undefined tf backend from tensorboard-only install are different dependency/import failures." + "reason": "Importing sentence-transformers with latest transformers is unrelated to TensorFlow backend installation behavior." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Attention kernel divergence on torch.compile is unrelated to ANSI codes in loading_report." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "Missing AutoImageProcessor import and Python version support are unrelated." + "reason": "Cannot import AutoImageProcessor has nothing to do with the Python 3.9/3.10 support issue." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Float8 storage lookup failure and Qwen3.5 chat crash are distinct problems." - }, - { - "left": "issue:44188", - "right": "issue:44336", - "accept": false, - "reason": "Torch.compile attention-kernel divergence vs ANSI codes in loading_report are unrelated." + "reason": "Float8 storage import error and Qwen3.5 chat crash are different failures." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "Parallel test race on captured_info.txt and stale num_return_sequences warning are different issues." + "reason": "pytest-xdist race on captured_info.txt is unrelated to the stale num_return_sequences warning." }, { "left": "issue:44336", "right": "issue:44373", "accept": false, - "reason": "ANSI escape handling in loading_report and a wrong docstring for position_ids are unrelated." + "reason": "ANSI output formatting issue and wrong docstring for position_ids are different problems." }, { - "left": "issue:44246", - "right": "issue:44496", + "left": "issue:42491", + "right": "issue:43299", "accept": false, - "reason": "Slow import intermittency and unrecognized model/config are different failure modes." + "reason": "Both concern Qwen3 MoE family regressions, but one is LoRA compatibility across versions and the other is model loading breakage in v5; not the same underlying bug." }, { - "left": "issue:42491", - "right": "issue:43299", + "left": "issue:44246", + "right": "issue:44496", "accept": false, - "reason": "Both involve Qwen3 MoE/version regressions, but they affect different models and loading paths." + "reason": "Import slowness is unrelated to unrecognized model/config model_type error." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "position_ids buffer access error vs Gemma4 zero3 from_pretrained crash are unrelated." + "reason": "Registering position_ids as a buffer causing read errors is unrelated to gemma-4 zero3 loading." }, { "left": "issue:44242", "right": "issue:45593", "accept": false, - "reason": "Missing load-balancing loss and missing auxiliary losses are separate model-training bugs." + "reason": "Load-balancing loss omission in MoE routing is unrelated to D-FINE auxiliary losses when denoising is off." }, { "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Qwen3-MoE sliding_window bug and Qwen3.5-35B chat crash are different issues." + "reason": "Sliding_window issue in Qwen3-MoE is unrelated to Qwen3.5-35B chat crashes." }, { "left": "issue:43526", "right": "issue:45003", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug and unsafe sys.modules access in modeling_utils are unrelated." + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline tokenizer cache loading failure is unrelated to AttentionInterface.register changing function behavior." } ] }, @@ -11521,6 +11585,7 @@ "issue:45003", "issue:45137", "issue:45200", + "issue:45229", "issue:45250", "issue:45292", "issue:45335", @@ -11534,11 +11599,9 @@ "issue:45987", "issue:46082", "issue:46123", - "issue:46132", "issue:46369" ], "soft_pairs": [ - "issue:42886|issue:46132", "issue:41950|issue:44220", "issue:44246|issue:44871", "issue:44336|issue:44485", @@ -11563,213 +11626,214 @@ "issue:45292|issue:45701", "issue:44877|issue:45375", "issue:44987|issue:45459", - "issue:45987|issue:46082", "issue:43232|issue:44336", + "issue:45987|issue:46082", "issue:44995|issue:45200", - "issue:44466|issue:45335" + "issue:44466|issue:45335", + "issue:42886|issue:45229" ], - "prepared_review_unit_hash": "246226257d5e5b6548fb4dff2fca2e9ba73ec6428b091f1e2b6cd23fbd21d628", + "prepared_review_unit_hash": "c3d7ea87c0637b033a1af5ae4e7c735c27452b146ce6121f2d957e703a3277bf", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36400, - "estimated_input_tokens": 9100, - "estimated_eval_tokens": 18456 + "serialized_chars": 36376, + "estimated_input_tokens": 9094, + "estimated_eval_tokens": 18444 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most pairs are only superficially similar (same subsystem, same error class, or same model family) but do not describe the same underlying bug or change.", + "summary": "This set is a heterogeneous mix of unrelated Transformers bug reports. None of the soft-edge pairs look like the same underlying defect, so no duplicate merges are justified. The most representative single issue is the high-activity Gemma4 CUDA OOM report.", "confidence": 0.96, - "canonical_issue_reason": "No single canonical duplicate stands out because the set spans unrelated areas (imports, config loading, generation/cache, pipelines, deepspeed, and model-specific bugs). If a cluster anchor is needed, issue 45701 is the broadest user-facing regression report.", + "canonical_issue_reason": "issue:45229 is the strongest standalone issue here: it is detailed, high-activity, and describes a concrete reproducible multi-GPU inference failure with clear user impact.", "canonical_pr_reason": null, - "best_issue_reason": "issue 45701 is the best global representative because it is broad, user-facing, open, and has multiple inbound references; it is still not a duplicate of the other issues.", + "best_issue_reason": "issue:45229 is the best representative issue overall because it has the clearest problem statement and the most discussion around a concrete, actionable bug.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42886", - "right": "issue:46132", - "accept": false, - "reason": "Different problems: offline cache loading vs AttentionInterface.register changing registered-function behavior." - }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Both are pipeline-related, but one is video-classification image-processor lookup and the other is audio feature extraction." + "reason": "Different subsystems and failure modes: video-classification processor lookup vs audio fbank feature extraction." }, { "left": "issue:44246", "right": "issue:44871", "accept": false, - "reason": "Unrelated: import performance vs a Gemma eos_token_id configuration mismatch." + "reason": "One is an intermittent import slowdown; the other is a Gemma eos_token_id config mismatch." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI color output in loading_report is unrelated to GLM-5 RoPE implementation concerns." + "reason": "Unrelated bugs: terminal ANSI output handling vs GLM-5 RoPE implementation." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but for different symbols and different root causes." + "reason": "Both are import errors, but they concern different symbols and different packages/modules." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Different missing imports in different modules; not the same bug." + "reason": "Different missing exports: Qwen2_5_VLImageProcessor vs HybridCache." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "Unsafe sys.modules access is a different code-path issue from pytest-xdist file races in testing utils." + "reason": "sys.modules access bug and pytest-xdist file race are unrelated issues." }, { "left": "issue:44779", "right": "issue:45292", "accept": false, - "reason": "Tokenizer correctness regression vs embedding resize not propagating to output embeddings." + "reason": "Tokenizer correctness regression is not the same as embedding resize not propagating." }, { "left": "issue:43475", "right": "issue:46369", "accept": false, - "reason": "Different SAM video-model failures: missing output attribute vs memory overflow." + "reason": "SAM3 attribute error and SAM3 memory overflow are distinct bugs." }, { "left": "issue:44242", "right": "issue:44857", "accept": false, - "reason": "MoE load-balancing loss logic is unrelated to a float16 AMP crash in LwDetrImageLoss." + "reason": "Different models and symptoms: MoE load-balancing loss vs LwDetrImageLoss AMP crash." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "A position_ids buffer error is not the same as a reported slowdown after version upgrade." + "reason": "A position_ids buffer access error is unrelated to a performance regression after upgrade." }, { "left": "issue:43352", "right": "issue:45250", "accept": false, - "reason": "One is a model-specific Flash Attention support error; the other is a generic Flash Attention feature issue." + "reason": "A model-specific Flash Attention support limitation is not the same as a generic Flash Attention issue." }, { "left": "issue:43673", "right": "issue:44561", "accept": false, - "reason": "Cache handling during chunked prefill is unrelated to removal of is_torch_fx_available breaking trust_remote_code models." + "reason": "Generation cache/chunked_prefill bug and trust_remote_code import breakage are separate failures." }, { "left": "issue:36246", "right": "issue:43596", "accept": false, - "reason": "ImportError for Qwen2.5 VL image processor is not the same as a Zero3 BertModel index error." + "reason": "ImportError for a missing processor and a DeepSpeed/BertModel IndexError are unrelated." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "Docs mismatch for return_all_scores is unrelated to sys.modules access safety." + "reason": "Documentation mismatch in a pipeline vs unsafe sys.modules access in modeling utils." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "Different model families and failures: ESM2 broken vs Gemma 4 mm_token_type_ids defaults." + "reason": "ESM2 breakage and Gemma4 mm_token_type_ids behavior are different model-specific bugs." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Unrecognized model/config loading is a different issue from unsafe module lookup." + "reason": "Model config loading failure and sys.modules access are not the same code-path problem." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Both involve deepspeed, but one is MoE weight conversion and the other is a Zero3 deque/indexing crash." + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed is distinct from a ZeRO3 deque underflow." }, { "left": "issue:43262", "right": "issue:44246", "accept": false, - "reason": "Audio chat-template sample-rate defaulting is unrelated to import latency." + "reason": "Audio chat-template sampling-rate bug is unrelated to import performance." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Both touch multimodal config handling, but they concern different fields and different code paths." + "reason": "Both involve composite Qwen configs, but one is a dropped config field and the other is missing text_config unwrapping." }, { "left": "issue:45292", "right": "issue:45584", "accept": false, - "reason": "Embedding resize propagation and Whisper empty-transcription generation are distinct bugs." + "reason": "Embedding resize propagation and Whisper empty-transcription generation failure are different issues." }, { "left": "issue:45412", "right": "issue:46123", "accept": false, - "reason": "Memory release/lifecycle behavior is unrelated to MaskGenerationPipeline's final-batch flag bug." + "reason": "Memory reclamation on deletion and batch-finalization logic in MaskGenerationPipeline are unrelated." }, { "left": "issue:45292", "right": "issue:45701", "accept": false, - "reason": "Embedding resize behavior is not the same as a tokenizer regression across versions." + "reason": "A token embedding resize bug is not the same as a broad tokenization version-change report." }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Both are config/schema issues, but for different models and different missing-field behavior." + "reason": "Both mention strict config handling, but they describe different models and different failure mechanisms." }, { "left": "issue:44987", "right": "issue:45459", "accept": false, - "reason": "Model loading failure under v5.1.0 is unrelated to protobuf decode error masking." + "reason": "Model loading regression in v5.1.0 is unrelated to protobuf exception masking in tokenizers." }, { - "left": "issue:45987", - "right": "issue:46082", + "left": "issue:43232", + "right": "issue:44336", "accept": false, - "reason": "Cache API return-type mismatch and LlamaConfig validation are different issues." + "reason": "Generation kwargs sync logic and ANSI logging output are unrelated." }, { - "left": "issue:43232", - "right": "issue:44336", + "left": "issue:45987", + "right": "issue:46082", "accept": false, - "reason": "Generation kwargs sync bug is unrelated to terminal ANSI leakage." + "reason": "Cache-length return type and Llama head_dim validation are different API/validation bugs." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "GlmMoeDsa second-pass cache crash is unrelated to Gemma 4 token-type defaults." + "reason": "Stale indexer cache crash on second forward pass is unrelated to mm_token_type_ids defaults." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Serialization of tied lm_head weights is not the same as resize_token_embeddings failing to update decoder embeddings." + "reason": "Serialization inconsistency for lm_head.weight is not the same bug as resize_token_embeddings propagation." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are completely different failures." } ] }, @@ -11793,7 +11857,8 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:42886", + "issue:28282", + "issue:30990", "issue:42994", "issue:43065", "issue:43232", @@ -11827,9 +11892,9 @@ "issue:44977", "issue:44995", "issue:45137", - "issue:45229", "issue:45230", "issue:45356", + "issue:45381", "issue:45397", "issue:45563", "issue:45584", @@ -11842,10 +11907,9 @@ "issue:46202" ], "soft_pairs": [ - "issue:42886|issue:45229", "issue:43065|issue:43408", - "issue:44479|issue:46170", "issue:45397|issue:45656", + "issue:44479|issue:46170", "issue:45137|issue:45600", "issue:43475|issue:43540", "issue:44060|issue:44077", @@ -11861,8 +11925,8 @@ "issue:43673|issue:44734", "issue:43065|issue:46123", "issue:44297|issue:44623", - "issue:44869|issue:45356", "issue:44361|issue:44749", + "issue:44869|issue:45356", "issue:43646|issue:44162", "issue:44623|issue:45656", "issue:44749|issue:45656", @@ -11872,229 +11936,237 @@ "issue:43673|issue:43827", "issue:45230|issue:45397", "issue:42994|issue:43482", + "issue:44038|issue:44623", "issue:43475|issue:43643", - "issue:44038|issue:44623" + "issue:28282|issue:30990", + "issue:43475|issue:45381" ], - "prepared_review_unit_hash": "43931810a47a44c5b78b749cbe0484ddd7d2c5961a6d66e135c0e148dc2ddfe0", + "prepared_review_unit_hash": "b814494854e9abfd32fb8bc14e314f8a5f6c3d5f5b5566327202789107d1bdcc", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 32, - "serialized_chars": 36554, - "estimated_input_tokens": 9139, - "estimated_eval_tokens": 18534 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37459, + "estimated_input_tokens": 9365, + "estimated_eval_tokens": 18986 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are unrelated bugs/regressions across tokenizers, generation, serving, model families, and training/deepspeed paths. The soft-similarity pairs mostly share broad keywords, not the same concrete bug or fix.", - "confidence": 0.96, - "canonical_issue_reason": "No single canonical issue stands out; the issues do not form one duplicate family, so any choice would be arbitrary.", + "summary": "This cluster is mostly a set of unrelated issue reports, with one clear duplicate-like pair around Qwen video input regression. The strongest anchor is the broad v5.3.0 Qwen video regression report.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44479 is the best representative because it names a concrete v5.3.0 regression, identifies the affected Qwen video model families, and is broad enough to anchor the closest related report.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong global representative because the cluster is not centered on one underlying defect.", + "best_issue_reason": "issue:44479 is the strongest global issue candidate: it is specific, high-signal, and covers the widest relevant surface among the listed issues.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:42886", - "right": "issue:45229", - "accept": false, - "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU CUDA OOM are unrelated." - }, { "left": "issue:43065", "right": "issue:43408", "accept": false, - "reason": "Both are SAM3-related, but one is a dummy Conv2d concern and the other is a model-type warning; different bugs." + "reason": "Both are SAM3-related, but one is a dummy Conv2d concern and the other is a model-type/config warning; different bug surfaces." }, { - "left": "issue:44479", - "right": "issue:46170", + "left": "issue:45397", + "right": "issue:45656", "accept": false, - "reason": "Qwen video regression and Windows/ROCm import dependency failure are different code paths." + "reason": "Different subsystems and symptoms: DeepSpeed zero3 from_pretrained vs optimizer stepping twice." }, { - "left": "issue:45397", - "right": "issue:45656", + "left": "issue:44479", + "right": "issue:46170", "accept": false, - "reason": "Gemma-4 Zero3 loading and double optimizer stepping with deepspeed are not the same problem." + "reason": "Qwen video regression and Windows ROCm/torch.distributed import failure are unrelated." }, { "left": "issue:45137", "right": "issue:45600", "accept": false, - "reason": "ZeRO3 deque failure and Sam3LiteText auto_mappings references are unrelated." + "reason": "ZeRO3 deque error vs stale Sam3LiteText auto-mapping references; not the same underlying defect." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "SAM3 vision encoder attribute error and Qwen3OmniMoe video-processing ValueError are different issues." + "reason": "Different model families and failures: missing SAM3 vision encoder output field vs Qwen3OmniMoe video input ValueError." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Incorrect tied-weights warning in Qwen3-Next and patchtsmixer post_init policy are unrelated." + "reason": "A tied-weights warning and PatchTSMixer post_init policy change are distinct issues." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "Chat/completions request handling and Gemma pipeline roundtrip on a special character are unrelated." + "reason": "API request rejection and Gemma pipeline roundtrip token failure do not share a concrete code path." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are different bugs." + "reason": "Tokenizer special-token registration bug vs ESM2 model breakage are different problems." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "A tied-embeddings warning during Qwen3.5 LoRA fine-tuning is not the same as a flash-attention generation failure." + "reason": "Warning about tie_word_embeddings and flash-attention generation failure are not the same bug." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "Stale indexer cache on second forward and flash_attention aux-None crash are distinct failures." + "reason": "Stale indexer cache on second forward pass vs flash_attention s_aux=None crash are different code paths." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "Qwen3.5 tied-embeddings warning and paged-generate stale warning are different issues." + "reason": "Both involve warnings, but one is about tied embeddings and the other about stale num_return_sequences validation." }, { "left": "issue:44188", "right": "issue:46202", "accept": false, - "reason": "torch.compile kernel divergence and Gemma pipeline roundtrip failure are unrelated." + "reason": "torch.compile branching divergence in attention kernels is unrelated to Gemma token roundtrip failure." }, { "left": "issue:44877", "right": "issue:46170", "accept": false, - "reason": "Strict granite_speech config loading and Windows/ROCm torch.distributed dependency are unrelated." + "reason": "Strict config loading for granite_speech and ROCm/Windows import dependency issues are unrelated." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch and Qwen3.5 tied-embeddings warning are different model-loading problems." + "reason": "Weight-shape mismatch during Qwen3-VL loading is not the same as a tied-embeddings warning." }, { "left": "issue:45584", "right": "issue:46170", "accept": false, - "reason": "Whisper empty transcription generation and Windows/ROCm import error are unrelated." + "reason": "Whisper empty transcription failure is unrelated to the Windows ROCm import/dependency issue." }, { "left": "issue:43673", "right": "issue:44734", "accept": false, - "reason": "Both mention cache, but chunked_prefill GenerationMixin cache missing and /v1/responses KV indexing crash are different code paths." + "reason": "Both are generation/cache-related, but one is a chunked_prefill cache gap and the other is a server-side tensor indexing crash; too different to merge." }, { "left": "issue:43065", "right": "issue:46123", "accept": false, - "reason": "SAM3 pixel decoder placeholder conv and MaskGenerationPipeline final-batch handling are unrelated." + "reason": "SAM3 decoder dummy conv and MaskGenerationPipeline final-batch handling are different bugs." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "tokenizer.save_pretrained class mismatch and processor.save_pretrained missing files are different save paths." + "reason": "Tokenizer_class mismatch and processor.save_pretrained missing files are both serialization-related but not the same concrete failure." }, { - "left": "issue:44869", - "right": "issue:45356", + "left": "issue:44361", + "right": "issue:44749", "accept": false, - "reason": "Whisper timestamp decode crash and Kimi-K2.5 tokenizer codec regression are not the same bug." + "reason": "MLukeTokenizer AttributeError and a Chinese report about data filtering slowdown are unrelated." }, { - "left": "issue:44361", - "right": "issue:44749", + "left": "issue:44869", + "right": "issue:45356", "accept": false, - "reason": "MLukeTokenizer task AttributeError and a training slowdown regression are unrelated." + "reason": "Whisper trailing replacement-character crash and Kimi-K2.5 codec-handling regression are different decode bugs." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Transformers 5 custom model init breakage and ESM2 breakage are too broad and not the same defect." + "reason": "Custom model initialization breakage in v5 and ESM2 breakage are separate issues." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Processor save_pretrained missing files and deepspeed optimizer double-step are unrelated." + "reason": "Processor save_pretrained file emission and DeepSpeed double optimizer step are unrelated." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "Inference slowdown and optimizer being stepped twice are different issues." + "reason": "Training slowdown report and duplicate optimizer step report do not match." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "Generation kwargs update after sync_gpus and missing cache during chunked_prefill are different generation bugs." + "reason": "Different generation internals: sync_gpus kwargs update vs missing cache during chunked_prefill." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma eos_token configuration inconsistency and AttentionInterface.register behavior change are unrelated." + "reason": "Gemma EOS-token config inconsistency and AttentionInterface.register behavior change are not the same bug." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "resume_from_checkpoint key mismatch and a data-processing slowdown are unrelated." + "reason": "resume_from_checkpoint key mismatch is unrelated to data filtering slowdown." }, { "left": "issue:43673", "right": "issue:43827", "accept": false, - "reason": "Missing generation cache and stale pipeline docs are not the same bug." + "reason": "Generation cache bug and outdated pipeline docs are separate concerns; same theme only, not same issue." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "Generic bug report and Gemma-4 Zero3 loading issue are unrelated." + "reason": "A generic bug report cannot be treated as the same concrete issue as gemma-4 zero3 loading failure." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized model saving failure and GGUF loading failure are different directions and formats." + "reason": "Quantized model saving failure and Qwen2.5-GGUF loading failure are different directions of the pipeline." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe v5.0 bug and processor save_pretrained missing files are unrelated." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "Missing fpn_position_embeddings in SAM3 output and trust_remote_code config field loss are unrelated." + "reason": "Missing fpn_position_embeddings on a vision output object is not the same as missing fields in AutoConfig with trust_remote_code." }, { - "left": "issue:44038", - "right": "issue:44623", + "left": "issue:45381", + "right": "issue:45656", "accept": false, - "reason": "Qwen3-VL-Moe bug and processor.save_pretrained missing files are unrelated." + "reason": "Qwen2.5-VL video position regression and duplicate optimizer stepping are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "These describe the same Qwen video-input regression family in v5.3.0, with 45381 giving a specific qwen2.5-vl symptom already covered by 44479." } ] }, @@ -12118,8 +12190,8 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:28282", "issue:30990", + "issue:35141", "issue:36010", "issue:41628", "issue:42175", @@ -12129,8 +12201,8 @@ "issue:42898", "issue:42994", "issue:43232", + "issue:43278", "issue:43377", - "issue:43475", "issue:43504", "issue:43525", "issue:43526", @@ -12139,6 +12211,7 @@ "issue:43723", "issue:43824", "issue:43873", + "issue:43976", "issue:44038", "issue:44188", "issue:44222", @@ -12156,7 +12229,6 @@ "issue:45137", "issue:45310", "issue:45375", - "issue:45381", "issue:45406", "issue:45478", "issue:45561", @@ -12168,11 +12240,9 @@ "issue:46326" ], "soft_pairs": [ - "issue:28282|issue:30990", - "issue:43475|issue:45381", "issue:45406|issue:46202", - "issue:41628|issue:44933", "issue:42371|issue:45092", + "issue:41628|issue:44933", "issue:45375|issue:45478", "issue:42994|issue:43525", "issue:43723|issue:46326", @@ -12189,240 +12259,249 @@ "issue:42994|issue:43824", "issue:44038|issue:45310", "issue:36010|issue:41628", - "issue:42831|issue:44222", "issue:43232|issue:44792", "issue:43673|issue:44188", - "issue:42175|issue:45712", + "issue:42831|issue:44222", "issue:44496|issue:45561", + "issue:42175|issue:45712", "issue:43526|issue:45561", "issue:43504|issue:45561", "issue:43377|issue:44242", "issue:42175|issue:44933", - "issue:42898|issue:44568" + "issue:42898|issue:44568", + "issue:43976|issue:44933", + "issue:43278|issue:45137", + "issue:30990|issue:35141" ], - "prepared_review_unit_hash": "83ebfa9e9ad4bc8d62319d0b76edbf2c7a10f5a843c48e78a19ec53754535ffc", + "prepared_review_unit_hash": "31387615e13846e05cb98bdf604eb67e5aae906e2ed0cd3de294e4506fdfd89b", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36887, - "estimated_input_tokens": 9222, - "estimated_eval_tokens": 18700 + "soft_pair_count": 32, + "serialized_chars": 37183, + "estimated_input_tokens": 9296, + "estimated_eval_tokens": 18848 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly separate bug reports, but there are two true duplicate signals in the packet: the Qwen3ForCausalLM VRAM-leak pair (42673/43704) and the exact-title duplicate pair 45310/45478 for the Qwen3.5 MoE from_pretrained error. The remaining soft pairs are theme-similar but not the same concrete defect.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 45310 is the best canonical anchor: it is concrete, sits in the Qwen3.5 MoE from_pretrained regression cluster, and has an exact duplicate in 45478.", + "summary": "This cluster is mostly a grab-bag of distinct Transformers bug reports; only one soft-edge pair is a clear duplicate. The best representative issue is a broad top-level import regression report.", + "confidence": 0.87, + "canonical_issue_reason": "issue:41628 is the most central and broadly applicable report here: a top-level `transformers` import failure with multiple nearby import-regression links.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45310 is the strongest single representative because it has the clearest duplicate signal in 45478 and captures the specific Qwen3.5 MoE from_pretrained failure more directly than the other nearby Qwen/MoE reports.", + "best_issue_reason": "issue:41628 has the widest apparent scope among the candidates and best matches the cluster\u2019s repeated theme of API/import regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:28282", - "right": "issue:30990", - "accept": false, - "reason": "Both are load-related, but one is a missing PyTorch dependency error and the other is a Sentence Transformers hang." - }, - { - "left": "issue:43475", - "right": "issue:45381", - "accept": false, - "reason": "Different models and failures: SAM3 attribute error vs qwen2.5-vl video position-id bug." - }, { "left": "issue:45406", "right": "issue:46202", "accept": false, - "reason": "Gemma4 processor attribute crash and Gemma pipeline roundtrip failure are different code paths." + "reason": "Different bugs: Gemma4Processor serving-time attribute error vs Gemma pipeline roundtrip tokenization failure." }, { - "left": "issue:41628", - "right": "issue:44933", + "left": "issue:42371", + "right": "issue:45092", "accept": false, - "reason": "Both are import issues, but they involve different missing symbols and unrelated root causes." + "reason": "TF32 API guidance is unrelated to the InternVL2 v5 meta-initialization incompatibility." }, { - "left": "issue:42371", - "right": "issue:45092", + "left": "issue:41628", + "right": "issue:44933", "accept": false, - "reason": "TF32 API guidance and InternVL2 meta-init incompatibility are unrelated defects." + "reason": "Both are import-related, but one is a missing top-level class and the other is a nonexistent image_utils symbol; not the same bug." }, { "left": "issue:45375", "right": "issue:45478", "accept": false, - "reason": "Same model family, but a missing config field is not the same as a from_pretrained loading error." + "reason": "Same model family, but one is a specific config-field-drop report and the other is only a generic from_pretrained failure; insufficiently identical." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized model saving and missing pad_token_id are unrelated bugs." + "reason": "Quantized model saving failure and missing pad_token_id in Llama4Config are different code paths." }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Tokenizer loading regression vs continue_final_message behavior are different code paths." + "reason": "Tokenizer loading regression and continue_final_message behavior for mistral_common are different tokenizer bugs." }, { "left": "issue:45310", "right": "issue:45375", "accept": false, - "reason": "Related Qwen3.5/MoE area, but from_pretrained failure and dropped config field are different issues." + "reason": "Related Qwen3.5 MoE area, but one is a generic load error and the other is a specific config-field omission." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Mistral tokenizer regex crash and auto_docstring annotation crash are unrelated." + "reason": "Tokenizer regex crash and auto_docstring annotation crash are unrelated subsystems." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Quantization/offloading behavior and pipeline-parallel incorrect output are different bugs." + "reason": "Quantization/offloading behavior and pipeline-parallel incorrect output are unrelated runtime bugs." }, { "left": "issue:46077", "right": "issue:46123", "accept": false, - "reason": "Attention-mask construction and final partial-batch handling are not the same defect." + "reason": "Gemma4 attention-mask construction and partial-batch pipeline handling are different problems." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Exact same title and symptom: Qwen3ForCausalLM VRAM leak in multiple dataloader threads." + "reason": "Same title and same Qwen3ForCausalLM VRAM leak symptom; these look like duplicate reports of the same bug." }, { "left": "issue:43232", "right": "issue:44188", "accept": false, - "reason": "Generation kwargs handling after sync_gpus and torch.compile attention divergence are unrelated." + "reason": "Generation kwargs/sync_gpus issue and torch.compile attention-kernel branching issue are different generation-path bugs." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Different missing imports: AutoImageProcessor versus HybridCache." + "reason": "Both are import errors, but they concern different exported symbols and likely different regressions." }, { "left": "issue:44964", "right": "issue:45137", "accept": false, - "reason": "Phi-4 multimodal loading failure and DeepSpeed ZeRO3 deque underflow are unrelated." + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Pipeline-parallel output corruption and a Qwen3_5 image loop are different bugs." + "reason": "Pipeline-parallel wrong output and Qwen3_5 infinite-loop-on-image are different failure modes." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "Both involve Qwen3.5 MoE/FP8, but save_pretrained failure and from_pretrained error are distinct code paths." + "reason": "FP8 save_pretrained MoE and Qwen3.5 MoE from_pretrained error are not the same concrete bug." }, { "left": "issue:42994", "right": "issue:43824", "accept": false, - "reason": "Quantized model saving and a missing Qwen2_5_VL import are unrelated." + "reason": "Quantized model saving and a Qwen2.5-VL import error are unrelated." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Related Qwen/MoE family, but Qwen3-VL-Moe and Qwen3.5 MoE from_pretrained failures are not the same bug." + "reason": "Qwen3-VL-Moe on transformers 5.0 and Qwen3.5 MoE load failure are different model/version issues." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they concern different objects and different root causes." - }, - { - "left": "issue:42831", - "right": "issue:44222", - "accept": false, - "reason": "FP8 accuracy regression is not the same as FP8 save_pretrained failure." + "reason": "Both are import errors, but GenerationMixin-from-generation and AutoImageProcessor-top-level are separate regressions." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Generation kwargs handling and a janus test failure are unrelated." + "reason": "Generation state-update bug and janus image-generation test failure are unrelated." }, { "left": "issue:43673", "right": "issue:44188", "accept": false, - "reason": "Missing GenerationMixin cache in chunked_prefill and torch.compile attention divergence are unrelated." + "reason": "Chunked-prefill cache handling and compile-time attention-kernel divergence are different code paths." }, { - "left": "issue:42175", - "right": "issue:45712", + "left": "issue:42831", + "right": "issue:44222", "accept": false, - "reason": "TensorFlow backend packaging and leftover dummy-class cleanup are unrelated." + "reason": "Both mention FP8, but one is an accuracy issue and the other is saving pretrained weights." }, { "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Model config recognition failure and pytest-xdist file race are unrelated." + "reason": "Model config loading failure and xdist file-race bug are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Backend dependency packaging and dummy-class cleanup are different maintenance issues." }, { "left": "issue:43526", "right": "issue:45561", "accept": false, - "reason": "BeitImageProcessorFast label reduction and xdist capture-file race are unrelated." + "reason": "BeitImageProcessorFast label reduction and xdist captured_info.txt race are unrelated." }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "Beit preset loading and xdist file race are unrelated." + "reason": "Legacy-field pretrained loading failure and xdist race are unrelated." }, { "left": "issue:43377", "right": "issue:44242", "accept": false, - "reason": "MIMI padding-mask batching bug and MoE load-balancing loss omission are different defects." + "reason": "MIMI padding-mask mismatch and MoE load-balancing loss gating are different model bugs." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "TensorFlow backend packaging and image_utils import failure are unrelated." + "reason": "Packaging/backend selection and image_utils import exposure are unrelated." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Tokenizer cleanup-space behavior change and BOS/EOS insertion bug are distinct regressions." + "reason": "Tokenizer cleanup-space behavior change and BOS/EOS insertion regression are different tokenizer issues." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility and image_utils import exposure are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift in eval and ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence Transformers loading stall and embedding reinit after resize are different problems." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The result is not fully grounded because the packet contains another exact-duplicate signal the summary omits: issues 45310 and 45478 have the same title (\u201c[BUG] transformers>=5.4.0, Qwen3.5 Moe from_pretrained error\u201d). Saying that only the Qwen3ForCausalLM VRAM-leak pair looks like a true duplicate overstates the evidence. The conservative verdicts on the listed soft pairs are mostly fine, but the overall summary needs to acknowledge the 45310/45478 duplicate as well." + "accept": true, + "feedback": "The summary stays within the packet evidence: it treats the cluster as mostly heterogeneous, highlights 41628 as a reasonable broad import-regression representative, and only accepts the clearly duplicated 42673/43704 pair. The other soft-edge rejections are conservative and grounded in the issue titles/descriptions." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -12440,9 +12519,9 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ + "issue:22355", "issue:28282", "issue:30333", - "issue:30990", "issue:33290", "issue:34689", "issue:35141", @@ -12450,14 +12529,13 @@ "issue:42222", "issue:42831", "issue:42898", - "issue:43278", + "issue:43122", "issue:43479", "issue:43644", "issue:43723", "issue:43784", "issue:43824", "issue:43874", - "issue:43976", "issue:44016", "issue:44060", "issue:44075", @@ -12473,10 +12551,10 @@ "issue:44779", "issue:44869", "issue:44933", + "issue:45071", "issue:45092", "issue:45103", "issue:45125", - "issue:45137", "issue:45200", "issue:45245", "issue:45310", @@ -12487,12 +12565,10 @@ "issue:45812", "issue:46082", "issue:46132", + "issue:46209", "issue:46231" ], "soft_pairs": [ - "issue:43278|issue:45137", - "issue:43976|issue:44933", - "issue:30990|issue:35141", "issue:42222|issue:45200", "issue:44623|issue:45310", "issue:43479|issue:44869", @@ -12508,255 +12584,271 @@ "issue:43874|issue:45092", "issue:44060|issue:45125", "issue:44295|issue:45656", - "issue:42898|issue:44448", "issue:44075|issue:46231", + "issue:42898|issue:44448", "issue:45405|issue:45712", "issue:44448|issue:46132", "issue:44222|issue:44297", "issue:44933|issue:45712", "issue:44297|issue:45656", "issue:43479|issue:43644", - "issue:30333|issue:34689", "issue:44222|issue:44623", + "issue:30333|issue:34689", "issue:42831|issue:44623", "issue:45245|issue:46132", "issue:30333|issue:36331", "issue:30333|issue:35141", + "issue:44190|issue:46082", "issue:28282|issue:30333", - "issue:44190|issue:46082" + "issue:43122|issue:46209", + "issue:22355|issue:28282", + "issue:45071|issue:46082", + "issue:44297|issue:45310" ], - "prepared_review_unit_hash": "7f36155e5cea7b9c92bc4409b092c14c67fc9572d403c7d3741d5453c1d6ee51", + "prepared_review_unit_hash": "07d1e10c132c981fadc4193f49f93dcef9db04732102a5d3bab4c3c737e6fa36", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 38007, - "estimated_input_tokens": 9502, - "estimated_eval_tokens": 19260 + "soft_pair_count": 35, + "serialized_chars": 38256, + "estimated_input_tokens": 9564, + "estimated_eval_tokens": 19384 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous. The soft links mostly point to broad Transformers v5/tokenizer/model-loading regressions, but they do not describe the same concrete bug, so I would not merge these pairs as duplicates. No PRs were provided.", - "confidence": 0.41, - "canonical_issue_reason": "issue:42898 is the best representative because it is a clear, high-signal v5 tokenizer-behavior regression and it has the broadest soft connectivity to other tokenizer-related items in this set.", + "summary": "These items are a heterogeneous mix of unrelated Transformers issues; the soft links mostly reflect superficial similarity around v5 regressions, tokenization, save_pretrained, and model-loading, not the same underlying bug. I would not merge any of the soft pairs.", + "confidence": 0.97, + "canonical_issue_reason": "issue:42898 is the most centrally connected item in the soft graph and is the broadest representative of the recurring v5/tokenization-regression theme, though the cluster is too mixed to treat it as a true duplicate hub.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the strongest global representative for the cluster\u2019s recurring v5/tokenization regression theme, even though the overall set is not truly one duplicate family.", + "best_issue_reason": "issue:42898 is the best anchor among these issues because it has the most soft similarities and reflects the most repeated theme in the set, but it is still not a suitable duplicate target for the others.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43278", - "right": "issue:45137", - "accept": false, - "reason": "Different failures: dtype drift in evaluation vs a DeepSpeed ZeRO3 deque/indexing error." - }, - { - "left": "issue:43976", - "right": "issue:44933", - "accept": false, - "reason": "Unrelated symptoms: Python version support regression vs a missing/nonexistent import from image_utils." - }, - { - "left": "issue:30990", - "right": "issue:35141", - "accept": false, - "reason": "Sentence-transformers loading hang and embedding reinitialization are different code paths and bugs." - }, { "left": "issue:42222", "right": "issue:45200", "accept": false, - "reason": "Vitpose model breakage and Gemma 4 token-type defaults are unrelated multimodal issues." + "reason": "Different model families and bugs: vitpose model breakage vs Gemma 4 multimodal token type IDs." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Processor save_pretrained file handling and Qwen3.5 Moe from_pretrained failure are different problems." + "reason": "Both involve model/processor loading, but one is missing save_pretrained files and the other is a Qwen3.5 MoE from_pretrained error." }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Config initialization bug vs Whisper timestamp decode crash; no shared underlying defect." + "reason": "Unrelated bugs: Phi4MultimodalConfig default initialization vs Whisper word-timestamp decode crash." }, { "left": "issue:30333", "right": "issue:33290", "accept": false, - "reason": "MLflow reporting on failed jobs is unrelated to an Adafactor/DeepSpeed OOM." + "reason": "Training-job MLflow reporting is unrelated to DeepSpeed/Adafactor OOM behavior." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Different object types and failure modes: multimodal config defaults vs ModelOutput key assignment." + "reason": "Config default handling and ModelOutput key assignment are different code paths." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different behaviors and code paths; not the same bug." + "reason": "Both are v5 tokenization regressions, but they affect different models and different failure modes." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Importing sentence-transformers with missing nn and missing Qwen2.5-VL class are unrelated import failures." + "reason": "Different import failures: sentence-transformers with latest transformers vs missing Qwen2.5-VL symbol." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "AutoTokenizer loading issue vs special-token insertion regression; related subsystem but not the same defect." + "reason": "Tokenizer loading failure is not the same as add_special_tokens not emitting BOS/EOS." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Both are v5 tokenizer regressions, but one is clean_up_tokenization_spaces behavior and the other is DeepSeek token IDs." + "reason": "Both are tokenizer regressions, but they concern different tokenizers/models and likely different fixes." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Different tokenizer regression: Kimi-K2.5 codec/regex handling is not the same as clean_up_tokenization_spaces changes." + "reason": "Different tokenizer regressions for different model families; not the same underlying bug." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "Serve KV-cache continuation crash and auto_docstring annotation handling are unrelated." + "reason": "Serve KV-cache indexing crash and auto-docstring AttributeError are unrelated." }, { "left": "issue:44016", "right": "issue:46231", "accept": false, - "reason": "Notebook syntax error and docs parameter-name mismatch are not duplicates." + "reason": "Notebook syntax error and documentation parameter mismatch are not the same issue." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Image processor patch-count method bug and meta-initialization incompatibility are different multimodal issues." + "reason": "Different multimodal-loading problems: missing image-processor method vs remote-code checkpoint/meta-init incompatibility." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Incorrect weight-tying warning and missing tensor-parallel plan are separate Qwen3 issues." + "reason": "Both are Qwen-related, but one is a tied-weights warning bug and the other is a missing tensor-parallel plan." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Buffer read failure and double optimizer-step behavior are distinct runtime bugs." + "reason": "Buffer read/registration bug is unrelated to optimizer step being called twice with DeepSpeed." }, { - "left": "issue:42898", - "right": "issue:44448", + "left": "issue:44075", + "right": "issue:46231", "accept": false, - "reason": "Tokenization behavior regression and Pegasus output drift are not the same underlying issue." + "reason": "Optimizer argument handling is unrelated to a docs-only parameter name mismatch." }, { - "left": "issue:44075", - "right": "issue:46231", + "left": "issue:42898", + "right": "issue:44448", "accept": false, - "reason": "SGD argument handling and a documentation parameter-name mismatch are unrelated." + "reason": "Different v5 regressions affecting different model/tokenizer behavior." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "Unsupported PEFT version bump and leftover dummy PT classes are separate repository hygiene/release issues." + "reason": "PEFT version pinning and leftover dummy classes are unrelated maintenance issues." }, { "left": "issue:44448", "right": "issue:46132", "accept": false, - "reason": "Pegasus v4/v5 output change is unrelated to AttentionInterface.register altering registered-function behavior." + "reason": "Pegasus output regression and AttentionInterface.register behavior change are different code paths." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "FP8 save_pretrained handling and tokenizer save_pretrained metadata mismatch are different save-path bugs." + "reason": "Both touch save_pretrained, but one is FP8 MoE saving and the other is tokenizer metadata mismatch." }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "A missing import from image_utils is not the same as dummy classes leaking into dir(transformers)." + "reason": "Nonexistent image_utils import and dummy_pt_objects leakage are different import/symbol problems." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer config mismatch and DeepSpeed double-step behavior are unrelated." + "reason": "Tokenizer save_pretrained metadata bug is unrelated to DeepSpeed optimizer stepping." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Default config initialization and junk-filled non-persistent buffers are separate v5 regressions." + "reason": "Config default initialization and non-persistent buffer junk are different regressions." }, { - "left": "issue:30333", - "right": "issue:34689", + "left": "issue:44222", + "right": "issue:44623", "accept": false, - "reason": "Failed-job MLflow reporting and Llama 3.2 vision model loading are unrelated issues." + "reason": "Both concern save_pretrained, but one is FP8 MoE artifacts and the other is processor missing files." }, { - "left": "issue:44222", - "right": "issue:44623", + "left": "issue:30333", + "right": "issue:34689", "accept": false, - "reason": "FP8 MoE save_pretrained and processor save_pretrained missing files are different save/export problems." + "reason": "MLflow reporting on failed training and Llama 3.2 vision model loading are unrelated." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FP8 accuracy issue and missing processor files do not share the same bug." + "reason": "FP8 accuracy problems and processor save_pretrained file omissions are distinct bugs." }, { "left": "issue:45245", "right": "issue:46132", "accept": false, - "reason": "Category-cardinality runtime error and AttentionInterface.register behavior change are unrelated." + "reason": "Category-count runtime error and AttentionInterface.register behavior change are unrelated." }, { "left": "issue:30333", "right": "issue:36331", "accept": false, - "reason": "MLflow reporting on failed jobs and CustomTrainer.compute_loss signature breakage are different trainer issues." + "reason": "Failed-job MLflow reporting and CustomTrainer.compute_loss signature change are different training issues." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "Failed-job reporting and embedding reinitialization are unrelated." + "reason": "MLflow reporting and embedding reinitialization on resize are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Dataset loading in a script and LlamaConfig head_dim validation are unrelated." }, { "left": "issue:28282", "right": "issue:30333", "accept": false, - "reason": "PyTorch-import error and MLflow reporting on failed jobs are unrelated." + "reason": "Missing torch dependency import error is unrelated to MLflow job reporting." }, { - "left": "issue:44190", + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenization behavior change and FP8 inference failure on B200 are unrelated." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Different import errors: transformers.onnx module missing vs AutoModel requiring PyTorch." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "PretrainedConfig type checking regression is not the same as LlamaConfig head_dim validation." + }, + { + "left": "issue:42898", "right": "issue:46082", "accept": false, - "reason": "Local dataset loading in an example script and LlamaConfig head_dim validation are unrelated." + "reason": "Tokenizer behavior regression and LlamaConfig constructor validation are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer metadata save bug and Qwen3.5 MoE from_pretrained error are different problems." } ] }, @@ -12780,41 +12872,42 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:22355", - "issue:28282", "issue:30990", "issue:38175", + "issue:39401", "issue:41628", "issue:42757", "issue:42831", + "issue:42886", "issue:42907", - "issue:43122", + "issue:42994", "issue:43232", "issue:43475", "issue:43479", "issue:43644", "issue:43723", "issue:43761", + "issue:43824", + "issue:44038", "issue:44206", "issue:44222", "issue:44261", "issue:44263", "issue:44279", "issue:44295", - "issue:44297", "issue:44360", "issue:44373", "issue:44483", "issue:44484", "issue:44485", "issue:44514", + "issue:44568", "issue:44623", "issue:44829", "issue:44871", "issue:44913", "issue:44977", "issue:44991", - "issue:45071", "issue:45092", "issue:45103", "issue:45310", @@ -12825,16 +12918,11 @@ "issue:46002", "issue:46050", "issue:46082", - "issue:46209", "issue:46305", "issue:46328" ], "soft_pairs": [ - "issue:45071|issue:46082", - "issue:22355|issue:28282", - "issue:43122|issue:46209", "issue:43475|issue:46305", - "issue:44297|issue:45310", "issue:41628|issue:42757", "issue:44206|issue:45103", "issue:43723|issue:44977", @@ -12849,219 +12937,230 @@ "issue:43723|issue:44871", "issue:44279|issue:46328", "issue:43723|issue:44991", - "issue:44514|issue:45103", "issue:42907|issue:46050", - "issue:30990|issue:38175", + "issue:44514|issue:45103", "issue:45092|issue:45357", "issue:44263|issue:44360", + "issue:30990|issue:38175", "issue:43644|issue:46305", "issue:44829|issue:46002", "issue:42831|issue:45310", "issue:44261|issue:45092", - "issue:43479|issue:44913" + "issue:43479|issue:44913", + "issue:41628|issue:43824", + "issue:42831|issue:44038", + "issue:44263|issue:44373", + "issue:39401|issue:44568", + "issue:42886|issue:42994" ], - "prepared_review_unit_hash": "fac06f7ec641d9b297a018c82936db047d03c184ba182ec777d003c442a58c14", + "prepared_review_unit_hash": "031e6ff4c54974c358fe7179482bc43e444acefe74d4fdcbcb0cf1fb00da6b13", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36136, - "estimated_input_tokens": 9034, - "estimated_eval_tokens": 18324 + "soft_pair_count": 30, + "serialized_chars": 36436, + "estimated_input_tokens": 9109, + "estimated_eval_tokens": 18474 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are mostly unrelated closed issues across many different Transformers subsystems. The shared wording is superficial; none of the soft pairs look like the same underlying bug or change, so all soft edges should be rejected.", - "confidence": 0.9, - "canonical_issue_reason": "issue:43723 is the broadest, most general v5 loading regression in the set and the best single anchor if one issue must represent the cluster.", + "summary": "No soft pair looks like a true duplicate. The set is mostly a grab-bag of unrelated Transformers issues; the closest matches are usually just shared keywords like tokenizer, save_pretrained, or config handling, not the same concrete bug.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 43723 is the broadest, most representative issue in this set: a tokenizer-loading regression in v5, which is the closest shared theme across the tokenizer-related items.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43723 is the most representative issue here because it describes a clear, broadly applicable tokenizer-loading regression rather than a narrow model-specific edge case.", + "best_issue_reason": "Issue 43723 is also the most actionable as a triage anchor: it is a concrete regression report with a clear failure mode and version context, while the others are either more model-specific or orthogonal.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45071", - "right": "issue:46082", + "left": "issue:43475", + "right": "issue:46305", "accept": false, - "reason": "Different config/type-checking bug vs. Llama head_dim validation bug; same area, but not the same failure mode." + "reason": "Different models and different bugs: missing attribute in SAM 3 video vs a streaming mask off-by-one." }, { - "left": "issue:22355", - "right": "issue:28282", + "left": "issue:41628", + "right": "issue:42757", "accept": false, - "reason": "One is a missing transformers.onnx module import; the other is an AutoModel PyTorch dependency error." + "reason": "Both are import errors, but for unrelated symbols and subsystems; not the same underlying fix." }, { - "left": "issue:43122", - "right": "issue:46209", + "left": "issue:44206", + "right": "issue:45103", "accept": false, - "reason": "Tokenizer behavior regression and FP8 inference failure are unrelated bugs." - }, - { - "left": "issue:43475", - "right": "issue:46305", - "accept": false, - "reason": "SAM 3 video attribute error and MoonshineStreaming mask off-by-one are different code paths." - }, - { - "left": "issue:44297", - "right": "issue:45310", - "accept": false, - "reason": "Tokenizer save metadata mismatch vs. Qwen3.5 from_pretrained failure; not the same underlying issue." - }, - { - "left": "issue:41628", - "right": "issue:42757", - "accept": false, - "reason": "AutoImageProcessor import problem and huggingface_hub is_offline_mode import problem are separate dependency/import bugs." - }, - { - "left": "issue:44206", - "right": "issue:45103", - "accept": false, - "reason": "Feature extractor center-arg crash and auto_docstring future-annotations crash are unrelated." + "reason": "Feature extractor crash vs auto-docstring annotation crash; unrelated code paths." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Generic tokenizer loading regression and Qwen3.5 flash-attention generation bug are different issues." + "reason": "Tokenizer loading regression vs flash-attention generation failure; not the same bug." }, { "left": "issue:45381", "right": "issue:46305", "accept": false, - "reason": "Qwen2.5-VL vision position_ids bug and MoonshineStreaming window mask bug are not the same fix." + "reason": "Both mention position/masking semantics, but one is Qwen2.5-VL video positions and the other is Moonshine streaming window logic." }, { "left": "issue:43232", "right": "issue:45834", "accept": false, - "reason": "Generation kwargs handling after sync_gpus is unrelated to Kosmos2.5 OCR index errors." + "reason": "Generation kwargs update bug vs long OCR input index error; different failure modes." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 save_pretrained bug and shard-size question are different topics." + "reason": "FP8 save_pretrained failure vs a question about default max_shard_size; not duplicates." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "GlmMoeDsaIndexer torch.split issue and InternVL2 meta-init incompatibility are distinct problems." + "reason": "Moe DSA torch.split issue vs remote-code meta-init incompatibility; different concrete problems." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Position_ids buffer access error and processor.save_pretrained missing files are different save/load bugs." + "reason": "Buffer registration/read error vs processor save_pretrained missing files; separate save/load paths." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Docstring mismatch for position_ids and GLM-5 RoPE implementation are unrelated." + "reason": "Wrong docstring vs RoPE implementation bug; documentation and implementation are not the same issue." }, { "left": "issue:43761", "right": "issue:46082", "accept": false, - "reason": "CLIP hidden_states regression and LlamaConfig head_dim validation are unrelated." + "reason": "CLIP hidden_states regression vs Llama head_dim validation; unrelated model behaviors." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Chat completions API rejection and Gemma4Processor missing _tokenizer are separate failures." + "reason": "Chat/completions request rejection vs Gemma4Processor missing _tokenizer; different layers." }, { "left": "issue:43723", "right": "issue:44871", "accept": false, - "reason": "Tokenizer load regression and Gemma-3 eos_token_id mismatch are not the same bug." + "reason": "Tokenizer load regression vs eos_token_id configuration mismatch; related area, but different bug." }, { "left": "issue:44279", "right": "issue:46328", "accept": false, - "reason": "General dependency issue and SequenceFeatureExtractor.pad performance regression are unrelated." + "reason": "Generic dependency complaint vs feature extractor pad performance issue; no shared concrete defect." }, { "left": "issue:43723", "right": "issue:44991", "accept": false, - "reason": "Both involve tokenizer loading, but the specific failure reports and models differ." - }, - { - "left": "issue:44514", - "right": "issue:45103", - "accept": false, - "reason": "Qwen2_5_VL batched chat-template crash and auto_docstring annotations crash are unrelated." + "reason": "Both concern tokenizer loading, but for different models and failure causes." }, { "left": "issue:42907", "right": "issue:46050", "accept": false, - "reason": "Dequantized save failure and sinq save/load breakage are different quantization-related bugs." + "reason": "Both involve quantized-model persistence, but dequantized Ministrals/Devstrals saving and sinq save/load breakage are distinct." }, { - "left": "issue:30990", - "right": "issue:38175", + "left": "issue:44514", + "right": "issue:45103", "accept": false, - "reason": "Loading hang in sentence-transformers and zero probabilities in siglip2-base are unrelated symptoms." + "reason": "Batched chat template crash vs auto-docstring annotation crash; unrelated subsystems." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Remote-code meta-init incompatibility and incorrect visual encoder keys on save_pretrained are different issues." + "reason": "Meta-init incompatibility in old InternVL2 checkpoints vs incorrect visual encoder keys on save; different save/load bugs." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Same broad DSA indexer area, but one is a torch.split bug and the other is a missing ReLU; not the same underlying change." + "reason": "Same component family, but one is a torch.split return-value bug and the other is a missing ReLU; not the same underlying change." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Stuck loading sentence-transformers vs zero-probability outputs in SigLIP2; unrelated symptoms." }, { "left": "issue:43644", "right": "issue:46305", "accept": false, - "reason": "Non-persistent buffer junk and MoonshineStreaming mask off-by-one are unrelated." + "reason": "Non-persistent buffer junk in v5 vs streaming window off-by-one; no shared code-path." }, { "left": "issue:44829", "right": "issue:46002", "accept": false, - "reason": "Flash-attention training degeneration and NaNs after checkout/init are different failure modes." + "reason": "Flash-attention training degeneration vs NaNs after checkout/init; both are training-quality issues but not the same bug." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FineGrainedFP8 accuracy issue and Qwen3.5 from_pretrained error are unrelated." + "reason": "FP8 accuracy issue vs Qwen3.5 MoE from_pretrained error; different parts of the stack." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "MLA rms_norm_eps precision issue and InternVL2 meta-init incompatibility are different bugs." + "reason": "MLA normalization epsilon precision issue vs checkpoint meta-init incompatibility; unrelated." }, { "left": "issue:43479", "right": "issue:44913", "accept": false, - "reason": "Phi4MultimodalConfig default-init bug and GPTNeoXConfig rotary_pct reload bug are separate config issues." + "reason": "Default config initialization bug vs config value not persisting on reload; both are config-related but distinct." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Two import errors for different exported classes; too broad to be the same bug." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FineGrainedFP8 accuracy problem vs Qwen3-VL-Moe bug report; no concrete overlap." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "Indexer logic bug vs wrong docstring; one is code behavior, the other is documentation." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer offset_mapping bug vs add_special_tokens/BOS-EOS regression; both tokenizer issues, but different behaviors and likely different fixes." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache loading failure vs quantized model saving failure; separate persistence paths." } ] }, @@ -13096,7 +13195,6 @@ "issue:42831", "issue:42886", "issue:42907", - "issue:42994", "issue:43066", "issue:43122", "issue:43232", @@ -13105,8 +13203,8 @@ "issue:43646", "issue:43756", "issue:43784", - "issue:43824", - "issue:44038", + "issue:43937", + "issue:43976", "issue:44060", "issue:44117", "issue:44188", @@ -13119,13 +13217,14 @@ "issue:44464", "issue:44488", "issue:44521", - "issue:44568", "issue:44671", "issue:44704", + "issue:44743", "issue:44945", "issue:45278", "issue:45290", "issue:45593", + "issue:45600", "issue:45753", "issue:46002", "issue:46082", @@ -13135,15 +13234,10 @@ "issue:46369" ], "soft_pairs": [ - "issue:41628|issue:43824", - "issue:42831|issue:44038", - "issue:44263|issue:44373", - "issue:39401|issue:44568", - "issue:42886|issue:42994", "issue:42886|issue:43066", "issue:43408|issue:46369", - "issue:44368|issue:46305", "issue:42757|issue:45278", + "issue:44368|issue:46305", "issue:42831|issue:45753", "issue:22355|issue:28218", "issue:43232|issue:45593", @@ -13160,206 +13254,211 @@ "issue:44188|issue:44373", "issue:42907|issue:44945", "issue:43122|issue:46326", - "issue:44704|issue:45290", + "issue:44261|issue:44373", "issue:39401|issue:44488", - "issue:44261|issue:44373" + "issue:42907|issue:43937", + "issue:44704|issue:45290", + "issue:41628|issue:45278", + "issue:43976|issue:44945", + "issue:43232|issue:44743", + "issue:43784|issue:45600" ], - "prepared_review_unit_hash": "1594b9ff86746f36e64b23dc84c7667ea0e8496dae8acdba0eaa929b184eb72e", + "prepared_review_unit_hash": "12e25f49cf472004709dbf4b541e5c1634de3f5a970367285251d3489c6ea9b6", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 35704, - "estimated_input_tokens": 8926, - "estimated_eval_tokens": 18108 + "serialized_chars": 35705, + "estimated_input_tokens": 8927, + "estimated_eval_tokens": 18110 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft pair looks like a true duplicate; the set is mostly broad semantic similarity across different v5 regressions (tokenization, imports, generation, model-specific bugs). I would use issue 43122 as the closest representative because it is the broadest tokenizer-regression report and overlaps the largest related subgroup, but it is not a real duplicate target for the others.", - "confidence": 0.74, - "canonical_issue_reason": "Issue 43122 is the best umbrella among these for version-related tokenizer behavior regressions; it is the broadest and most reusable representative of the tokenizer-heavy subset.", + "summary": "This cluster is heterogeneous and mostly contains unrelated Transformers bugs/errors that only share generic library/version breakage language. There are no pull requests in the set. I would treat the cluster as low-quality for duplicate consolidation; if forced to pick an issue representative, the broad upgrade/import-breakage report is the closest umbrella, but it is still not a true duplicate anchor.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella-style report in the set (\u201cmany import errors after update\u201d), so it best matches the cluster\u2019s superficial similarity around API/import breakage. It is still not a strong duplicate canonical because the other items are about distinct subsystems and symptoms.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43122 is the most useful stand-in for triage because it captures a general cross-version tokenization regression pattern, even though the cluster itself is not duplicate-heavy.", + "best_issue_reason": "Issue 45278 is the best available representative only because it is the most generic and can subsume several import-related complaints. It is not a good true canonical for the cluster\u2019s mixed contents.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:41628", - "right": "issue:43824", - "accept": false, - "reason": "Both are import errors, but they are for different missing symbols/models and do not share the same concrete failure path." - }, - { - "left": "issue:42831", - "right": "issue:44038", - "accept": false, - "reason": "One is a FineGrainedFP8 accuracy issue; the other is a Qwen3-VL-Moe bug. Similar area, different bug." - }, - { - "left": "issue:44263", - "right": "issue:44373", - "accept": false, - "reason": "Unrelated: torch.split return handling vs a docstring mistake for position_ids." - }, - { - "left": "issue:39401", - "right": "issue:44568", - "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and different behaviors (offset mapping vs BOS/EOS insertion)." - }, - { - "left": "issue:42886", - "right": "issue:42994", - "accept": false, - "reason": "Offline cache loading failure and quantized-model saving failure are different code paths and symptoms." - }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "One is cache loading in offline mode; the other is an incorrect tokenizer decoder type in v5." + "reason": "Offline cache loading failure and a tokenizer decoder-type mismatch in v5 are different bugs in different code paths." }, { "left": "issue:43408", "right": "issue:46369", "accept": false, - "reason": "Different SAM3 model issues with different symptoms; not the same underlying bug." + "reason": "A model-type warning for sam3_tracker is unrelated to a memory overflow in sam3-litetext." }, { - "left": "issue:44368", - "right": "issue:46305", + "left": "issue:42757", + "right": "issue:45278", "accept": false, - "reason": "A warning about tie_word_embeddings is unrelated to a sliding-window mask off-by-one error." + "reason": "One is a specific missing import from huggingface_hub; the other is a broad upgrade regression with many import errors, not the same underlying bug." }, { - "left": "issue:42757", - "right": "issue:45278", + "left": "issue:44368", + "right": "issue:46305", "accept": false, - "reason": "45278 is a broad import-regression report, but this pair does not point to the same concrete missing import or root cause." + "reason": "A tie_word_embeddings warning in Qwen3.5 LoRA fine-tuning is unrelated to MoonshineStreaming\u2019s sliding-window off-by-one mask bug." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "Completely different bugs: FP8 accuracy vs an infinite-loop multimodal generation issue." + "reason": "FineGrainedFP8 accuracy degradation and a Qwen3_5 infinite loop on a specific image are distinct model/runtime failures." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "Module import failure and tokenizer spacing behavior are unrelated bugs." + "reason": "Missing transformers.onnx module import and extra tokenizer spacing after an added token are unrelated tokenizer/import issues." }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Generation kwarg/update logic and D-FINE auxiliary-loss behavior are unrelated." + "reason": "Generation kwargs handling after sync_gpus and missing auxiliary losses when denoising is off affect different generation/training paths." }, { "left": "issue:44464", "right": "issue:46082", "accept": false, - "reason": "Chunked generation inconsistency under compile and LlamaConfig validation are different problems." + "reason": "Chunked generation inconsistency under compiled forward is not the same as LlamaConfig rejecting an explicit head_dim value." }, { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Different model-specific issues: RoPE layer dropping vs a tied-weights warning." + "reason": "Smollm3 RoPE-layer dropping and an incorrect tied-weights warning in Qwen3-Next are different architecture/config bugs." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "A broken CLI/env command is not the same as a NameError during sentence-transformers import." + "reason": "A broken `transformers env` command and a NameError when importing sentence-transformers are unrelated failure modes." }, { "left": "issue:43646", "right": "issue:46002", "accept": false, - "reason": "Custom model initialization breakage and NaNs in classification heads are distinct failures." + "reason": "Custom model initialization breakage in Transformers 5.0 and NaNs in classification heads after checkout/init are separate problems." }, { "left": "issue:44671", "right": "issue:46132", "accept": false, - "reason": "Masked LM prediction quality and AttentionInterface.register behavior are separate issues." + "reason": "CamemBERT masked-LM prediction regression and AttentionInterface.register behavior change are different code-path issues." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "Tokenizer mapping returning None and AutoProcessor kwargs forwarding are different call sites and bugs." + "reason": "A tokenizer mapping assumption bug and AutoProcessor not forwarding kwargs to cached_file are unrelated API defects." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "Reading a buffer-backed position_ids tensor and saving the wrong tokenizer_class are unrelated." + "reason": "Reading position_ids after buffer registration and tokenizer_class metadata mismatch on save_pretrained are not the same bug." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Stuck loading Sentence Transformers and a MacOS bus error on a CLIP model do not share the same root cause." + "reason": "Sentence Transformers hanging on load and a MacOS bus error with a CLIP model are distinct runtime failures." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Both touch tokenizer-related APIs, but one is general tokenization drift and the other is assistant-mask generation for multimodal chat templates." + "reason": "Version-dependent tokenization differences and zero assistant masks for multimodal chat templates are different tokenizer/template issues." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 settings guidance and torch.split return values are unrelated." + "reason": "TF32 configuration guidance and a torch.split return-value issue in GlmMoeDsaIndexer have no shared underlying bug." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "A torch.compile-dependent attention-kernel divergence is not the same as a docstring typo." + "reason": "Torch.compile attention-kernel divergence and a wrong docstring are unrelated; one is a runtime correctness bug, the other is documentation." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Quantized/dequantized model saving and pipeline-parallel output correctness are different bugs." + "reason": "Saving dequantized models and incorrect output under pipeline parallelism are different failures in different subsystems." }, { "left": "issue:43122", "right": "issue:46326", "accept": false, - "reason": "Both are tokenizer behavior regressions, but they affect different mechanisms: general tokenization drift vs continue_final_message logic for mistral_common." + "reason": "Tokenization drift across versions is not the same as `continue_final_message` failing for mistral_common tokenizers." }, { - "left": "issue:44704", - "right": "issue:45290", + "left": "issue:44261", + "right": "issue:44373", "accept": false, - "reason": "Processor kwargs not reaching cached_file is different from apply_chat_template crashing on tool-call assistant messages." + "reason": "A missing rms_norm_eps causing precision error and a wrong docstring are unrelated." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Different tokenizer-loading failures for different models; not the same bug." + "reason": "Qwen3 offset_mapping errors and failure to load a specific model checkpoint are different tokenizer/model-loading bugs." }, { - "left": "issue:44261", - "right": "issue:44373", + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Dequantized model save failures and an invalid GenerationConfig for GLM-5 are separate issues." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "Passing kwargs to cached_file and apply_chat_template crashing on tool-call assistant messages involve different API paths." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Missing AutoImageProcessor import and broad import failures after upgrade are related only at a high level; they are not the same concrete bug." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version incompatibility and pipeline-parallelism output corruption are unrelated regressions." + }, + { + "left": "issue:43232", + "right": "issue:44743", "accept": false, - "reason": "A precision/config issue in MLA q_a_layernorm is unrelated to a docstring error." + "reason": "Generation kwargs update logic and Qwen3.5 recurrent-state reset under cache are different generation bugs." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Importing sentence-transformers causing NameError and stale auto_mappings references to removed configs are unrelated import-time failures." } ] }, @@ -13385,11 +13484,12 @@ "nodes": [ "issue:22355", "issue:28218", + "issue:29127", "issue:29159", "issue:30990", "issue:36296", + "issue:38617", "issue:39401", - "issue:41628", "issue:42175", "issue:42673", "issue:42757", @@ -13397,7 +13497,6 @@ "issue:43066", "issue:43116", "issue:43122", - "issue:43232", "issue:43278", "issue:43408", "issue:43452", @@ -13405,19 +13504,19 @@ "issue:43493", "issue:43784", "issue:43824", - "issue:43937", - "issue:43976", + "issue:43992", "issue:44038", "issue:44222", "issue:44295", "issue:44297", "issue:44351", + "issue:44355", "issue:44360", "issue:44485", "issue:44514", "issue:44521", "issue:44625", - "issue:44743", + "issue:44704", "issue:44779", "issue:44829", "issue:44938", @@ -13429,31 +13528,26 @@ "issue:45278", "issue:45446", "issue:45478", - "issue:45600", - "issue:45753" + "issue:45753", + "issue:46231" ], "soft_pairs": [ - "issue:42907|issue:43937", - "issue:41628|issue:45278", - "issue:43976|issue:44945", - "issue:43232|issue:44743", - "issue:43784|issue:45600", - "issue:39401|issue:44521", "issue:42907|issue:44964", - "issue:43116|issue:45200", + "issue:39401|issue:44521", "issue:43122|issue:44514", + "issue:43116|issue:45200", "issue:44625|issue:44829", "issue:43408|issue:43475", "issue:44038|issue:44297", "issue:44360|issue:44485", "issue:44038|issue:44295", - "issue:28218|issue:30990", "issue:42907|issue:43452", + "issue:28218|issue:30990", "issue:43122|issue:44779", "issue:44038|issue:44222", "issue:45216|issue:45478", - "issue:43824|issue:44351", "issue:36296|issue:43116", + "issue:43824|issue:44351", "issue:43784|issue:45446", "issue:22355|issue:29159", "issue:42757|issue:44351", @@ -13461,210 +13555,208 @@ "issue:42175|issue:44938", "issue:43493|issue:45753", "issue:42673|issue:45072", - "issue:43066|issue:44779" + "issue:43066|issue:44779", + "issue:22355|issue:29127", + "issue:43992|issue:44704", + "issue:44355|issue:46231", + "issue:38617|issue:45278" ], - "prepared_review_unit_hash": "73aaf723eaf9de91017f237de43e5832db39117744d366f52a28f82e67d1a416", + "prepared_review_unit_hash": "cc228e264968f8474e836bc2d407463a7ce9ba516fffe9476859413397d601ad", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 35938, - "estimated_input_tokens": 8985, - "estimated_eval_tokens": 18226 + "soft_pair_count": 28, + "serialized_chars": 35702, + "estimated_input_tokens": 8926, + "estimated_eval_tokens": 18108 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: it mixes tokenizer regressions, import errors, model-specific bugs, and generation/runtime issues. I do not see any soft pair that clearly represents the same underlying bug, so all candidate duplicate links should be rejected.", - "confidence": 0.72, - "canonical_issue_reason": "No single true canonical duplicate emerges from the set; if one issue must anchor the import-error subcluster, issue 45278 is the broadest umbrella because it explicitly reports many import failures after upgrading.", + "summary": "This cluster is mostly a grab bag of unrelated Hugging Face Transformers issues; the soft pairs are mostly keyword-similarity false positives rather than true duplicates.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 43122 is the broadest tokenizer-regression report here and the closest thing to a reusable duplicate target, though the overall set is heterogeneous.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best overall issue candidate only for the import-error family, since it is the most general and could reasonably absorb related symbol/import failures; however, it still does not represent the whole cluster.", + "best_issue_reason": "Issue 43122 is the best representative issue because it describes a user-facing tokenizer behavior regression across versions, which is broad enough to absorb the closest tokenizer-related complaints.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42907", - "right": "issue:43937", - "accept": false, - "reason": "Different bugs: save_pretrained/dequantization failure versus invalid GenerationConfig on GLM-5." - }, - { - "left": "issue:41628", - "right": "issue:45278", - "accept": false, - "reason": "Both are import-related, but they fail on different missing symbols and different code paths." - }, - { - "left": "issue:43976", - "right": "issue:44945", - "accept": false, - "reason": "Python-version load failure and incorrect pipeline-parallel output are unrelated." - }, - { - "left": "issue:43232", - "right": "issue:44743", - "accept": false, - "reason": "Generation kwargs after sync_gpus and recurrent-state reset with cache are distinct generation bugs." - }, - { - "left": "issue:43784", - "right": "issue:45600", + "right": "issue:44964", "accept": false, - "reason": "Sentence-transformers import NameError and Sam3LiteText auto-mapping CI break are different failures." + "reason": "Different models and failure modes: dequantized save failure vs loading a multimodal checkpoint." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Both involve Qwen3 tokenization/chat templating, but one is offset mapping and the other assistant masks in multimodal input." + "reason": "Both touch tokenizer/chat-template behavior, but one is wrong offset mapping and the other is assistant mask generation for multimodal inputs." }, { - "left": "issue:42907", - "right": "issue:44964", + "left": "issue:43122", + "right": "issue:44514", "accept": false, - "reason": "Different model families and different failure modes: dequantized save versus model load error." + "reason": "Both are tokenizer-related, but one is general tokenization drift across versions and the other is a batched chat-template crash." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Example-script multi-label output bug is unrelated to Gemma 4 mm_token_type_ids propagation." - }, - { - "left": "issue:43122", - "right": "issue:44514", - "accept": false, - "reason": "Both touch tokenization/chat templates, but one is a version-to-version tokenization change and the other is a batched template crash." + "reason": "Different tasks and code paths: example script empty multi-label outputs vs mm_token_type_ids handling in Gemma 4." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "num_labels propagation and flash_attention_3 degenerate training are unrelated." + "reason": "Distinct bugs: config propagation for Qwen3.5 vs degenerate training with flash_attention_3." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Same SAM3 Video area, but a config/model-type warning is not the same bug as a missing output attribute." + "reason": "Related to SAM 3 naming, but one is a model-type warning and the other is a missing attribute on an output object." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe runtime bug and tokenizer_class metadata mismatch are unrelated." + "reason": "One is a Qwen3-VL-Moe bug; the other is tokenizer save_pretrained writing the wrong tokenizer_class." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "DSA indexer ReLU discussion and GLM-5 RoPE implementation are unrelated topics." + "reason": "Different components: DSA indexer activation choice vs GLM-5 RoPE implementation." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Different Qwen3/MoE-related symptoms: model bug versus position_ids buffer access." + "reason": "Unrelated symptoms: Qwen3-VL-Moe bug vs position_ids buffer read error." }, { - "left": "issue:28218", - "right": "issue:30990", + "left": "issue:42907", + "right": "issue:43452", "accept": false, - "reason": "Tokenizer spacing bug and Sentence Transformers loading hang are different problems." + "reason": "Different failure paths: dequantized saving vs gguf_file breaking AutoTokenizer/AutoModel loading." }, { - "left": "issue:42907", - "right": "issue:43452", + "left": "issue:28218", + "right": "issue:30990", "accept": false, - "reason": "Dequantized save failure and gguf_file load breakage are separate serialization/loading issues." + "reason": "Tokenizer spacing bug and Sentence Transformers loading hang are not the same underlying issue." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and different behaviors." + "reason": "Both are tokenizer regressions in v5, but they affect different models and reproduce differently." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Qwen3-VL-Moe bug and FP8 save_pretrained moe bug are different model/features and code paths." + "reason": "Same broad family name, but one is Qwen3-VL-Moe behavior and the other is FP8 save_pretrained for moe models." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Qwen3.5 save_pretrained regression and Qwen3.5 from_pretrained error are related only loosely; they are not the same concrete bug." + "reason": "Both involve Qwen3.5 and version 5.4+, but one is save_pretrained corruption and the other is a from_pretrained load error; not enough evidence they're the same bug." }, { - "left": "issue:43824", - "right": "issue:44351", + "left": "issue:36296", + "right": "issue:43116", "accept": false, - "reason": "Missing Qwen2.5-VL export/import symbol and missing HybridCache export are different import errors." + "reason": "Tensor parallel training bug vs example script empty results; different subsystems and symptoms." }, { - "left": "issue:36296", - "right": "issue:43116", + "left": "issue:43824", + "right": "issue:44351", "accept": false, - "reason": "Tensor-parallel training bug and example script empty multi-label results are unrelated." + "reason": "Both are import errors, but they are for different missing symbols from different modules." }, { "left": "issue:43784", "right": "issue:45446", "accept": false, - "reason": "Sentence-transformers import failure and flex_attention version-check bug are unrelated." + "reason": "Importing sentence-transformers breaks due to a missing nn symbol, while the other is a PyTorch version check bug in flex_attention." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "Missing transformers.onnx module and slow/fast tokenizer inconsistency are different issues." + "reason": "Unrelated issues: missing transformers.onnx module vs slow/fast tokenizer inconsistency." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they reference different missing names and likely need separate fixes." + "reason": "Both are import errors, but they concern different libraries and symbols, so they are not the same defect." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Embedding dtype drift during evaluate and incorrect pipeline-parallel output are different defects." + "reason": "Different runtime problems: dtype shift between train/eval vs incorrect output under pipeline parallelism." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "Backend packaging omission for torch install and Python 3.14 load failure are unrelated." + "reason": "Backend packaging issue vs Python 3.14 load failure; too broad and not the same code path." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "Different multimodal bugs in different models; no shared concrete code path." + "reason": "Both are Qwen3_5-related, but one is SigLIP2/JAX parity and the other is an image-triggered infinite loop." }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "VRAM leak in Qwen3ForCausalLM threads and bfloat16 dtype mismatches are unrelated." + "reason": "Different bugs in different areas: VRAM leak in Qwen3_5 threading vs bfloat16 inference dtype mismatch in unrelated models." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both mention tokenizer v5 regressions, but they are model-specific and not the same bug." + "reason": "Both mention tokenizer behavior in v5, but one is decoder-type assignment and the other is Deepseek tokenization output regression." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Completely different problem domains: missing onnx module vs LayoutLMv3 error-message clarity." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Separate from-pretrained/cached-file behaviors affecting different model families and different missing weights/kwargs." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled Python load errors are unrelated to the documentation mismatch in set_training." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "Both are import problems after upgrading, but the missing symbol and broad import errors are not the same specific bug." } ] }, @@ -13688,16 +13780,14 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:22355", - "issue:29127", "issue:33453", - "issue:38617", "issue:39401", "issue:42222", "issue:42371", "issue:42673", "issue:43064", "issue:43116", + "issue:43122", "issue:43441", "issue:43450", "issue:43493", @@ -13710,266 +13800,282 @@ "issue:43937", "issue:43976", "issue:43986", - "issue:43992", "issue:44038", "issue:44075", "issue:44117", "issue:44188", "issue:44291", - "issue:44355", "issue:44380", + "issue:44410", "issue:44466", "issue:44485", - "issue:44704", "issue:44779", "issue:44936", "issue:44977", "issue:45071", + "issue:45081", "issue:45083", - "issue:45278", + "issue:45310", + "issue:45341", "issue:45357", + "issue:45405", "issue:45446", "issue:45478", + "issue:45561", "issue:45563", "issue:45588", "issue:46002", "issue:46077", - "issue:46231", + "issue:46298", "issue:46328" ], "soft_pairs": [ - "issue:22355|issue:29127", - "issue:43992|issue:44704", - "issue:44355|issue:46231", - "issue:38617|issue:45278", "issue:43937|issue:44485", - "issue:43611|issue:46002", "issue:44188|issue:46328", + "issue:43611|issue:46002", "issue:43116|issue:43493", "issue:42222|issue:43441", "issue:44075|issue:44936", "issue:42371|issue:43874", "issue:45357|issue:45478", - "issue:43976|issue:45446", "issue:44038|issue:45478", + "issue:43976|issue:45446", "issue:43116|issue:43825", - "issue:45083|issue:45446", "issue:43064|issue:43673", + "issue:45083|issue:45446", "issue:43937|issue:45563", "issue:33453|issue:39401", "issue:44380|issue:46077", - "issue:44117|issue:45588", "issue:39401|issue:44779", + "issue:44117|issue:45588", "issue:42673|issue:43906", "issue:43827|issue:45446", "issue:43937|issue:44977", "issue:43450|issue:43986", "issue:43673|issue:46328", "issue:44291|issue:45071", - "issue:44466|issue:46002" + "issue:44466|issue:46002", + "issue:45405|issue:45446", + "issue:43122|issue:46298", + "issue:44117|issue:45081", + "issue:45341|issue:45561", + "issue:44410|issue:45310", + "issue:45341|issue:46328" ], - "prepared_review_unit_hash": "b9fa2d7c7530aa1d0773d6cc5a379e72f00a6ba452017957be47bdf296cbc4a8", + "prepared_review_unit_hash": "30bca90f019b743423e329fef998e0442698d08a8565936fc670805d747a8e8d", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 29, - "serialized_chars": 36210, - "estimated_input_tokens": 9053, - "estimated_eval_tokens": 18362 + "soft_pair_count": 31, + "serialized_chars": 36787, + "estimated_input_tokens": 9197, + "estimated_eval_tokens": 18650 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous issues with a few lexical near-matches; the soft pairs generally point to different models, APIs, or bug classes rather than true duplicates. No PRs are present.", - "confidence": 0.34, - "canonical_issue_reason": "issue:45278 is the broadest umbrella item here: a cross-version upgrade report about many import errors, which is the closest thing to a cluster-level regression theme among otherwise narrow reports.", + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports across unrelated subsystems; the soft edges are all superficial and should be rejected. The most representative issue is the broad tokenizer regression report, with no PRs present.", + "confidence": 0.72, + "canonical_issue_reason": "issue:33453 is the broadest, earliest tokenizer-regression report in the set and best serves as an umbrella representative for the tokenizer-related subset.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the most suitable representative because it is broad, regression-oriented, and could plausibly subsume related import/loading breakages better than the model-specific or doc-only issues.", + "best_issue_reason": "issue:33453 is also the best global representative because its title is generic enough to cover the tokenizer-loading regression theme and it has the strongest discussion signal among the tokenizer cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:22355", - "right": "issue:29127", - "accept": false, - "reason": "Different failures: missing `transformers.onnx` module vs a LayoutLMv3 validation/error-message issue." - }, - { - "left": "issue:43992", - "right": "issue:44704", - "accept": false, - "reason": "Different code paths: UMT5 weight loading vs AutoProcessor/cached_file kwargs forwarding." - }, - { - "left": "issue:44355", - "right": "issue:46231", - "accept": false, - "reason": "A runtime error in compiled Python files is unrelated to a docs parameter-name mismatch." - }, - { - "left": "issue:38617", - "right": "issue:45278", - "accept": false, - "reason": "Both are import-related, but one is a specific missing symbol while the other is a broad upgrade report; not clearly the same bug." - }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "Different GLM-5 problems: invalid GenerationConfig vs RoPE implementation discussion." + "reason": "Both mention GLM-5, but one is a GenerationConfig validation failure and the other is a RoPE implementation discussion; different code paths and symptoms." }, { - "left": "issue:43611", - "right": "issue:46002", + "left": "issue:44188", + "right": "issue:46328", "accept": false, - "reason": "Model-loading `base_model_prefix` breakage is unrelated to NaNs in classification heads." + "reason": "torch.compile attention-kernel branching is unrelated to a pad() performance inefficiency in SequenceFeatureExtractor." }, { - "left": "issue:44188", - "right": "issue:46328", + "left": "issue:43611", + "right": "issue:46002", "accept": false, - "reason": "Attention-kernel divergence under `torch.compile` is unrelated to a padding performance issue." + "reason": "Model loading with base_model_prefix and NaNs in classification heads are different bugs in different parts of the stack." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Multi-label example-script output bug is unrelated to SigLIP2 implementation drift." + "reason": "Empty multi-label results in an example script and SigLIP2 implementation discrepancies are unrelated issues." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "VitPose import breakage and Ministral-3 FlashAttention failure are unrelated model-specific issues." + "reason": "Vitpose model breakage and Ministral FlashAttention failure involve different models and failure modes." }, { "left": "issue:44075", "right": "issue:44936", "accept": false, - "reason": "Optimizer argument handling is not the same as `trainer.evaluate()` failing after training." + "reason": "SGD arguments being ignored is distinct from trainer.evaluate() failing after training; they are separate trainer bugs." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 API guidance and a missing image-processor method are unrelated." + "reason": "A TF32 settings API warning is unrelated to a missing get_number_of_image_patches method in a processor." }, { "left": "issue:45357", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but one is save_pretrained key serialization and the other is from_pretrained loading failure; not enough to treat as one bug." + "reason": "Both are Qwen3.5-related, but one is save_pretrained key serialization and the other is a from_pretrained loading error; not the same bug." }, { - "left": "issue:43976", - "right": "issue:45446", + "left": "issue:44038", + "right": "issue:45478", "accept": false, - "reason": "Python version support breakage is unrelated to a flex_attention import-version check." + "reason": "These target different model families/symptoms: Qwen3-VL-Moe behavior versus Qwen3.5 Moe loading failure." }, { - "left": "issue:44038", - "right": "issue:45478", + "left": "issue:43976", + "right": "issue:45446", "accept": false, - "reason": "Different Qwen variants and different failure modes; too broad to merge." + "reason": "Python-version support mismatch and a flex_attention import version check are separate dependency issues." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Example-script multi-label behavior is unrelated to a pipeline() error-message regression." + "reason": "Example-script multi-label output failure is unrelated to a pipeline() error message about translation support." }, { - "left": "issue:45083", - "right": "issue:45446", + "left": "issue:43064", + "right": "issue:43673", "accept": false, - "reason": "qwen3_omni_moe helper behavior is unrelated to the flex_attention PyTorch version check." + "reason": "FSDP2/PEFT optimizer-state corruption and missing GenerationMixin cache in chunked_prefill are different code paths." }, { - "left": "issue:43064", - "right": "issue:43673", + "left": "issue:45083", + "right": "issue:45446", "accept": false, - "reason": "Distributed optimizer-state corruption and generation cache absence are separate training vs inference bugs." + "reason": "qwen3_omni_moe feature-length calculation and flex_attention PyTorch version gating are unrelated bugs." }, { "left": "issue:43937", "right": "issue:45563", "accept": false, - "reason": "Both touch generation, but invalid GenerationConfig and stale warning text are different defects." + "reason": "Invalid GenerationConfig and a stale num_return_sequences warning are different generation issues." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Tokenizer loading regression and Qwen3 offset_mapping are different tokenizer bugs." + "reason": "Both are tokenizer-related, but one is loading regression and the other is wrong offset_mapping; not the same concrete bug." }, { "left": "issue:44380", "right": "issue:46077", "accept": false, - "reason": "Both involve attention behavior, but GPT2 backend scaling and Gemma4 bidirectional masking are distinct problems." + "reason": "GPT2 attention scaling under SDPA/FlashAttention and Gemma4 bidirectional-mask construction are different attention bugs." }, { - "left": "issue:44117", - "right": "issue:45588", + "left": "issue:39401", + "right": "issue:44779", "accept": false, - "reason": "Tokenizer mapping assumptions and flash_attention sinkless-model crash are unrelated." + "reason": "Tokenizer offset_mapping regressions for Qwen3 and incorrect Deepseek tokenization are similar in theme but distinct model/code-path failures." }, { - "left": "issue:39401", - "right": "issue:44779", + "left": "issue:44117", + "right": "issue:45588", "accept": false, - "reason": "Both are tokenizer regressions, but for different models and different symptoms; not the same underlying bug." + "reason": "TOKENIZER_MAPPING_NAMES returning None is unrelated to a flash_attention crash on s_aux=None." }, { "left": "issue:42673", "right": "issue:43906", "accept": false, - "reason": "VRAM leakage in multi-dataloader-thread use does not match an isolated reproduction of another issue." + "reason": "A VRAM leak across dataloader threads is not the same as an isolated reproduction of an older unrelated issue." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still referencing removed pipeline() APIs is unrelated to a flex_attention import guard bug." + "reason": "Docs still referencing pipeline() and a flex_attention import version check are unrelated documentation vs runtime issues." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "GLM-5 generation config validation and Qwen3.5 flash-attention generation failure are different model issues." + "reason": "Both are generation-related, but one is GenerationConfig validation and the other is flash-attention generation failure for a different model." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Video processor batching shape bug is not the same as the AutoProcessor/torchvision crash." + "reason": "Batched video-processor shape bugs and missing torchvision crash handling are both video-related but not the same failure." }, { "left": "issue:43673", "right": "issue:46328", "accept": false, - "reason": "Generation-cache missing and SequenceFeatureExtractor.pad performance are unrelated." + "reason": "Missing cache in chunked_prefill and inefficient numpy-to-list conversion in pad() are unrelated issues." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "`init_empty_weights` argument handling and `PretrainedConfig` type checking are both v5 regressions but not the same code-path problem." + "reason": "Unexpected _is_hf_initialized handling and PretrainedConfig type-checking regressions are separate v5 breakages." }, { "left": "issue:44466", "right": "issue:46002", "accept": false, - "reason": "Serialization of tied weights is unrelated to NaNs in classification heads." + "reason": "Inconsistent lm_head serialization and NaNs in classification heads are different bugs affecting different behaviors." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "A PEFT minimum-version bump and a flex_attention import version check are unrelated dependency problems." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Tokenizer behavior changes and a noisy logging message do not look like the same underlying issue." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping None and a Mistral regex patch crashing on backend_tokenizer are both tokenizer-adjacent but distinct failures." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "A small testing_utils bug and a pytest-xdist race in captured_info.txt are separate testing-related issues." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing qwen3next projections and a Qwen3.5 Moe from_pretrained error are different model-implementation/loading problems." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "testing_utils cleanup bug and SequenceFeatureExtractor.pad inefficiency are unrelated." } ] }, @@ -14005,56 +14111,51 @@ "issue:42831", "issue:43012", "issue:43066", + "issue:43097", "issue:43122", "issue:43408", "issue:43421", "issue:43577", "issue:43582", + "issue:43673", "issue:43742", + "issue:43874", "issue:43937", "issue:44075", - "issue:44117", "issue:44220", "issue:44261", + "issue:44263", "issue:44410", "issue:44483", "issue:44484", + "issue:44485", + "issue:44493", "issue:44496", "issue:44829", "issue:44877", "issue:44933", - "issue:45081", + "issue:45071", "issue:45216", "issue:45278", "issue:45310", "issue:45341", "issue:45362", - "issue:45405", - "issue:45446", "issue:45464", "issue:45478", "issue:45542", - "issue:45561", "issue:45656", "issue:46178", "issue:46231", - "issue:46298", "issue:46328" ], "soft_pairs": [ - "issue:45405|issue:45446", - "issue:44117|issue:45081", - "issue:43122|issue:46298", - "issue:45341|issue:45561", - "issue:44410|issue:45310", - "issue:45341|issue:46328", "issue:44829|issue:45216", "issue:42222|issue:46178", "issue:38617|issue:41628", "issue:42617|issue:45341", - "issue:42673|issue:43582", "issue:43408|issue:43742", "issue:45362|issue:46328", + "issue:42673|issue:43582", "issue:39401|issue:43421", "issue:42490|issue:42673", "issue:44877|issue:46231", @@ -14065,215 +14166,228 @@ "issue:44933|issue:45278", "issue:43066|issue:43122", "issue:42222|issue:45542", - "issue:44483|issue:45464", - "issue:42371|issue:42831", "issue:44410|issue:45478", - "issue:43012|issue:43577", + "issue:42371|issue:42831", + "issue:44483|issue:45464", "issue:44075|issue:45656", - "issue:42175|issue:42222" + "issue:42175|issue:42222", + "issue:43012|issue:43577", + "issue:43097|issue:44493", + "issue:43673|issue:45341", + "issue:41628|issue:45071", + "issue:44263|issue:44485", + "issue:45071|issue:45542", + "issue:43012|issue:43408", + "issue:43874|issue:44261" ], - "prepared_review_unit_hash": "86147d9d6cbf2ec390bec1d195a89e0b7003b9c8a3347da8b6f8b4d1d00c469c", + "prepared_review_unit_hash": "fc394dbc5f902adabb1b49bb9ff66b3fc00e3e4f2900e644a271dc536314bd55", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35531, - "estimated_input_tokens": 8883, - "estimated_eval_tokens": 18022 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36276, + "estimated_input_tokens": 9069, + "estimated_eval_tokens": 18394 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, with one clear duplicate pair: the Qwen3.5 MoE `from_pretrained` error. Most other pairs only share a broad subsystem or symptom class, not the same underlying bug.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 44484 is the clearest anchor: it states the concrete Qwen3.5 MoE `from_pretrained` failure, and it has an exact duplicate in 45478 with the same title and bug.", + "summary": "This cluster is heterogeneous: it mixes import regressions, tokenizer/decoder behavior changes, model-specific loading/runtime bugs, and backend/optimizer issues. None of the soft pairs look like strong duplicates of the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.22, + "canonical_issue_reason": "Issue 45278 is the best umbrella representative because it most broadly captures the import-regression theme seen in several nearby reports, but the overall cluster is too mixed for a single true duplicate center.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44484 is the best representative because it is the earliest, most specific report for the only exact-match duplicate pair in this set.", + "best_issue_reason": "Issue 45278 is the most suitable global representative: it is broad, user-facing, and closest to the repeated import-error complaints in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:45405", - "right": "issue:45446", - "accept": false, - "reason": "Unrelated version/dependency bump vs a PyTorch version-check bug; different failure modes and code paths." - }, - { - "left": "issue:44117", - "right": "issue:45081", - "accept": false, - "reason": "Both are tokenizer-related, but one is a mapping assumption bug and the other is a Mistral regex patch crash; not the same defect." - }, - { - "left": "issue:43122", - "right": "issue:46298", - "accept": false, - "reason": "Different tokenization regression vs noisy logging; no shared underlying bug." - }, - { - "left": "issue:45341", - "right": "issue:45561", - "accept": false, - "reason": "Both mention testing utilities, but one is a vague small bug and the other is a specific xdist file race; not the same issue." - }, - { - "left": "issue:44410", - "right": "issue:45310", - "accept": false, - "reason": "Different Qwen model bugs: missing projections in one layer vs a `from_pretrained` load error." - }, - { - "left": "issue:45341", - "right": "issue:46328", - "accept": false, - "reason": "Testing-utils bug vs a `SequenceFeatureExtractor.pad` performance issue; unrelated." - }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "Both touch model behavior, but one is FA3 training degeneration and the other is a broken save/load checkpoint regression." + "reason": "Both involve Qwen-related regressions, but one is a flash-attention training degradation and the other is a save_pretrained checkpoint regression; different code paths and symptoms." }, { "left": "issue:42222", "right": "issue:46178", "accept": false, - "reason": "Different backend/model errors: vitpose model breakage vs missing `tensorflow_text` backend mapping." + "reason": "VitPose model breakage and a tensorflow_text backend-mapping error are unrelated failures." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but for different missing symbols and unrelated APIs." + "reason": "Both are import errors, but they concern different missing symbols in different modules and do not point to the same bug." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "A 3D parallel script failure and a testing-utils bug are unrelated." - }, - { - "left": "issue:42673", - "right": "issue:43582", - "accept": false, - "reason": "VRAM leak in Qwen3 thread usage vs Apple Silicon allocator TypeError; different bugs." + "reason": "A failure to run 3d_parallel.py and a small testing_utils bug are not the same issue." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "A model-type warning and a key error on loading a different model are not the same problem." + "reason": "Model-type warning for sam3_tracker vs a KeyError loading MobileLLM-125M are unrelated." }, { "left": "issue:45362", "right": "issue:46328", "accept": false, - "reason": "Qwen chat crash vs a sequence feature extractor performance issue; unrelated." + "reason": "Qwen chat crashes and SequenceFeatureExtractor.pad performance are different subsystems with no shared underlying defect." + }, + { + "left": "issue:42673", + "right": "issue:43582", + "accept": false, + "reason": "A VRAM leak in Qwen3ForCausalLM threads and an Apple Silicon TypeError in caching_allocator_warmup are unrelated." }, { "left": "issue:39401", "right": "issue:43421", "accept": false, - "reason": "Both are tokenizer internals, but offset mapping corruption and post-processor update behavior are different concrete bugs." + "reason": "Both affect tokenization, but one is wrong offset_mapping and the other is runtime post-processor updates; not the same concrete bug." }, { "left": "issue:42490", "right": "issue:42673", "accept": false, - "reason": "Local-path import warning vs VRAM leak in multi-threaded Qwen3 usage; no overlap." + "reason": "Local-path import warning and Qwen3 VRAM leak are unrelated." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Strict config loading bug vs a documentation parameter-name mismatch." + "reason": "Strict config loading and a docs parameter-name mismatch are unrelated." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "TF32 warning/configuration issue vs MLA layer precision/config bug; different code paths." + "reason": "TF32 API warning and MLA rms_norm_eps accuracy issue do not share the same root cause." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "Different Qwen3.5 MoE load errors with different reports and symptoms; not enough evidence they are the same bug." + "reason": "max_shard_size behavior and Qwen3.5 MoE from_pretrained errors are different issues." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "ImportError in configuration utils vs a fbank feature-extraction issue; unrelated." + "reason": "ImportError for layer_type_validation and _torch_extract_fbank_features are unrelated." }, { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "GenerationConfig validation on GLM-5 vs unrecognized model/config loading; distinct failures." + "reason": "Invalid GenerationConfig and unrecognized model/config.json errors are separate loading problems." }, { "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "Both involve imports, but one is a specific nonexistent import and the other is a broad upgrade-time import breakage." + "reason": "The first is a specific missing import from image_utils; the second is a broad upgrade-time import regression, so they are not the same bug." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both are tokenizer version regressions, but decoder-type mismatch and tokenization output changes are different concrete bugs." + "reason": "Both are tokenizer-related v5 regressions, but one is decoder-type selection and the other is changed tokenization output; too little evidence of one concrete shared defect." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "Vitpose model breakage vs missing TensorFlow backend when only tensorboard is installed; unrelated." + "reason": "VitPose breakage and missing TensorFlow backend after installing tensorboard are unrelated." }, { - "left": "issue:44483", - "right": "issue:45464", + "left": "issue:44410", + "right": "issue:45478", "accept": false, - "reason": "Both are chat/completions failures, but one is a general API rejection and the other is a specific streaming inference crash on Qwen3.5-0.8B." + "reason": "Missing attn projections in qwen3next and Qwen3.5 MoE from_pretrained errors are different model-loading bugs." }, { "left": "issue:42371", "right": "issue:42831", "accept": false, - "reason": "TF32 settings warning vs FineGrainedFP8 accuracy regression; different issues." + "reason": "TF32 behavior control and FineGrainedFP8 accuracy issues are not the same problem." }, { - "left": "issue:44410", - "right": "issue:45478", + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is a general API acceptance failure and the other is a model-specific streaming inference failure." + }, + { + "left": "issue:44075", + "right": "issue:45656", "accept": false, - "reason": "Missing projections in qwen3next vs the Qwen3.5 MoE `from_pretrained` error; unrelated." + "reason": "Both are optimizer-related, but one says SGD args are ignored and the other says optimizer.step is called twice under deepspeed; distinct bugs." + }, + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "Missing TensorFlow in a torch install and VitPose model breakage are different issues." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "PyTorch warning on bfloat16 compilation vs BLIP2 dtype propagation bug; not the same defect." + "reason": "A compile-time float32 warning for bf16 and a loading-time dtype mismatch in Blip2 are not the same underlying bug." }, { - "left": "issue:44075", - "right": "issue:45656", + "left": "issue:43097", + "right": "issue:44493", "accept": false, - "reason": "Optimizer args ignored vs optimizer step called twice under deepspeed; different optimizer bugs." + "reason": "Removed tie_embeddings behavior and unexpected position_id keys are unrelated configuration/model-loading regressions." }, { - "left": "issue:42175", - "right": "issue:42222", + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "GenerationMixin cache handling and a testing_utils bug have no shared code path." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "ImportError for AutoImageProcessor and PreTrainedConfig type-checking breakage are unrelated regressions." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "A torch.split return-value issue in GlmMoeDsaIndexer and a RoPE implementation discussion are not the same concrete bug." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Config type checking and TensorFlow backend resolution are unrelated." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "These are both warnings, but they concern totally different models and mechanisms." + }, + { + "left": "issue:43874", + "right": "issue:44261", "accept": false, - "reason": "TensorFlow not included in torch install vs vitpose model breakage; unrelated." + "reason": "Missing get_number_of_image_patches in a GLM image processor and an MLA rms_norm_eps precision issue are unrelated." } ] }, @@ -14298,19 +14412,18 @@ "cluster_id": "cluster-10000-445", "nodes": [ "issue:38617", - "issue:41628", "issue:42757", "issue:42890", - "issue:43012", "issue:43064", "issue:43097", "issue:43335", - "issue:43408", "issue:43441", "issue:43452", "issue:43502", "issue:43531", "issue:43673", + "issue:43824", + "issue:43873", "issue:43874", "issue:43931", "issue:43986", @@ -14321,22 +14434,23 @@ "issue:44261", "issue:44263", "issue:44315", + "issue:44351", "issue:44360", "issue:44410", "issue:44484", "issue:44485", - "issue:44493", + "issue:44704", "issue:44829", "issue:44933", "issue:44987", "issue:45003", "issue:45071", + "issue:45103", "issue:45341", "issue:45372", "issue:45375", "issue:45440", "issue:45478", - "issue:45542", "issue:45588", "issue:45753", "issue:46050", @@ -14347,13 +14461,6 @@ "issue:46369" ], "soft_pairs": [ - "issue:43097|issue:44493", - "issue:43673|issue:45341", - "issue:41628|issue:45071", - "issue:44263|issue:44485", - "issue:45071|issue:45542", - "issue:43874|issue:44261", - "issue:43012|issue:43408", "issue:44263|issue:46172", "issue:44360|issue:46172", "issue:44484|issue:45478", @@ -14362,16 +14469,16 @@ "issue:43335|issue:45071", "issue:43931|issue:44038", "issue:43097|issue:43452", - "issue:43673|issue:46123", "issue:45003|issue:45341", + "issue:43673|issue:46123", "issue:43986|issue:45341", "issue:44933|issue:46231", "issue:45753|issue:46369", "issue:44117|issue:44987", "issue:43986|issue:44484", "issue:43874|issue:44263", - "issue:43986|issue:45372", "issue:45440|issue:46172", + "issue:43986|issue:45372", "issue:43531|issue:45341", "issue:44261|issue:44485", "issue:43097|issue:46050", @@ -14382,252 +14489,245 @@ "issue:44410|issue:45375", "issue:44038|issue:44410", "issue:42890|issue:46298", - "issue:43502|issue:43992" + "issue:43502|issue:43992", + "issue:38617|issue:44351", + "issue:43097|issue:43873", + "issue:44704|issue:45103", + "issue:43986|issue:44117", + "issue:43824|issue:43986" ], - "prepared_review_unit_hash": "17d7e9ffef6e1033402a66402a3640115f2d6cf2badbf3a2c5c28efbd15188d0", + "prepared_review_unit_hash": "3743df9bdc065dd39a4ef37810a13af196b622c1d9da394faf4042a99a7c219b", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 36, - "serialized_chars": 38314, - "estimated_input_tokens": 9579, - "estimated_eval_tokens": 19414 + "soft_pair_count": 34, + "serialized_chars": 37751, + "estimated_input_tokens": 9438, + "estimated_eval_tokens": 19132 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly unrelated Transformers bug reports spanning imports, model loading, attention, configs, and training; the similarity is largely lexical rather than the same underlying defect.", - "confidence": 0.98, - "canonical_issue_reason": "issue:43502 is the clearest core bug report: it has a concrete, reproducible incorrect behavior in loading semantics and broad user impact.", + "summary": "This cluster is highly heterogeneous: mostly unrelated Transformers regressions spanning imports, model loading, config/schema issues, training, attention, and utility bugs. None of the soft candidate pairs look like true duplicates or mergeable PR-equivalent changes, so all soft edges should be rejected.", + "confidence": 0.93, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43502 is the strongest standalone issue for triage because it is specific, actionable, and clearly describes the failing code path.", + "best_issue_reason": "Issue 43097 is the broadest and most impactful regression report in the set, with discussion and inbound references, but it still does not represent the rest of the unrelated issues.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43097", - "right": "issue:44493", - "accept": false, - "reason": "Different problems: one is an API removal/compatibility complaint, the other is a position-id/key mismatch in model outputs." - }, - { - "left": "issue:43673", - "right": "issue:45341", - "accept": false, - "reason": "Unrelated areas: generation cache behavior vs a small testing utility bug." - }, - { - "left": "issue:41628", - "right": "issue:45071", - "accept": false, - "reason": "Both are import/type issues, but they concern different symbols and code paths." - }, - { - "left": "issue:44263", - "right": "issue:44485", - "accept": false, - "reason": "Different GLM internals: torch.split handling vs RoPE implementation." - }, - { - "left": "issue:45071", - "right": "issue:45542", - "accept": false, - "reason": "One is config type checking, the other is backend detection when TensorFlow is absent." - }, - { - "left": "issue:43874", - "right": "issue:44261", - "accept": false, - "reason": "Different failures in different modules: image processor method missing vs MLA norm epsilon precision." - }, - { - "left": "issue:43012", - "right": "issue:43408", - "accept": false, - "reason": "Unrelated symptoms: PyTorch bfloat16 warning vs a model-type mismatch warning." - }, { "left": "issue:44263", "right": "issue:46172", "accept": false, - "reason": "Different model math/code paths; no shared defect beyond both being tensor ops." + "reason": "Different model families and different bugs: GlmMoeDsaIndexer split behavior vs DeepSeek v4 tensor transpose/cast logic." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "Distinct implementation bugs in separate model families." + "reason": "DSA indexer activation bug vs DeepSeek v4 comb transformation bug; no shared code path." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "A save_pretrained shard-size question is unrelated to a Qwen3.5 from_pretrained failure." + "reason": "save_pretrained shard-size question is unrelated to Qwen3.5 MoE loading failures." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Different shape/loading mismatch vs missing projections in another model." + "reason": "Both are Qwen-related, but one is a VL weight-shape mismatch and the other is missing projections in qwen3next; different defects." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Both mention flash attention, but one is a model-specific crash and the other is degenerate training behavior." + "reason": "Both mention flash attention, but one is a model-specific failure and the other is degenerate training behavior in classification." }, { "left": "issue:43335", "right": "issue:45071", "accept": false, - "reason": "SwitchTransformers config construction bug is unrelated to PretrainedConfig type checking." + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to PretrainedConfig type checking regressions." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Different Qwen VL/MoE issues with different failure modes." + "reason": "Same broad model family, but different failures: loading shape mismatch vs version-specific Qwen3-VL-Moe breakage." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "API removal vs gguf_file handling in tokenizer/model loading are separate issues." + "reason": "Embeddings tie-setting removal and gguf_file loading failure are unrelated from_pretrained/save-pretrained issues." }, { - "left": "issue:43673", - "right": "issue:46123", + "left": "issue:45003", + "right": "issue:45341", "accept": false, - "reason": "Both involve generation behavior, but cache absence and pipeline partial-batch dropping are different bugs." + "reason": "Unsafely accessing sys.modules and a testing_utils bug are different code paths and symptoms." }, { - "left": "issue:45003", - "right": "issue:45341", + "left": "issue:43673", + "right": "issue:46123", "accept": false, - "reason": "Unsafe sys.modules access is unrelated to a testing utility bug." + "reason": "Cache handling during chunked_prefill vs pipeline partial-batch is_last logic are unrelated generation/pipeline bugs." }, { "left": "issue:43986", "right": "issue:45341", "accept": false, - "reason": "Video-model processor crash without torchvision is unrelated to testing utils." + "reason": "Processor loading without torchvision has nothing to do with the testing_utils issue." }, { "left": "issue:44933", "right": "issue:46231", "accept": false, - "reason": "Missing import from image_utils and a docs parameter-name mismatch are not the same issue." + "reason": "Missing import from image_utils is unrelated to a documentation mismatch in set_training()." }, { "left": "issue:45753", "right": "issue:46369", "accept": false, - "reason": "An infinite-loop generation bug and a memory overflow report are distinct failure modes." + "reason": "Infinite-loop generation for a specific image is not the same as memory overflow in sam3-litetext." }, { "left": "issue:44117", "right": "issue:44987", "accept": false, - "reason": "Tokenizer mapping None handling is different from a specific model loading failure in transformers>=5.1.0." + "reason": "Tokenizer mapping None handling and a specific model load failure are different from_pretrained problems, not the same bug." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "A processor dependency crash is unrelated to a save_pretrained shard-size discussion." + "reason": "Video-model processor crash without torchvision is unrelated to shard-size behavior." }, { "left": "issue:43874", "right": "issue:44263", "accept": false, - "reason": "Missing image-patch helper and torch.split return handling are unrelated model subsystems." + "reason": "Missing image-patch count method in an image processor is unrelated to torch.split return handling in a DSA indexer." }, { - "left": "issue:43986", - "right": "issue:45372", + "left": "issue:45440", + "right": "issue:46172", "accept": false, - "reason": "Different missing-dependency/import failures in separate processor stacks." + "reason": "DeepseekV3MoE parity mismatch and DeepSeek v4 tensor manipulation are distinct model implementation issues." }, { - "left": "issue:45440", - "right": "issue:46172", + "left": "issue:43986", + "right": "issue:45372", "accept": false, - "reason": "Native DeepseekV3 divergence is a broader implementation mismatch, not the same as the comb transpose bug." + "reason": "Different missing dependency/import problems on different processor-loading paths; not one underlying bug." }, { "left": "issue:43531", "right": "issue:45341", "accept": false, - "reason": "A Qwen3-MoE sliding-window issue is unrelated to testing utilities." + "reason": "Qwen3-MoE sliding_window bug is unrelated to a testing utility fix." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Precision/epsilon handling in MLA q_a_layernorm is not the same as GLM-5 RoPE implementation." + "reason": "RMS norm epsilon precision issue and GLM-5 RoPE implementation are separate model math bugs." }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "API deprecation/behavior change is unrelated to quantized model save/load breakage." + "reason": "Removed tie_embeddings behavior and quantized model save/load corruption are different persistence regressions." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "Removed tie_embeddings API and flash_attention s_aux=None crash are different code paths." + "reason": "Tie-embedding removal is unrelated to flash_attention.py crashing on sink-less models." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "Wrong optimizer state under FSDP2/PEFT is unrelated to Liger Kernel not applying with model_init." + "reason": "Trainer optimizer-state corruption under FSDP2/PEFT is unrelated to Liger Kernel application with model_init." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but for different modules and symbols with different root causes." + "reason": "Both are import errors, but they involve different missing symbols in different modules and likely different fixes." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "AddedToken keyword duplication and tokenizer-mapping None are unrelated tokenizer bugs." + "reason": "AddedToken special-argument collision and tokenizer mapping None handling are unrelated tokenizer bugs." }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Missing attention projections in qwen3next and a missing config field in Qwen3_5MoeVisionConfig are different defects." + "reason": "Missing qwen3next projections and a missing Qwen3_5MoeVisionConfig field are separate configuration/model-definition problems." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Both involve Qwen models, but one is a Transformers v5/Qwen3-VL-Moe failure and the other is missing layer projections." + "reason": "Qwen3-VL-Moe version failure and qwen3next missing projections are not the same concrete code-path problem." }, { "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "A flaky integration test due to missing set_seed and a noisy logging message are not the same bug." + "reason": "Seed-related flaky test failure and noisy logging are completely unrelated." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "local_files_only still causing API calls is unrelated to a missing embed_tokens.weight in UMT5Encoder." + "reason": "local_files_only networking violation and missing embed token weights are different from_pretrained bugs." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import-name errors, but they concern different APIs and likely different export paths; not one duplicate bug." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "A removed config behavior and quantization offloading failure are unrelated regressions." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and auto_docstring annotation parsing crash are separate utility issues." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video processor dependency crash and tokenizer mapping None handling are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43986", + "accept": false, + "reason": "Different import/loading failures for different multimodal model paths; no shared underlying bug." } ] }, @@ -14666,17 +14766,17 @@ "issue:43526", "issue:43550", "issue:43582", + "issue:43618", + "issue:43761", "issue:43824", "issue:43825", - "issue:43873", "issue:43906", "issue:43976", - "issue:43986", "issue:43992", - "issue:44117", + "issue:44038", "issue:44162", "issue:44222", - "issue:44351", + "issue:44230", "issue:44368", "issue:44380", "issue:44483", @@ -14684,10 +14784,9 @@ "issue:44514", "issue:44655", "issue:44683", - "issue:44704", "issue:44936", "issue:44938", - "issue:45103", + "issue:45290", "issue:45478", "issue:45479", "issue:45600", @@ -14701,11 +14800,6 @@ "issue:46365" ], "soft_pairs": [ - "issue:38617|issue:44351", - "issue:43097|issue:43873", - "issue:44704|issue:45103", - "issue:43986|issue:44117", - "issue:43824|issue:43986", "issue:44938|issue:46178", "issue:43066|issue:43906", "issue:43208|issue:43550", @@ -14725,242 +14819,240 @@ "issue:44380|issue:46132", "issue:44222|issue:45478", "issue:44483|issue:44514", - "issue:43208|issue:43526", "issue:36683|issue:43582", + "issue:43208|issue:43526", "issue:44936|issue:46365", "issue:43012|issue:44683", "issue:42490|issue:43012", "issue:42757|issue:43824", "issue:38617|issue:43502", "issue:45600|issue:46298", - "issue:45698|issue:46172" + "issue:45698|issue:46172", + "issue:43618|issue:44380", + "issue:44038|issue:44230", + "issue:43618|issue:43761", + "issue:44483|issue:45290" ], - "prepared_review_unit_hash": "aad3c03151cb4a1d7bfb5977fb7eac7e280b3f5569c74cbe9dd63f8b18996a63", + "prepared_review_unit_hash": "bb0a4cd3f1508d60071b02ea82125dea177c94e2066fa8acc0b4256b2c09ea7c", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37372, - "estimated_input_tokens": 9343, - "estimated_eval_tokens": 18942 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36431, + "estimated_input_tokens": 9108, + "estimated_eval_tokens": 18472 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The soft pairs are mostly unrelated and should not be merged; they span different import failures, model-specific bugs, save/load issues, and warnings. I\u2019d treat issue 43873 as the closest canonical issue because it is open, broadly scoped, and has the strongest discussion/inbound-reference signal among the set.", - "confidence": 0.36, - "canonical_issue_reason": "issue:43873 is the most broadly framed and actionable item here, and it has the highest apparent activity/inbound references, making it the best representative issue.", + "summary": "These items are a heterogeneous set of unrelated bug reports that only share broad Transformers/HF vocabulary. None of the soft-edge pairs look like the same concrete underlying issue, so all should be rejected as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the best overall issue candidate because it is still open, user-facing, and general enough to anchor a cluster more naturally than the highly specific bug reports.", + "best_issue_reason": "issue:43012 is the cleanest standalone bug report in the set: concrete symptoms, likely reproducible, and a focused code-path problem; however, it does not meaningfully subsume the other items.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:38617", - "right": "issue:44351", - "accept": false, - "reason": "Both are import errors, but they target different missing symbols and likely different dependency/version breakages." - }, - { - "left": "issue:43097", - "right": "issue:43873", - "accept": false, - "reason": "One is a v5 API removal complaint; the other is an offloading/quantization behavior bug. Different code paths." - }, - { - "left": "issue:44704", - "right": "issue:45103", - "accept": false, - "reason": "AutoProcessor kwargs forwarding and an auto-docstring annotations crash are unrelated problems." - }, - { - "left": "issue:43986", - "right": "issue:44117", - "accept": false, - "reason": "Video-model processor failure without torchvision and tokenizer mapping None-handling are different failure modes." - }, - { - "left": "issue:43824", - "right": "issue:43986", - "accept": false, - "reason": "Qwen2_5_VL import failure and torchvision-missing processor crash are distinct issues." - }, { "left": "issue:44938", "right": "issue:46178", "accept": false, - "reason": "Python 3.14 import/load failure is unrelated to a backend mapping error for tensorflow_text." + "reason": "Unrelated problems: Python 3.14 import failure vs an invalid backend name error in a backend mapping." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "The reproduction issue does not clearly match the tokenizer-decoder-type bug." + "reason": "Different issues: tokenizer decoder type regression vs an isolated reproduction of another external bug." }, { "left": "issue:43208", "right": "issue:43550", "accept": false, - "reason": "xLSTM training blockers and Bamba torch.compile/SDPA failure are different model-specific bugs." + "reason": "Separate model-specific failures in different code paths; not the same training/compile bug." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Missing embed token weights on load and max shard size behavior on save_pretrained are different code paths." + "reason": "Missing weight loading vs save_pretrained shard-size behavior; no shared underlying defect." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "A precision warning during compile is not the same as a model failure with torch.compile and SDPA." + "reason": "A bfloat16 compile warning is not the same as a model compile failure under SDPA." }, { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "Pipeline task messaging and saving Pipeline objects are unrelated." + "reason": "Pipeline task support messaging vs inability to save pipeline objects are different bugs." }, { "left": "issue:43012", "right": "issue:44368", "accept": false, - "reason": "Both mention warnings, but they concern different models and different causes." + "reason": "Unrelated warning reports: PyTorch precision warning vs tie_word_embeddings config warning." }, { "left": "issue:45712", "right": "issue:46298", "accept": false, - "reason": "Dummy PT classes leaking into dir() and a noisy logging message are not the same bug." + "reason": "Left is dummy-class cleanup/CI leakage; right is a noisy logging message. Not the same bug." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Dequantized model saving and Pipeline save_pretrained failures affect different objects and paths." + "reason": "Saving dequantized models and saving pipeline objects involve different save paths and failures." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "A save failure and an infinite-loop inference bug are unrelated." + "reason": "Different model families and symptoms; dequantized save failure is not an infinite-image-loop bug." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "Unsupported Flash Attention 2.0 for NemotronH and GPT2 attention scaling being ignored are different issues." + "reason": "FlashAttention support gating for Nemotron vs GPT2 attention scaling ignored under SDPA/FlashAttention." }, { "left": "issue:43097", "right": "issue:46298", "accept": false, - "reason": "API-removal behavior and noisy logging are unrelated." + "reason": "A removed API warning/compat issue is not the same as a noisy logging message." }, { "left": "issue:44162", "right": "issue:46172", "accept": false, - "reason": "ESM2 being broken and a DeepSeek tensor operation bug are unrelated." + "reason": "ESM2 breakage and a DeepSeek tensor operation failure are distinct code paths and models." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Flash Attention support for a model and a Python-version compatibility problem are different." + "reason": "Model attention backend support vs Python-version compatibility are unrelated." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Attention-backend config loss and a degenerate classification loss bug are unrelated." + "reason": "Attention backend scaling bug is unrelated to degenerate classification loss with num_labels=1." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "FSDP2/PEFT optimizer state corruption and GPT2 attention scaling are different code paths." + "reason": "Distributed optimizer-state corruption is not the same as attention scaling being ignored." }, { "left": "issue:44380", "right": "issue:46132", "accept": false, - "reason": "Attention scaling ignored and AttentionInterface.register changing behavior are not the same underlying bug." + "reason": "Attention scaling bug vs register() changing callback behavior are different issues." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen3.5 MoE, but one is save_pretrained FP8 behavior and the other is a from_pretrained error." + "reason": "FP8 save_pretrained and from_pretrained errors in related models, but not the same concrete failure." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Chat completions API rejection and batched apply_chat_template padding handling are unrelated." + "reason": "Chat completions API acceptance failure is different from apply_chat_template crashing on batched input." }, { - "left": "issue:43208", - "right": "issue:43526", + "left": "issue:36683", + "right": "issue:43582", "accept": false, - "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels behavior are different subsystems." + "reason": "Gemma3Config missing vocab_size and an Apple Silicon allocator TypeError are unrelated." }, { - "left": "issue:36683", - "right": "issue:43582", + "left": "issue:43208", + "right": "issue:43526", "accept": false, - "reason": "A Gemma config AttributeError and an Apple Silicon allocator TypeError are unrelated." + "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels behavior are different subsystems." }, { "left": "issue:44936", "right": "issue:46365", "accept": false, - "reason": "trainer.evaluate() failing after train() and import failure caused by datasets are different issues." + "reason": "trainer.evaluate() failure after train is unrelated to importing Trainer conflicting with datasets." }, { "left": "issue:43012", "right": "issue:44683", "accept": false, - "reason": "A warning on bfloat16 compile is not the same as flex_attention failing on torch >= 2.9." + "reason": "A compile-time precision warning is not the same as compiled flex_attention failing on torch>=2.9." }, { "left": "issue:42490", "right": "issue:43012", "accept": false, - "reason": "Trailing-slash local model loading warnings and bfloat16 compile precision warnings are unrelated." + "reason": "Path-trailing-slash import warning vs bfloat16 compile warning are different concerns." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Different missing imports from different packages/symbols; not the same root cause." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "ImportError for layer_type_validation is unrelated to local_files_only still making API requests." + }, + { + "left": "issue:45600", + "right": "issue:46298", + "accept": false, + "reason": "Broken auto_mappings references are a CI/import bug, not a logging-noise issue." + }, + { + "left": "issue:45698", + "right": "issue:46172", + "accept": false, + "reason": "Wrong custom module loading after save_pretrained is unrelated to DeepSeek tensor conversion failure." }, { - "left": "issue:42757", - "right": "issue:43824", + "left": "issue:43618", + "right": "issue:44380", "accept": false, - "reason": "Missing is_offline_mode and missing Qwen2_5_VLForConditionalGeneration are different import errors." + "reason": "Missing CLIPOutput attentions assignment and GPT2 attention scaling under SDPA are different regressions." }, { - "left": "issue:38617", - "right": "issue:43502", + "left": "issue:44038", + "right": "issue:44230", "accept": false, - "reason": "Importing a removed configuration helper and unwanted API requests despite local_files_only=True are different problems." + "reason": "Qwen3-VL-Moe v5 incompatibility and fp8 inference support are different bugs/features." }, { - "left": "issue:45600", - "right": "issue:46298", + "left": "issue:43618", + "right": "issue:43761", "accept": false, - "reason": "Broken auto_mappings references and noisy logging message do not match the same underlying bug." + "reason": "One is attentions not assigned; the other is hidden_states not returned. Similar area, different defect." }, { - "left": "issue:45698", - "right": "issue:46172", + "left": "issue:44483", + "right": "issue:45290", "accept": false, - "reason": "Wrong custom module loading after save_pretrained and a DeepSeek tensor operation bug are unrelated." + "reason": "Chat/completions request acceptance and tool-call chat template crashes are distinct paths." } ] }, @@ -14986,6 +15078,7 @@ "nodes": [ "issue:36010", "issue:36683", + "issue:41669", "issue:42175", "issue:42757", "issue:42994", @@ -14999,32 +15092,30 @@ "issue:43611", "issue:43618", "issue:43756", - "issue:43761", "issue:43867", "issue:43901", "issue:43931", "issue:43937", "issue:43976", "issue:43992", - "issue:44038", "issue:44075", "issue:44077", - "issue:44230", "issue:44242", + "issue:44246", "issue:44355", "issue:44380", - "issue:44483", "issue:44485", "issue:44492", "issue:44625", "issue:44655", "issue:44855", "issue:44938", + "issue:45071", "issue:45216", "issue:45276", "issue:45278", - "issue:45290", "issue:45310", + "issue:45375", "issue:45584", "issue:45593", "issue:46077", @@ -15034,273 +15125,273 @@ "issue:46365" ], "soft_pairs": [ - "issue:43618|issue:44380", - "issue:44038|issue:44230", - "issue:43618|issue:43761", - "issue:44483|issue:45290", "issue:43756|issue:44485", "issue:43299|issue:43867", - "issue:36683|issue:43475", "issue:43066|issue:46209", + "issue:36683|issue:43475", "issue:42994|issue:44655", "issue:45276|issue:46077", "issue:43901|issue:44655", "issue:44077|issue:44380", - "issue:36683|issue:44077", "issue:36010|issue:42757", - "issue:43611|issue:44655", "issue:43352|issue:43937", + "issue:43611|issue:44655", "issue:44625|issue:46077", + "issue:36683|issue:44077", "issue:43867|issue:45216", "issue:46077|issue:46082", "issue:45584|issue:46365", - "issue:42175|issue:46365", "issue:43976|issue:44938", + "issue:42175|issue:46365", "issue:43352|issue:45593", "issue:36683|issue:43525", "issue:43618|issue:45216", "issue:44077|issue:46365", - "issue:36683|issue:44242", "issue:43352|issue:43931", + "issue:36683|issue:44242", "issue:44075|issue:46365", - "issue:43064|issue:44492", "issue:43992|issue:45310", + "issue:43064|issue:44492", "issue:43010|issue:45278", "issue:44355|issue:44855", - "issue:43867|issue:46202" + "issue:43867|issue:46202", + "issue:43867|issue:45375", + "issue:41669|issue:44246", + "issue:44938|issue:45071", + "issue:36683|issue:43064" ], - "prepared_review_unit_hash": "43e13ca2656a29efd0070ab0900fd5dbbec773b44cbd3240ca66cc543e297280", + "prepared_review_unit_hash": "4e4297c34b1df79ac716a1182eca2c0f05e6c5c93709bff7671ecfdfe48898ed", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 34, - "serialized_chars": 37427, - "estimated_input_tokens": 9357, - "estimated_eval_tokens": 18970 + "serialized_chars": 36819, + "estimated_input_tokens": 9205, + "estimated_eval_tokens": 18666 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly a grab-bag of unrelated Transformers issues; several are broad v5 regressions, but the soft pairs do not look like the same underlying bug or change.", - "confidence": 0.52, - "canonical_issue_reason": "issue:45278 is the broadest, most umbrella-like report in the set, covering multiple import breakages after upgrading to Transformers 5.5.0, so it is the best single issue to represent the cluster if one must be chosen.", + "summary": "This cluster is mostly heterogeneous Transformers issues with only superficial similarity in a few titles. I reject all soft edges; none look like the same underlying bug/change. The broadest representative issue is the 4.57.0\u21925.5.0 import-error regression.", + "confidence": 0.28, + "canonical_issue_reason": "issue:45278 is the broadest umbrella here: it describes widespread import failures after upgrading to 5.5.0, which is the closest match to the cluster\u2019s general theme of breakage/regression reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the clearest global representative because it is a wide regression report rather than a narrow model-specific symptom, and it can subsume the general import-error theme better than the other items.", + "best_issue_reason": "issue:45278 is the best single issue to represent this set because it is the most general regression report and could plausibly subsume several import/load breakage reports, even though the cluster is not truly duplicate-like.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43618", - "right": "issue:44380", - "accept": false, - "reason": "Different models and different failures: CLIP attentions missing vs GPT-2 attention scaling ignored." - }, - { - "left": "issue:44038", - "right": "issue:44230", - "accept": false, - "reason": "One is a general Qwen3-VL-Moe regression, the other is fp8 inference support; not the same bug." - }, - { - "left": "issue:43618", - "right": "issue:43761", - "accept": false, - "reason": "Both are CLIP regressions, but they affect different outputs and appear to be separate code paths." - }, - { - "left": "issue:44483", - "right": "issue:45290", - "accept": false, - "reason": "API request acceptance and chat template tokenization crashes are different issues." - }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Different models and different RoPE-related concerns; not one concrete bug." + "reason": "Different models and different bugs: Smollm3 RoPE layer count vs GLM-5 RoPE implementation discussion." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Both are loading failures, but the titles point to distinct failure modes and models." + "reason": "Both are loading-related, but one is Qwen3VL MoE loading and the other is a generic state_dict sorting failure; different failure modes." }, { - "left": "issue:36683", - "right": "issue:43475", + "left": "issue:43066", + "right": "issue:46209", "accept": false, - "reason": "Unrelated attribute errors in different model families." + "reason": "Wrong tokenizer decoder type and FP8 inference on B200 are unrelated code paths." }, { - "left": "issue:43066", - "right": "issue:46209", + "left": "issue:36683", + "right": "issue:43475", "accept": false, - "reason": "Tokenizer decoder type issue and FP8 inference failure are unrelated." + "reason": "Gemma3 config missing vocab_size and SAM 3 Video missing fpn_position_embeddings are different model/config bugs." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Model quantization save failure and pipeline save_pretrained failure are different save paths." + "reason": "Quantized model saving and saving Pipeline objects are both save-related, but they are not the same underlying problem." }, { "left": "issue:45276", "right": "issue:46077", "accept": false, - "reason": "Both involve Gemma4, but one is embedding resize behavior and the other is attention mask construction." + "reason": "Same model family, but resize_token_embeddings and bidirectional attention masks are separate subsystems." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Docs mismatch for TextClassificationPipeline is unrelated to pipeline serialization." + "reason": "Docs mentioning return_all_scores is a documentation mismatch; save_pretrained on Pipeline objects is a runtime serialization bug." }, { "left": "issue:44077", "right": "issue:44380", "accept": false, - "reason": "Different model behavior: patchtsmixer init rules vs GPT-2 attention scaling." + "reason": "PatchTSMixer post_init handling and GPT2 attention scaling under SDPA/FlashAttention are unrelated." }, { - "left": "issue:36683", - "right": "issue:44077", + "left": "issue:36010", + "right": "issue:42757", "accept": false, - "reason": "Different models and unrelated attribute/configuration bugs." + "reason": "Both are import errors, but they involve different missing symbols from different dependencies." }, { - "left": "issue:36010", - "right": "issue:42757", + "left": "issue:43352", + "right": "issue:43937", "accept": false, - "reason": "Both are import errors, but for different symbols from different packages." + "reason": "Flash Attention 2 support error for Nemotron and GLM-5 GenerationConfig validation are different problems." }, { "left": "issue:43611", "right": "issue:44655", "accept": false, - "reason": "Model loading with base_model_prefix and pipeline saving are not the same bug." + "reason": "Base_model_prefix loading regression is not the same as Pipeline save_pretrained serialization." }, { - "left": "issue:43352", - "right": "issue:43937", + "left": "issue:44625", + "right": "issue:46077", "accept": false, - "reason": "FlashAttention support for Nemotron is unrelated to GLM-5 generation config validation." + "reason": "Qwen3.5 num_labels propagation and Gemma4 bidirectional attention are unrelated issues." }, { - "left": "issue:44625", - "right": "issue:46077", + "left": "issue:36683", + "right": "issue:44077", "accept": false, - "reason": "Different Gemma/Qwen config propagation vs attention-mask behavior." + "reason": "Different models and unrelated attributes: Gemma3 vocab_size vs PatchTSMixer post_init." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "State-dict ordering load error and Qwen3.5 save_pretrained regression are different serialization problems." + "reason": "Both mention loading/saving, but one is generic state_dict sorting and the other is a Qwen3.5 checkpoint regression." }, { "left": "issue:46077", "right": "issue:46082", "accept": false, - "reason": "Different Llama/Gemma config validation and attention behavior issues." + "reason": "Both are Gemma4/Llama configs, but the actual bugs are different config semantics." }, { "left": "issue:45584", "right": "issue:46365", "accept": false, - "reason": "Whisper generation failure and Trainer import conflict are unrelated." + "reason": "Whisper empty-transcription generation failure is unrelated to importing Trainer being affected by datasets." }, { - "left": "issue:42175", - "right": "issue:46365", + "left": "issue:43976", + "right": "issue:44938", "accept": false, - "reason": "Backend installation packaging issue and Trainer import conflict are different problems." + "reason": "Both are Python-version compatibility complaints, but they are about different versions and likely different causes." }, { - "left": "issue:43976", - "right": "issue:44938", + "left": "issue:42175", + "right": "issue:46365", "accept": false, - "reason": "Python version compatibility issues, but for different versions and failure causes." + "reason": "TensorFlow backend packaging and Trainer import interference from datasets are not the same issue." }, { "left": "issue:43352", "right": "issue:45593", "accept": false, - "reason": "FlashAttention support and missing auxiliary losses are unrelated model behaviors." + "reason": "Unsupported Flash Attention 2 and D-FINE auxiliary loss omission are unrelated model behaviors." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Different config attributes on different models; not the same underlying bug." + "reason": "Different missing config attributes in different models." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "Different CLIP output regression versus Qwen3.5 checkpoint saving regression." + "reason": "CLIPOutput attentions regression and Qwen3.5 checkpoint regression are unrelated." }, { "left": "issue:44077", "right": "issue:46365", "accept": false, - "reason": "Patchtsmixer post_init policy and Trainer import conflict are unrelated." + "reason": "PatchTSMixer post_init and Trainer import shadowing by datasets are unrelated." }, { - "left": "issue:36683", - "right": "issue:44242", + "left": "issue:43352", + "right": "issue:43931", "accept": false, - "reason": "Gemma3 vocab_size and MoE load-balancing loss are distinct issues." + "reason": "Nemotron Flash Attention support and Qwen3-VL weight-shape mismatch are different loading/runtime failures." }, { - "left": "issue:43352", - "right": "issue:43931", + "left": "issue:36683", + "right": "issue:44242", "accept": false, - "reason": "Nemotron FlashAttention support and Qwen3-VL weight-shape mismatch are different failures." + "reason": "Gemma3 config attribute bug and load-balancing loss omission are unrelated." }, { "left": "issue:44075", "right": "issue:46365", "accept": false, - "reason": "Optimizer argument handling and Trainer import conflict are not the same bug." + "reason": "SGD args not used is an optimizer bug, not an import/dependency issue." }, { - "left": "issue:43064", - "right": "issue:44492", + "left": "issue:43992", + "right": "issue:45310", "accept": false, - "reason": "Distributed optimizer-state corruption and a cache-strategy typo are unrelated." + "reason": "Both involve Qwen3.5 loading/saving, but one misses an encoder weight and the other is a from_pretrained error after 5.4.0; not the same bug." }, { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:43064", + "right": "issue:44492", "accept": false, - "reason": "Both concern Qwen3.5 loading, but one misses a weight and the other has a from_pretrained error; not enough to merge." + "reason": "FSDP2/PEFT optimizer state corruption and a cache-strategy typo are unrelated." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "No-no_grad decorator for cache updates and broad import failures are unrelated." + "reason": "A cache update no_grad decorator bug is unrelated to widespread import failures after upgrading Transformers." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Generic compiled-file errors and a specific Python 3.13 TorchScript parsing failure are different bugs." + "reason": "Compiled Python file errors and the Python 3.13 torch.jit decorator parsing bug overlap only loosely; not the same underlying issue." }, { "left": "issue:43867", "right": "issue:46202", "accept": false, - "reason": "Load-model error with sorted state_dict and Gemma pipeline roundtrip failure are unrelated." + "reason": "State_dict sorting load failure and Gemma pipeline roundtrip on \u2581 are different loading/serialization bugs." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Generic state_dict ordering failure and Qwen3_5MoeVisionConfig field dropping are different problems." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Import-time slowness from removing import * and intermittent import slowness are too broad to treat as the same concrete bug." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different compatibility issues." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3 config attribute error and Trainer FSDP2 optimizer-state corruption are unrelated." } ] }, @@ -15326,29 +15417,30 @@ "nodes": [ "issue:28282", "issue:30333", + "issue:30990", + "issue:31795", "issue:33453", "issue:36296", - "issue:36683", - "issue:41669", "issue:42548", + "issue:42831", "issue:42890", "issue:43010", - "issue:43064", "issue:43065", "issue:43232", + "issue:43299", "issue:43502", "issue:43575", "issue:43576", "issue:43618", "issue:43723", "issue:43867", + "issue:43874", "issue:43976", "issue:43992", "issue:44062", "issue:44077", "issue:44112", "issue:44222", - "issue:44246", "issue:44351", "issue:44355", "issue:44360", @@ -15358,9 +15450,8 @@ "issue:44625", "issue:44749", "issue:44857", - "issue:44938", + "issue:44961", "issue:44964", - "issue:45071", "issue:45125", "issue:45357", "issue:45375", @@ -15374,10 +15465,6 @@ "issue:46365" ], "soft_pairs": [ - "issue:43867|issue:45375", - "issue:41669|issue:44246", - "issue:36683|issue:43064", - "issue:44938|issue:45071", "issue:42890|issue:45561", "issue:43010|issue:45850", "issue:44625|issue:46365", @@ -15404,239 +15491,254 @@ "issue:45125|issue:45478", "issue:44355|issue:44749", "issue:30333|issue:33453", + "issue:43010|issue:43232", "issue:45125|issue:45375", - "issue:43010|issue:43232" + "issue:30990|issue:31795", + "issue:43502|issue:44222", + "issue:43299|issue:44556", + "issue:43502|issue:44961", + "issue:30333|issue:36296", + "issue:42831|issue:43874" ], - "prepared_review_unit_hash": "52cf60d236d51c33887e033f30e661bda370fc6a89c28aedc544c1e26e9bcd61", + "prepared_review_unit_hash": "ebdf204fc2aadd08b797ff98e3a46dd29b75f1a16a8fde653d604f578ec2c1a2", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36839, - "estimated_input_tokens": 9210, - "estimated_eval_tokens": 18676 + "soft_pair_count": 34, + "serialized_chars": 37400, + "estimated_input_tokens": 9350, + "estimated_eval_tokens": 18956 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a loose Transformers v5 regression cluster centered on import/export breakages, with a separate import-time slowdown pair and many unrelated model/config/test issues. The Qwen3.5 MoE pair should not be merged; it looks like two different bugs in the same family, not one fixable change.", - "confidence": 0.34, - "canonical_issue_reason": "Issue 42548 is the clearest representative of the cluster\u2019s main theme: a concrete top-level transformers import/export breakage, and it has a close duplicate-like sibling in 44351.", + "summary": "This cluster is highly heterogeneous: import errors, loading regressions, tensor-parallel/FP8 issues, CI flakes, and docs problems. None of the soft-similarity pairs look like true duplicates or the same concrete fix path.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue cleanly represents the set; the items span unrelated bugs, regressions, docs, and CI/test flakes.", "canonical_pr_reason": null, - "best_issue_reason": "42548 is the best anchor because it is specific, user-facing, and matches the strongest duplicate signal in the set: missing public names from the transformers package.", + "best_issue_reason": "issue:43502 is the most concrete and self-contained bug report in the set, but it is not a true duplicate hub for the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43867", - "right": "issue:45375", - "accept": false, - "reason": "Different bugs: state_dict/model loading failure vs missing Qwen3.5 vision config field." - }, - { - "left": "issue:41669", - "right": "issue:44246", - "accept": true, - "reason": "Both describe import-time slowdown in transformers, with wildcard imports and large import surface as the likely shared cause." - }, - { - "left": "issue:36683", - "right": "issue:43064", - "accept": false, - "reason": "Unrelated failures: Gemma3 config attribute error vs FSDP2/PEFT optimizer-state corruption." - }, - { - "left": "issue:44938", - "right": "issue:45071", - "accept": false, - "reason": "Python-version load failure and config type-checking breakage are different failure modes." - }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "Different CI/test problems: seed flakiness vs xdist captured_info.txt race." + "reason": "Both are test failures, but one is a missing set_seed in SAM HQ integration tests and the other is an xdist race in captured_info.txt; different bugs." }, { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "Specific cache/generation API behavior vs a broad complaint that v5 breaks many models." + "reason": "Specific no_grad decoration request vs a broad complaint that 5.0 breaks many models; not the same code-path." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "Different areas entirely: Qwen3.5 num_labels propagation vs Trainer import depending on datasets." + "reason": "Qwen3.5 num_labels propagation bug vs a Trainer import conflict tied to datasets; unrelated failures." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "Both are model loading/saving issues, but the concrete failure paths are different." + "reason": "Model load failure on sorted state_dict vs incorrect visual encoder keys in save_pretrained; opposite directions and different causes." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Both involve Qwen-ish loading problems, but one is missing weights in from_pretrained and the other is a generic Qwen3.5 MoE load error." + "reason": "UMT5Encoder missing embed_tokens.weight vs Qwen3.5 MoE from_pretrained error; different model families and code paths." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "Cache/generation behavior change vs Python compatibility issue are unrelated." + "reason": "No_grad/cache update bug vs Python-version compatibility regression; unrelated." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Different tensor-parallel bugs: a training bug vs OOM while loading a specific large model." + "reason": "Generic tensor-parallel training bug vs OOM while loading a specific large Qwen2 model; not the same defect." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "Different model-specific API changes, not the same bug." + "reason": "Missing CLIPOutput attentions vs patchtsmixer post_init validation; different subsystems." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "PyTorch-missing ImportError vs tokenizer loading regression are unrelated." + "reason": "PyTorch missing import error vs tokenizer loading regression; unrelated import/load issues." }, { "left": "issue:43618", "right": "issue:46285", "accept": false, - "reason": "Both CLIP-related, but one is missing attentions assignment while the other is a structural class inconsistency." + "reason": "Both mention CLIP, but one is an output field regression and the other is a model structure inconsistency; different fixes." }, { "left": "issue:43618", "right": "issue:44857", "accept": false, - "reason": "Unrelated CLIP output issue vs an AMP/CUDA crash in a different model loss." + "reason": "CLIP output regression vs LwDetrImageLoss AMP crash; unrelated." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "DSA/ReLU discussion is unrelated to DeepseekV3MoE implementation divergence." + "reason": "DSA indexer ReLU discussion vs DeepseekV3MoE implementation divergence; different bugs." }, { "left": "issue:43502", "right": "issue:44062", "accept": false, - "reason": "Local-files-only network leakage vs AddedToken argument duplication are different code paths." + "reason": "local_files_only still making API calls vs tokenizers.AddedToken keyword conflict; unrelated." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Test seeding flakiness vs Phi-4 multimodal load failure." + "reason": "SAM HQ test seeding issue vs Phi-4 multimodal load failure; different models and failure modes." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Different CI/test failures with different causes." + "reason": "Both are CI/test issues, but one is seeding in SAM HQ and the other is a stale device override in GraniteSpeech; not the same bug." }, { "left": "issue:42890", "right": "issue:45600", "accept": false, - "reason": "Different CI/test failures; one is about seeding, the other about stale auto_mappings references." + "reason": "SAM HQ flakiness vs auto_mappings referencing removed configs; unrelated." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Tokenizer loading and checkpoint reload are both from_pretrained-related, but not the same concrete bug." + "reason": "Tokenizer loading regression vs checkpoint reload incompatibility after upgrade; both are loading-related but not the same concrete path." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Int4 memory regression vs checkpoint reload regression are different issues." + "reason": "int4 quantization memory growth/OOM vs generic reload failure after version upgrade; different root causes." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "FP8 save_pretrained bug vs missing tensor-parallel plan are different code paths." + "reason": "FP8 save_pretrained issue for MoE vs missing _tp_plan for tensor parallelism; distinct model/export paths." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "Cache-update decoration issue vs dummy Conv2d in Sam3PixelDecoder." + "reason": "Cache/layer no_grad decoration vs dummy Conv2d in Sam3PixelDecoder; unrelated." }, { "left": "issue:43576", "right": "issue:46132", "accept": false, - "reason": "Broken env command is unrelated to AttentionInterface.register behavior change." + "reason": "Broken transformers env command vs AttentionInterface.register behavior change; different APIs." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "Local-files-only network requests vs GLM-5 RoPE implementation are unrelated." + "reason": "local_files_only network leak vs GLM-5 RoPE implementation discussion; not the same issue." }, { "left": "issue:42548", "right": "issue:44351", - "accept": true, - "reason": "Both are top-level 'cannot import name ... from transformers' regressions and look like the same export-surface breakage." + "accept": false, + "reason": "Different top-level import errors for different symbols; likely separate export/regression issues." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Same model family, but one is a missing _tp_plan tensor-parallel issue and the other is a generic from_pretrained error; too different to treat as one bug." + "reason": "Missing _tp_plan vs Qwen3.5 MoE from_pretrained error; related model line, but not the same concrete bug." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled Python file errors vs a performance regression after upgrade." + "reason": "Compiled Python file errors vs a performance regression after upgrade; unrelated." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLFlow reporting failure is unrelated to tokenizer-loading regression." + "reason": "MLFlow reporting on failed training jobs vs tokenizer loading regression; different functionality." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "no_grad on cache updates vs generation kwargs update after sync_gpus; different generation internals." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Both Qwen3.5-related, but one is tensor-parallel planning and the other is missing vision-config metadata." + "reason": "Tensor-parallel plan missing vs vision config field dropped by strict config handling; different bug classes." }, { - "left": "issue:43010", - "right": "issue:43232", + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Loading hang in Sentence Transformers vs documentation confusion in model.forward; not duplicates." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "local_files_only network issue vs FP8 MoE save_pretrained bug; unrelated." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3VL MoE loading regression vs generic checkpoint reload failure after upgrading from v4.57; too broad to be the same fix." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "API calls despite local_files_only vs unrelated 'racoon' issue; no overlap." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLFlow failure reporting vs tensor-parallel training bug; unrelated." + }, + { + "left": "issue:42831", + "right": "issue:43874", "accept": false, - "reason": "Both generation-related, but they concern different methods and failure modes." + "reason": "FineGrainedFP8 accuracy issue vs GLM46V image processor missing method; different components and failure modes." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The cluster summary is mostly grounded, but one soft-edge acceptance is too speculative: issue:45125 vs issue:45478 only clearly share the Qwen3.5 MoE family, while one is a missing `_tp_plan` tensor-parallel problem and the other is a generic `from_pretrained` error. That is not conservative enough to count as the same bugfix PR. The import/export pair at issue:42548 vs issue:44351 is plausible, but the Qwen3.5 pair makes the overall verdict too strong." - }, - "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -15653,25 +15755,23 @@ "model": "gpt-5.4-mini?service_tier=flex", "cluster_id": "cluster-10000-445", "nodes": [ - "issue:30333", - "issue:30990", "issue:31795", "issue:33453", "issue:34689", "issue:36296", "issue:38617", "issue:41628", + "issue:41669", + "issue:42371", "issue:42548", "issue:42757", - "issue:42831", - "issue:43299", - "issue:43502", + "issue:43525", "issue:43526", + "issue:43572", "issue:43576", "issue:43673", "issue:43824", "issue:43827", - "issue:43874", "issue:43901", "issue:44038", "issue:44222", @@ -15691,11 +15791,13 @@ "issue:44961", "issue:45003", "issue:45071", + "issue:45092", "issue:45125", "issue:45230", "issue:45362", "issue:45468", "issue:45542", + "issue:45753", "issue:45850", "issue:46095", "issue:46143", @@ -15703,12 +15805,6 @@ "issue:46209" ], "soft_pairs": [ - "issue:30990|issue:31795", - "issue:43502|issue:44222", - "issue:43299|issue:44556", - "issue:43502|issue:44961", - "issue:30333|issue:36296", - "issue:42831|issue:43874", "issue:44355|issue:45071", "issue:43827|issue:44961", "issue:44355|issue:44655", @@ -15720,12 +15816,12 @@ "issue:43576|issue:43824", "issue:31795|issue:38617", "issue:44961|issue:45468", - "issue:44222|issue:44230", "issue:43576|issue:44938", - "issue:44908|issue:44961", - "issue:41628|issue:42548", + "issue:44222|issue:44230", "issue:46143|issue:46144", + "issue:41628|issue:42548", "issue:44485|issue:44961", + "issue:44908|issue:44961", "issue:44315|issue:46209", "issue:36296|issue:43901", "issue:36296|issue:43526", @@ -15734,249 +15830,250 @@ "issue:44623|issue:45230", "issue:44273|issue:45542", "issue:44038|issue:45125", - "issue:44273|issue:45003", - "issue:44393|issue:45071", "issue:44230|issue:46095", - "issue:43576|issue:44556" + "issue:44393|issue:45071", + "issue:44273|issue:45003", + "issue:43576|issue:44556", + "issue:44393|issue:45753", + "issue:43525|issue:45850", + "issue:41669|issue:43673", + "issue:44556|issue:45092", + "issue:43572|issue:45850", + "issue:44492|issue:45850", + "issue:42371|issue:44492" ], - "prepared_review_unit_hash": "eecfb4ebeb0fc31a2e62a3a23c0efa71d12faed719aac2c75f65173fd8211382", + "prepared_review_unit_hash": "67ba8efbe2f62c696f93e2e54186fab1f20f201e3bfd471fcf6d72789f9a18bb", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 35, - "serialized_chars": 37324, - "estimated_input_tokens": 9331, - "estimated_eval_tokens": 18918 + "soft_pair_count": 36, + "serialized_chars": 37610, + "estimated_input_tokens": 9403, + "estimated_eval_tokens": 19062 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a mixed set of mostly unrelated Transformers regressions, import errors, and docs issues. The soft pairs share only broad surface similarity (often \"v5 broke X\" or \"loading failed\") and do not look like the same underlying bug/change.", - "confidence": 0.93, - "canonical_issue_reason": "issue:45850 is the broadest umbrella complaint about widespread v5 regressions, so it is the closest representative of this mixed cluster.", + "summary": "This cluster is a mixed bag of Transformers v5 regressions, import/export breakages, and unrelated model-specific bugs. None of the soft pairs look like safe duplicate merges, so I reject them all.", + "confidence": 0.74, + "canonical_issue_reason": "issue:44556 is the clearest concrete regression in the set: reloading a v4.57 checkpoint after upgrading to v5.2/v5.3. It is specific, actionable, and representative of the recurring v5 loading-breakage theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 is the best global representative of the common theme, but it is still too vague to serve as a true duplicate target.", + "best_issue_reason": "issue:44556 is the most actionable issue here because it names a concrete failure mode and affected versions, unlike the more generic or broad complaints.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:30990", - "right": "issue:31795", - "accept": false, - "reason": "Different problems: a loading hang vs confusing forward-argument docs." - }, - { - "left": "issue:43502", - "right": "issue:44222", - "accept": false, - "reason": "Different code paths: offline-network behavior vs FP8 save_pretrained support." - }, - { - "left": "issue:43299", - "right": "issue:44556", - "accept": false, - "reason": "Both are version-regression loading reports, but for different models and failure modes." - }, - { - "left": "issue:43502", - "right": "issue:44961", - "accept": false, - "reason": "Unrelated: network requests despite local_files_only vs a dummy/nonspecific issue." - }, - { - "left": "issue:30333", - "right": "issue:36296", - "accept": false, - "reason": "MLFlow reporting on failed jobs is unrelated to tensor-parallel training bugs." - }, - { - "left": "issue:42831", - "right": "issue:43874", - "accept": false, - "reason": "Different bugs: FP8 accuracy regression vs missing processor method causing an AttributeError." - }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled-file loading errors are unrelated to PretrainedConfig type-checking breakage." + "reason": "Different bugs: compiled Python file loading errors vs a v5.4 PretrainedConfig/type-checking regression." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Docs still mentioning pipeline() is unrelated to the second issue." + "reason": "One is a docs issue about removed `pipeline()` usage; the other is an unrelated placeholder issue." }, { "left": "issue:44355", "right": "issue:44655", "accept": false, - "reason": "Compiled Python file errors and save_pretrained pipeline issues are separate failures." + "reason": "Different failure modes: compiled `.py` errors vs `save_pretrained` on Pipeline objects." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel application and DSA indexer ReLU are unrelated subsystems." + "reason": "Liger Kernel application via `model_init` is unrelated to the DSA indexer/ReLU discussion." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Both involve loading, but one is a tokenizer regression and the other is a specific model-loading breakage." + "reason": "Both concern loading, but one is tokenizer loading and the other is Llama 3.2 Vision model loading; not the same bug." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Importing is_offline_mode and the transformers env command are different failures." + "reason": "`huggingface_hub` import failure is distinct from the broken `transformers env` command." }, { "left": "issue:33453", "right": "issue:36296", "accept": false, - "reason": "Tokenizer loading regression is unrelated to tensor-parallel training." + "reason": "Tokenizer loading regression and tensor-parallel training bug are different code paths and symptoms." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "Broken env CLI and PretrainedConfig type checking are unrelated." + "reason": "CLI/environment command breakage is not the same as a `PretrainedConfig` typing regression." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "CLI breakage is unrelated to the missing Qwen2.5-VL import." + "reason": "Broken `transformers env` and missing `Qwen2_5_VLForConditionalGeneration` import are separate import/export issues." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Documentation confusion and an ImportError are not the same bug." + "reason": "Documentation confusion for `forward` arguments is unrelated to the `layer_type_validation` import error." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "No meaningful relation: a nonspecific issue vs Gemma-4 audio positional encoding." + "reason": "No shared bug: a placeholder issue versus a Gemma-4 audio positional encoding bug." }, { - "left": "issue:44222", - "right": "issue:44230", + "left": "issue:43576", + "right": "issue:44938", "accept": false, - "reason": "Same FP8 theme, but one is save/export behavior and the other is inference support." + "reason": "A broken CLI command is not the same as Python 3.14 load/import compatibility." }, { - "left": "issue:43576", - "right": "issue:44938", + "left": "issue:44222", + "right": "issue:44230", "accept": false, - "reason": "Broken env CLI is unrelated to Python 3.14 import/load failures." + "reason": "Both mention fp8/moe, but one is about `save_pretrained` and the other is about inference support; too different to merge." }, { - "left": "issue:44908", - "right": "issue:44961", + "left": "issue:46143", + "right": "issue:46144", "accept": false, - "reason": "Scheduler kwargs handling is unrelated to the other issue." + "reason": "Closely related RoFormer topics, but one is missing `**kwargs` passthrough while the other is attention-interface usage; not clearly the same concrete bug." }, { "left": "issue:41628", "right": "issue:42548", "accept": false, - "reason": "Both are import errors, but for different exported symbols and likely different fixes." + "reason": "Missing `AutoImageProcessor` and missing `PreTrainedModel` are separate import surface regressions." }, { - "left": "issue:46143", - "right": "issue:46144", + "left": "issue:44485", + "right": "issue:44961", "accept": false, - "reason": "Both are about RoFormer, but one is kwargs forwarding and the other is attention implementation design." + "reason": "GLM-5 RoPE implementation bug is unrelated to the placeholder issue." }, { - "left": "issue:44485", + "left": "issue:44908", "right": "issue:44961", "accept": false, - "reason": "RoPE implementation discussion is unrelated to the nonspecific issue." + "reason": "Scheduler kwarg handling has nothing to do with the placeholder issue." }, { "left": "issue:44315", "right": "issue:46209", "accept": false, - "reason": "Liger Kernel application and FP8 inference on B200 are unrelated bugs." + "reason": "`model_init`/Liger Kernel behavior is unrelated to fp8 inference failure on B200." }, { "left": "issue:36296", "right": "issue:43901", "accept": false, - "reason": "Tensor-parallel training and pipeline docs mentioning return_all_scores are unrelated." + "reason": "Tensor-parallel training bug and pipeline docs mentioning `return_all_scores` are unrelated." }, { "left": "issue:36296", "right": "issue:43526", "accept": false, - "reason": "Tensor-parallel training bug is unrelated to BeitImageProcessorFast label handling." + "reason": "Tensor-parallel training and BeitImageProcessorFast label reduction are different model/component bugs." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "Chunked-prefill cache missing and a cache-strategy typo are not the same issue." + "reason": "A missing generation cache in chunked prefill is not the same as a cache-strategy typo." }, { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "The former is a specific Qwen3.5 chat crash; the latter is only a broad meta-report about many breakages." + "reason": "A specific chat crash in Qwen3.5-35B is not the same as the broad claim that v5 breaks too many models." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "Missing processor save files and a generic bug report are not the same underlying problem." + "reason": "`processor.save_pretrained(...)` missing files is concrete; `Bug report` is too vague and not the same issue." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "Lazy loading failure and the tensorboard-without-TF backend error are unrelated." + "reason": "Lazy-loading behavior and a TensorBoard-without-TF backend error are unrelated." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Both mention Qwen3-Moe/v5, but they describe different missing functionality and should not be merged." + "reason": "A broad Qwen3-VL-Moe v5 bug report is not the same as missing `_tp_plan` for tensor parallelism." }, { - "left": "issue:44273", - "right": "issue:45003", + "left": "issue:44230", + "right": "issue:46095", "accept": false, - "reason": "Lazy loading behavior and unsafe sys.modules access are different code-path issues." + "reason": "FP8 inference support for Qwen3.5/moe is unrelated to a DeepSeekV4 weight-conversion question." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Qwen3-VL bounding-box errors are unrelated to type-checking breakage." + "reason": "A multimodal bounding-box output bug is unrelated to `PretrainedConfig` type-checking breakage." }, { - "left": "issue:44230", - "right": "issue:46095", + "left": "issue:44273", + "right": "issue:45003", "accept": false, - "reason": "FP8 inference support for Qwen models is unrelated to DeepSeek weight-conversion guidance." + "reason": "Lazy loading problems are unrelated to unsafe `sys.modules` access in `modeling_utils`." }, { "left": "issue:43576", "right": "issue:44556", "accept": false, - "reason": "Broken env CLI and checkpoint reload regressions are separate failures." + "reason": "Broken `transformers env` and checkpoint reload failures after version upgrade are different bugs." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Bounding-box hallucination/errors and infinite looping on a specific image are different Qwen3-VL failures." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_id` in `Llama4Config` is a specific config bug, not the same as the generic v5 breakage complaint." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_idx` in `StableLmConfig` is a distinct config regression from the broad v5 complaint." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo in cache strategy text is not the same as a generic \"v5 breaks too many models\" report." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 behavior API settings and cache-strategy typo are unrelated issues." } ] }, @@ -16007,18 +16104,19 @@ "issue:41669", "issue:42371", "issue:42890", + "issue:43298", "issue:43352", "issue:43408", + "issue:43502", "issue:43519", - "issue:43525", - "issue:43572", "issue:43576", - "issue:43673", + "issue:43606", "issue:43704", "issue:43827", "issue:43976", "issue:44077", "issue:44246", + "issue:44263", "issue:44273", "issue:44295", "issue:44297", @@ -16028,35 +16126,27 @@ "issue:44393", "issue:44485", "issue:44492", - "issue:44556", + "issue:44623", "issue:44655", "issue:44704", "issue:44829", "issue:44861", "issue:44908", "issue:45071", - "issue:45092", "issue:45230", "issue:45278", "issue:45310", "issue:45341", "issue:45468", "issue:45478", - "issue:45753", "issue:45850", "issue:46018", + "issue:46095", "issue:46172", "issue:46218", "issue:46328" ], "soft_pairs": [ - "issue:44393|issue:45753", - "issue:43525|issue:45850", - "issue:41669|issue:43673", - "issue:44556|issue:45092", - "issue:43572|issue:45850", - "issue:44492|issue:45850", - "issue:42371|issue:44492", "issue:45278|issue:45850", "issue:41669|issue:46218", "issue:42890|issue:44908", @@ -16092,294 +16182,278 @@ "issue:31795|issue:34689", "issue:31515|issue:31795", "issue:44485|issue:45468", - "issue:43704|issue:44485" + "issue:43704|issue:44485", + "issue:43298|issue:46095", + "issue:45850|issue:46218", + "issue:43502|issue:43519", + "issue:42371|issue:43704", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:43606|issue:45468", + "issue:44263|issue:45468" ], - "prepared_review_unit_hash": "7b4c2fcb21c5538a61282ef8586b03b932efbc21f51051e0e9f594c44d312067", + "prepared_review_unit_hash": "4a35f86f71fd29ba987499e41e03be7c4d30964d0d596f9ec2ccba54c2928cbf", "component_budget": { "node_count": 445, "item_count": 445, "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 43, - "serialized_chars": 39856, - "estimated_input_tokens": 9964, - "estimated_eval_tokens": 20184 + "soft_pair_count": 44, + "serialized_chars": 40175, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These soft pairs are mostly false positives: they share broad Transformers/v5/model-loading language, but each points to a different concrete bug or model-specific regression. I would not merge any of the candidate pairs as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "issue:45850 is the broadest representative of the recurring v5 regression theme, but it is still only an umbrella report rather than an exact duplicate of the others.", + "summary": "This cluster is mostly heterogeneous: loading/import/model-setup regressions, docs issues, and unrelated model-specific bugs. I don\u2019t see any soft pair that is clearly the same underlying defect or change, so all soft edges should be rejected.", + "confidence": 0.82, + "canonical_issue_reason": "issue:43502 is the best anchor because it is a concrete, user-facing loading/fetch bug with clear scope (`local_files_only=True` still triggering network access).", "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 best captures the widest shared complaint surface across the cluster (general v5 breakage), making it the most useful canonical issue placeholder.", + "best_issue_reason": "issue:43502 is the strongest representative of the cluster\u2019s loading/import regression theme while still being specific enough to triage against.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44393", - "right": "issue:45753", - "accept": false, - "reason": "Both are Qwen vision/model issues, but one is bounding-box output hallucination and the other is an infinite-loop case on a specific image; different failure modes." - }, - { - "left": "issue:43525", - "right": "issue:45850", - "accept": false, - "reason": "Llama4Config pad_token_id AttributeError is a specific config regression, not the same bug as the broad 'many models break since 5.0' report." - }, - { - "left": "issue:41669", - "right": "issue:43673", - "accept": false, - "reason": "Import-time slowdown from wildcard imports is unrelated to the GenerationMixin cache missing during chunked prefill." - }, - { - "left": "issue:44556", - "right": "issue:45092", - "accept": false, - "reason": "Both are checkpoint-loading regressions after v5, but they involve different model families and different root causes (general reload failure vs remote-code/meta-init incompatibility)." - }, - { - "left": "issue:43572", - "right": "issue:45850", - "accept": false, - "reason": "StableLmConfig missing pad_token_idx is a model config field regression, not the same as the generic v5 breakage complaint." - }, - { - "left": "issue:44492", - "right": "issue:45850", - "accept": false, - "reason": "A typo in cache strategy docs/API is unrelated to broad model loading regressions." - }, - { - "left": "issue:42371", - "right": "issue:44492", - "accept": false, - "reason": "TF32 behavior API migration and a cache-strategy typo are different documentation/API issues." - }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "General import errors after upgrade are broader, but this is still a different failure class than the specific 'many models break' report." + "reason": "Both are broad release-breakage reports, but one is about import errors after upgrade and the other about many models breaking; not the same bug." }, { "left": "issue:41669", "right": "issue:46218", "accept": false, - "reason": "Wildcard-import startup overhead is unrelated to the missing Qwen3_5ForCausalLM import error." + "reason": "One is about removing `import *` to speed imports; the other is a missing Qwen3.5 import symbol. Different fixes and code paths." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "A flaky integration test due to missing set_seed has nothing to do with inverse_sqrt scheduler kwargs handling." + "reason": "A flaky integration test seed issue is unrelated to the inverse_sqrt scheduler ignoring kwargs." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Model type mismatch warnings for sam3_video/sam3_tracker are unrelated to Liger Kernel application via model_init." + "reason": "Wrong model type warning for sam3 tracker vs Liger kernel not applied with `model_init`; unrelated." }, { "left": "issue:44246", "right": "issue:44273", "accept": false, - "reason": "Import-time slowness and lazy loading not working are related at a high level, but not clearly the same underlying bug." + "reason": "Import-time slowness and lazy loading failure are both performance/loading-adjacent, but not the same concrete defect." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained mismatch is a concrete tokenizer serialization bug; 'Bug report' provides no basis for the same root cause." + "reason": "Tokenizer class mismatch in save_pretrained is unrelated to a generic bug report." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention 2.0 unsupported for Nemotron is unrelated to Qwen3-VL 2D box output behavior." + "reason": "Flash Attention 2.0 unsupported and Qwen3-VL bbox hallucination are different model-specific problems." }, { "left": "issue:44704", "right": "issue:44908", "accept": false, - "reason": "AutoProcessor cached_file kwargs forwarding and scheduler kwargs propagation are different code paths." + "reason": "AutoProcessor kwargs propagation and scheduler kwargs handling are separate API bugs." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "Reading position_ids after buffer registration is a specific tensor/buffer bug, not a match for an unspecified bug report." + "reason": "Reading `position_ids` after buffer registration is unrelated to the generic bug report." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "Broken 'transformers env' command is unrelated to scheduler configuration handling." + "reason": "`transformers env` being broken and scheduler kwargs being ignored are unrelated." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python-version compatibility breakage is not the same as a Qwen3-VL bounding-box issue." + "reason": "Python version compatibility and Qwen3-VL bbox output are unrelated." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs handling and PretrainedConfig type-checking are unrelated regressions." + "reason": "Scheduler kwargs passthrough and `PretrainedConfig` type checking are different regressions." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Import-star performance regression is unrelated to the HybridCache import failure." + "reason": "Import cleanup for startup performance is not the same as a missing `HybridCache` export." }, { "left": "issue:44492", "right": "issue:46328", "accept": false, - "reason": "A cache-strategy typo and a SequenceFeatureExtractor.pad performance issue are different problems." + "reason": "A cache-strategy typo and a `SequenceFeatureExtractor.pad` performance issue are unrelated." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Import-time slowdown from wildcard imports is unrelated to a testing_utils bug." + "reason": "Import-style cleanup and a testing_utils bug do not share the same underlying problem." }, { "left": "issue:43976", "right": "issue:45850", "accept": false, - "reason": "Python 3.9+ compatibility breakage is a separate release-compatibility issue from generic v5 model-loading failures." + "reason": "Python support matrix issues and broad model breakage after v5 are not the same defect." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Different models and different symptoms: Qwen3-VL box output vs Qwen3.5 Moe from_pretrained error." + "reason": "Qwen3-VL bounding-box output issue and Qwen3.5 MoE `from_pretrained` failure are unrelated." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "patchtsmixer post_init policy and _get_tied_weight_keys crashing on list input are distinct implementation issues." + "reason": "Optional `post_init` enforcement and `_get_tied_weight_keys` AttributeError are separate code paths." }, { "left": "issue:43976", "right": "issue:46218", "accept": false, - "reason": "Python version support and missing Qwen3_5ForCausalLM import are unrelated." + "reason": "Python version compatibility and missing `Qwen3_5ForCausalLM` import are unrelated." }, { "left": "issue:43827", "right": "issue:44908", "accept": false, - "reason": "Docs still referencing pipeline() after removals is unrelated to scheduler kwargs forwarding." + "reason": "Docs still referencing `pipeline()` and scheduler kwargs propagation are unrelated." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "A documentation confusion issue and a tokenizer-loading regression are not the same bug." + "reason": "Documentation confusion in `forward` and tokenizer loading regression are different issues." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger Kernel not being applied with model_init and degenerate training under flash_attention_3 are different training-path issues." + "reason": "Liger kernel not applied under `model_init` and degenerate training with flash_attention_3 are unrelated." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "A broken env command and lazy loading not working are different release-v5 problems, not the same concrete bug." + "reason": "Broken env CLI and lazy loading behavior are different bugs." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "Pipeline save_pretrained failure and tied-weights key handling are unrelated." + "reason": "Pipeline save_pretrained support and tied-weight-key crashes are unrelated serialization vs utility bugs." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "DSA indexer ReLU discussion has nothing to do with Gemma-4 audio relative positional encoding." + "reason": "DSA indexer ReLU discussion and Gemma-4 audio positional encoding are unrelated model internals." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "An unspecified bug report cannot be treated as the same issue as a concrete Qwen3.5 Moe from_pretrained error." + "reason": "A generic bug report is not the same as the specific Qwen3.5 MoE from_pretrained error." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading malfunction and inverse_sqrt scheduler kwargs are unrelated." + "reason": "Lazy loading problems and scheduler kwargs handling are unrelated." }, { "left": "issue:43576", "right": "issue:46218", "accept": false, - "reason": "Broken env command and missing Qwen3_5ForCausalLM import are different failures." + "reason": "Env command breakage and missing Qwen3.5 import symbol are unrelated." }, { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "DeepSeek-V4 shared-expert gating and a tensor transpose/dtype bug are not the same underlying model defect." + "reason": "DeepSeek-V4 gating logic and a weight conversion/dtype-transpose bug are different model-specific issues." }, { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are unrelated model-specific bugs." + "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are separate model bugs." }, { - "left": "issue:43519", - "right": "issue:44908", + "left": "issue:43704", + "right": "issue:44485", "accept": false, - "reason": "Timestamp calculation in a processor is unrelated to scheduler kwargs handling." + "reason": "VRAM leak in multithreaded dataloading and RoPE implementation are unrelated." }, { - "left": "issue:41669", + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Continuous batching support for audio models and a DeepSeek-V4 weight conversion question are not the same bug." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "Broad v5 breakage and a specific missing import are not the same underlying issue." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "Network requests despite `local_files_only=True` and Qwen3VL timestamp calculation are unrelated." + }, + { + "left": "issue:42371", "right": "issue:43704", "accept": false, - "reason": "Import-time wildcard-import slowdown is unrelated to VRAM leakage in Qwen3ForCausalLM." + "reason": "TF32 settings warning and VRAM leak in dataloader threads are unrelated." }, { - "left": "issue:31795", - "right": "issue:34689", + "left": "issue:43519", + "right": "issue:44623", "accept": false, - "reason": "Documentation confusion in forward args is unrelated to model loading breakage for Llama 3.2 Vision." + "reason": "Timestamp calculation bug and missing files in `processor.save_pretrained` are separate processor issues." }, { - "left": "issue:31515", - "right": "issue:31795", + "left": "issue:43519", + "right": "issue:43704", "accept": false, - "reason": "Slow from_pretrained checkpoint loading and forward-pass documentation confusion are different topics." + "reason": "Qwen3VL timestamp bug and multithreaded VRAM leak are unrelated." }, { - "left": "issue:44485", + "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "GLM-5 RoPE implementation and Gemma-4 audio positional encoding are distinct architecture bugs." + "reason": "CPU offload device mismatch and Gemma-4 audio positional encoding are different subsystems." }, { - "left": "issue:43704", - "right": "issue:44485", + "left": "issue:44263", + "right": "issue:45468", "accept": false, - "reason": "VRAM leak in Qwen3ForCausalLM and GLM-5 RoPE implementation are unrelated issues." + "reason": "DSA indexer `torch.split` behavior and Gemma-4 audio positional encoding are unrelated." } ] }, @@ -16409,52 +16483,37 @@ "issue:36296", "issue:37934", "issue:38617", - "issue:42371", "issue:42913", "issue:42936", "issue:43066", "issue:43240", "issue:43298", "issue:43352", - "issue:43502", - "issue:43519", "issue:43525", "issue:43572", - "issue:43606", "issue:43618", "issue:43668", - "issue:43704", "issue:43717", "issue:43813", "issue:44031", "issue:44230", "issue:44261", - "issue:44263", "issue:44351", "issue:44492", - "issue:44556", "issue:44623", "issue:44861", "issue:44908", + "issue:44938", "issue:44998", "issue:45290", "issue:45310", "issue:45468", "issue:45850", "issue:46018", - "issue:46095", "issue:46144", "issue:46218" ], "soft_pairs": [ - "issue:43298|issue:46095", - "issue:45850|issue:46218", - "issue:43502|issue:43519", - "issue:42371|issue:43704", - "issue:43519|issue:44623", - "issue:43519|issue:43704", - "issue:43606|issue:45468", - "issue:44263|issue:45468", "issue:44261|issue:45468", "issue:44351|issue:46218", "issue:43352|issue:46218", @@ -16464,7 +16523,7 @@ "issue:43618|issue:44998", "issue:44998|issue:45850", "issue:44861|issue:45290", - "issue:44556|issue:44998", + "issue:44938|issue:44998", "issue:43525|issue:44861", "issue:43572|issue:44861", "issue:31515|issue:38617", @@ -16487,268 +16546,214 @@ "issue:37934|issue:43668", "issue:42936|issue:43668" ], - "prepared_review_unit_hash": "055364d7052097054e621f9838228dbcc712dc215668f83d837d3108b6d1aee9", + "prepared_review_unit_hash": "0213c9cc2ca43b965682ed5aca99f53ff5e9d611e60faad1f79ef03663356c00", "component_budget": { - "node_count": 445, - "item_count": 445, - "soft_pair_count": 1647, - "serialized_chars": 852354, - "estimated_input_tokens": 213089, - "estimated_eval_tokens": 426434 - }, - "budget": { - "node_count": 42, - "item_count": 42, - "soft_pair_count": 39, - "serialized_chars": 35172, - "estimated_input_tokens": 8793, - "estimated_eval_tokens": 17842 - }, - "overflow_policy": "truncate_then_skip", - "trimmed": false, - "aggressively_trimmed": false, - "split": true, - "analyst_result": { - "summary": "This cluster is highly heterogeneous: most items are unrelated model-specific bugs, tokenizer/config regressions, or infra issues. None of the soft pairs look like the same underlying defect/change, so I would not merge any of them as duplicates.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated subsystems, so there is no reliable duplicate anchor.", - "canonical_pr_reason": null, - "best_issue_reason": "If one issue must serve as a loose umbrella, issue 45850 is the broadest version-regression complaint and is the closest thing to a cluster anchor, though it is still not a true duplicate representative.", - "best_pr_reason": null, - "soft_edge_verdicts": [ - { - "left": "issue:43298", - "right": "issue:46095", - "accept": false, - "reason": "Audio continuous batching support vs DeepSeek-V4 weight-conversion question; unrelated topics." - }, - { - "left": "issue:45850", - "right": "issue:46218", - "accept": false, - "reason": "Broad v5 breakage complaint vs a specific Qwen3_5 import error; different problems and code paths." - }, - { - "left": "issue:43502", - "right": "issue:43519", - "accept": false, - "reason": "local_files_only network leak vs Qwen3VL timestamp bug; not the same bug." - }, - { - "left": "issue:42371", - "right": "issue:43704", - "accept": false, - "reason": "TF32 settings API complaint vs VRAM leak in Qwen3ForCausalLM; unrelated." - }, - { - "left": "issue:43519", - "right": "issue:44623", - "accept": false, - "reason": "Timestamp calculation bug vs processor.save_pretrained missing files; different failure modes." - }, - { - "left": "issue:43519", - "right": "issue:43704", - "accept": false, - "reason": "Timestamp math vs multi-thread VRAM leak; distinct issues." - }, - { - "left": "issue:43606", - "right": "issue:45468", - "accept": false, - "reason": "CPU offload device mismatch in bark-small vs Gemma-4 audio positional encoding; unrelated models." - }, - { - "left": "issue:44263", - "right": "issue:45468", - "accept": false, - "reason": "torch.split return handling vs Gemma-4 audio positional encoding; different code paths." - }, + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 35, + "item_count": 35, + "soft_pair_count": 31, + "serialized_chars": 28702, + "estimated_input_tokens": 7176, + "estimated_eval_tokens": 14608 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: mostly unrelated model/config/tokenizer regressions plus a few broad v5 complaints. No soft duplicate pair is close enough to merge into one concrete fix.", + "confidence": 0.94, + "canonical_issue_reason": "Issue 45850 is the broadest umbrella-style report in the set (\u201cSince 5.0 version it breaks too many models\u201d), so it is the closest thing to a cluster anchor, even though it is too vague to be a strong duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44261 is the most concrete and actionable bug report: it names a specific model path, describes the precision regression, and has the strongest discussion signal among the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "Layernorm epsilon precision issue vs Gemma-4 audio encoding bug; unrelated." + "reason": "Different models and failure modes: MLA q_a_layernorm eps handling vs Gemma-4 audio positional encoding." }, { "left": "issue:44351", "right": "issue:46218", "accept": false, - "reason": "HybridCache import failure vs Qwen3_5 model import failure; different symbols and likely different fixes." + "reason": "Separate import failures for different symbols/models; not the same code-path bug." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Flash Attention support gating vs import error; unrelated." + "reason": "Flash Attention support error for Nemotron vs Qwen3.5 import-module failure are unrelated." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both tokenizer-related, but one is a v4\u2192v5 behavior change and the other is a wrong decoder type; not the same defect." + "reason": "Both are tokenizer/v5-related, but one is a behavior regression and the other is a wrong decoder-type issue." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs bug vs unrelated spam issue." + "reason": "Inverse-sqrt scheduler kwargs bug is unrelated to the unrelated placeholder issue." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "Cache strategy typo vs unrelated issue; no overlap." + "reason": "Cache typo report does not match the other issue\u2019s subject." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression vs unrelated spam." + "reason": "CLIPOutput attentions regression is unrelated to the other issue." }, { "left": "issue:44998", "right": "issue:45850", "accept": false, - "reason": "Unemployment vs broad model-breakage complaint; no technical relation." + "reason": "One is a nonspecific complaint, the other is a broad v5 breakage report; not the same concrete bug." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weights utility crash vs chat-template tool-call crash; unrelated." + "reason": "Different failures in different helpers: tied weights vs chat template/tool-call crash." }, { - "left": "issue:44556", + "left": "issue:44938", "right": "issue:44998", "accept": false, - "reason": "Checkpoint reload regression vs unrelated spam; different domains." + "reason": "Python 3.14 import/load compatibility issue is unrelated to the other issue." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_id in config vs _get_tied_weight_keys crash; unrelated bugs." + "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight-key handling." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_idx in StableLmConfig vs tied-weights utility crash; different issues." + "reason": "StableLm pad_token_idx regression is a different config issue." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Slow from_pretrained checkpoint loading vs import error for layer_type_validation; unrelated." + "reason": "Checkpoint loading slowdown vs missing layer_type_validation import are unrelated." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "fp8 inference support request vs processor.save_pretrained missing files; not the same change." + "reason": "FP8 Qwen3.5 support request is unrelated to processor.save_pretrained missing files." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow reporting on failed training jobs vs slow checkpoint loading; unrelated." + "reason": "MLflow reporting on failed training jobs is unrelated to checkpoint loading speed." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Checkpoint loading slowdown vs tensor-parallel training bug; different subsystems." + "reason": "Checkpoint load slowdown and tensor-parallel training bug are different problems." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "fp8 support request vs Qwen3.5 MoE from_pretrained error; not the same bug." + "reason": "Both mention Qwen3.5/MoE, but one is fp8 inference support and the other is a from_pretrained import error." }, { "left": "issue:31515", "right": "issue:33453", "accept": false, - "reason": "Both involve loading, but one is slow checkpoint loading and the other is tokenizer loading regression; different fixes." + "reason": "Loading slowdown and tokenizer-loading regression are not the same code-path." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Audio continuous batching support vs fp8 support request; both feature asks, but not the same feature or code path." + "reason": "Continuous batching audio-model support is unrelated to Qwen3.5 FP8 inference support." }, { "left": "issue:43717", "right": "issue:46144", "accept": false, - "reason": "Mamba init_weights behavior vs RoFormer attention interface usage; unrelated models." + "reason": "Mamba init_weights behavior and RoFormer attention interface use are different model implementations." }, { "left": "issue:42936", "right": "issue:46018", "accept": false, - "reason": "Mask2former ignore_value regression vs DeepSeek-V4 shared-expert gating bug; unrelated." + "reason": "Mask2former ignore_value handling is unrelated to DeepSeek-V4 shared-expert gating." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "ModernBERT norm_eps type hint vs tokenizer regex warning regression; unrelated." + "reason": "Incorrect type hint and tokenizer regex warning are different classes of issues." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "ignore_value not used in Mask2former vs kwargs not passed to loss function; different code paths." + "reason": "Model loss-parameter handling does not match a Mask2former ignore_value bug." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Loss kwargs propagation bug vs RoFormer attention interface issue; unrelated." + "reason": "Loss kwargs propagation and RoFormer attention interface are unrelated." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2former ignore_value bug vs tokenizer regex warning; no overlap." + "reason": "Mask2former ignore_value bug and tokenizer regex warning are unrelated." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "Mask2former loss parameter bug vs RoFormer attention implementation; unrelated." + "reason": "No shared underlying bug; one is Mask2former-specific, the other is RoFormer-specific." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Mask2former ignore_value bug vs quantization-operation typo; unrelated." - }, - { - "left": "issue:43240", - "right": "issue:43668", - "accept": false, - "reason": "Loss kwargs propagation vs ModernBERT type hint issue; different layers and symptoms." + "reason": "Mask2former ignore_value and a typo in quantization_operation are unrelated." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "Type-hint problem vs typo in quantization label; unrelated." + "reason": "ModernBERT type-hint issue and quantization typo are unrelated." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "RMSNorm question for Llama4TextL2Norm vs ModernBERT norm_eps type hint; both norm-related but not the same issue." + "reason": "Llama4 RMS norm question is unrelated to ModernBERT norm_eps typing." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2former ignore_value regression vs ModernBERT norm_eps type hint; unrelated." + "reason": "Mask2former ignore_value handling is unrelated to ModernBERT config typing." } ] }, @@ -16781,52 +16786,52 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "0dca315a8b74eb5c0b1a996d5ddc6326937cacf3c06eebf554372cf3110b20c3", + "prepared_review_unit_hash": "0c6208b5ca0a2b9cf6e47ca2ecb97b81a684925d123f836f3502dfaceb88b478", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, - "estimated_input_tokens": 719, - "estimated_eval_tokens": 1694 + "serialized_chars": 2878, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2876, - "estimated_input_tokens": 719, - "estimated_eval_tokens": 1694 + "serialized_chars": 2878, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These three issues are not duplicates of one another: one is about truncation behavior, one asks for ReduceLROnPlateau-like functionality, and one asks for Wav2Vec2 word boundary offsets. They appear to be unrelated feature requests/questions rather than one underlying bug.", + "summary": "The three issues are not duplicates: one is about truncation strategy in tokenization, one about ReduceLROnPlateau-like scheduler behavior, and one about Wav2Vec2 word time offsets.", "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the topics are unrelated and do not describe the same underlying problem or feature.", + "canonical_issue_reason": "No canonical issue fits this cluster because the issues are unrelated feature requests.", "canonical_pr_reason": null, - "best_issue_reason": "No single best issue for deduplication: none of the three is representative of the others.", + "best_issue_reason": "No single issue is a good global representative; 10519 is the most self-contained, but it does not match the other two topics.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Different domains and requests: text truncation behavior vs Wav2Vec2 time offsets. No shared underlying bug or change." + "reason": "Different domains: truncation behavior vs Wav2Vec2 word offset extraction; no shared underlying bug/change." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "Unrelated features: truncation-from-beginning option vs scheduler-style ReduceLROnPlateau behavior." + "reason": "Unrelated requests: truncation-from-beginning option vs ReduceLROnPlateau-like functionality." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "Different functionality entirely: learning-rate scheduling request vs extracting word start/end offsets from audio model outputs." + "reason": "Different features in different components: learning-rate scheduling vs speech model word timing offsets." } ] }, @@ -16858,12 +16863,12 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "7b7af35de1ebded84a1e7b53743583188d8cda6c7e70dd021d91b46201b6fd58", + "prepared_review_unit_hash": "b8cccd8b3bde5b1566cc710955767d5b2b81778a8496361a0f97bbae6df460ae", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2477, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -16871,7 +16876,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2477, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -16880,24 +16885,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are not duplicates of one another: one is about TAPAS tokenization mismatching TensorFlow, one is about an auth/error-message problem when downloading private models, and one is a generic failure loading bert-base-uncased.", - "confidence": 0.95, - "canonical_issue_reason": "issue #16618 is the broadest and most central-looking model-loading report in the set, but it is still not the same underlying bug as the other two.", + "summary": "These three issues do not appear to be duplicates of one another: one is about Tapas tokenization differences, one about a bad authentication-related download error, and one about failing to load bert-base-uncased. The only broad overlap is general model/package loading behavior, but the concrete problems differ.", + "confidence": 0.88, + "canonical_issue_reason": "Issue #16618 is the broadest and most representative of the cluster because it concerns a model-loading failure. The other issues are narrower and clearly different in scope: #15980 is specifically about private-model auth messaging, and #13244 is about tokenization behavior.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the best cluster anchor only in the sense that it is the most general and appears most representative of a model-loading failure; the other issues are clearly about different code paths.", + "best_issue_reason": "#16618 is the best issue to keep as the cluster representative because it has the most generic user-facing symptom and closest overlap with the other model-loading complaint, while still being distinct from the tokenization and login-message issues.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "TAPAS tokenization mismatch versus a bert-base-uncased loading failure; different subsystem and root cause." + "reason": "Different underlying problem: Tapas tokenization mismatch vs. a bert-base-uncased model loading failure. They share only a broad NLP/model context." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Private-model download/auth error-message issue versus generic model loading failure; not the same bug or fix path." + "reason": "#15980 is about the error message when downloading a private model without login; #16618 is about loading a public model failing. Related area, but not the same bug." } ] }, @@ -16927,12 +16932,12 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "8386e9d4dd57e448f98db38291f026dfc48484ea9537121c896ec3d13dd2b3f5", + "prepared_review_unit_hash": "6cce52829ae05b6f26bf483401a285a35819216e91a64c69b6d165d687651e95", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -16940,7 +16945,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1589, + "serialized_chars": 1590, "estimated_input_tokens": 398, "estimated_eval_tokens": 1052 }, @@ -16949,18 +16954,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated feature-tracking issues: one for Flash Attention 2 support across more architectures, the other for GGUF support across more architectures. They share the \u201ccommunity contribution\u201d framing but not the same underlying change or code path.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 26350 is the more active and discussion-heavy thread, so it is the better representative issue for this loose cluster, though it is not a duplicate of 33260.", + "summary": "The two issues are both community contribution requests about adding architecture support, but they target different features: Flash Attention 2 vs GGUF. They are not duplicates and should not be clustered as the same underlying request.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 26350 is the strongest standalone anchor because it has substantially higher discussion activity and clearer community traction.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different features and different integration targets: Flash Attention 2 support versus GGUF support. They are both architecture-support requests, but not the same bug or change." + "reason": "Different feature requests and code paths: one asks for Flash Attention 2 support across more architectures, the other asks for GGUF support across more architectures. They do not describe the same bug or change." } ] }, @@ -16990,7 +16995,7 @@ "soft_pairs": [ "issue:26413|issue:31501" ], - "prepared_review_unit_hash": "cc7da97faa3980c438273e532264c94764789ce47abd47ed34ef155fec2c4150", + "prepared_review_unit_hash": "0959fe5237863835eeb8f39f7f00d7d1f4d0bc07a90c576e8b49e30c1e88a628", "component_budget": { "node_count": 2, "item_count": 2, @@ -17012,18 +17017,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are superficially similar training-start failures, but they do not clearly describe the same underlying bug. The first is a specific `resume_from_checkpoint`/`epoch_iterator` failure; the second is a broader FineWeb SLM training startup problem with no concrete shared failure mode shown.", - "confidence": 0.58, - "canonical_issue_reason": "Issue 26413 is the more specific and diagnostic report, with an explicit error message and a concrete failure path (`resume_from_checkpoint` / `epoch_iterator`).", + "summary": "The two issues both concern training startup failures, but only one describes a specific root-cause error. They do not appear to be confidently the same underlying bug.", + "confidence": 0.61, + "canonical_issue_reason": "issue #26413 is the better canonical issue because it names a concrete failure mode and error message: `resume_from_checkpoint` fails with `There seems to be not a single sample in your epoch_iterator`.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 26413 is the best representative because it names the exact exception and likely root cause, making it the most actionable reference point for triage.", + "best_issue_reason": "issue #26413 is the best representative issue overall since it is specific, reproducible, and tied to an explicit code-path failure rather than a broad symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26413", "right": "issue:31501", "accept": false, - "reason": "Both concern training not starting, but the shared evidence is only broad similarity. The first is a checkpoint-resume iterator error; the second is an unspecific FineWeb startup issue, so they do not clearly look like the same underlying bug." + "reason": "Both involve training not starting, but #31501 is too broad and does not clearly point to the same checkpoint-resume/epoch_iterator bug. The overlap is only at the symptom level." } ] }, @@ -17082,187 +17087,187 @@ "issue:44370|issue:46226", "issue:46222|issue:46226" ], - "prepared_review_unit_hash": "c8bda664f16ab0bff926145eb6fd83d9f2542ab7d5af9d96b911b639ca0acd09", + "prepared_review_unit_hash": "8fc41a402ca28af86e515adaf8abd23ca00b53f78f5846de8e574296c7732b22", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13521, - "estimated_input_tokens": 3381, - "estimated_eval_tokens": 7018 + "serialized_chars": 13531, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13521, - "estimated_input_tokens": 3381, - "estimated_eval_tokens": 7018 + "serialized_chars": 13531, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This is a mixed cluster, not one true duplicate set: Turkish/Spanish/Romanian translation issues are separate language requests, while the placeholder-title issues form a near-duplicate subcluster, and the two Thai-titled issues look like an exact duplicate pair.", - "confidence": 0.88, - "canonical_issue_reason": "issue:43834 is the best anchor for the placeholder-template subcluster because it is the most central generic translation issue and has the broadest exact-title similarity to the other placeholder entries.", + "summary": "This cluster is mostly false-positive similarity across unrelated language translation requests. The only clearly supported duplicate subset is the exact placeholder-title trio 43834/44365/44370; the Thai-language pair 46222/46226 is not strong enough to fold into that group without body-level evidence.", + "confidence": 0.84, + "canonical_issue_reason": "43834 is the earliest issue in the exact placeholder-title trio, which is the only clearly supported duplicate set here.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43834 is the most representative single issue in the duplicate-like placeholder group, but it should not be used to represent the language-specific translation requests.", + "best_issue_reason": "43834 is the cleanest representative of the exact placeholder-title trio and predates the other clearly matching copies.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Different target language/content; these are separate translation requests, not the same issue." + "reason": "Different language request vs generic placeholder issue; similarity is only superficial." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Different target language/content; only the template matches, not the underlying task." + "reason": "Turkish docs translation request is distinct from the placeholder-template issue." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Turkish-specific request vs generic placeholder issue; not the same underlying item." + "reason": "Different target language request; not the same underlying issue." }, { "left": "issue:27088", "right": "issue:38435", "accept": false, - "reason": "Different languages (Turkish vs Romanian), so these are distinct translation issues." + "reason": "Turkish and Romanian translation requests are separate issues." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Different languages (Turkish vs Spanish); similar template, but not the same request." + "reason": "Turkish and Spanish translation requests are not duplicates." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Spanish-specific translation request vs generic placeholder title; not a true duplicate." + "reason": "Spanish translation request is distinct from the generic placeholder issue." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Spanish request and generic placeholder issue are not the same underlying task." + "reason": "Different language-specific request; template similarity alone is not enough." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Spanish vs Romanian translation requests are distinct issues." + "reason": "Spanish and Romanian translation requests are separate issues." }, { "left": "issue:28936", "right": "issue:44370", "accept": false, - "reason": "Different language request; shared template text is not enough to merge." + "reason": "Spanish request and generic placeholder issue are not the same change." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Romanian-specific request vs generic placeholder issue; not the same underlying change." + "reason": "Romanian translation request is distinct from the placeholder-template issue." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Romanian request and placeholder issue are not the same issue." + "reason": "Romanian translation request does not match the generic placeholder issue." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Romanian translation request vs generic placeholder title; different underlying work." + "reason": "Romanian-specific request is not the same as the placeholder-template issue." }, { "left": "issue:38435", "right": "issue:46222", "accept": false, - "reason": "Romanian translation request is unrelated to the Thai-titled issue." + "reason": "Romanian translation request and opaque Thai-text issue are not clearly the same underlying task." }, { "left": "issue:38435", "right": "issue:46226", "accept": false, - "reason": "Romanian translation request is unrelated to the Thai-titled issue." + "reason": "Romanian translation request and opaque Thai-text issue are not clearly the same underlying task." }, { "left": "issue:43834", "right": "issue:44370", "accept": true, - "reason": "Both are the same generic placeholder translation issue with near-identical content and title." + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." }, { "left": "issue:43834", "right": "issue:44365", "accept": true, - "reason": "Same placeholder-title translation issue; looks like a duplicate entry." + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." }, { "left": "issue:43834", "right": "issue:46222", "accept": false, - "reason": "Very similar template text, but the Thai-titled issue appears to be a different underlying request." + "reason": "High soft similarity is not enough here; the title is different and there is no body evidence tying it to the placeholder trio." }, { "left": "issue:43834", "right": "issue:46226", "accept": false, - "reason": "Very similar template text, but the Thai-titled issue appears to be a different underlying request." + "reason": "High soft similarity is not enough here; the title is different and there is no body evidence tying it to the placeholder trio." }, { "left": "issue:44365", "right": "issue:44370", "accept": true, - "reason": "Identical placeholder translation issue; strong duplicate signal." + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." }, { "left": "issue:44365", "right": "issue:46222", "accept": false, - "reason": "Placeholder translation issue vs Thai-titled issue; not enough evidence of same underlying task." + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." }, { "left": "issue:44365", "right": "issue:46226", "accept": false, - "reason": "Placeholder translation issue vs Thai-titled issue; not enough evidence of same underlying task." + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." }, { "left": "issue:44370", "right": "issue:46222", "accept": false, - "reason": "Placeholder translation issue vs Thai-titled issue; different underlying content." + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." }, { "left": "issue:44370", "right": "issue:46226", "accept": false, - "reason": "Placeholder translation issue vs Thai-titled issue; different underlying content." + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." }, { "left": "issue:46222", "right": "issue:46226", - "accept": true, - "reason": "These two are text-identical and appear to be the same issue duplicated." + "accept": false, + "reason": "The Thai titles are identical, but without body content or explicit linkage the evidence is still too thin to assert the same underlying issue conservatively." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays conservative: the Turkish/Spanish/Romanian items are clearly language-specific by title, the placeholder-title issues are tightly similar, and 46222/46226 have identical titles and perfect similarity. I don\u2019t see any soft-edge verdicts that overclaim beyond the evidence." + "accept": false, + "feedback": "The writeup overstates the evidence for the duplicate subset. The packet clearly supports treating 43834/44365/44370 as near-identical placeholder-title issues, but it does not ground the claim that 46222 and 46226 are the same placeholder issue: their titles are completely different (Thai text), and only soft similarity is available. That makes those accept verdicts non-conservative. The summary should be narrowed to the explicitly matching placeholder-title trio unless there is stronger body-level evidence." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -17335,142 +17340,142 @@ "issue:43837|issue:45177", "issue:45322|issue:46133" ], - "prepared_review_unit_hash": "b40fef4bdfdcacb48f2dc627f5211e83ee37b9409cd28c88331b47fdb2a74ac4", + "prepared_review_unit_hash": "11d1bae104e9ff19dbb33363ce18cf8015696268ad9f1307d785e10e402dd4c5", "component_budget": { "node_count": 16, "item_count": 16, "soft_pair_count": 36, - "serialized_chars": 21661, - "estimated_input_tokens": 5416, - "estimated_eval_tokens": 11088 + "serialized_chars": 21658, + "estimated_input_tokens": 5415, + "estimated_eval_tokens": 11086 }, "budget": { "node_count": 16, "item_count": 16, "soft_pair_count": 36, - "serialized_chars": 21661, - "estimated_input_tokens": 5416, - "estimated_eval_tokens": 11088 + "serialized_chars": 21658, + "estimated_input_tokens": 5415, + "estimated_eval_tokens": 11086 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a loose title-similarity group of unrelated model-addition requests across text, vision, audio, and OCR. No soft edge looks like the same underlying issue/change, so none should be merged as duplicates.", - "confidence": 0.97, - "canonical_issue_reason": "No true canonical issue: the items span distinct model requests and modalities, so there is no single duplicate target.", + "summary": "This cluster is mostly a set of distinct model-request issues that only share generic \"add/support model\" wording. The soft links are driven by superficial similarity in model names and request style, not the same underlying change, so none should be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "No single true canonical issue: these requests target different models, tasks, and code paths, so the cluster is not a duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43671 is the closest thematic hub among the speech/audio requests, but it is still only an umbrella anchor, not a real duplicate representative.", + "best_issue_reason": "issue:43837 is the best representative hub for the cluster because it is one of the most connected support requests and sits near several of the other model-addition issues, but it still is not a duplicate of them.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model requests: EuroBert vs DEIMv2; same broad \u201cadd model\u201d pattern only." + "reason": "EuroBert config vs DEIMv2 are different model requests with unrelated architectures." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "BERT config request vs Qwen3-TTS support; unrelated model families and modalities." + "reason": "EuroBert config and Qwen3-TTS are unrelated text vs audio model additions." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "EuroBert config and RF-DETR are separate architectures with no shared code-path." + "reason": "EuroBert vs RF-DETR are different models and different tasks." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are distinct vision models; not the same change." + "reason": "RF-DETR and DEIMv2 are distinct vision-model requests, not the same change." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "Object detection model request vs LFM2.5 Audio request; different subsystem." + "reason": "RF-DETR vs LFM2.5 Audio are different model families and modalities." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR and sparse transformer are unrelated model families." + "reason": "RF-DETR and OpenAI Weight-Sparse Transformer are unrelated requests." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR vs Qwen3-ASR; different task and implementation path." + "reason": "RF-DETR and Qwen3-ASR are different models and code paths." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are unrelated vision vs speech requests." + "reason": "RF-DETR and Qwen3-TTS are unrelated model-support requests." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different model types and changes." + "reason": "NeoBERT vs Qwen3-ASR are different model families and tasks." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT vs sparse transformer request; no shared underlying bug or fix." + "reason": "NeoBERT and the sparse transformer request describe different architectures." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT and LFM2.5 Audio target different model families." + "reason": "NeoBERT and LFM2.5 Audio are unrelated text vs audio requests." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are unrelated requests." + "reason": "RF-DETR vs NeoBERT are distinct model additions." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are different BERT-like models, not the same addition." + "reason": "NeoBERT and ModernVBERT are both BERT-like names but refer to different concrete model additions." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture vs Qwen3-ASR support; different models and modalities." + "reason": "MiniCPM3 and Qwen3-ASR are different models and different support paths." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 architecture request is unrelated to the sparse-transformer model request." + "reason": "MiniCPM3 and the sparse transformer request are unrelated changes." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are separate model integrations." + "reason": "MiniCPM3 and ModernVBERT are separate model requests, not the same issue." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 vs Voxtral TTS; different code paths and model types." + "reason": "MiniCPM3 vs Voxtral TTS are different models and modalities." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are unrelated vision and speech requests." + "reason": "DEIMv2 and Qwen3-TTS are unrelated model-support requests." }, { "left": "issue:42503", @@ -17482,85 +17487,85 @@ "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT vs Qwen3-TTS; not the same underlying model support." + "reason": "ModernVBERT and Qwen3-TTS are unrelated requests." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT is a specific model request, while BERT-like models with RoPE is a broad architectural discussion." + "reason": "One is a specific model request, the other is a broad RoPE/BERT-like architecture request; not the same underlying change." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE discussion and Voxtral TTS are unrelated." + "reason": "BERT-like models with RoPE and Voxtral TTS are unrelated." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are related by vendor name but are different tasks and likely separate implementations." + "reason": "Qwen3-TTS and Qwen3-ASR are related in family but require different concrete support paths, so they are not the same change." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS support vs Voxtral TTS support are different model integrations." + "reason": "Qwen3-TTS vs Voxtral TTS are different model-support requests." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Speech/TTS support request vs DeepSeek-OCR2; different modality and code path." + "reason": "TTS support and OCR2 are different modalities and code paths." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS and sparse transformer request are unrelated." + "reason": "Qwen3-TTS and the sparse transformer request are unrelated." }, { "left": "issue:43671", "right": "issue:43909", "accept": false, - "reason": "TTS support and LFM2.5 Audio are distinct audio-model requests." + "reason": "Qwen3-TTS vs LFM2.5 Audio are different model additions." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are not the same model or fix." + "reason": "Qwen3-ASR and LFM2.5 Audio are different models and support implementations." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "ASR support request vs EUPE encoder request; different model classes." + "reason": "Qwen3-ASR and EUPE are unrelated model requests." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Qwen3-ASR and sparse transformer are unrelated." + "reason": "ASR support and a sparse transformer request do not describe the same code change." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "Audio model request vs SAM3-LiteText; different domains and implementations." + "reason": "LFM2.5 Audio and SAM3-LiteText are different modalities and model families." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Sparse transformer request and EUPE encoder request are unrelated model additions." + "reason": "Weight-Sparse Transformer and EUPE are separate research-model requests." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are different speech models and likely separate work." + "reason": "Qwen3-ASR and Voxtral TTS are different audio-model support requests." }, { "left": "issue:37015", @@ -17572,13 +17577,13 @@ "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "ASR support and OCR2 support are different modalities and code paths." + "reason": "ASR support and DeepSeek-OCR2 are different model tasks." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are different research model requests; same broad request style only." + "reason": "EUPE and TIPSv2 are distinct vision-language model requests, not the same underlying feature." } ] }, @@ -17608,12 +17613,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "b53779adef30cd236b7090d583916ec94b3e28e2691e21ad386f83ed006c9e8b", + "prepared_review_unit_hash": "d6decbfccab1f1c064c6e6c051634d0cd25e40744a61e6793393b8ee10feaba1", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1479, + "serialized_chars": 1480, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -17621,7 +17626,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1479, + "serialized_chars": 1480, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -17630,18 +17635,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are not duplicates: one is about block decoding, the other about DeepSeek v3.2 model support. There is no shared concrete bug or change target.", + "summary": "The two issues are not duplicates: one is about block decoding, the other about DeepSeek v3.2 model support. They describe different problem domains with no explicit linkage or shared implementation target.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "No canonical issue: the items address unrelated topics and should not be merged as duplicates.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "No best issue for a duplicate cluster; neither issue clearly subsumes the other.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different problem domains with no shared code-path or explicit link; similarity is only superficial." + "reason": "Different underlying problems: block decoding vs DeepSeek v3.2 support. No shared code path, filenames, or explicit references tying them together." } ] }, @@ -17671,12 +17676,12 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "f102506a1f35bf71a61b1f4d627a48bbdf3d0e1dbec54caf15f83364c18b41b5", + "prepared_review_unit_hash": "28fce62e6d25aeb54d34302fd1dad5d1876f9bcf791c008a3a77da075e3c3ea3", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1582, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -17684,7 +17689,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, + "serialized_chars": 1582, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -17693,18 +17698,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related only at a high level (both mention Trainer customization), but they describe different problems: one is about `remove_unused_column`, the other about FLOPS estimation for custom trainers. Not duplicates.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These two closed issues are related thematically (custom trainers and Trainer behavior) but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue fits both; the titles point to separate Trainer concerns with different expected fixes.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "If one issue needs to represent the cluster, #43598 is the better umbrella because it is more general about Trainer customizability, but it still does not cover FLOPS estimation.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Different underlying problems and likely different fixes: customizability of `remove_unused_column` vs FLOPS estimation for custom trainers." + "reason": "Only a broad topical overlap around custom trainers; the concrete bugs differ (`remove_unused_column` handling vs FLOPS estimation), so they are not the same underlying issue." } ] }, @@ -17734,12 +17739,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "9a5ccf354199a754155fd04096b8ac7b9c5b6bd350c52af974376e144fb901b5", + "prepared_review_unit_hash": "7dc5f32f2631106dfaac72056f96a27d8951aa5fe396836c2df1f1f1bc38c46c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1612, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -17747,7 +17752,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1612, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -17756,18 +17761,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicates: one asks for Qwen3-Omni auto-class registration, while the other requests AutoModelForSequenceClassification support for Qwen3.5/Qwen3_5Config. They target different model families and different auto-model behaviors.", - "confidence": 0.96, - "canonical_issue_reason": "Issue 43684 is the broader, more central registration request and is the better representative issue for this cluster.", + "summary": "These are related only in that both request AutoModel-family support for Qwen variants, but they target different models and different auto classes, so they are not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 43684 is the broader registration request and the more canonical representative of the cluster, but it does not subsume the Qwen3.5 sequence-classification support request.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43684 is the clearest standalone representative because it focuses on the general auto-class registration gap rather than a task-specific extension.", + "best_issue_reason": "Issue 43684 is the cleaner cluster representative because it asks for model registration in core auto classes and is phrased as a general support gap rather than a single downstream task.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different underlying requests: Qwen3-Omni registration vs Qwen3.5 sequence-classification support. Same broad ecosystem, but not the same bug/change." + "reason": "Different underlying requests: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 support for AutoModelForSequenceClassification. Same broad subsystem, but not the same bug/change." } ] }, @@ -17804,76 +17809,76 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "e55c39a0a74cb989afa137a7264a0cbf251c08d72a6b608f74d7909651131930", + "prepared_review_unit_hash": "ed750937ccaf09a913164f540dc2d4b1704531d87ad17ef8345c0cbff6ec38ef", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4263, - "estimated_input_tokens": 1066, - "estimated_eval_tokens": 2388 + "serialized_chars": 4267, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 }, "budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4263, - "estimated_input_tokens": 1066, - "estimated_eval_tokens": 2388 + "serialized_chars": 4267, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three issues titled \"huggingface\" look like duplicates of the same report; the \"Fett\" issue appears unrelated. Among the huggingface issues, the earliest one is the best canonical anchor, while the one with discussion activity is the strongest representative for triage context.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface\" reports, making it a reasonable stable anchor for the duplicate set.", + "summary": "Three issues titled \"huggingface \" look like duplicates of the same report; \"Fett\" appears unrelated. I\u2019d use issue 43845 as the representative canonical issue among the duplicate set.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43845 is the earliest of the three nearly identical \"huggingface \" issues and is the cleanest representative for that duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43859 has the same title as the other huggingface reports and at least some discussion activity, so it is the most informative representative of that duplicate group.", + "best_issue_reason": "43845 best represents the cluster because it matches the other duplicate issues\u2019 title/profile and is earliest among them.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles; \"Fett\" does not look like the same underlying issue as \"huggingface\"." + "reason": "Different title and no evidence of the same underlying problem; similarity is too weak to treat as duplicates." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles; no evidence these concern the same bug or request." + "reason": "Different title and no concrete shared bug/change signal beyond generic soft similarity." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no shared concrete details beyond vague similarity." + "reason": "Different title and insufficient evidence that it reports the same issue." }, { "left": "issue:43845", "right": "issue:43846", "accept": true, - "reason": "Same title and nearly simultaneous creation strongly suggest duplicate reports of the same issue." + "reason": "Same title, same general metadata profile, and created close together; these look like the same underlying issue." }, { "left": "issue:43845", "right": "issue:43859", "accept": true, - "reason": "Same title and close topic with no differentiating evidence; likely the same underlying report." + "reason": "Same title and clearly overlapping report pattern; likely duplicate of the same issue." }, { "left": "issue:43846", "right": "issue:43859", "accept": true, - "reason": "Same title and same apparent report pattern; fits a duplicate issue cluster." + "reason": "Same title and close temporal proximity strongly suggest duplicate reports of the same problem." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded and conservative overall. The analyst correctly excludes issue 43835 as unrelated, and the three \"huggingface\" issues are reasonably grouped based on matching titles, close creation times, and the soft-similarity signal. The choice of 43859 as the representative is also supported by its discussion activity." + "feedback": "The clustering is grounded in the packet: 43845/43846/43859 share the same title and have higher soft-similarity scores, while 43835 has a different title and much weaker similarity. The canonical choice of 43845 as the earliest issue in that duplicate set is also supported by the timestamps. The soft-edge verdicts are conservative overall." }, "evaluator_used": true, "retried": false, @@ -17911,88 +17916,88 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "d2c988db67b9eb0993d682ed42ff57ae69d470cd9d6d8685a2da0239d3d1f80e", + "prepared_review_unit_hash": "c1684091d65ce2b2542f988dedfa44f36fb5e952c84f126f3ea4c557878f06d7", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5794, - "estimated_input_tokens": 1449, - "estimated_eval_tokens": 3154 + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5794, - "estimated_input_tokens": 1449, - "estimated_eval_tokens": 3154 + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are not duplicates of one another; they look like unrelated spam or placeholder issues, with one generic documentation suggestion mixed in.", - "confidence": 0.96, - "canonical_issue_reason": "No clear canonical duplicate exists. If one issue must represent the cluster, #44418 is the most substantive title, but it does not describe the same underlying problem as the others.", + "summary": "These are all low-signal, unrelated issue reports dominated by spammy or placeholder titles; there is no clear shared underlying bug or request.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 44418 is the most neutral, issue-like entry in an otherwise spam/placeholder cluster, so it is the least-bad representative.", "canonical_pr_reason": null, - "best_issue_reason": "#44418 is the only non-obvious spam entry and is the closest to a real issue title, so it is the best representative among this set.", + "best_issue_reason": "Issue 44418 is the best global representative only because it looks like a real issue title; the others are more obviously spam or promotional noise.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are just '' placeholders; no evidence of the same underlying issue." + "reason": "Both titles are just \"\"; identical placeholder text is not enough to show the same underlying issue." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders with no concrete shared bug or change." + "reason": "Both are placeholder spam issues, but there is no concrete bug, request, or target proving they are duplicates." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is spam and the other is a generic documentation suggestion; they do not point to the same issue." + "reason": "One is spam placeholder text and the other is a generic documentation suggestion; they do not describe the same problem." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs. unrelated 'Google LLC' text; no shared bug/change." + "reason": "\"\" and \"Check out \\\"Google LLC\\\"\" are clearly different spam-like posts, not the same issue." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "Generic documentation suggestion vs. spam placeholder; unrelated." + "reason": "A documentation enhancement suggestion and a spam placeholder issue are not the same underlying request." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Generic documentation suggestion vs. spam placeholder; unrelated." + "reason": "The titles do not describe the same bug or change; one is a generic doc request, the other is spam." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders, but there is no evidence they describe the same concrete issue." + "reason": "Matching placeholder spam titles alone are insufficient to merge them as duplicates without shared concrete content." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs. unrelated company-name text; no shared underlying bug." + "reason": "Two different spammy titles with no shared technical substance; no evidence of the same underlying issue." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder vs. unrelated company-name text; no shared underlying bug." + "reason": "No shared bug, feature, or target; these are just separate low-content spam issues." } ] }, @@ -18022,7 +18027,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "f3798b08c4dee45d1296b21a5405541a54e97b5cfc9009c26f3221e789c28aee", + "prepared_review_unit_hash": "6ad466947253d50830dc473b87618cbd6394d5f573c80dd22ccbf65ed760aca2", "component_budget": { "node_count": 2, "item_count": 2, @@ -18044,28 +18049,28 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are highly similar and likely duplicates, but the packet does not provide enough hard linkage to treat them as confirmed duplicates. #45145 remains the best canonical issue because it was opened first and already has an inbound reference.", - "confidence": 0.83, - "canonical_issue_reason": "#45145 is the better canonical issue because it predates #45151 and already has an inbound reference, making it the stronger anchor record.", + "summary": "These two closed issues have the same title and highly similar content, indicating they are duplicate reports of the same energy-reduction idea. Issue #45145 is the better canonical choice because it was created earlier and already has an inbound reference.", + "confidence": 0.99, + "canonical_issue_reason": "#45145 is the earlier report and has an inbound reference, so it is the best canonical issue for this duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the strongest issue representative for the cluster due to being earlier and slightly more established in the discussion graph.", + "best_issue_reason": "It is the earliest instance of the duplicated title/body and is therefore the most stable issue to keep as the canonical tracker.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", - "accept": false, - "reason": "Reject as a confirmed duplicate edge: the titles are identical and the bodies are very similar, but the packet only shows soft similarity with no explicit links, no shared targets, and deterministic_accept is false. This supports likely-duplicate or closely-related status, not a definitive merge." + "accept": true, + "reason": "Same title, very similar body size/content, and no distinguishing metadata suggests they are duplicates of the same issue." } ] }, "evaluator_result": { - "accept": false, - "feedback": "The summary is mostly grounded, but the soft-edge verdict is too strong for the evidence available. The packet only shows a soft-similarity signal (jaccard 0.63) with no explicit links or shared targets, and `deterministic_accept` is false. Saying the issues are \u201cnear-certain duplicates\u201d and accepting the edge goes beyond what the packet supports; a more conservative assessment would be that they are likely duplicates or closely related, not confirmed duplicates." + "accept": true, + "feedback": "The pair is reasonably grounded as a duplicate: the issues have identical titles, very similar body sizes, and the packet flags soft similarity between them. The canonical choice of issue #45145 is conservative and supported by earlier creation time plus an inbound reference. No clear overstatement beyond the available evidence." }, "evaluator_used": true, - "retried": true, - "accepted_nontrivial_soft_edge": false, + "retried": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -18142,12 +18147,12 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "7d3946f739f1c5b2eec1522a82e55cd2eca362287ab34d643cc4006e98f809e6", + "prepared_review_unit_hash": "02f172400c17462b4a37e2d4b7da9c472dc4513e93d7c86137252a7cabe57b78", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1671, + "serialized_chars": 1670, "estimated_input_tokens": 418, "estimated_eval_tokens": 1092 }, @@ -18155,7 +18160,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1671, + "serialized_chars": 1670, "estimated_input_tokens": 418, "estimated_eval_tokens": 1092 }, @@ -18164,18 +18169,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate 4-bit inference bugs: one is a Gemma4/audio-module uint8 `torch.finfo()` TypeError, the other is a `BitsAndBytesConfig` regression where `llm_int8_skip_modules` drops the default `lm_head` exclusion and triggers an AssertionError. They share only broad quantization/inference context, not the same code path.", - "confidence": 0.94, + "summary": "These are both 4-bit inference reports, but they describe different bugs: one is a Gemma4/audio-module torch.finfo() TypeError on uint8 weights, the other is a BitsAndBytesConfig skip-module regression that clears the default lm_head exclusion and triggers an AssertionError. Shared topic, different underlying defects.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 45674 is the clearer canonical bug report for triage within this pair: it has a specific config regression, a concrete failure mode, and a direct path to reproducing/fixing the default exclusion behavior.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Different underlying failures and code paths: one is a dtype/finfo TypeError in Gemma4 audio modules, the other is a BitsAndBytesConfig skip-module regression causing an AssertionError. They are related only by 4-bit inference." + "reason": "Same broad quantized inference area, but different failure modes and code paths: Gemma4 audio-module uint8/tensor dtype handling vs BitsAndBytesConfig lm_head exclusion behavior. Not the same underlying bug." } ] }, @@ -18205,40 +18210,40 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "3e00336d81045f73485476da1d729f3c94314659de905e4b5a8a36603f7486ba", + "prepared_review_unit_hash": "ba73de174d39eef1d6c0979a33a7e3338fc7f8205bc58c57c011b815ea5950a1", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1477, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1477, - "estimated_input_tokens": 370, - "estimated_eval_tokens": 996 + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are both empty, unrelated personal-name/date titles and do not appear to describe the same underlying bug or request. Treat them as separate low-signal issues, not duplicates.", - "confidence": 0.97, - "canonical_issue_reason": "Issue #45762 is marginally better only because it was created first; neither issue contains any actual problem description or actionable content.", + "summary": "The two issues are unrelated-looking title-only entries with different names and dates; there is not enough evidence that they describe the same underlying problem.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #45762 is the best available issue by tie-breaker only; both are equally poor candidates for canonicalization due to missing context.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "The titles reference different people and different dates, and both bodies are empty. There is no evidence they report the same bug or request." + "reason": "Different person names and different dates; no shared bug report, code path, or explicit linkage indicating a duplicate." } ] }, @@ -18268,7 +18273,7 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "41b6617d5d1e318bf36a5140053e0f5370ed8cd3e68844be2dd54915ced9c2bc", + "prepared_review_unit_hash": "b0094e1332dbafad69640bb20c4d604ad6dc86b5fb0ec033297b9bea06720991", "component_budget": { "node_count": 2, "item_count": 2, @@ -18290,18 +18295,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues are empty and have unrelated personal-name titles with no technical content, so they do not look like duplicates or the same underlying bug.", - "confidence": 0.98, + "summary": "These two issues are just unrelated personal-name/date entries with no project context or shared bug/feature. They do not look like duplicates and should not be clustered together as the same underlying issue.", + "confidence": 0.96, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a meaningful canonical candidate; both are content-free and unrelated, so there is no clear duplicate target.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "Different names and dates, no body, no shared technical details, and no indication they describe the same bug or request." + "reason": "Titles only contain different names and birth dates; there is no shared code-path, error, or issue description indicating the same underlying problem." } ] }, @@ -18331,7 +18336,7 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "e9477da91bc0f626ffb98aee75f07fca427557ff7fc8df160784a772d5a9c7b3", + "prepared_review_unit_hash": "7e40d7f5d54309b2d72e201f6541349c3365589be4205a7ffa70e105b5ce404b", "component_budget": { "node_count": 2, "item_count": 2, @@ -18353,18 +18358,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one requests a new model head for token classification, while the other asks for FP8 training support. They concern different features and different code paths.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "The two issues are not duplicates: one requests a new model class for token classification/value modeling, while the other asks for FP8 training support. They concern different features and different code paths.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: these requests are about unrelated functionality, so neither can represent the other as a duplicate target.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Neither issue is a good global best for the cluster because the cluster does not appear to represent one underlying bug or feature request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different underlying requests: model/classification head support vs training precision support. No shared code path or concrete bug match." + "reason": "Different feature requests in different areas: model/class support for token classification vs FP8 training support. They do not look like the same underlying issue." } ] }, @@ -18404,88 +18409,88 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "7d84e2eccf287bda3769b847a79f1f8c9e45249a998d75f371626e209b6cb9f2", + "prepared_review_unit_hash": "1a5cd63eae185948bef63ce0743018df85b39709149f853919fbeeeeaf5e1dab", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5729, - "estimated_input_tokens": 1433, - "estimated_eval_tokens": 3122 + "serialized_chars": 5727, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5729, - "estimated_input_tokens": 1433, - "estimated_eval_tokens": 3122 + "serialized_chars": 5727, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster contains two exact duplicate issue pairs and one unrelated marketing/dataset post. The two duplicate groups are 45870/45871 and 45872/45873; 45869 is separate.", + "summary": "Two clear duplicate issue pairs emerge: 45870/45871 are exact duplicates, and 45872/45873 are exact duplicates. Issue 45869 is only loosely related by finance/data wording and is not the same underlying request.", "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue fits the whole cluster because it splits into two distinct duplicate pairs with different requests; if a representative is needed, 45870 is the cleanest canonical for its pair because it is the earlier of the exact duplicates.", + "canonical_issue_reason": "Issue 45870 is the best canonical representative of the exact duplicate pair 45870/45871: it is slightly earlier and has the only discussion activity, while 45871 appears identical.", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the best issue to anchor deduping for one subgroup: it is an exact duplicate target for 45871 and appears earlier, but it should not be used as the sole representative for the entire cluster.", + "best_issue_reason": "Issue 45870 is the strongest overall issue candidate among the set because it is part of an exact duplicate pair and is the earliest, most canonical wording. 45872 is equally a valid canonical for its own separate duplicate pair, but 45870 is the cleanest representative here.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Different subjects: a dataset promotion versus a financial time-series example." + "reason": "Different underlying requests: 45869 is a financial dataset offer, while 45870 is an example/tutorial about financial time series prediction with DataForge." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Different subjects: a dataset promotion versus a financial time-series example." + "reason": "Same as above; the dataset bundle request is not the same issue as the DataForge prediction example." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Exact duplicate title and matching body shape; same underlying example issue." + "reason": "These are exact duplicates: same title, same body size, and no distinguishing evidence." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Related theme, but different request: prediction example versus scraping service." + "reason": "Both are finance-related, but one is a prediction example and the other is a scraping-service feature request; they are not the same concrete issue." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Related theme, but different request: prediction example versus scraping service." + "reason": "The titles describe different requests with different product focus, so they should not be merged as duplicates." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Related theme, but different request: prediction example versus scraping service." + "reason": "The example/tutorial issue and the scraping-service issue are distinct underlying requests." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Related theme, but different request: prediction example versus scraping service." + "reason": "Not the same bug or change; only broad financial/time-series similarity is present." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Exact duplicate title and matching body shape; same underlying example issue." + "reason": "These are exact duplicates: same title, same body size, and no distinguishing evidence." } ] }, "evaluator_result": { "accept": true, - "feedback": "The grouping is grounded in the packet: 45870/45871 and 45872/45873 are the two clear duplicate pairs, while 45869 is separate from both example-request clusters. The canonical choice is conservative because no single issue cleanly represents all five items, and 45870 is a reasonable anchor for its duplicate pair." + "feedback": "The assessment is grounded in the packet: 45870/45871 and 45872/45873 have identical titles, matching body lengths, and the strongest similarity signal available, while 45869 is only loosely related by theme. The canonical choice for 45870 over 45871 is also supported by the earlier timestamp and discussion activity." }, "evaluator_used": true, "retried": false, @@ -18547,172 +18552,184 @@ "pull_request:44827|pull_request:46048", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "688022e355a25b3d0defb50090ed1fd7048d73d687d1f92d65216dbf18d9163e", + "prepared_review_unit_hash": "b396088bbe5aaccede7b265f84346086cedde74404fb8c2c6fb7e56e5740d6f1", "component_budget": { "node_count": 16, "item_count": 16, "soft_pair_count": 22, - "serialized_chars": 42945, - "estimated_input_tokens": 10737, - "estimated_eval_tokens": 21730 + "serialized_chars": 42942, + "estimated_input_tokens": 10736, + "estimated_eval_tokens": 21728 }, "budget": { "node_count": 16, "item_count": 16, "soft_pair_count": 22, - "serialized_chars": 42945, - "estimated_input_tokens": 10737, - "estimated_eval_tokens": 21730 + "serialized_chars": 42942, + "estimated_input_tokens": 10736, + "estimated_eval_tokens": 21728 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated PRs across new models, docs, export tests, and backend/integration tweaks. The only true duplicate pair is the two `fixed_cross_entropy` PRs for issue 43240.", + "summary": "Only one pair looks like the same underlying fix: the two `fixed_cross_entropy` PRs in `loss_utils.py`. The rest are unrelated feature/model additions, docs, or subsystem-specific changes and should not be merged as duplicates.", "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43251 is the best canonical PR for the duplicate loss fix: it explicitly targets issue 43240, updates `fixed_cross_entropy` to forward supported kwargs, and includes tests. It is the clearest representative of the underlying change.", + "canonical_pr_reason": "PR 43254 is the clearest representative of the `fixed_cross_entropy` change: it updates the shared loss helper in the same file, adds `weight` and `label_smoothing`, and preserves `**kwargs` while matching the same bug fix as the other PR.", "best_issue_reason": null, - "best_pr_reason": "PR 43251 is the strongest overall representative in the cluster because it is issue-linked, narrowly scoped, and the duplicate-loss fix is the only clearly overlapping code-path change here.", + "best_pr_reason": "PR 43254 is the best standalone representative of the cluster\u2019s only true duplicate pair because it cleanly captures the concrete `cross_entropy` API support change without extra issue-specific framing.", "soft_edge_verdicts": [ { "left": "pull_request:39895", "right": "pull_request:43424", "accept": false, - "reason": "Different features: VideoPrism model/docs addition vs executorch dynamic-shape export test." + "reason": "Unrelated PRs: one adds a new VideoPrism model and docs, the other adds Executorch export tests for dynamic shapes. Different code paths and different kinds of change." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Unrelated: Mistral tokenizer API additions vs executorch export testing." + "reason": "Different targets: Mistral tokenizer helper methods vs Executorch dynamic-shape export tests. No shared underlying bug or change." }, { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Different models and fixes: tokenizer methods vs adding MobileLLM." + "reason": "Tokenizer utility fix vs new MobileLLM model scaffolding. Same repo area only in the broad sense, not the same fix." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Different areas: processor loading robustness vs new MobileLLM model implementation." + "reason": "Processor robustness changes across existing models vs adding a new MobileLLM model. Different underlying work." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Loss-function kwarg forwarding is unrelated to adding MobileLLM." + "reason": "Loss-helper API fix vs new model implementation. No overlap in concrete bug or code path." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Both touch training/model infra, but they address different code paths and cannot be merged as one fix." + "reason": "`fixed_cross_entropy` kwargs support and Mistral4 test fixes are unrelated changes in different subsystems." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "Executorch export tests and Mistral4 test/model fixes are separate changes." + "reason": "Executorch export tests and Mistral4 test/model fixes do not address the same underlying issue." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Executorch export test vs new model implementation; no shared bug/change." + "reason": "Dynamic export testing vs adding MobileLLM. Different features and no shared bug." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "New model addition vs config-resolution order fix; unrelated." + "reason": "New MobileLLM model addition is unrelated to the config resolution-order fix in `modeling_utils.py`." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "MobileLLM model addition is unrelated to Falcon output-collection refactor." + "reason": "New model scaffold vs Falcon output-collection refactor. Different model families and different change intent." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Falcon refactor and model output-capture work are unrelated to the model-creation skill PR." + "reason": "Falcon refactor and model output-capture work is unrelated to the model-creation skill PR." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Falcon refactor vs Mistral4 test/model fixes; different code paths." + "reason": "Falcon standardized output interface refactor and Mistral4 test fixes are separate model-specific changes." }, { "left": "pull_request:44827", "right": "pull_request:46344", "accept": false, - "reason": "Mistral4 test fixes are unrelated to VitPose graph-break handling." + "reason": "Mistral4 test/model changes vs VitPose `flip_back` graph-break fix are unrelated code paths." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Mistral4 test fixes vs CLIP-like `layrnorm` renaming; different underlying issues." + "reason": "Mistral4 test fixes and `layrnorm` renaming/conversion-mapping cleanup are different bugs." }, { "left": "pull_request:44827", "right": "pull_request:46267", "accept": false, - "reason": "Mistral4 tests are unrelated to chat parsing updates." + "reason": "Mistral4 test fixes vs serve chat parsing update are unrelated features/fixes." }, { "left": "pull_request:44827", "right": "pull_request:46367", "accept": false, - "reason": "Mistral4 test/model fixes are unrelated to DeepGEMM availability changes." + "reason": "Mistral4 test/model work is unrelated to DeepGEMM availability and XPU enablement." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "Mistral4 fixes and the model-creation skill are unrelated." + "reason": "Mistral4 fixes and the model-creation skill PR do not share a concrete bug or change." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "New MobileLLM model implementation is unrelated to the model-creation skill PR." + "reason": "Adding MobileLLM and adding a model-creation skill are different artifacts with no shared underlying fix." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Mistral4 fixes and config-resolution-order fix are different issues." + "reason": "Mistral4-specific fixes and `cls.config` resolution-order logic in `modeling_utils.py` are unrelated." }, { "left": "pull_request:46048", "right": "pull_request:46367", "accept": false, - "reason": "TDT loss kernel integration and DeepGEMM/XPU availability are distinct integrations, not the same change." + "reason": "TDT loss-kernel integration and DeepGEMM/XPU enablement are different kernel integrations, not the same change." }, { "left": "pull_request:44827", "right": "pull_request:46048", "accept": false, - "reason": "Mistral4 test/model fixes do not overlap with TDT loss kernel work." + "reason": "Mistral4 test/model work is unrelated to the TDT loss-kernel integration PR." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46344", + "accept": false, + "reason": "Mistral4 test fixes and VitPose graph-break changes do not fix the same bug." + }, + { + "left": "pull_request:46048", + "right": "pull_request:46367", + "accept": false, + "reason": "Different integrations: TDT loss kernel vs DeepGEMM availability helper and XPU path. No shared concrete code-path issue." }, { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both modify `fixed_cross_entropy` in the same file to forward supported kwargs (`weight`/`label_smoothing`) for the same issue 43240; they are effectively duplicate fixes and could plausibly be merged into one PR." + "reason": "Both PRs change `src/transformers/loss/loss_utils.py` in the same `fixed_cross_entropy` function to add `weight` and `label_smoothing` support for the same bug/fix. The differences are minor API-shape details, so they are duplicate implementations of the same underlying change." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the one accepted soft-edge pair (43251/43254) is the only clearly mergeable same-bugfix duplicate: both touch `src/transformers/loss/loss_utils.py`, both target issue 43240, and both forward `weight`/`label_smoothing` through `fixed_cross_entropy`. The other verdicts are conservative and do not overstate overlap." + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted pair, 43251/43254, is well supported by the shared file/function, the same issue target, and near-identical `fixed_cross_entropy` API changes. The other rejections also read as appropriately cautious and do not overclaim duplication." }, "evaluator_used": true, "retried": false, @@ -18741,12 +18758,12 @@ "pull_request:40546|pull_request:42781", "pull_request:42781|pull_request:46331" ], - "prepared_review_unit_hash": "7298cb89be78f730ec787c1b5f4ea318df98f10dcaa515493ccafd9f568ed8b4", + "prepared_review_unit_hash": "ed65ea101bec4f5d64fc27686cbdf8f875d714b90d510df5fc43858298a8281c", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10379, + "serialized_chars": 10380, "estimated_input_tokens": 2595, "estimated_eval_tokens": 5446 }, @@ -18754,7 +18771,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10379, + "serialized_chars": 10380, "estimated_input_tokens": 2595, "estimated_eval_tokens": 5446 }, @@ -18763,24 +18780,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "No duplicate pair is present here. PR 40546 is the core VibeVoice implementation, PR 42781 is a draft VibeVoice Realtime extension with extra tokenizer/runtime changes, and PR 46331 is a separate Parakeet-RNNT model addition that only overlaps in shared auto-registration plumbing.", - "confidence": 0.96, + "summary": "This cluster is not a duplicate set: one PR adds base VibeVoice, another adds VibeVoice Realtime as an extension, and the third is unrelated Parakeet-RNNT work that only overlaps on shared auto-registration plumbing. None of the soft edges should be merged.", + "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": "No issue artifacts are present; among the PRs, 40546 is the most central and complete VibeVoice implementation, with broad model/docs/auto wiring and substantial review activity.", + "canonical_pr_reason": "PR 40546 is the best canonical representative because it is the original, focused VibeVoice implementation and the clearest anchor for the model family in this cluster.", "best_issue_reason": null, - "best_pr_reason": "40546 is the best representative PR in this cluster because it covers the base VibeVoice feature set and is the most developed non-draft implementation; the others are either a narrower realtime follow-up or an unrelated model PR.", + "best_pr_reason": "PR 40546 is the best overall PR here: it is the most self-contained and central change, while the other PRs are either a larger derivative feature set (42781) or a different model family altogether (46331).", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "Same broad model family, but not the same underlying change: 40546 adds core VibeVoice, while 42781 is a draft realtime variant with acoustic-tokenizer and generation changes. They are separate features, not duplicates." + "reason": "Related model family, but not the same change: 40546 adds base VibeVoice, while 42781 adds VibeVoice Realtime plus acoustic-tokenizer and generation changes. They touch overlapping files, but the concrete code-path work is different and would not plausibly collapse into one duplicate PR." }, { "left": "pull_request:42781", "right": "pull_request:46331", "accept": false, - "reason": "Different model families and different code paths. The overlap in auto files is incidental shared registration plumbing, not evidence of the same bug or change." + "reason": "These are different models with only generic auto-mapping overlap. 42781 is VibeVoice Realtime, while 46331 is Parakeet-RNNT; they do not fix the same bug or implement the same feature." } ] }, @@ -18810,7 +18827,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "2ed61f3a1a53bee2de403cb70f21bba7a794a4729c8a8c9fd3b18d27f2702340", + "prepared_review_unit_hash": "70af54203d28a10e9288d263f7bc719f179190ae3d29663bdadc40048aa54d3f", "component_budget": { "node_count": 2, "item_count": 2, @@ -18832,18 +18849,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These pull requests are unrelated: one adds TimesFM forecasting with covariates, the other adds DINOv3 image classification support. They do not fix the same code path or represent the same change.", - "confidence": 0.98, + "summary": "The two pull requests are unrelated: one adds covariate forecasting to TimesFM, the other adds DINOv3 image classification support. They touch different models, code paths, and user-facing features, so they should not be clustered as duplicates.", + "confidence": 0.99, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR #41224 is the cleaner representative of this cluster because it is a narrower, more self-contained model capability addition with a smaller diff and clear scoped filenames.", + "best_pr_reason": "No single PR is a good canonical representative because the items address different features in different model families and are not duplicates.", "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different domains and code paths: TimesFM time-series covariates versus DINOv3 image classification. They are not the same underlying bug or change." + "reason": "Different underlying changes: TimesFM time-series covariate forecasting versus DINOv3 image classification support. They do not fix the same bug or implement the same change and would not plausibly be merged into one PR." } ] }, @@ -18938,12 +18955,12 @@ "pull_request:41797|pull_request:45186", "pull_request:45186|pull_request:45630" ], - "prepared_review_unit_hash": "979585c798cef078dfdab0908eb41ada91c77d341199d0359365927eca2aaf20", + "prepared_review_unit_hash": "0b14c91c651c93129d42b56fdb3db46cbf6418695142e43dd2a5b379be05508f", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10436, + "serialized_chars": 10435, "estimated_input_tokens": 2609, "estimated_eval_tokens": 5474 }, @@ -18951,7 +18968,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 10436, + "serialized_chars": 10435, "estimated_input_tokens": 2609, "estimated_eval_tokens": 5474 }, @@ -18960,24 +18977,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three items are separate new-model addition PRs, not duplicates: DeepSeek OCR, Isaac, and Kimi2-6 each introduce different model families with shared Hugging Face boilerplate and auto-registration plumbing only.", - "confidence": 0.95, + "summary": "These are three separate model-addition PRs for different models (DeepSeek OCR, Isaac, and Kimi2-6). They share boilerplate registration/docs files, but they do not appear to fix the same code-path problem or be mergeable into one PR.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: these are distinct model-onboarding changes for different architectures and cannot plausibly be merged as one duplicate PR.", + "canonical_pr_reason": "No duplicate PR stands out here; if a single representative is needed, PR #45630 is the strongest cluster exemplar because it is the newest and most substantial model addition, with broad integration changes and high review activity.", "best_issue_reason": null, - "best_pr_reason": "PR 45630 (Kimi2-6) is the strongest representative of the cluster because it has the broadest integration surface and the most review activity, but it is still not a duplicate of the others.", + "best_pr_reason": "PR #45630 is the best representative of the cluster\u2019s general theme because it is a complete, heavily reviewed model-addition PR. The other two are similar in shape but target different models and different implementations.", "soft_edge_verdicts": [ { "left": "pull_request:41797", "right": "pull_request:45186", "accept": false, - "reason": "Different model families (DeepSeek OCR vs Isaac); overlap is limited to standard docs/auto-registry scaffolding, not the same code-path fix." + "reason": "Different models and implementations: DeepSeek OCR vs Isaac. Shared registry/docs updates are boilerplate for adding a new model, not evidence of the same underlying change." }, { "left": "pull_request:45186", "right": "pull_request:45630", "accept": false, - "reason": "Different model families (Isaac vs Kimi2-6); shared files are generic model-addition plumbing, so they are not the same underlying change." + "reason": "Different models and code paths: Isaac vs Kimi2-6. The overlap is only in generic auto-registration and docs, so these are not duplicates." } ] }, @@ -19007,12 +19024,12 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "f1603ad3494064960418d2da1b4286c5c2d35bf3924b3083974fc00524e7ba6f", + "prepared_review_unit_hash": "f74641d847753619d527c323eded87943ce2c15121a77509e414c9cb5f838e8c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -19020,7 +19037,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5934, + "serialized_chars": 5935, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -19029,18 +19046,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests update GitHub Actions dependencies, but they are not the same underlying change: one is a broad Node 24 compatibility refresh across many workflow actions, while the other is a separate bump of docker/build-push-action and a few related workflow actions.", - "confidence": 0.92, + "summary": "Both PRs are GitHub Actions dependency bumps in workflow files, but they target different upgrades and different action families: Node 24 compatibility vs. docker/build-push-action v6. They are related in theme, not duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #42975 is the better representative because it has a clear, specific goal (\u201cNode 24 compatibility\u201d) and a broader coordinated workflow-action upgrade rather than a single-action version bump.", + "canonical_pr_reason": "PR 42975 is the broader and more representative workflow-actions upgrade, spanning many workflows and matching the cluster theme more closely than the narrower Docker action bump.", "best_issue_reason": null, - "best_pr_reason": "PR #42975 is the strongest standalone representative in this cluster due to its explicit compatibility motive and wider workflow coverage; PR #42976 is a distinct dependency update path.", + "best_pr_reason": "PR 42975 is the best single representative because it is the more comprehensive GitHub Actions compatibility update across the repository.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Reject: these are related only at the level of GitHub Actions maintenance. #42975 upgrades multiple workflow actions for Node 24 compatibility, while #42976 specifically bumps docker/build-push-action and a few other workflow dependencies. They do not appear to fix the same concrete problem or be mergeable as one PR." + "reason": "Different underlying changes: 42975 upgrades several GitHub Actions versions for Node 24 compatibility, while 42976 only bumps docker/build-push-action to v6 in Docker image workflows. They do not appear to fix the same concrete problem or be mergeable into one PR." } ] }, @@ -19077,70 +19094,70 @@ "pull_request:43448|pull_request:45490", "pull_request:43451|pull_request:45490" ], - "prepared_review_unit_hash": "ea732f8aef889a98b78f174caa29a3cc4657048a546f36f68b5e5976fa554514", + "prepared_review_unit_hash": "19e7d47ee328d9c72fa152ef7c6cda7a3124b70c4d867d8220c55efbb5d49c66", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 14252, - "estimated_input_tokens": 3563, - "estimated_eval_tokens": 7382 + "serialized_chars": 14253, + "estimated_input_tokens": 3564, + "estimated_eval_tokens": 7384 }, "budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 14252, - "estimated_input_tokens": 3563, - "estimated_eval_tokens": 7382 + "serialized_chars": 14253, + "estimated_input_tokens": 3564, + "estimated_eval_tokens": 7384 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All four pull requests are separate model-addition changes. They share boilerplate docs/auto-registration updates, but they target different model families and code paths, so none should be merged as duplicates.", - "confidence": 0.98, + "summary": "All four pull requests are separate model-addition PRs, not duplicates. They share common Transformers scaffolding (docs, auto-mapping registrations, tests), but each targets a different model family and code path: ViT NEPA, Molmo, Molmo2, and CTSM.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: these are distinct feature PRs (ViT NEPA, Molmo, Molmo2, CTSM) with different model implementations and no shared underlying bug/change.", + "canonical_pr_reason": "PR 43451 (Molmo2) is the broadest and most complete model-addition PR in this set, with the highest review activity and the most extensive model/processor/video support; it best represents the cluster as a whole if one PR must be chosen.", "best_issue_reason": null, - "best_pr_reason": "No single PR is a good duplicate representative; 43451 has the broadest discussion, but it still addresses a different model addition than the others.", + "best_pr_reason": "PR 43451 is the strongest standalone PR to anchor on because it adds a distinct model family with multiple components and substantial implementation/test coverage, whereas the other PRs are also independent model additions but narrower or less developed.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "ViT NEPA vs Molmo are different model families; overlap is only in generic model-registration/docs plumbing." + "reason": "Different model additions (ViT NEPA vs Molmo). Similarity is only in shared integration files and docs, not the same underlying change." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "ViT NEPA and Molmo2 are unrelated model additions with different implementations and files; shared auto/docs changes are boilerplate." + "reason": "Different model additions (ViT NEPA vs Molmo2). Shared auto/docs boilerplate does not indicate the same bug or feature." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "Vision model addition vs time-series model addition; completely different subsystems and code paths." + "reason": "Different model additions (ViT NEPA vs CTSM). They touch shared registry/docs files but implement unrelated models and code paths." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are related in name but are distinct model families with separate model, processor, and integration changes." + "reason": "Molmo and Molmo2 are related by naming but appear to be separate model families with different configs, processors, and test sets; not mergeable as one PR." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo and CTSM are unrelated model additions; shared files are generic registry/docs updates only." + "reason": "Molmo and CTSM are unrelated model additions; overlap is limited to common auto-registration and documentation updates." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 and CTSM fix/add different models in different domains, so they are not the same underlying change." + "reason": "Molmo2 and CTSM are distinct model implementations with different filenames and functionality; shared framework files are insufficient for duplication." } ] }, @@ -19177,70 +19194,70 @@ "pull_request:43665|pull_request:46380", "pull_request:44815|pull_request:46380" ], - "prepared_review_unit_hash": "c8007fb5f80012e61cd009df588765e2ebc66782347b05fb9eff172192fe3d24", + "prepared_review_unit_hash": "d45176e2ae7de1d8a50a28979496b6dc471c1cf21d608734c3dd83e12f6a43a6", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 11145, - "estimated_input_tokens": 2787, - "estimated_eval_tokens": 5830 + "serialized_chars": 11142, + "estimated_input_tokens": 2786, + "estimated_eval_tokens": 5828 }, "budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 11145, - "estimated_input_tokens": 2787, - "estimated_eval_tokens": 5830 + "serialized_chars": 11142, + "estimated_input_tokens": 2786, + "estimated_eval_tokens": 5828 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are not duplicates of one another: they target different models and different code paths (Whisper decoding, CLIP/ViT tests, FP8 dequant/loading, and generation config validation).", - "confidence": 0.98, + "summary": "These PRs are not duplicates: they target different code paths (Whisper tokenization, test-only edits, FP8 dequantization, and generation watermarking/config validation).", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46380 is the clearest standalone bug fix: it addresses a concrete validation issue around dict-valued watermarking_config and includes related tests, making it more substantive than the test-only or broadly scoped alternatives.", + "canonical_pr_reason": "No strong duplicate cluster exists; if one PR must be chosen as the reference, 46380 is the most substantive and self-contained production fix, with real runtime logic plus tests.", "best_issue_reason": null, - "best_pr_reason": "PR 46380 is the strongest overall candidate because it fixes a specific runtime bug in generation config validation with an obvious code-path correction; the others are either unrelated or appear to be test scaffolding / broader refactors.", + "best_pr_reason": "46380 is the strongest standalone PR in the set because it addresses a concrete runtime validation bug and includes broad test coverage, while the others are either unrelated or look like test-focused noise.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Whisper tokenizer decode logic vs CLIP/ViT test edits; different models, different behavior, and no shared underlying bug." + "reason": "Whisper tokenization/timestamp decoding vs. test-only edits in CLIP/ViT; no shared underlying bug or change." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper timestamp decoding vs FP8 dequant/loading changes; unrelated subsystems and code paths." + "reason": "Whisper decode logic vs. FP8 dequantization/Mistral4; different subsystems and failure modes." }, { "left": "pull_request:43291", "right": "pull_request:46380", "accept": false, - "reason": "Whisper tokenizer/modeling changes vs generation watermarking validation; these fix different bugs in different areas." + "reason": "Whisper tokenization fix vs. generation watermarking/config validation; the shared Whisper filename is incidental, not the same bug." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "The first is a test-only CLIP/ViT change, while the second is a runtime FP8 dequant/loading fix; they are not the same change." + "reason": "Bogus test edits vs. FP8 dequant fix; unrelated." }, { "left": "pull_request:43665", "right": "pull_request:46380", "accept": false, - "reason": "Test stubs in CLIP/ViT do not match the generation config and Whisper-related runtime fixes in the other PR." + "reason": "Test-only noise in vision tests vs. generation/watermarking runtime fix; no duplicate overlap." }, { "left": "pull_request:44815", "right": "pull_request:46380", "accept": false, - "reason": "FP8 quantization/dequant loading fixes are unrelated to generation watermarking validation and Whisper model changes." + "reason": "FP8 kernel padding/dequant path vs. generation config watermarking validation; distinct concrete issues." } ] }, @@ -19270,7 +19287,7 @@ "soft_pairs": [ "pull_request:43488|pull_request:43532" ], - "prepared_review_unit_hash": "d69742e26c9f4b59944c0395b4ef4a1148c3f63955f0a2e2da2cc0bdb77ca003", + "prepared_review_unit_hash": "3764f8271bab9f93a174563d4bab67fa29d491a1ced29c81e87f69a785e00f09", "component_budget": { "node_count": 2, "item_count": 2, @@ -19292,18 +19309,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs with only superficial overlap around a shared Vit test file; they address different changes and are not duplicates.", - "confidence": 0.94, + "summary": "The two PRs only overlap on a shared Vit test file, but they address unrelated changes: one tweaks repo-bot/setup metadata, the other changes GitHub workflows and notification/testing plumbing. They should not be merged as duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43488 is the more specific artifact: a small repo-bot/version-format test tweak with setup/dependency adjustments. PR #43532 is broader workflow/test plumbing and a different change set.", + "canonical_pr_reason": "No canonical PR: the overlap is only a shared test file, while the underlying changes are different and not mergeable into one fix.", "best_issue_reason": null, - "best_pr_reason": "PR #43488 is the best representative PR in this cluster because it is narrower and centered on a single concrete behavior change, whereas PR #43532 spans CI workflow restructuring and unrelated utilities.", + "best_pr_reason": "Neither PR is a good cluster representative; if forced, 43532 is the broader change, but it still does not match 43488's repo-bot/version dependency tweak.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "They do not fix the same underlying bug or implement the same change; one is a repo-bot/version-format test PR, the other rewrites CI/workflow behavior and test utilities. Shared Vit test filenames are incidental." + "reason": "Shared filename alone is not enough; 43488 is a setup/version/dependency and bot-format change, while 43532 is workflow/notification/test infrastructure work. Different code paths and different fixes." } ] }, @@ -19333,7 +19350,7 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "971a1e0c6fd9853e0106f3b712d31de5af01a7a4bfb14ef543d1317dfd832c97", + "prepared_review_unit_hash": "6bad0104c754d12e3a97153940461097f5b891a18beae61442a0bde35cb84e71", "component_budget": { "node_count": 2, "item_count": 2, @@ -19355,18 +19372,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two related but distinct PRs: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. They share some scaffolding and docs, but they target different code paths, different model integrations, and different tests, so they are not duplicates.", - "confidence": 0.97, + "summary": "Two related but distinct PRs add separate promptable segmentation pipelines. They overlap in shared docs/registration plumbing, but one implements concept segmentation and the other visual segmentation with different models/tests, so they are not duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 43613 is the better representative of the cluster because it is the more comprehensive segmentation-pipeline change and touches a broader set of model integrations and tests.", "best_issue_reason": null, - "best_pr_reason": "PR #43613 is the broader and more representative of the shared segmentation-pipeline effort, but it still addresses a different feature than #43612, so it should not be treated as the same change.", + "best_pr_reason": "PR 43613 is the strongest representative of the set: it is a concrete feature PR with substantial implementation, docs, and test coverage for a specific pipeline.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Reject: despite overlapping pipeline/doc scaffolding, these PRs implement different features (concept vs visual segmentation) with different model and test coverage, so they are not the same underlying change and would not be merged into one PR." + "reason": "Reject: they are adjacent features, not the same fix. 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation and integrates different underlying models/tests. Shared docs and pipeline registry changes are incidental." } ] }, @@ -19396,40 +19413,40 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "91873eb6eddc809561710bf915f6f7b6e8d0b93cb5d3a49ec36be5e36880ce14", + "prepared_review_unit_hash": "f82a1394dda9bb7dd3b38871728f29759afc2d75cf21d41dd347f2cbe166116d", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4293, - "estimated_input_tokens": 1074, - "estimated_eval_tokens": 2404 + "serialized_chars": 4292, + "estimated_input_tokens": 1073, + "estimated_eval_tokens": 2402 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4293, - "estimated_input_tokens": 1074, - "estimated_eval_tokens": 2404 + "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": "Two open Trainer enhancement PRs touch the same file but implement different extension hooks: one changes gradient-accumulation loss scaling, the other refactors signature-column filtering for PEFT/custom trainers. They are related by subsystem, not the same underlying change.", - "confidence": 0.94, + "summary": "Two separate Trainer extensibility PRs: one adds a hook for gradient-accumulation loss scaling, the other adds a hook for signature-column filtering. They share a file and theme but not the same underlying change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical duplicate PR: the two changes address different Trainer behaviors and should not be merged as one representative fix.", + "canonical_pr_reason": "PR #43663 is a clean, self-contained API extraction in `trainer.py` that introduces a reusable override point for column filtering and cleanly refactors existing logic.", "best_issue_reason": null, - "best_pr_reason": "43651 is the more concrete standalone trainer behavior change, but it is not a better duplicate representative because it targets a different code path than 43663.", + "best_pr_reason": "PR #43663 is the better representative because it is a focused trainer API change with a clear override surface and minimal surrounding scope.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Reject: both modify trainer.py, but one is about gradient-accumulation loss scaling and the other about model signature column filtering; they are separate features with different code paths and not plausibly one merged PR." + "reason": "Both touch `trainer.py` and aim to make custom trainers more extensible, but they address different code paths: loss scaling during gradient accumulation vs. dataset column filtering. They are not the same fix and would not plausibly be merged as one PR." } ] }, @@ -19459,12 +19476,12 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "aa88bc6236bd66ff94dd0aef65399136852591fc86d90f9730bd95262596c631", + "prepared_review_unit_hash": "2c82f11e768e35c0d3892f8c25617a3e2a8a9886deed3038f0fd8e6170c26dd6", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3664, + "serialized_chars": 3663, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -19472,7 +19489,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3664, + "serialized_chars": 3663, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -19481,18 +19498,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs touch `src/transformers/cli/serve.py`, but they fix different problems: one is a broader typing/`__future__` annotations cleanup, the other is a specific `TypeAdapter` import/forward-reference issue when `pydantic` is absent.", - "confidence": 0.91, + "summary": "Both PRs modify `src/transformers/cli/serve.py` and mention `TypeAdapter`, but they address different typing/import failure modes rather than the same concrete fix.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43842 is the clearer, more specific fix for the `TypeAdapter` runtime NameError and directly matches the stated bug.", + "canonical_pr_reason": "PR 43656 is the cleaner representative of the cluster: it directly targets the CLI `TypeAdapter` NameError, is explicitly linked to issue 43576, and appears to fix the runtime annotation-evaluation problem in one focused change.", "best_issue_reason": null, - "best_pr_reason": "PR #43842 is the best representative of the cluster because it targets the concrete `TypeAdapter` failure mode, whereas PR #43656 is a separate annotations-related change.", + "best_pr_reason": "PR 43656 is the best overall representative because its title and issue link match the shared symptom most directly, while the alternate PR is a narrower pydantic-availability fix.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "They are not the same underlying fix: #43656 changes typing behavior via `__future__ annotations` and annotation cleanup, while #43842 explicitly handles `TypeAdapter` when `pydantic` is not installed. Shared file alone is not enough to merge them." + "reason": "Same file and similar symptom, but different root causes: 43656 uses `from __future__ import annotations` to avoid eager annotation evaluation, while 43842 guards the `TypeAdapter` import behind `TYPE_CHECKING` for optional pydantic. They are not the same concrete change." } ] }, @@ -19651,16 +19668,16 @@ "pull_request:44073|pull_request:44101", "pull_request:44059|pull_request:44071", "pull_request:44073|pull_request:44161", - "pull_request:44018|pull_request:44059", "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", "pull_request:44059|pull_request:44068", "pull_request:44101|pull_request:44161", - "pull_request:44017|pull_request:44018", + "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:44018", "pull_request:43997|pull_request:44161", "pull_request:44074|pull_request:44161", "pull_request:44018|pull_request:44161", @@ -19684,382 +19701,406 @@ "pull_request:44044|pull_request:44129", "pull_request:44044|pull_request:44154" ], - "prepared_review_unit_hash": "122615ea5aa1672def08057fdf5946b0af1bebf64d7ad1e71c9b796b30f19e25", + "prepared_review_unit_hash": "de0b38e89a81f997d39964a77ebf027ba47d03f1fb770afe7321c02b2c0afb74", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84893, - "estimated_input_tokens": 21224, - "estimated_eval_tokens": 42704 + "serialized_chars": 84892, + "estimated_input_tokens": 21223, + "estimated_eval_tokens": 42702 }, "budget": { "node_count": 27, "item_count": 27, "soft_pair_count": 72, - "serialized_chars": 75089, - "estimated_input_tokens": 18773, - "estimated_eval_tokens": 37802 + "serialized_chars": 75088, + "estimated_input_tokens": 18772, + "estimated_eval_tokens": 37800 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is dominated by model-by-model refactors migrating output collection to the new capture_outputs/can_return_tuple pattern, all broadly tied to umbrella issue #43979. Most pairings are only thematically similar across different models; the only clear duplicate-like overlaps are within the GPT-Neo set and the GPT-J/CodeGen set.", - "confidence": 0.72, - "canonical_issue_reason": "Issue #43979 is the umbrella tracker for the standardized output-tracing migration across many model implementations, which is the common thread connecting the cluster.", - "canonical_pr_reason": "PR #44066 is a strong representative of the cluster: it explicitly targets #43979 and makes the standardized output-tracing refactor in the GPT-J/CodeGen code paths concrete and easy to identify.", - "best_issue_reason": "#43979 best represents the shared underlying work because it is the coordination point for the entire output-tracing migration rather than a single model-specific instance.", - "best_pr_reason": "#44066 is the clearest cluster exemplar because it has an explicit issue link, a focused title, and a concrete cross-file GPT-J/CodeGen output-tracing refactor.", + "summary": "This cluster is mostly a broad issue-driven refactor campaign for standardized output tracing across many different models. Most pairs are not true duplicates because they touch different model families and different code paths; the only clear duplicate sets are the GPT-J/CodeGen PRs and the GPT-Neo PRs.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44066 is the best canonical representative for the GPT-J/CodeGen duplicate set: it is explicitly issue-linked, touches the same concrete output-tracing code paths in both files, and is the clearest umbrella refactor among the overlapping GPT-J variants.", + "best_issue_reason": null, + "best_pr_reason": "PR 44066 is the strongest single representative in the cluster because it covers the core standardized output-tracing change with explicit linkage to the tracking issue and the broadest direct overlap with the other true duplicates.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Both relate to output tracing, but they target different model families and code paths; shared issue #43979 is not enough to make them duplicates." + "reason": "Different model families and code paths; same high-level refactor theme is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different models and different implementation paths; this is the same refactor theme, not the same bug/change." + "reason": "Different model families and code paths; only the output-tracing theme is shared." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both are GPT-J output-tracing refactors in modeling_gptj.py; 44066 is just a broader variant that also includes CodeGen." + "reason": "Both refactor GPT-J output tracing in modeling_gptj.py toward the same standardized decorator-based interface." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "Different model families (ResNet/RegNet vs EfficientNet/other CNN refactor work); same umbrella issue does not make them the same change." + "reason": "Different model families and implementations; same issue target alone does not make them duplicates." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "These are unrelated model-specific refactors in different architectures; they share a migration pattern, not the same code-path fix." + "reason": "EfficientNet vs GPT-J/CodeGen are different concrete code paths despite similar refactor wording." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J/CodeGen output tracing versus EfficientNet output tracing are separate model changes." + "reason": "GPT-J/CodeGen and EfficientNet are unrelated model implementations." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa v2 are unrelated implementations; same standardized-output theme only." + "reason": "MobileNetV2 and DeBERTa are unrelated model code paths." }, { - "left": "pull_request:44066", - "right": "pull_request:44086", + "left": "pull_request:44086", + "right": "pull_request:44722", "accept": false, - "reason": "GPT-J/CodeGen versus MGP-STR are different code paths and different model families." + "reason": "MGP-STR and GPT-J/CodeGen are different model families with no shared bug fix." }, { "left": "pull_request:44066", - "right": "pull_request:44071", + "right": "pull_request:44086", "accept": false, - "reason": "Different model families (GPT-J/CodeGen vs MPT); only the refactor pattern is shared." + "reason": "Different model families and different output-handling implementations." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both are GPT-Neo output-tracing decorator migrations touching the same model/test files, so they look like the same underlying change." + "reason": "Both are GPT-Neo output-tracing refactors in the same model file and appear to implement the same decorator-based capture change." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J are separate models and separate code paths, despite similar output-tracing refactors." + "reason": "GPT-Neo and GPT-J/CodeGen are different model families and code paths." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J/CodeGen and GPT-Neo are different model implementations; not the same concrete fix." + "reason": "GPT-J/CodeGen and GPT-Neo are separate implementations, not the same duplicate change." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet refactor work is unrelated to GPT-J/CodeGen output tracing." + "reason": "ResNet and GPT-J/CodeGen are unrelated model code paths." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J/CodeGen are different model paths; only the standardized tracing mechanism is shared." + "reason": "MGP-STR and GPT-J/CodeGen are unrelated model implementations." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet versus GPT-J output tracing are different model-specific changes." + "reason": "ResNet and GPT-J are unrelated code paths." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated architectures; this is a generic refactor similarity only." + "reason": "GPT-Neo and ResNet are different models with different output-tracing logic." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are different model families and code paths." + "reason": "ResNet and MPT are different model families." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo output tracing and Flaubert output tracing are separate model-specific changes." + "reason": "GPT-Neo and Flaubert/XLM are unrelated model code paths." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert are unrelated model implementations; same migration theme only." + "reason": "GPT-Neo and Flaubert/XLM are different model families." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo versus GPT-J/CodeGen are different code paths, so these are not duplicates." + "reason": "GPT-Neo and GPT-J/CodeGen are different model refactors." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J/CodeGen are separate model refactors." + "reason": "MPT and GPT-J/CodeGen are unrelated implementations." }, { - "left": "pull_request:44085", + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model families." + }, + { + "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J output tracing and MGP-STR output tracing are unrelated model changes." + "reason": "GPT-Neo and MGP-STR are unrelated code paths." }, { "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 and MGP-STR are different model families and different output-handling paths." + "reason": "GPT-2 and MGP-STR are different model families." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert/XLM are unrelated code paths; same refactor style does not imply duplicate." + "reason": "ViLT and Flaubert/XLM are unrelated model code paths." }, { "left": "pull_request:44059", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 output tracing is separate from GPT-J/CodeGen output tracing." + "reason": "GPT-2 and GPT-J/CodeGen are different model implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are unrelated model families." }, { "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "SegFormer and GPT-J/CodeGen are distinct model-specific refactors." + "reason": "SegFormer and GPT-J/CodeGen are different code paths." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and LongT5 are different model families; they only share the standardized tracing theme." + "reason": "SegFormer and LongT5 are unrelated model implementations." }, { - "left": "pull_request:43997", + "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "RegNet and VisualBERT are unrelated model code paths." + "reason": "SegFormer and VisualBERT are different model families." }, { - "left": "pull_request:44073", - "right": "pull_request:44101", + "left": "pull_request:44068", + "right": "pull_request:44071", "accept": false, - "reason": "VisualBERT and Flaubert/XLM are different model implementations; not the same underlying change." + "reason": "GPT-Neo and MPT are unrelated models." }, { - "left": "pull_request:44073", - "right": "pull_request:44161", + "left": "pull_request:43997", + "right": "pull_request:44073", "accept": false, - "reason": "VisualBERT and LongT5 are separate model-specific output-tracing migrations." + "reason": "RegNet and VisualBERT are unrelated code paths." }, { - "left": "pull_request:44101", - "right": "pull_request:44161", + "left": "pull_request:43996", + "right": "pull_request:44017", "accept": false, - "reason": "Flaubert/XLM and LongT5 are unrelated model paths." + "reason": "CVT/FNet and SegFormer are different model families." }, { - "left": "pull_request:44017", - "right": "pull_request:44018", + "left": "pull_request:44010", + "right": "pull_request:44071", "accept": false, - "reason": "SegFormer and GPT-Neo are different models; shared issue target is not enough." + "reason": "SqueezeBERT and MPT are unrelated implementations." }, { - "left": "pull_request:44056", - "right": "pull_request:44161", + "left": "pull_request:44073", + "right": "pull_request:44101", "accept": false, - "reason": "MPNet and LongT5 are separate code paths and separate refactors." + "reason": "VisualBERT and Flaubert/XLM are different model code paths." }, { - "left": "pull_request:43997", - "right": "pull_request:44161", + "left": "pull_request:44059", + "right": "pull_request:44071", "accept": false, - "reason": "RegNet and LongT5 are unrelated model implementations." + "reason": "GPT-2 and MPT are unrelated models." }, { - "left": "pull_request:44074", + "left": "pull_request:44073", "right": "pull_request:44161", "accept": false, - "reason": "TextNet and LongT5 are different model-specific changes." + "reason": "VisualBERT and LongT5 are different model families." }, { - "left": "pull_request:43997", - "right": "pull_request:44018", + "left": "pull_request:44056", + "right": "pull_request:44101", "accept": false, - "reason": "RegNet and GPT-Neo are unrelated model families." + "reason": "MPNet and Flaubert/XLM are unrelated code paths." }, { - "left": "pull_request:44056", - "right": "pull_request:44101", + "left": "pull_request:44018", + "right": "pull_request:44059", "accept": false, - "reason": "MPNet and Flaubert/XLM are distinct code paths; same migration theme only." + "reason": "GPT-Neo and GPT-2 are different models." }, { - "left": "pull_request:43997", - "right": "pull_request:44056", + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-2 and GPT-Neo are different model implementations." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44019", "accept": false, - "reason": "RegNet and MPNet are different model implementations." + "reason": "GPT-Neo and ResNet are unrelated code paths." }, { "left": "pull_request:44071", - "right": "pull_request:44076", + "right": "pull_request:44086", "accept": false, - "reason": "MPT and ImageGPT are separate model-specific refactors." + "reason": "MPT and MGP-STR are different model families." }, { - "left": "pull_request:44074", - "right": "pull_request:44101", + "left": "pull_request:44010", + "right": "pull_request:44073", "accept": false, - "reason": "TextNet and Flaubert/XLM are unrelated models." + "reason": "SqueezeBERT and VisualBERT are unrelated models." }, { - "left": "pull_request:43997", - "right": "pull_request:44074", + "left": "pull_request:44017", + "right": "pull_request:44161", "accept": false, - "reason": "RegNet and TextNet are different code paths and different model families." + "reason": "SegFormer and LongT5 are different code paths." }, { - "left": "pull_request:44010", - "right": "pull_request:44071", + "left": "pull_request:44074", + "right": "pull_request:44101", "accept": false, - "reason": "SqueezeBERT and MPT are unrelated model implementations." + "reason": "TextNet and Flaubert/XLM are unrelated model implementations." }, { - "left": "pull_request:44018", - "right": "pull_request:44129", + "left": "pull_request:44056", + "right": "pull_request:44161", "accept": false, - "reason": "GPT-Neo and SpeechT5 are different model families; same issue umbrella is too broad." + "reason": "MPNet and LongT5 are different model families." }, { - "left": "pull_request:44010", - "right": "pull_request:44073", + "left": "pull_request:43997", + "right": "pull_request:44018", "accept": false, - "reason": "SqueezeBERT and VisualBERT are separate models and separate fixes." + "reason": "RegNet and GPT-Neo are unrelated code paths." }, { "left": "pull_request:43997", - "right": "pull_request:44129", + "right": "pull_request:44161", "accept": false, - "reason": "RegNet and SpeechT5 are unrelated output-tracing refactors." + "reason": "RegNet and LongT5 are unrelated models." }, { - "left": "pull_request:44056", + "left": "pull_request:44074", "right": "pull_request:44161", "accept": false, - "reason": "MPNet and LongT5 are different models and different code paths." + "reason": "TextNet and LongT5 are different implementations." }, { - "left": "pull_request:43997", - "right": "pull_request:44071", + "left": "pull_request:44018", + "right": "pull_request:44161", "accept": false, - "reason": "RegNet and MPT are unrelated model-specific changes." + "reason": "GPT-Neo and LongT5 are unrelated model families." }, { - "left": "pull_request:44010", - "right": "pull_request:44018", + "left": "pull_request:43997", + "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBERT and GPT-Neo are distinct model implementations; only the refactor motif overlaps." + "reason": "RegNet and MPT are unrelated code paths." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBERT and DeBERTa v2 are different code paths and different model families." + "reason": "SqueezeBERT and DeBERTa are different model families." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa v2 and TextNet are separate model-specific output-tracing migrations." + "reason": "DeBERTa and TextNet are unrelated implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa and VITS are different model families." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J/CodeGen standardized output-tracing change touching the same files; these look like duplicate variants of one refactor." + "reason": "Same GPT-J/CodeGen output-tracing refactor, touching the same concrete files and code path." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both modify GPT-J output tracing in modeling_gptj.py; 44722 is a broader follow-up that still targets the same code path." + "reason": "Both affect the GPT-J output-tracing path in modeling_gptj.py and look like the same underlying change." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are different model paths; no shared concrete fix." + "reason": "GPT-J and SpeechT5 are unrelated model code paths." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "CVT/FNet versus RegNet/ResNet/RT-DETR ResNet are different model families; the shared issue only reflects the umbrella migration." + "reason": "CVT/FNet and RegNet/ResNet are different model families." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "CVT/FNet and SpeechT5 are unrelated model-specific refactors." + "reason": "CVT/FNet and SpeechT5 are unrelated output-tracing changes." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa v2 and SpeechT5 are different model implementations." + "reason": "DeBERTa and SpeechT5 are different model implementations." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa v2 and VITS are different code paths and different models." + "reason": "DeBERTa and VITS are unrelated code paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary stays within the packet: #43979 is clearly the umbrella tracker, and #44066 is a reasonable canonical PR exemplar because it explicitly links that issue and covers the GPT-J/CodeGen refactor. The soft-edge decisions are mostly conservative and supported by the file/model overlap shown in the diffs. The true pairs also look mergeable as the same underlying bugfix within the same model family (GPT-Neo and GPT-J/CodeGen), not just thematic similarity." + "feedback": "The summary is grounded in the packet and the duplicate verdicts stay conservative. The accepted soft edges are supported by shared files/issue targets and clearly overlapping GPT-J/CodeGen or GPT-Neo refactors, while the cross-family pairs are correctly rejected rather than over-merged." }, "evaluator_used": true, "retried": false, @@ -20097,14 +20138,14 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "ab0e32c3c2ecbb68df72b61c2c873bce4af792d31f8f38f569caacaad567f359", + "prepared_review_unit_hash": "37d1d30e42fcded831f906349f7eb0e9e33d7c223dc729c17a8b1822ded386d9", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 78, - "serialized_chars": 84893, - "estimated_input_tokens": 21224, - "estimated_eval_tokens": 42704 + "serialized_chars": 84892, + "estimated_input_tokens": 21223, + "estimated_eval_tokens": 42702 }, "budget": { "node_count": 8, @@ -20119,48 +20160,48 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all related to a shared framework-wide effort to standardize output tracing/capture across different model implementations, but they touch different models and are not duplicates of one another.", - "confidence": 0.91, + "summary": "This cluster is a set of related but model-specific refactors to standardized output tracing/capture; they are not duplicates of one underlying bug or one PR.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44073 is a clear, representative example of the cluster: it updates one model to use the new capture_outputs/can_return_tuple pattern and includes the core output-tracing refactor elements seen throughout the group.", + "canonical_pr_reason": "No single PR is a true canonical duplicate here; each PR targets a different model and code path, even though they share the same refactor theme.", "best_issue_reason": null, - "best_pr_reason": "PR 44073 is a strong cluster representative because its title and diff directly reflect the standardized output-tracing migration, making it a good anchor for the broader family of similar refactors.", + "best_pr_reason": "pull_request:44073 is the clearest representative of the shared refactor pattern, with an explicit standardized output-tracing title and direct capture_outputs/can_return_tuple changes.", "soft_edge_verdicts": [ { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "Both are output-tracing refactors, but they affect different model code paths (FocalNet vs. ImageGPT) and are independent model-specific migrations, not the same underlying change." + "reason": "Different models and code paths (FocalNet vs ImageGPT); same refactor theme, but not the same concrete fix." }, { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "VisualBert and TextNet are separate model implementations with different forward/output signatures; this is the same general refactor theme, but not one mergeable PR for the same concrete bug/change." + "reason": "Both are output-tracing refactors, but VisualBert and TextNet are separate implementations with different model-specific changes." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are distinct models with different output-recording setups; they share the migration pattern but not the same underlying fix." + "reason": "MPNet and VisualBert are distinct model modules; these are parallel migrations, not one mergeable PR." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "These are separate model-specific refactors in different files and do not fix the same concrete code-path problem." + "reason": "Different models and differing output-capture plumbing; not the same underlying bug or change." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are unrelated model implementations; the PRs share a modernization pattern but are not duplicates or one combined fix." + "reason": "RegNet and SqueezeBert are unrelated concrete code paths despite similar standardized-output refactoring." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and SpeechT5 target different architectures and output paths; they are independent refactors despite both touching standardized output capture." + "reason": "SqueezeBert vs SpeechT5 are separate model-specific output tracing changes; the shared theme is too broad for duplication." } ] }, @@ -20239,268 +20280,268 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "78447474720b1f2759306b32ec38dd23fc98970380abe73a0ee3a056955431de", + "prepared_review_unit_hash": "114d3cb6aa07f0f46d9d765b0f4d361ace01601de3680adcc0204059de8493b4", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 39, - "serialized_chars": 39789, - "estimated_input_tokens": 9948, - "estimated_eval_tokens": 20152 + "serialized_chars": 39787, + "estimated_input_tokens": 9947, + "estimated_eval_tokens": 20150 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 39, - "serialized_chars": 39789, - "estimated_input_tokens": 9948, - "estimated_eval_tokens": 20152 + "serialized_chars": 39787, + "estimated_input_tokens": 9947, + "estimated_eval_tokens": 20150 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This is a false-positive similarity cluster: every PR is a separate model-specific output-tracing refactor, not the same underlying bug or change.", - "confidence": 0.97, + "summary": "These are mostly model-specific refactors for the same output-tracing pattern, but they are not duplicates of one another: each PR targets a different model or code path, so they should stay separate.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "No true canonical PR: the items are separate model-specific output-tracing refactors, so there isn\u2019t one PR that represents the same underlying fix.", + "canonical_pr_reason": "No single PR is a true duplicate anchor here; the closest representative of the shared theme is PR 44025 because it is a substantive output-tracing refactor spanning two related DepthAnything files.", "best_issue_reason": null, - "best_pr_reason": "No single best PR for duplicate consolidation; the set is a broad template-similarity cluster rather than one mergeable change.", + "best_pr_reason": "PR 44025 is the best representative of the cluster theme because it is a broader, concrete output-tracing change across closely related files, but it still is not a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Different model files and code paths; only the refactor theme matches." + "reason": "Different model families and files (`timm_backbone` vs `upernet`); same refactor theme, but not the same code-path or change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different model implementations; same output-tracing wording is not enough for duplication." + "reason": "Both are output-tracing refactors, but they touch unrelated models (`timm_backbone` vs `vision_text_dual_encoder`)." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "Separate model-specific refactors, not one concrete bug fix." + "reason": "Separate model implementations (`timm_backbone` vs `univnet`); no shared concrete bug or mergeable single PR." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "Different architectures and files; these are not the same change." + "reason": "`timm_backbone` and `mamba/falcon_mamba` are distinct code paths; similarity is only at the refactor label level." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "Different model code paths; shared refactor pattern only." + "reason": "Different architectures (`timm_backbone` vs `rwkv`); not the same underlying change." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "Mobilenet_v1 and univnet are unrelated code paths, so not duplicates." + "reason": "`mobilenet_v1` and `univnet` are unrelated models; same tracing refactor pattern, but not duplicate work." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different models and filenames; same refactor theme but not the same change." + "reason": "Different model code paths (`mobilenet_v1` vs `upernet`); too broad to be the same change." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "Different model families; these PRs could not plausibly be merged into one fix." + "reason": "`mobilenet_v1` and `depth_anything` are unrelated implementations, even though both refactor output handling." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "Separate model implementations; only the generic output-tracing refactor overlaps." + "reason": "Not the same code path: `mobilenet_v1` vs `vision_encoder_decoder`." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "Different code paths and model files; not the same underlying issue." + "reason": "Different subsystems (`mobilenet_v1` vs `speech_encoder_decoder`); similar refactor, not duplicate." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "Vision-text dual encoder and univnet are unrelated changes despite similar wording." + "reason": "`vision_text_dual_encoder` and `univnet` are unrelated models; same broad theme only." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "Different model code paths; not one shared bug or fix." + "reason": "Different model families and code paths (`vision_text_dual_encoder` vs `upernet`)." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "These touch different models and cannot be merged as one PR." + "reason": "`vision_text_dual_encoder` and `mamba/falcon_mamba` do not share a concrete bug or change." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Separate model-specific refactors; no shared concrete bug." + "reason": "Different models (`vision_text_dual_encoder` vs `depth_anything`); too generic to merge as one duplicate PR." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "Univnet and UperNet are different model implementations, so not duplicates." + "reason": "`univnet` and `upernet` are separate model code paths; only the refactor style matches." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "Different code paths and files; only the refactor motif matches." + "reason": "Distinct architectures (`univnet` vs `mamba/falcon_mamba`); not the same underlying fix." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "Different model families; not the same underlying change." + "reason": "`univnet` and `depth_anything/prompt_depth_anything` are unrelated model implementations." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "UperNet and RWKV are unrelated model changes, so reject as duplicate." + "reason": "`upernet` and `rwkv` are different model families; same refactor theme is insufficient for duplication." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "Different models and code paths; not a single fix." + "reason": "Separate model code paths (`upernet` vs `mamba/falcon_mamba`) with no shared concrete bug." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "Falcon/Mamba vs CodeGen are unrelated implementations; only title similarity matches." + "reason": "`mamba/falcon_mamba` and `codegen` are unrelated internals; not plausibly one PR." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "Different models and files; too broad to be the same change." + "reason": "`timm_backbone` and `codegen` affect totally different model paths; only the output-tracing refactor label is shared." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "Unrelated model code paths; not mergeable into one PR." + "reason": "`univnet` and `codegen` are unrelated changes, despite both touching output handling." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "Different architectures and change sites; not duplicates." + "reason": "`upernet` and `codegen` do not share the same concrete bug or implementation path." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "Different model implementations; same refactor template only." + "reason": "`vision_text_dual_encoder` and `codegen` are different models and different fix scopes." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "Depth Anything and Vision Encoder-Decoder are different code paths; not the same concrete change." + "reason": "Both are encoder-decoder wrappers, but they target different modalities (`depth_anything` vs `vision_encoder_decoder`) and separate code paths." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "Separate model families and files; reject as duplicate." + "reason": "`depth_anything` and `speech_encoder_decoder` are unrelated models; not the same underlying change." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "Different models and implementations; only the generic refactor overlaps." + "reason": "Different model families (`depth_anything` vs `superpoint`); same refactor pattern only." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "Depth Anything-related refactor vs RWKV refactor; not the same bug or change." + "reason": "`depth_anything` and `rwkv` are unrelated code paths." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "Different model code paths; similar output-tracing cleanup only." + "reason": "Different models (`depth_anything` vs `dpr`); not a duplicate change." }, { "left": "pull_request:44026", - "right": "pull_request:44029", + "right": "pull_request:44027", "accept": false, - "reason": "Vision Encoder-Decoder and RWKV are unrelated changes." + "reason": "Both are encoder-decoder wrappers, but vision and speech variants are distinct code paths and should not be merged as duplicates." }, { "left": "pull_request:44026", - "right": "pull_request:44027", + "right": "pull_request:44029", "accept": false, - "reason": "Different encoder-decoder wrappers; not one concrete fix." + "reason": "`vision_encoder_decoder` and `rwkv` are unrelated; only the refactor style is similar." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "Different model implementations; shared refactor theme is insufficient." + "reason": "`vision_encoder_decoder` and `superpoint` are different model implementations." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "DPR and Speech Encoder-Decoder are separate changes, not duplicates." + "reason": "`vision_encoder_decoder` and `dpr` are separate code paths with no shared concrete bug." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "Different model files and code paths; no same underlying bug." + "reason": "`speech_encoder_decoder` and `superpoint` are unrelated models; not the same change." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "Speech Encoder-Decoder vs RWKV are unrelated implementations." + "reason": "`speech_encoder_decoder` and `rwkv` are different architectures and fixes." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "Different models and code paths; not mergeable into one PR." + "reason": "`speech_encoder_decoder` and `dpr` do not share a concrete code-path problem." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are different subsystems; reject as duplicate." + "reason": "`superpoint` and `rwkv` are unrelated; the overlap is only in refactor wording." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "Different model implementations; only the refactor pattern is shared." + "reason": "`superpoint` and `dpr` are distinct models and should remain separate." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR are unrelated code paths, so they are not the same change." + "reason": "`rwkv` and `dpr` are different model code paths; not a duplicate PR pair." } ] }, @@ -20530,12 +20571,12 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "ffd9e3a3a9fa8d8064bd815f09816322e20f58afa49e0e09f2c9ab38bdef5f56", + "prepared_review_unit_hash": "9a678589e6d0f910bc0f690ea5baac79d8e70fcfcf6db24c3a233e5c0d44d455", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6551, + "serialized_chars": 6552, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -20543,7 +20584,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6551, + "serialized_chars": 6552, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -20552,18 +20593,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated model-addition PRs: one adds PenguinVL, the other adds Voxtral TTS. They share some boilerplate auto-registry/doc files, but they do not appear to fix the same code path or belong in one merged PR.", - "confidence": 0.99, + "summary": "These are not duplicates: one PR adds PenguinVL, a vision-language model, while the other adds Voxtral TTS. The overlap is limited to shared Transformers plumbing and docs scaffolding.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45401 is the better representative because it is a focused model-specific addition with an explicit issue target and a clear, self-contained scope.", + "canonical_pr_reason": "No single canonical PR: they implement different model families and different code paths, so they should stay separate.", "best_issue_reason": null, - "best_pr_reason": "PR #45401 is the strongest standalone PR in the cluster: it has a direct issue reference, a narrower feature scope, and a coherent implementation for one model family.", + "best_pr_reason": "45401 is the cleaner representative of the cluster because it is more narrowly scoped, has an explicit linked issue, and focuses on one concrete TTS integration.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Different models and different features: PenguinVL implementation vs. Voxtral TTS support. Shared registry/doc scaffolding is incidental, not evidence of the same underlying change." + "reason": "Reject: different underlying features and model code paths. PenguinVL is a multimodal vision-language model; Voxtral TTS is a text-to-speech model. Shared auto-registration and docs changes are not enough to indicate the same change." } ] }, @@ -20597,12 +20638,12 @@ "pull_request:45363|pull_request:46339", "pull_request:45666|pull_request:46339" ], - "prepared_review_unit_hash": "b0fab21f3968645b411e5b8d256d2958766ce02e68e52599e992cef43ae16c5e", + "prepared_review_unit_hash": "d09e75066c01b900dbff85c338ad201f1035a4ee81515d0d95156208209b1334", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 3, - "serialized_chars": 9633, + "serialized_chars": 9635, "estimated_input_tokens": 2409, "estimated_eval_tokens": 5074 }, @@ -20610,7 +20651,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 3, - "serialized_chars": 9633, + "serialized_chars": 9635, "estimated_input_tokens": 2409, "estimated_eval_tokens": 5074 }, @@ -20619,36 +20660,36 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three closely related PR iterations around KernelConfig-based n-to-1 fusion, plus one adjacent but broader Module Fusion API PR. The strongest canonical representative is the newest simplified KernelConfig fusion PR.", - "confidence": 0.79, + "summary": "This cluster is centered on the n-to-1 kernel fusion / KernelConfig feature line. The later PRs (45363, 45666, 46339) look like iterations of the same concrete integration work, while 44979 is a separate lower-level module fusion API proposal.", + "confidence": 0.73, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46339 is the latest refinement of the same KernelConfig n-to-1 fusion codepath, and it consolidates the shared core loading / hub kernel / modeling changes seen in the earlier PRs.", + "canonical_pr_reason": "PR 46339 is the best representative because it appears to be the latest, most consolidated iteration of the same KernelConfig-based n-to-1 fusion work, with shared core-loading, hub-kernel, modeling, and test changes.", "best_issue_reason": null, - "best_pr_reason": "PR 46339 is the best representative of the cluster because it is the most recent and most consolidated implementation of the same feature family, with overlapping files and the same concrete fusion path.", + "best_pr_reason": "PR 46339 most completely captures the feature family in this cluster and overlaps the most with the later KernelConfig fusion implementation work.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Related fusion work, but 44979 is a generic Module Fusion API while 45363 is the specific KernelConfig n-to-1 kernel fusion implementation. They are adjacent abstractions rather than the same underlying change." + "reason": "Related theme, but not the same concrete change: 44979 introduces a generic Module Fusion API, while 45363 implements KernelConfig-based kernel fusion integration and example plumbing." }, { "left": "pull_request:45363", "right": "pull_request:46339", "accept": true, - "reason": "Both target the same KernelConfig n-to-1 fusion path and modify the same core loading / kernel integration areas; 46339 reads as a refinement of the earlier implementation rather than a separate change." + "reason": "Both are the same n-to-1 KernelConfig kernel-fusion feature line, with overlapping integration files and very similar goal; 46339 looks like a refinement/simplification of the same change." }, { "left": "pull_request:45666", "right": "pull_request:46339", "accept": true, - "reason": "These appear to be the same concrete kernel-fusion feature evolving across iterations, with heavily overlapping files and identical KernelConfig fusion semantics in the same codepath." + "reason": "These target the same concrete KernelConfig-based n-to-1 fusion path, share core files, and 46339 appears to be a later cleanup/extension of the same work." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the three KernelConfig PRs share the same feature family by title and overlapping files, and 44979 is clearly the broader Module Fusion API PR. The soft-edge decisions are conservative enough, with the 44979\u219445363 rejection reasonably justified by the broader vs. specific scope." + "feedback": "The summary and soft-edge calls are grounded in the packet: 44979 is a separate Module Fusion API, while 45363/45666/46339 are clearly KernelConfig n-to-1 fusion iterations with overlapping filenames and consistent titles. The accept/false and accept/true pair verdicts look conservative rather than overstated." }, "evaluator_used": true, "retried": false, @@ -20675,7 +20716,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "566210ec6fa078a0a92b8ec93540db3bbaf4132608ec586870f4aff360f17d86", + "prepared_review_unit_hash": "a431808eea965ba87d67b6f386afb5df32cf361647cf3fa45555d98e06bb97e2", "component_budget": { "node_count": 2, "item_count": 2, @@ -20697,18 +20738,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same Mistral pre-tokenizer AttributeError in `tokenization_utils_tokenizers.py` and target the same issue. They differ mostly by completeness, with one adding tests, so they are duplicates for triage purposes.", - "confidence": 0.98, - "canonical_issue_reason": null, - "canonical_pr_reason": "`pull_request:45317` is the stronger canonical PR: it fixes the same `_patch_mistral_regex` AttributeError and also adds test coverage, making it the more complete implementation.", - "best_issue_reason": null, - "best_pr_reason": "`pull_request:45317` is the best representative because it contains the same code-path fix as `45086` plus validation, and is more likely to be the final merged form.", + "summary": "Both pull requests are near-identical fixes for the same Mistral regex pre-tokenizer AttributeError in `src/transformers/tokenization_utils_tokenizers.py`, and both explicitly target issue #45081. They should be clustered together, with PR #45317 as the stronger canonical PR because it adds tests and has the clearer, more complete change set.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #45081 is the shared tracker for both PRs; no separate issue artifact is present in the cluster.", + "canonical_pr_reason": "PR #45317 is the better canonical PR because it implements the same fix as #45086 and includes additional test coverage and a clearer title.", + "best_issue_reason": "If anchoring to an issue, #45081 is the only and shared issue target for the cluster.", + "best_pr_reason": "PR #45317 is the best representative PR: same underlying fix as #45086, plus tests and more explicit intent.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying bug fix in the same file and same target issue; the second PR appears to be an expanded version of the first, not a separate change." + "reason": "Same code path, same file, same AttributeError fix, and both target the same issue; the diffs are effectively the same change." } ] }, @@ -20738,40 +20779,40 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "a34056b1bf970788c1c0ff21194e70c4c5b44ed35b39c3bbf2b30c6a171109c5", + "prepared_review_unit_hash": "53558b51ea2a824b1172f1829c2e248291ecd820fc36eb73cef1bd4c15e1d90b", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6062, - "estimated_input_tokens": 1516, - "estimated_eval_tokens": 3288 + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6062, - "estimated_input_tokens": 1516, - "estimated_eval_tokens": 3288 + "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 related NemotronH pull requests, but they are not duplicates: one is a narrow checkpoint conversion-mapping update, the other is a broader architectural split into NemotronHDense/NemotronHSparse with docs, auto-mapping, and tests.", - "confidence": 0.95, + "summary": "Adjacent NemotronH work, but the PRs are not duplicates: one is a refactor to align NemotronH with GraniteMoeHybrid and conversion mapping, while the other splits NemotronH into dedicated dense/sparse model variants with docs/tests.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45569 is the better representative of the cluster because it covers the larger, end-user-facing model refactor and the new dense/sparse model families, whereas 45115 is only a focused conversion-mapping helper.", + "canonical_pr_reason": "No canonical PR: these are related NemotronH changes, but they target different concrete code paths and different deliverables.", "best_issue_reason": null, - "best_pr_reason": "PR 45569 is the strongest standalone PR: it bundles the actual NemotronH split, configuration/model wiring, documentation, and tests. PR 45115 is narrower and more of an internal conversion-maintenance change.", + "best_pr_reason": "45569 is the more complete representative of the NemotronH family rework because it adds the dedicated dense/sparse implementations plus docs/tests, whereas 45115 is a narrower refactor.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Not the same underlying change. 45115 updates weight-conversion mappings and rename rules, while 45569 restructures NemotronH into separate dense/sparse model implementations with docs and tests. They touch related files but solve different problems and would not plausibly be merged as one PR." + "reason": "Different changes: 45115 refactors NemotronH inheritance/checkpoint conversion toward GraniteMoeHybrid; 45569 introduces separate NemotronHDense/NemotronHSparse packages and documentation. Shared files are incidental, not the same underlying fix." } ] }, @@ -20801,7 +20842,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "e0b18217db2687efc42ae8bf6d960a47e01491b728858f6a54710d687db7b7a2", + "prepared_review_unit_hash": "b944dd7836e66b4cb8f80dc6055a178b8849b51ffcaf5b507f5dc22f160517ab", "component_budget": { "node_count": 2, "item_count": 2, @@ -20823,18 +20864,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two related but distinct feature PRs: one adds heterogeneous per-layer config plumbing, the other extends that into modeling/cache behavior for heterogeneous models. They share infrastructure, but they do not look like the same concrete change and should not be merged as duplicates.", - "confidence": 0.91, + "summary": "These PRs are related but not duplicates: #45333 adds heterogeneous per-layer configuration plumbing, while #45332 builds broader heterogeneous model/runtime support on top of that, including cache/modeling changes. They share groundwork files, but they are different changesets and should not be merged as the same PR.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45332 is the best cluster representative because it covers the broader end-to-end heterogeneous model support, including modeling and cache/masking changes, not just config plumbing.", + "canonical_pr_reason": "PR #45332 is the broader, more central change: it covers heterogeneous model support end-to-end and incorporates the runtime/modeling pieces that would likely depend on the config groundwork.", "best_issue_reason": null, - "best_pr_reason": "PR 45332 is the strongest standalone PR in the set: it is broader, more user-visible, and closer to the actual heterogeneous-model feature surface than the config-only follow-up.", + "best_pr_reason": "PR #45332 is the better representative for the cluster because it is the larger feature umbrella and touches the main code paths beyond configuration plumbing.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "Related feature work, but not the same underlying change. 45333 is limited to per-layer config support, while 45332 adds downstream modeling/cache integration. They could be sequenced, but they are not one mergeable fix/PR." + "reason": "Related foundation and follow-on work, but not the same underlying change. #45333 is config-only per-layer support; #45332 extends into modeling/cache behavior and thus fixes a different, broader problem." } ] }, @@ -20864,12 +20905,12 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "e759039006fbf1f67c587a852033118721be7fe08d18e5dafa36738e86dfb0b5", + "prepared_review_unit_hash": "565c33728705fb6e261cd56348904739c5649317fefc521cdc497352bb76ac5a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5513, + "serialized_chars": 5514, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -20877,7 +20918,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5513, + "serialized_chars": 5514, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -20886,18 +20927,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are both typing-related PRs, but they address different problems: one is a broad repo-wide type-checking cleanup, the other adds concrete typing protocol helpers and model-specific support. Not duplicates.", - "confidence": 0.95, + "summary": "Both PRs are typing-related, but they address different scopes: #45415 is a broad repo-wide type-checking sweep with config changes, while #45425 adds specific typing protocols/helpers for a narrower set of runtime code paths. They are not the same underlying change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45425 is the more concrete, self-contained change: it adds typing aliases/protocols in `_typing.py` and updates a small set of call sites. PR 45415 is a broad mechanical type-checking sweep with config changes and many file edits, so it is less suitable as the canonical representative.", + "canonical_pr_reason": "#45425 is the more concrete and self-contained typing change; #45415 is a much broader sweep across many files plus a pyproject typing-ignore update.", "best_issue_reason": null, - "best_pr_reason": "45425 is the better representative of a distinct, focused typing improvement. 45415 is a wide-ranging type-checking pass rather than a single underlying bug or feature change.", + "best_pr_reason": "#45425 is the better representative of the cluster because it targets specific typing gaps in implementation code, whereas #45415 is a general typing cleanup pass.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Both touch typing, but they are not the same underlying change: 45415 is a broad type-checking cleanup across many files, while 45425 introduces specific typing protocols/aliases and associated model integration. They would not plausibly be merged as one PR." + "reason": "Different changes: #45415 is a wide type-checking pass and config adjustment, while #45425 adds specific typing protocols/helpers for modeling/PEFT-related code. Shared filenames alone do not indicate the same fix." } ] }, @@ -20927,12 +20968,12 @@ "soft_pairs": [ "pull_request:45513|pull_request:46084" ], - "prepared_review_unit_hash": "466c1dae75259bda8c22968fdf5f46453ffd2ce93b52f82a6e2603d74d0ffb7f", + "prepared_review_unit_hash": "6bc690e0ae5e6ed3c6cd11f5e2c58a583306787a5170d1ae8d21e6ccac59d282", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4857, + "serialized_chars": 4858, "estimated_input_tokens": 1215, "estimated_eval_tokens": 2686 }, @@ -20940,7 +20981,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4857, + "serialized_chars": 4858, "estimated_input_tokens": 1215, "estimated_eval_tokens": 2686 }, @@ -20949,18 +20990,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs both adjust cached forward paths for seq_len > 1, but they target different model families and different bugs: Qwen3.5/GDN linear attention vs Mamba2 CUDA kernel cache handling. They should not be treated as duplicates.", - "confidence": 0.98, + "summary": "The two pull requests are both cache-related, but they fix different model families and different code paths: Qwen3.5/OLMo hybrid linear-attention cached multi-token forward versus Mamba2 use_cache handling for seq_len > 1. They are not duplicates of the same underlying change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46084 is the better canonical representative because it has an explicit linked issue, a narrowly defined Mamba2 cache bug, and a concrete fix path in one model implementation.", + "canonical_pr_reason": "PR #46084 is the stronger canonical PR: it has an explicit issue target (#46032), a narrowly defined Mamba2 cache bug, and a small focused diff that clearly addresses one concrete failure mode.", "best_issue_reason": null, - "best_pr_reason": "PR #46084 is the best representative overall because it is more self-contained and specific; PR #45513 spans multiple model files and appears to be a broader cross-model fix rather than the clearest single bug instance.", + "best_pr_reason": "PR #46084 is the best representative of the cluster because it is the most precisely scoped and best-documented fix; PR #45513 is a broader multi-model cache correction without an explicit target issue.", "soft_edge_verdicts": [ { "left": "pull_request:45513", "right": "pull_request:46084", "accept": false, - "reason": "Reject: they involve different architectures and code paths. #45513 fixes multi-token cached forward in Qwen3.5/related models for GDN linear attention, while #46084 fixes Mamba2 use_cache behavior for seq_len > 1. Similar caching terminology is not enough to make them the same change." + "reason": "Both involve cached forward passes, but they affect different architectures and different implementation paths. #45513 fixes multi-token cached forward in Qwen3.5/OLMo hybrid linear attention, while #46084 fixes Mamba2 use_cache when seq_len > 1. They are not the same bug and would not be merged as one PR." } ] }, @@ -20990,7 +21031,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "4a6ea75973abb215144199ea2a7112b783bcbe38758a5019c5540f591c3e8678", + "prepared_review_unit_hash": "503a6dacf26904bda8c58efabbe15bbbaf79ad8e50ad075153a8b473c47bd35b", "component_budget": { "node_count": 2, "item_count": 2, @@ -21012,18 +21053,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related by expert-parallel loading, but they fix different bugs: one preserves distributed config metadata during serialization/loading, while the other prevents expert parameters from being overwritten during post-shard wrapping under FSDP2. They should not be merged as duplicates.", - "confidence": 0.89, + "summary": "These two PRs both touch expert-parallel loading paths, but they fix different bugs: #45548 is about making distributed_config survive serialization and exposing has_ep for EP + ZeRO-3 loading via accelerate, while #45662 fixes expert parameters being overwritten by rank-0 broadcast under EP + FSDP2. They are related but not the same change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45662 is the better representative of the cluster because it targets a concrete runtime corruption path in the model-loading flow, with explicit code changes around sharding/wrapping and expert parameters.", + "canonical_pr_reason": "No single canonical PR for a duplicate cluster: the PRs target different concrete failure modes and should stay separate; if forced, #45662 is the more specific bug fix.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the strongest standalone candidate: it fixes a specific EP/FSDP2 overwrite bug in the loading path and includes the most direct code-path correction.", + "best_pr_reason": "#45662 is the stronger representative because it fixes a concrete overwrite bug in the loading path with explicit code changes across the affected flow, whereas #45548 is a different serialization/loading compatibility fix.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Different underlying failures: #45548 is about serializing distributed_config / exposing has_ep for EP + ZeRO-3 loading, while #45662 fixes experts being overwritten by rank-0 broadcast during FSDP2 post-shard handling. They touch overlapping EP machinery but not the same bug." + "reason": "Both concern expert-parallel loading, but the bugs and code paths differ: one preserves distributed_config / has_ep for ZeRO-3 loading, the other prevents rank-0 broadcast from overwriting experts in FSDP2." } ] }, @@ -21053,12 +21094,12 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "2d56df83a7d29e0d3675e780ceb8a6f9c3c62a02c13153d45dff7a11b1e4092c", + "prepared_review_unit_hash": "10f06d274aa97d82e877f3f8e8d87b08991763ba490b4024543e2a669b32ae6d", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4590, + "serialized_chars": 4589, "estimated_input_tokens": 1148, "estimated_eval_tokens": 2552 }, @@ -21066,7 +21107,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4590, + "serialized_chars": 4589, "estimated_input_tokens": 1148, "estimated_eval_tokens": 2552 }, @@ -21075,18 +21116,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same xdist collision problem for captured_info artifacts; 45645 appears to extend the same core fix with CI-side log collection changes.", - "confidence": 0.82, + "summary": "Both PRs address the same xdist collision around `captured_info` debug artifacts in testing workflows. PR 45645 is the broader, more complete fix; PR 45639 is a narrower subset focused on the `testing_utils` path handling.", + "confidence": 0.83, "canonical_issue_reason": null, - "canonical_pr_reason": "45645 is the more complete change set: it includes the worker-safe captured_info file handling in testing_utils/tests and also updates CI to collect multiple captured_info*.txt files.", + "canonical_pr_reason": "PR 45645 is the better canonical PR because it covers the core `testing_utils` xdist-safe change plus the CI/workflow handling needed to preserve and aggregate the multiple `captured_info*.txt` outputs.", "best_issue_reason": null, - "best_pr_reason": "45645 is the strongest representative of the cluster because it fixes the underlying artifact collision and the downstream log-consumption path in one place.", + "best_pr_reason": "PR 45645 is the best overall PR since it appears to fully address the collision problem end-to-end, while PR 45639 only fixes the artifact write path in `testing_utils`.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Same underlying xdist captured_info collision bug; 45645 looks like a broader version of the same fix and could plausibly be merged together with 45639." + "reason": "Same underlying bug: xdist workers colliding on `captured_info` debug output. PR 45645 is a superset that includes the same core fix and related CI log handling, so they plausibly belong to one merged change." } ] }, @@ -21116,7 +21157,7 @@ "soft_pairs": [ "pull_request:45670|pull_request:46320" ], - "prepared_review_unit_hash": "0bbd9cd5617a087f15c694ff193eb762bc60110be487d720af172c2b36c0591a", + "prepared_review_unit_hash": "50ab93664d2dbec8de4a421b401a3fc39388b3032421f732db5c22ec44d1d146", "component_budget": { "node_count": 2, "item_count": 2, @@ -21138,18 +21179,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are unrelated: one is a small AutoModel registry/nit change for GLM ASR/Dia, while the other adds RecurrentGemma attention-dispatch support across model code, conversion, docs, and tests.", - "confidence": 0.97, + "summary": "The two PRs are unrelated: one adds a model class registration entry for glmasr in AutoModel, while the other implements RecurrentGemma attn_implementation dispatch and related model/conversion/test updates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46320 is the more substantial and self-contained code change, centered on a concrete RecurrentGemma implementation update rather than a minor model registry tweak.", + "canonical_pr_reason": "PR #46320 is the more substantive, self-contained change and a better representative of this cluster; PR #45670 is just a narrow AutoModel mapping tweak.", "best_issue_reason": null, - "best_pr_reason": "PR 46320 is the best representative of the cluster because it contains the broader, functional model-path change and associated validation work; PR 45670 is a narrow auto-mapping adjustment.", + "best_pr_reason": "PR #46320 is the best PR candidate because it fixes a concrete code path in the model implementation and includes supporting conversion and test updates, unlike the small registry-only tweak in #45670.", "soft_edge_verdicts": [ { "left": "pull_request:45670", "right": "pull_request:46320", "accept": false, - "reason": "Different models and different purposes: 45670 updates AutoModel mappings for GLM ASR/Dia, while 46320 adds attn_implementation dispatch for RecurrentGemma. They do not fix the same bug or change the same code path." + "reason": "Different models and different fixes: glmasr AutoModel registration versus RecurrentGemma attention-dispatch support. They do not look like the same underlying change and would not plausibly be merged as one PR." } ] }, @@ -21179,40 +21220,40 @@ "soft_pairs": [ "pull_request:45720|pull_request:46259" ], - "prepared_review_unit_hash": "457c4514bbf3620f648b9b2a603927a698e6ad7aba69b4f295ad2e1b61af10ab", + "prepared_review_unit_hash": "2b8a419a9051fc973d99f8d1da463126d8433e8ebe92da7b069513a3c6dabc91", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7188, - "estimated_input_tokens": 1797, - "estimated_eval_tokens": 3850 + "serialized_chars": 7189, + "estimated_input_tokens": 1798, + "estimated_eval_tokens": 3852 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7188, - "estimated_input_tokens": 1797, - "estimated_eval_tokens": 3850 + "serialized_chars": 7189, + "estimated_input_tokens": 1798, + "estimated_eval_tokens": 3852 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated modeling/linter PRs that only overlap superficially in repository area. PR #45720 implements TRF003 for tuple-return handling in several model forwards, while PR #46259 adds TRF018 for validating dead do_* flags in image/video processors.", + "summary": "Two open PRs are both linter/modeling-rule work, but they target different problems: TRF003 return-tuple handling in model forwards vs TRF018 dead do_* flag consumption in image/video processors. The shared model filenames are incidental overlap, not the same change.", "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: the two PRs target different rules and different code paths, so neither is a duplicate of the other.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "Neither PR is a good cluster representative for deduplication because they solve different underlying problems; if forced to pick one, #45720 is the more code-path-specific change, but it should not be treated as a duplicate of #46259.", + "best_pr_reason": "PR 46259 is the clearer representative because it introduces a new rule with accompanying documentation and a focused set of model fixes; however, it is not a duplicate of PR 45720.", "soft_edge_verdicts": [ { "left": "pull_request:45720", "right": "pull_request:46259", "accept": false, - "reason": "Different changes: TRF003 updates model forward return behavior across text/audio models, while TRF018 adds a rule about unused do_* preprocessing flags in image/video processors. Shared framework files do not indicate the same underlying bug or mergeable PR." + "reason": "Different underlying bugs/rules: 45720 implements TRF003 for can_return_tuple/return_dict behavior in model forwards, while 46259 adds TRF018 for unused do_* flags in image/video processor overrides. Overlapping files do not make the code-path problem the same." } ] }, @@ -21308,7 +21349,7 @@ "soft_pairs": [ "pull_request:45932|pull_request:46100" ], - "prepared_review_unit_hash": "d1c70742712534dae0604d31b46191cea192e5bb9cd834b9228554bfd619eaaa", + "prepared_review_unit_hash": "0ca5063a745ce6545de7620e7b4b9b99f50b9199e3c6e35b3d7979d18f511877", "component_budget": { "node_count": 2, "item_count": 2, @@ -21330,18 +21371,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are not duplicates: one is a broad, likely internal testing/Grafana-related change touching CI and several model files, while the other is a small test file edit with intentional failing assertions for migration/testing. They do not appear to fix the same concrete bug or implement the same change.", - "confidence": 0.95, + "summary": "These two PRs are unrelated: one is a broad model/config/workflow change with many files touched, while the other is a tiny ViT test-only PR adding intentionally failing assertions. They should not be clustered as duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45932 is the better canonical PR if one must be chosen because it contains the substantive code/config changes and broader implementation scope; PR 46100 is just a narrow test-only change with no sign of the same underlying fix.", + "canonical_pr_reason": "PR 45932 is the stronger canonical PR candidate because it contains the substantive code/config/workflow changes across multiple model files, whereas PR 46100 is only a test-only placeholder/migration change and does not represent the same underlying change.", "best_issue_reason": null, - "best_pr_reason": "PR 45932 is the best representative of the cluster because it is the more substantial change and has clearer code-path impact; PR 46100 is too trivial and unrelated in purpose to serve as the cluster representative.", + "best_pr_reason": "PR 45932 is the best representative of the cluster only in the sense that it is the more substantive artifact; PR 46100 is not suitable as a representative because it is a narrow test file edit with no overlap in bug/change intent.", "soft_edge_verdicts": [ { "left": "pull_request:45932", "right": "pull_request:46100", "accept": false, - "reason": "Reject: the PRs target different work. 45932 changes CI secrets and multiple model/config files, while 46100 only edits a ViT test with deliberate failing assertions. They do not appear to address the same underlying bug or merge into one PR." + "reason": "Different intent and code paths: 45932 touches multiple model implementations plus workflow/config files, while 46100 only edits a ViT test file with dummy failing assertions. They do not appear to fix the same underlying problem and would not plausibly merge into one PR." } ] }, @@ -21374,12 +21415,12 @@ "pull_request:46073|pull_request:46196", "pull_request:46154|pull_request:46196" ], - "prepared_review_unit_hash": "8383942c95e80f2182f3d155285faf4fcbaf8fe640593a0ec3b380d8a31b543e", + "prepared_review_unit_hash": "e6e08d8e0c23b8757b62fc8391a1d211944fec65b3c0fdc5481da2dd031d5826", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 8081, + "serialized_chars": 8082, "estimated_input_tokens": 2021, "estimated_eval_tokens": 4298 }, @@ -21387,7 +21428,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 8081, + "serialized_chars": 8082, "estimated_input_tokens": 2021, "estimated_eval_tokens": 4298 }, @@ -21396,30 +21437,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are related by the same SDPA-dispatch theme, but they target different model families and code paths, so they are not duplicates.", - "confidence": 0.98, + "summary": "These are related only at a high level: all add SDPA dispatch support, but each PR targets different model families and code paths, so they are not duplicates and should not be merged together.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: each change is model-specific (CTRL, T5, wav2vec2/SeamlessM4T) and fixes a different attention implementation path.", + "canonical_pr_reason": "PR 46196 is the best representative of the cluster because it covers the broadest set of model implementations and tests, making it the most general SDPA-dispatch change here.", "best_issue_reason": null, - "best_pr_reason": "46154 is the cleanest representative of the pattern: it is a focused T5 SDPA dispatch implementation with targeted tests, unlike the broader multi-model PRs.", + "best_pr_reason": "PR 46196 is the strongest standalone PR in this group due to its broader model coverage and more complete implementation surface compared with the single-model PRs.", "soft_edge_verdicts": [ { "left": "pull_request:46073", "right": "pull_request:46154", "accept": false, - "reason": "Same high-level SDPA theme, but CTRL and T5 use different architectures and attention code paths; not the same underlying change." + "reason": "Both are SDPA-dispatch PRs, but they fix different model-specific implementations (CTRL vs T5) and are not the same concrete code-path change." }, { "left": "pull_request:46073", "right": "pull_request:46196", "accept": false, - "reason": "CTRL vs wav2vec2/SeamlessM4T are distinct model families with separate dispatch implementations; not mergeable into one fix." + "reason": "They share the same theme, but target different architectures and files; this is not one mergeable bugfix/change." }, { "left": "pull_request:46154", "right": "pull_request:46196", "accept": false, - "reason": "Both add SDPA dispatch, but for different models and different concrete paths; they are related work, not duplicates." + "reason": "T5-only SDPA dispatch and the wav2vec2/SeamlessM4T SDPA dispatch work are separate model-family changes, not duplicates." } ] }, @@ -21449,12 +21490,12 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "4dcee48d5b99e151fdfd8806bbd77880d0bab31b89802084c7f087fb947b2b52", + "prepared_review_unit_hash": "af4a40adc6c6fcac91eb0a09264ab175f03c5be38d3bdf2cb94d71f8f4404d37", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5572, + "serialized_chars": 5571, "estimated_input_tokens": 1393, "estimated_eval_tokens": 3042 }, @@ -21462,7 +21503,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5572, + "serialized_chars": 5571, "estimated_input_tokens": 1393, "estimated_eval_tokens": 3042 }, @@ -21471,18 +21512,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs overlap in distributed code, but they are not the same change set: #46126 is a focused EP + Trainer integration feature, while #46269 is a broad distributed branch base touching many subsystems. The shared files alone are not enough to treat them as duplicates.", - "confidence": 0.88, + "summary": "These are not duplicates: one PR is a focused EP/Trainer integration on top of DistributedConfig, while the other is a broad distributed-branch base that bundles many unrelated distributed changes (FSDP, tensor parallel, integrations, MoE, CI). The shared filenames reflect subsystem overlap, not the same concrete change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "#46126 is the more concrete, narrowly scoped change: it targets EP routing plus Trainer integration and has a clear feature intent.", + "canonical_pr_reason": "PR #46126 is the more focused, concrete change: EP router masking plus Trainer/config integration on a specific distributed code path.", "best_issue_reason": null, - "best_pr_reason": "#46126 is the best representative of the cluster because it has a specific end-user feature scope, whereas #46269 is an umbrella distributed base PR rather than a single concrete fix.", + "best_pr_reason": "PR #46126 is the best representative PR for this cluster because it is narrowly scoped and clearly describes the shared EP/distributed functionality, whereas #46269 is an umbrella branch base with many unrelated edits.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Reject: the PRs overlap in distributed modules, but #46126 is a targeted EP/Trainer integration change and #46269 is a much broader distributed-base umbrella PR. They do not appear to fix the same concrete code-path problem and would not plausibly merge into one PR." + "reason": "Different underlying changes. #46126 targets EP router behavior and Trainer integration; #46269 is a broad distributed branch base touching many disparate areas, so they should not be merged as duplicates." } ] }, @@ -21512,7 +21553,7 @@ "soft_pairs": [ "pull_request:46152|pull_request:46207" ], - "prepared_review_unit_hash": "d9cd07de62e923d00034a376afa8e750edd6fa0738b87e91c2879fb9d6faa5bb", + "prepared_review_unit_hash": "d9addb867b6e4593c984b2011eddefbcd7dcb8b594b644f4250b94e5e9c8de6e", "component_budget": { "node_count": 2, "item_count": 2, @@ -21534,18 +21575,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs share a high-level theme of replacing direct file-open source inspection, but they address different code paths and different bugs: one in import utilities/resource resolution, the other in attention-implementation detection. They should not be clustered as duplicates.", - "confidence": 0.95, + "summary": "Both PRs replace file-open-based source access with safer source introspection, but they fix different code paths: one in import utilities, the other in attention-implementation detection. They are related in theme, not the same underlying change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46207 is the cleaner representative: it targets one concrete heuristic in `modeling_utils.py`, includes tests, and has a narrowly defined behavior change.", + "canonical_pr_reason": "PR 46207 is the cleaner representative because it has a well-scoped, concrete fix in a single behavioral path (`_can_set_attn_implementation`) and includes tests, making it the more specific duplicate-triage anchor.", "best_issue_reason": null, - "best_pr_reason": "PR #46207 is the best representative for the cluster because it is specific, self-contained, and has the clearest user-visible behavior and test coverage.", + "best_pr_reason": "PR 46207 is the best overall representative since it addresses a clearly defined bug pattern with an explicit code path and validation, whereas PR 46152 is a separate utility-level refactor/fix.", "soft_edge_verdicts": [ { "left": "pull_request:46152", "right": "pull_request:46207", "accept": false, - "reason": "Same broad motif (avoid `open()` on source files), but they modify different subsystems and solve different problems; they would not plausibly be merged into one PR." + "reason": "Both avoid reading source via `open()`, but they target different functions and different failure modes; they are not the same concrete bug or mergeable as one PR." } ] }, @@ -21580,12 +21621,12 @@ "pull_request:46362|pull_request:46368", "pull_request:46362|pull_request:46366" ], - "prepared_review_unit_hash": "9e36a3d27efe80e57c4b448578a09ee7f4d60914cfeec372135f3de1c22dbde5", + "prepared_review_unit_hash": "cdb7b6c2670d0fad6ff6807c3516ee02d1961c8fecc557a6328f73cf03100406", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 8226, + "serialized_chars": 8225, "estimated_input_tokens": 2057, "estimated_eval_tokens": 4370 }, @@ -21593,7 +21634,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 4, - "serialized_chars": 8226, + "serialized_chars": 8225, "estimated_input_tokens": 2057, "estimated_eval_tokens": 4370 }, @@ -21602,36 +21643,36 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs only share superficial test-maintenance similarity. They touch different model test files and fix different concrete problems; none are true duplicates.", - "confidence": 0.93, + "summary": "The cluster is mostly a set of small, model-specific test adjustments. Two PRs touch DeepSeek v4, but one skips quant-cache tests while the other extends a distributed test to XPU/CUDA; they do not fix the same concrete bug. The other PRs are unrelated model-specific skips in different files.", + "confidence": 0.83, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No single canonical PR stands out: the DeepSeek v4 PRs address different test failures, and the Cohere/Nemotron PRs are separate model-specific skips with no shared underlying change.", "best_issue_reason": null, - "best_pr_reason": "46362 is the most self-contained and concrete fix in the set, and it is the closest match to the only overlapping theme (a model-specific quant-cache test skip).", + "best_pr_reason": "pull_request:46362 is the clearest standalone change: it directly skips a single invalid DeepSeek v4 quant-cache test with a specific incompatibility explanation. But it is not a good cluster representative because the rest of the items are about different models or different failure modes.", "soft_edge_verdicts": [ { "left": "pull_request:46337", "right": "pull_request:46368", "accept": false, - "reason": "Different models and different fixes: Cohere2 MoE test threshold tuning versus NemotronH quant-cache test skipping." + "reason": "Both are model-test skips, but for different architectures and different cache behavior; they are not the same underlying bug or fix." }, { "left": "pull_request:46337", "right": "pull_request:46362", "accept": false, - "reason": "Unrelated changes: Cohere2 MoE accelerate/test-setup adjustments versus DeepSeek V4 skipping a quant-cache test." + "reason": "Cohere2 MoE offload/parallelism test adjustments are unrelated to DeepSeek v4 quant-cache incompatibility." }, { "left": "pull_request:46362", "right": "pull_request:46368", "accept": false, - "reason": "Both skip quant-cache tests, but they are model-specific incompatibilities in different architectures, not the same underlying bug or mergeable change." + "reason": "Both skip quant-cache tests, but the incompatibilities are model-specific (DeepSeek v4 vs NemotronH) and not mergeable as one PR." }, { "left": "pull_request:46362", "right": "pull_request:46366", "accept": false, - "reason": "Same file, but different concrete fixes: one skips a test, the other changes distributed worker backend selection for CUDA/XPU." + "reason": "DeepSeek v4 quant-cache skipping is unrelated to DeepSeek v4 distributed-worker backend selection for CUDA/XPU." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 9e134508955bd204b78509f7351db76155d658dd..ef6a82db7f1789276be611b3a4700a53ce27cada 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260603t180028z", + "analysis_id": "hybrid-model-20260604t000026z", "archived_artifacts": { - "hybrid": "snapshots/20260603T180028Z/analysis-runs/hybrid-model-20260603t180028z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260603T180028Z/analysis-runs/hybrid-model-20260603t180028z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-06-03T18:43:19Z", + "published_at": "2026-06-04T00:08:55Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260603T180028Z", + "snapshot_id": "20260604T000026Z", "variant": "hybrid" } diff --git a/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..b7be96a867e6d4449729f9e8cf68326db93bde07 --- /dev/null +++ b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json @@ -0,0 +1,2365 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260604T000026Z", + "generated_at": "2026-06-04T00:08:53Z", + "evidence_quality": "full", + "llm_enrichment": true, + "meta_bugs": [ + { + "cluster_id": "cluster-43979-11", + "summary": "Cluster of 1 issues and 10 PRs centered on issue #43979.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 43979, + "canonical_pr_number": 44007, + "issue_numbers": [ + 43979 + ], + "pr_numbers": [ + 43996, + 44007, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 43996, + "right_pr_number": 44007, + "code_similarity": 0.179, + "size_similarity": 0.576, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.429, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44013, + "code_similarity": 0.122, + "size_similarity": 0.318, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.392, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44044, + "code_similarity": 0.245, + "size_similarity": 0.864, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.479, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44066, + "code_similarity": 0.225, + "size_similarity": 0.818, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.408, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44072, + "code_similarity": 0.14, + "size_similarity": 0.303, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.528, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44085, + "code_similarity": 0.216, + "size_similarity": 0.783, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44129, + "code_similarity": 0.163, + "size_similarity": 0.643, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.229, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44154, + "code_similarity": 0.153, + "size_similarity": 0.535, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.31, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 43996, + "right_pr_number": 44722, + "code_similarity": 0.225, + "size_similarity": 0.848, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.368, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44013, + "code_similarity": 0.19, + "size_similarity": 0.553, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.531, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44044, + "code_similarity": 0.186, + "size_similarity": 0.667, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.354, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44066, + "code_similarity": 0.188, + "size_similarity": 0.704, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.315, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44072, + "code_similarity": 0.212, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.708, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44085, + "code_similarity": 0.195, + "size_similarity": 0.735, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.318, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44129, + "code_similarity": 0.103, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.191, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44154, + "code_similarity": 0.238, + "size_similarity": 0.93, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.344, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44007, + "right_pr_number": 44722, + "code_similarity": 0.178, + "size_similarity": 0.679, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.28, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44044, + "code_similarity": 0.126, + "size_similarity": 0.368, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.351, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44066, + "code_similarity": 0.127, + "size_similarity": 0.389, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.325, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44072, + "code_similarity": 0.29, + "size_similarity": 0.952, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.667, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44085, + "code_similarity": 0.131, + "size_similarity": 0.406, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.329, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44129, + "code_similarity": 0.07, + "size_similarity": 0.205, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.192, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44154, + "code_similarity": 0.177, + "size_similarity": 0.594, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.389, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44013, + "right_pr_number": 44722, + "code_similarity": 0.118, + "size_similarity": 0.375, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.287, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44066, + "code_similarity": 0.25, + "size_similarity": 0.947, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.404, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44072, + "code_similarity": 0.136, + "size_similarity": 0.351, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.442, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44085, + "code_similarity": 0.24, + "size_similarity": 0.906, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.394, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44129, + "code_similarity": 0.147, + "size_similarity": 0.555, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.243, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44154, + "code_similarity": 0.17, + "size_similarity": 0.62, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.306, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44044, + "right_pr_number": 44722, + "code_similarity": 0.257, + "size_similarity": 0.982, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.402, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44072, + "code_similarity": 0.133, + "size_similarity": 0.37, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.393, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44085, + "code_similarity": 0.763, + "size_similarity": 0.957, + "file_overlap": 0.5, + "area_overlap": 0.825, + "patch_similarity": 0.887, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ] + } + ] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44129, + "code_similarity": 0.145, + "size_similarity": 0.526, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.263, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44154, + "code_similarity": 0.174, + "size_similarity": 0.654, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.286, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44066, + "right_pr_number": 44722, + "code_similarity": 0.907, + "size_similarity": 0.964, + "file_overlap": 1.0, + "area_overlap": 0.808, + "patch_similarity": 0.874, + "shared_filenames": [ + "src/transformers/models/codegen/modeling_codegen.py", + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/codegen/modeling_codegen.py", + "left_ranges": [ + [ + 245, + 266 + ] + ], + "right_ranges": [ + [ + 228, + 234 + ] + ] + }, + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 41 + ], + [ + 174, + 181 + ], + [ + 250, + 257 + ], + [ + 398, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 466 + ], + [ + 471, + 488 + ], + [ + 518, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 606 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44085, + "code_similarity": 0.137, + "size_similarity": 0.387, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.398, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44129, + "code_similarity": 0.074, + "size_similarity": 0.195, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.231, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44154, + "code_similarity": 0.175, + "size_similarity": 0.566, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.414, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44072, + "right_pr_number": 44722, + "code_similarity": 0.124, + "size_similarity": 0.357, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.347, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44129, + "code_similarity": 0.141, + "size_similarity": 0.503, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.272, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44154, + "code_similarity": 0.18, + "size_similarity": 0.684, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.289, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44085, + "right_pr_number": 44722, + "code_similarity": 0.728, + "size_similarity": 0.923, + "file_overlap": 0.5, + "area_overlap": 0.791, + "patch_similarity": 0.78, + "shared_filenames": [ + "src/transformers/models/gptj/modeling_gptj.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gptj/modeling_gptj.py", + "left_ranges": [ + [ + 33, + 40 + ], + [ + 173, + 180 + ], + [ + 249, + 256 + ], + [ + 397, + 405 + ], + [ + 408, + 419 + ], + [ + 425, + 434 + ], + [ + 459, + 465 + ], + [ + 470, + 489 + ], + [ + 517, + 539 + ], + [ + 553, + 559 + ], + [ + 565, + 574 + ], + [ + 579, + 597 + ], + [ + 600, + 611 + ], + [ + 633, + 639 + ], + [ + 645, + 652 + ], + [ + 657, + 673 + ], + [ + 716, + 728 + ], + [ + 737, + 743 + ], + [ + 748, + 771 + ], + [ + 789, + 794 + ] + ], + "right_ranges": [ + [ + 33, + 42 + ], + [ + 175, + 182 + ], + [ + 251, + 258 + ], + [ + 399, + 420 + ], + [ + 426, + 435 + ], + [ + 460, + 467 + ], + [ + 472, + 489 + ], + [ + 519, + 540 + ], + [ + 554, + 560 + ], + [ + 566, + 575 + ], + [ + 580, + 598 + ], + [ + 601, + 612 + ], + [ + 634, + 640 + ], + [ + 646, + 653 + ], + [ + 658, + 663 + ], + [ + 666, + 674 + ], + [ + 717, + 722 + ], + [ + 738, + 744 + ], + [ + 749, + 772 + ], + [ + 790, + 795 + ] + ] + } + ] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44154, + "code_similarity": 0.099, + "size_similarity": 0.344, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.199, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44129, + "right_pr_number": 44722, + "code_similarity": 0.146, + "size_similarity": 0.545, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.247, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 44154, + "right_pr_number": 44722, + "code_similarity": 0.164, + "size_similarity": 0.631, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.255, + "shared_filenames": [], + "shared_file_areas": [] + } + ] + }, + { + "cluster_id": "cluster-41211-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41211.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41211, + "canonical_pr_number": 44339, + "issue_numbers": [ + 41211 + ], + "pr_numbers": [ + 41356, + 44339 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41356, + "right_pr_number": 44339, + "code_similarity": 0.155, + "size_similarity": 0.094, + "file_overlap": 0.273, + "area_overlap": 0.078, + "patch_similarity": 0.183, + "shared_filenames": [ + "docs/source/en/model_doc/deimv2.md", + "src/transformers/models/deimv2/__init__.py", + "src/transformers/models/deimv2/configuration_deimv2.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "tests/models/deimv2/__init__.py", + "tests/models/deimv2/test_modeling_deimv2.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/deimv2.md", + "left_ranges": [ + [ + 1, + 132 + ] + ], + "right_ranges": [ + [ + 1, + 65 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/__init__.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 29 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/configuration_deimv2.py", + "left_ranges": [ + [ + 1, + 74 + ] + ], + "right_ranges": [ + [ + 1, + 266 + ] + ] + }, + { + "filename": "tests/models/deimv2/test_modeling_deimv2.py", + "left_ranges": [ + [ + 1, + 15 + ] + ], + "right_ranges": [ + [ + 1, + 1734 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 + ], + "pr_numbers": [ + 43251, + 43254 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, + "shared_filenames": [ + "src/transformers/loss/loss_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/loss/loss_utils.py", + "left_ranges": [ + [ + 30, + 50 + ] + ], + "right_ranges": [ + [ + 30, + 43 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-41115-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #41115.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 41115, + "canonical_pr_number": 45613, + "issue_numbers": [ + 41115 + ], + "pr_numbers": [ + 41116, + 45613 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41116, + "right_pr_number": 45613, + "code_similarity": 0.797, + "size_similarity": 0.885, + "file_overlap": 0.75, + "area_overlap": 0.831, + "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": [ + [ + 730, + 737 + ] + ] + }, + { + "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": [ + [ + 720, + 726 + ], + [ + 1331, + 1337 + ] + ] + }, + { + "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-45363-3", + "summary": "Cluster of 3 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 45666, + "issue_numbers": [], + "pr_numbers": [ + 45363, + 45666, + 46339 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45363, + "right_pr_number": 45666, + "code_similarity": 0.392, + "size_similarity": 0.298, + "file_overlap": 0.571, + "area_overlap": 0.329, + "patch_similarity": 0.306, + "shared_filenames": [ + "fused_qwen_example.py", + "src/transformers/integrations/__init__.py", + "src/transformers/integrations/hub_kernels.py", + "src/transformers/modeling_utils.py" + ], + "shared_file_areas": [ + { + "filename": "fused_qwen_example.py", + "left_ranges": [ + [ + 1, + 30 + ] + ], + "right_ranges": [ + [ + 1, + 49 + ] + ] + }, + { + "filename": "src/transformers/integrations/__init__.py", + "left_ranges": [ + [ + 67, + 81 + ], + [ + 226, + 240 + ] + ], + "right_ranges": [ + [ + 67, + 73 + ], + [ + 222, + 228 + ] + ] + }, + { + "filename": "src/transformers/integrations/hub_kernels.py", + "left_ranges": [ + [ + 11, + 17 + ], + [ + 22, + 35 + ], + [ + 505, + 716 + ] + ], + "right_ranges": [ + [ + 11, + 53 + ], + [ + 286, + 295 + ], + [ + 521, + 1077 + ] + ] + }, + { + "filename": "src/transformers/modeling_utils.py", + "left_ranges": [ + [ + 3738, + 3747 + ] + ], + "right_ranges": [ + [ + 1330, + 1338 + ], + [ + 3763, + 3789 + ], + [ + 4223, + 4234 + ], + [ + 4628, + 4641 + ] + ] + } + ] + }, + { + "left_pr_number": 45363, + "right_pr_number": 46339, + "code_similarity": 0.504, + "size_similarity": 0.842, + "file_overlap": 0.286, + "area_overlap": 0.591, + "patch_similarity": 0.291, + "shared_filenames": [ + "src/transformers/integrations/hub_kernels.py", + "src/transformers/modeling_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/hub_kernels.py", + "left_ranges": [ + [ + 11, + 17 + ], + [ + 22, + 35 + ], + [ + 505, + 716 + ] + ], + "right_ranges": [ + [ + 14, + 51 + ], + [ + 284, + 293 + ], + [ + 520, + 744 + ] + ] + }, + { + "filename": "src/transformers/modeling_utils.py", + "left_ranges": [ + [ + 3738, + 3747 + ] + ], + "right_ranges": [ + [ + 1346, + 1354 + ], + [ + 3785, + 3811 + ], + [ + 4245, + 4257 + ], + [ + 4651, + 4664 + ] + ] + } + ] + }, + { + "left_pr_number": 45666, + "right_pr_number": 46339, + "code_similarity": 0.423, + "size_similarity": 0.354, + "file_overlap": 0.667, + "area_overlap": 0.283, + "patch_similarity": 0.356, + "shared_filenames": [ + "src/transformers/core_model_loading.py", + "src/transformers/integrations/hub_kernels.py", + "src/transformers/modeling_utils.py", + "tests/kernels/test_kernels.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/core_model_loading.py", + "left_ranges": [ + [ + 19, + 25 + ], + [ + 80, + 86 + ] + ], + "right_ranges": [ + [ + 19, + 25 + ], + [ + 80, + 86 + ] + ] + }, + { + "filename": "src/transformers/integrations/hub_kernels.py", + "left_ranges": [ + [ + 11, + 53 + ], + [ + 286, + 295 + ], + [ + 521, + 1077 + ] + ], + "right_ranges": [ + [ + 14, + 51 + ], + [ + 284, + 293 + ], + [ + 520, + 744 + ] + ] + }, + { + "filename": "src/transformers/modeling_utils.py", + "left_ranges": [ + [ + 1330, + 1338 + ], + [ + 3763, + 3789 + ], + [ + 4223, + 4234 + ], + [ + 4628, + 4641 + ] + ], + "right_ranges": [ + [ + 1346, + 1354 + ], + [ + 3785, + 3811 + ], + [ + 4245, + 4257 + ], + [ + 4651, + 4664 + ] + ] + }, + { + "filename": "tests/kernels/test_kernels.py", + "left_ranges": [ + [ + 19, + 54 + ], + [ + 228, + 281 + ], + [ + 294, + 317 + ], + [ + 605, + 995 + ] + ], + "right_ranges": [ + [ + 21, + 28 + ], + [ + 33, + 39 + ], + [ + 90, + 103 + ], + [ + 217, + 301 + ], + [ + 314, + 337 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45561-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45561, + "canonical_pr_number": 45645, + "issue_numbers": [ + 45561 + ], + "pr_numbers": [ + 45639, + 45645 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45639, + "right_pr_number": 45645, + "code_similarity": 0.494, + "size_similarity": 0.619, + "file_overlap": 0.5, + "area_overlap": 0.434, + "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": [ + [ + 3595, + 3602 + ], + [ + 3819, + 3853 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 86 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + } + ] + } + ] + }, + { + "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-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-45363-3", + "canonical_pr_number": 45666, + "duplicate_pr_numbers": [ + 45363, + 46339 + ], + "target_issue_number": null, + "reason": "PRs in cluster-45363-3 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45561-3", + "canonical_pr_number": 45645, + "duplicate_pr_numbers": [ + 45639 + ], + "target_issue_number": 45561, + "reason": "PRs in cluster-45561-3 are treated as duplicates because they converge on issue #45561 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-45647-3", + "canonical_pr_number": 45738, + "duplicate_pr_numbers": [ + 45737 + ], + "target_issue_number": 45647, + "reason": "PRs in cluster-45647-3 are treated as duplicates because they converge on issue #45647 with closing_reference, shared_issue_target evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-41211-3", + "issue_number": 41211, + "reason": "Issue #41211 is the strongest global issue candidate because it is open, belongs to a cluster with 3 artifacts, and carries 6 discussion comments plus 6 inbound references.", + "score": 148.0 + }, + "best_pr": { + "cluster_id": "cluster-43979-11", + "pr_number": 44007, + "reason": "PR #44007 is the strongest global PR candidate because it is open, belongs to a cluster with 11 artifacts, links to 1 issues, and carries 2 review/discussion events.", + "score": 238.886 + } +} diff --git a/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..1f2c90c2172b4989be9f7d13e7e5de093368c16b --- /dev/null +++ b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,21687 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260604T000026Z", + "generated_at": "2026-06-04T00:08:53Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 1, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:24643", + "issue:30277", + "issue:34634", + "issue:34928", + "issue:35545", + "issue:36010", + "issue:36331", + "issue:39290", + "issue:41720", + "issue:41762", + "issue:42915", + "issue:43039", + "issue:43296", + "issue:43425", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43827", + "issue:43828", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44589", + "issue:44805", + "issue:44918", + "issue:45081", + "issue:45161", + "issue:45237", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:45874", + "issue:45901", + "issue:45902", + "issue:46015", + "issue:46165", + "issue:46294", + "issue:46304" + ], + "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:45901|issue:46015", + "issue:44509|issue:44512", + "issue:35545|issue:36010", + "issue:44918|issue:46165", + "issue:44805|issue:45823", + "issue:43638|issue:44805", + "issue:43638|issue:46165", + "issue:43596|issue:44805", + "issue:44805|issue:46294", + "issue:44805|issue:45161", + "issue:44560|issue:46165", + "issue:43596|issue:46165", + "issue:43572|issue:46165", + "issue:34634|issue:35545", + "issue:30277|issue:43638", + "issue:43596|issue:43638", + "issue:43039|issue:45161", + "issue:43975|issue:44457", + "issue:44292|issue:44457", + "issue:34928|issue:44805", + "issue:24643|issue:30277", + "issue:45859|issue:45902", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:45874|issue:46294", + "issue:39290|issue:45081", + "issue:45823|issue:45874", + "issue:43425|issue:44292", + "issue:34928|issue:41720", + "issue:43828|issue:45161", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:43716|issue:45685", + "issue:44918|issue:45161", + "issue:45507|issue:45685", + "issue:44292|issue:44918", + "issue:43716|issue:44292", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43716|issue:44560", + "issue:43828|issue:45685", + "issue:44918|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:45237|issue:46015", + "issue:43039|issue:44805", + "issue:45800|issue:46304", + "issue:43296|issue:46304", + "issue:41762|issue:43596", + "issue:39290|issue:41720", + "issue:44589|issue:46304", + "issue:43901|issue:44512", + "issue:45823|issue:46294", + "issue:45237|issue:45901" + ], + "prepared_review_unit_hash": "61dc05edbad2a00ea7b6d6364739455596e81ef987b902e44894774533998ea1", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 67, + "serialized_chars": 47823, + "estimated_input_tokens": 11956, + "estimated_eval_tokens": 24168 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous bug reports across unrelated subsystems. The only clear duplicate subgroup is the ZeRO-3 \"index 0 is out of bounds\" loading failures around BERT/Gemma3.", + "confidence": 0.84, + "canonical_issue_reason": "issue:43596 is the clearest representative of the shared DeepSpeed ZeRO-3 loading failure; issue:43638 is a near-duplicate reproduction, and issue:41762 is close enough to the same empty-tensor/indexing path.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43596 is the best single anchor for this cluster because it names the concrete failure mode and is the most generic of the near-duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different failures: Qwen3.5 config attribute error vs streaming chat/completions bug." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated hardware/backend issues: AMD GPU failure vs MPS histogram int limitation." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different code paths: image preprocessor dtype mismatch vs AMD GPU runtime failure." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Autocast dtype mismatch is not the same bug as the AMD GPU assertion failure." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different model/backend problem; not the same underlying failure." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Qwen-3-8B NVFP4 runtime error is distinct from the AMD GPU issue." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch and ONNX export error are unrelated." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ONNX export failure vs FineGrainedFP8Config failure; different bugs." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Export-time ONNX issue and auto device mapping cuda assert are different problems." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Table QA crash and outdated RAG examples are not the same issue." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are stale docs, but they refer to different removed v5 documentation items." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "Model export error vs GenerationMixin import error; different code paths." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "TRL SFT embeddings issue is unrelated to resuming training with trust_remote_code." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask shape mismatch and Gemma4 PLE device mismatch are different failures." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both are IndexErrors, but the reported shapes and loading paths differ enough to avoid merging." + }, + { + "left": "issue:43638", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3 index error is unrelated to resuming training with a local model." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Different symptom and code path: ZeRO-3 empty tensor indexing vs mask/tensor shape mismatch." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Mask indexing error and Qwen3.5 video token mismatch are unrelated." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Different bug classes: tensor mask mismatch vs GPT-OSS tensor-parallel issue." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Video StopIteration bug is unrelated to the training resume problem." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3 Bert loading failure and resume-training issue are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:46165", + "accept": false, + "reason": "Missing pad_token_idx in config is unrelated to model-resume behavior." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice preset bug is unrelated to ONNX export." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Collective mismatch and ZeRO-3 index error are different distributed-training bugs." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same underlying ZeRO-3 loading/index-out-of-bounds failure; 43638 looks like another reproduction on a Bert variant." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross_entropy call is unrelated to GPT-OSS MoE tensor-parallel failure." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "Detokenization bug and LoRA merge/save inconsistency are different issues." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen NVFP4 runtime error is unrelated to LoRA merge/save mismatch." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Activation checkpointing/FSDP recompute mismatch is not the same as the mask shape error." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "DeepSpeed weight-dimension runtime error and collective mismatch are distinct failures." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Same model family, but different concrete bugs: CPU offload incompatibility vs uninitialized inv_freq after materialization." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 config attribute missing and Mistral regex patch crash are unrelated." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Device_map auto PLE device mismatch is not the same as reverse embedding lookup overhead." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen NVFP4 runtime failure are different problems." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS int histogram limitation are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43716", + "accept": false, + "reason": "Mixtral grouped_mm tracing error and Mistral-3 dtype mismatch are distinct." + }, + { + "left": "issue:43828", + "right": "issue:44292", + "accept": false, + "reason": "Different failures: autocast dtype mismatch vs Qwen NVFP4 runtime issue." + }, + { + "left": "issue:43716", + "right": "issue:45685", + "accept": false, + "reason": "Mistral-3 dtype mismatch is unrelated to the MPS histogram limitation." + }, + { + "left": "issue:44918", + "right": "issue:45161", + "accept": false, + "reason": "TRL input-embedding unpacking bug is unrelated to GPT-OSS MoE tensor-parallel failure." + }, + { + "left": "issue:45507", + "right": "issue:45685", + "accept": false, + "reason": "GraniteMoEHybrid invalid method call and MPS histogram issue are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions, but they cover different removed v5 content and should not be merged." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "dtype mismatch in Mistral-3 is unrelated to the Qwen3-vl video StopIteration bug." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram int issue are different bugs." + }, + { + "left": "issue:44918", + "right": "issue:45685", + "accept": false, + "reason": "SFT trainer embedding issue is unrelated to the MPS backend error." + }, + { + "left": "issue:43039", + "right": "issue:43828", + "accept": false, + "reason": "Liger kernel cross_entropy call and autocast dtype mismatch are different issues." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and compute_loss signature error are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Activation checkpointing/FSDP mismatch is unrelated to GenerationMixin import failure." + }, + { + "left": "issue:45237", + "right": "issue:46015", + "accept": false, + "reason": "AMD GPU runtime issue and outdated RAG examples are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger kernel cross_entropy usage bug is unrelated to the mask/tensor shape error." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Torch 2.4.1 compatibility issue and USE_DISTRIBUTED=0 breakage are distinct versioning problems." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "PaddleOCR-VL/vLLM load failure and distributed-off breakage are different issues." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": true, + "reason": "Both report the same ZeRO-3 index-out-of-bounds loading failure with size-0 tensors; 41762 is another model-specific reproduction." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Different documentation regressions about different removed v5 features." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3TextConfig missing attribute and Qwen3 auto-device-map cuda assert are unrelated." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Float8 storage lookup error and USE_DISTRIBUTED=0 support breakage are different problems." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "TextClassificationPipeline docs and transformers run-command docs are separate stale-doc issues." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Gemma4 device mismatch and Qwen3.5 video token mismatch are unrelated." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "AMD GPU failure and table-question-answering crash are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays conservative. The ZeRO-3 zero-sized tensor / index-out-of-bounds subgroup is the only clearly coherent duplicate set, and the accepted soft-edge pairs (43596\u219443638, 41762\u219443596) are well supported by the issue titles. The remaining non-duplicate verdicts are generally cautious and do not overstate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 2, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36010", + "issue:39290", + "issue:41093", + "issue:41762", + "issue:43039", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43792", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43872", + "issue:43975", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44479", + "issue:44560", + "issue:44589", + "issue:44805", + "issue:44841", + "issue:44863", + "issue:44918", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45084", + "issue:45161", + "issue:45362", + "issue:45464", + "issue:45685", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45901", + "issue:45920", + "issue:46015", + "issue:46294", + "issue:46304" + ], + "soft_pairs": [ + "issue:44841|issue:45823", + "issue:43039|issue:44560", + "issue:43975|issue:46015", + "issue:45812|issue:45920", + "issue:44863|issue:46015", + "issue:45020|issue:46304", + "issue:43866|issue:44863", + "issue:43872|issue:46304", + "issue:44292|issue:46015", + "issue:39290|issue:43531", + "issue:43854|issue:43866", + "issue:41093|issue:43638", + "issue:43792|issue:45901", + "issue:43425|issue:45070", + "issue:43596|issue:45800", + "issue:43975|issue:45901", + "issue:43792|issue:45070", + "issue:45362|issue:45464", + "issue:44292|issue:45901", + "issue:39290|issue:45823", + "issue:43404|issue:43866", + "issue:45464|issue:45685", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:44841|issue:45084", + "issue:36010|issue:39290", + "issue:45800|issue:45823", + "issue:41093|issue:41762", + "issue:43296|issue:43366", + "issue:43572|issue:45070", + "issue:44387|issue:45005", + "issue:44805|issue:44918", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:44291|issue:45070", + "issue:43366|issue:45464", + "issue:43531|issue:45070", + "issue:43716|issue:46294", + "issue:45020|issue:45800", + "issue:43541|issue:44589", + "issue:44479|issue:46294", + "issue:41093|issue:43596" + ], + "prepared_review_unit_hash": "6ab904eb1c28d9efa5315d0f7c04bbbcdee754913769e20a07a93bb1d448bbe3", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 42, + "serialized_chars": 39565, + "estimated_input_tokens": 9892, + "estimated_eval_tokens": 20040 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a set of several small duplicate subgroups rather than one unified bug. The clearest grouping is the AutoTokenizer regression family, with issue:45920 as the broad representative; the video-input regression pair (issue:44479 / issue:46294) and the init-empty-weights TypeError family also look coherent. Most other soft matches are only superficial, and issue:41093 vs issue:43638 should not be accepted without stronger shared evidence.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45920 is the best canonical issue for the tokenizer-regression subgroup: it covers multiple affected model families, describes the broadest concrete symptom, and has inbound references that make it the most representative report in this set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 is the strongest overall issue candidate because it captures a broad, concrete regression affecting several model families and is more representative than the narrower single-model reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Different model families and different failure modes; both are multimodal, but not the same bug." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Unrelated topics: Liger Kernel cross-entropy call vs Qwen video-template failure." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Different subsystems: detokenization bug vs outdated RAG docs/examples." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Same underlying AutoTokenizer wrong-token-IDs regression; issue:45812 is a narrower Granite instance of the broader multi-model issue." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "Checkpoint-loading bug vs documentation issue; no shared code-path." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Both are broad breakages, but they point to different compatibility problems and different code paths." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Both are load failures, but for different models and likely unrelated causes." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "The _is_hf_initialized TypeError is a specific init-empty-weights/bitsandbytes compatibility issue, not the USE_DISTRIBUTED=0 breakage." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "Unrelated: model-loading TypeError vs docs/examples issue." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding_window, but they affect different models and likely different root causes." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Different models and different load failures; not the same bug." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Both are ZeRO-3/BERT-related indexing failures, but the signatures differ (mask-shape mismatch vs out-of-bounds with size 0) and there is no shared issue text or code evidence here." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Different tasks and failures: Whisper model loading vs table-question-answering crash." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Tensor/torch version incompatibility vs pydantic PretrainedConfig regression; unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "Different issues: ZeRO-3/BERT index error vs torch/transformers version incompatibility." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Detokenization bug is unrelated to table-question-answering crashes." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper load failure and PretrainedConfig pydantic regression are different problems." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Same model family, but one is transformers chat and the other is vLLM streaming API; not enough evidence of one underlying bug." + }, + { + "left": "issue:44292", + "right": "issue:45901", + "accept": false, + "reason": "Different failures and code paths; no shared underlying bug." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Unrelated model/config issues; only superficial similarity in being config-related." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Different model families and symptoms; not a duplicate." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Chat/streaming failure vs MPS histogram-on-int error; unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Different failures in different execution paths: torch dynamo grouped_mm vs MPS histogram." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Both involve dtype-ish symptoms, but they are different models and distinct code paths." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Different subsystems: processor loading vs compile/template-node error." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to the Gemma3 sliding_window config issue." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Version incompatibility vs model-device mismatch; not the same bug." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are DeepSpeed-related indexing failures, but the error signatures and affected paths differ." + }, + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "Different model load/use cases: PaddleOCR-VL load failure vs GGUF gpt-oss support request." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is separate from the PretrainedConfig pydantic regression." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Int4 OOM regression and tied-weights issues are different problems." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Similar words around shape errors, but these are different failures in different training/loading flows." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Different model families and unrelated MoE/TP behavior." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Autocast dtype mismatch and chat-streaming API failure are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument is a model-loading compatibility bug, not the pydantic PretrainedConfig regression." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF gpt-oss support request vs Qwen3.5 streaming failure; unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Sliding_window model config issue is unrelated to the pydantic PretrainedConfig regression." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Different multimodal models and different symptoms; not enough evidence of a shared root cause." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Both mention compatibility regressions, but they target different version conflicts and code paths." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Grouped_mm tracing error and Float8 storage lookup error are different runtime failures." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": true, + "reason": "Same Qwen video-input regression family; issue:46294 is a specific manifestation of the broader v5.3.0 video token/features mismatch." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are indexing errors, but one is a mask-shape mismatch and the other is a zero-sized tensor path; not clearly the same bug." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "Most of the clustering is conservative and grounded, but one soft-edge verdict overstates the evidence: issue:41093 vs issue:43638 are both ZeRO-3/BERT-related indexing failures, yet the titles describe different signatures (mask-shape mismatch vs out-of-bounds with size 0). With no shared issue text or code evidence in the packet, treating them as the same underlying bug is too strong. The rest of the summary looks reasonable, but this pair should stay tentative or negative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 3, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36331", + "issue:38175", + "issue:39186", + "issue:39692", + "issue:40990", + "issue:41553", + "issue:41762", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43404", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43866", + "issue:43872", + "issue:43873", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:43994", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44403", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44534", + "issue:44589", + "issue:44863", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:45005", + "issue:45399", + "issue:45507", + "issue:45542", + "issue:45632", + "issue:45663", + "issue:45698", + "issue:45701", + "issue:45759", + "issue:45800", + "issue:45859", + "issue:45902" + ], + "soft_pairs": [ + "issue:43054|issue:43994", + "issue:45759|issue:45902", + "issue:45542|issue:45800", + "issue:39186|issue:44918", + "issue:43716|issue:44918", + "issue:40990|issue:44960", + "issue:45632|issue:45698", + "issue:43645|issue:45399", + "issue:43404|issue:45005", + "issue:42915|issue:45399", + "issue:44530|issue:45005", + "issue:43873|issue:45005", + "issue:44863|issue:45005", + "issue:44360|issue:44512", + "issue:44918|issue:45507", + "issue:44492|issue:44512", + "issue:43975|issue:44292", + "issue:44292|issue:45507", + "issue:43950|issue:44534", + "issue:39692|issue:43054", + "issue:43644|issue:43950", + "issue:43828|issue:44928", + "issue:36331|issue:41762", + "issue:43872|issue:44589", + "issue:43950|issue:45902", + "issue:43866|issue:44403", + "issue:43054|issue:45701", + "issue:38175|issue:43054", + "issue:41553|issue:43883", + "issue:45663|issue:45902", + "issue:44910|issue:45859", + "issue:42947|issue:44387" + ], + "prepared_review_unit_hash": "09765f6647f8495553c7c9393aa3466cf3f174e5ed882ba7712e08da394200be", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37146, + "estimated_input_tokens": 9287, + "estimated_eval_tokens": 18830 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The strongest duplicate cluster is the Transformers v5 non-persistent-buffer regression around 43950/43644/44534. The SigLIP2 pair looks like a plausible but tentative duplicate based on titles alone. Most other candidate pairs are unrelated.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 43950 is the clearest anchor for the non-persistent-buffer regression cluster because its title explicitly names the v5 regression, and it matches the related buffer-corruption reports.", + "canonical_pr_reason": null, + "best_issue_reason": "43950 is the best representative issue overall because it most explicitly states the underlying v5 non-persistent-buffer regression and aligns with the related buffer-corruption reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43054", + "right": "issue:43994", + "accept": true, + "reason": "Tentative duplicate: both titles describe SigLIP2 base-patch16-224 producing bad or nonsensical outputs, suggesting the same underlying model issue, though this is title-level evidence only." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "Different bugs: composite multimodal config unwrapping versus Qwen3_5MoeTextRotaryEmbedding uninitialized-memory behavior after materialization." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "Unrelated failure modes: TensorBoard/TensorFlow backend detection versus a torch/transformers version incompatibility." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "Different concrete code paths and symptoms: FSDP weight-shape error versus TRL input-embedding unpacking failure." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Different model families and failures: Mistral-3 dtype mismatch versus Qwen3.5 trainer/unpacking behavior." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "No shared underlying bug is evident from the titles; they concern GPT-OSS perplexity and GLM5 separately." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": false, + "reason": "Both involve loading custom code, but the concrete bugs differ: cache-path collision versus wrong custom module loaded after save/load." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Unrelated: Jupyter custom-model initialization issue versus flash-attn2 fallback gating." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but one is a specific Mistral3 lm_head-tie bug while the other is a broader translation-model v5 report." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Different issues: FP8 configuration failure versus flash-attn2 fallback checks." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash is unrelated to tied-weight behavior." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior is not the same bug as tied weights in translation models." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is unrelated to tied-weight issues." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "Different areas: DSA indexer behavior versus documentation about a removed command." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Different concrete bugs: TRL unpacking failure versus GraniteMoEHybrid invalid-method call." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A cache-strategy typo is not the same as outdated run-command documentation." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Detokenization regression and Qwen-3-8B-NVFP4 runtime error are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:45507", + "accept": false, + "reason": "Different models and failure modes; no shared code-path bug is evident." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same Transformers 5.x non-persistent-buffer corruption regression, described from slightly different angles." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Docs-example errors and model-quality regression are not the same underlying bug." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same Transformers 5.x non-persistent-buffer corruption regression." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Both mention dtype/position-related issues in Qwen-family models, but the actual failures are distinct." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Trainer signature mismatch and ZeRO-3 loading IndexError are unrelated." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Different incompatibilities: bitsandbytes constructor breakage versus missing Float8 storage type." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": false, + "reason": "Both involve loading/initialization, but one is the broad non-persistent-buffer regression and the other is a specific Qwen rotary-embedding memory bug." + }, + { + "left": "issue:43866", + "right": "issue:44403", + "accept": false, + "reason": "Checkpoint corruption and loading noise are not the same bug." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "SigLIP2 output-quality regression is unrelated to tokenization changes in a later transformers version." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both are SigLIP/SigLIP2-related, but one is zero probabilities and the other is degraded text embeddings; not the same bug." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Different model-loading problems: bad AutoTokenizer error for Voxtral versus missing all_tied_weights_keys on MolmoForCausalLM." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states rebuilding and Qwen rotary-embedding uninitialized memory are unrelated." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Both involve Qwen3.5 and position/rotary behavior, but one is FlashAttention position_ids handling and the other is CPU offload incompatibility." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Gradient checkpointing with PEFT LoRA is unrelated to int4 quantization memory growth/OOM." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The output is mostly conservative, but one key reason overstates what is supported by the packet: `best_issue_reason` claims 43950 \u201chas a precise reproduction,\u201d yet the packet does not include issue bodies, only titles and metadata. That makes the claim unsupported. The SigLIP2 pair can be treated as a tentative duplicate, but it should stay explicitly tentative rather than implying stronger evidence than the packet provides." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 4, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36331", + "issue:41093", + "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:43828", + "issue:43856", + "issue:43866", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43940", + "issue:43950", + "issue:43975", + "issue:44220", + "issue:44291", + "issue:44368", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44509", + "issue:44661", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45237", + "issue:45305", + "issue:45313", + "issue:45356", + "issue:45399", + "issue:45685", + "issue:45715", + "issue:45823", + "issue:45923", + "issue:46315" + ], + "soft_pairs": [ + "issue:43054|issue:43493", + "issue:44403|issue:44451", + "issue:41553|issue:43881", + "issue:43975|issue:45237", + "issue:44403|issue:44488", + "issue:43643|issue:45020", + "issue:42947|issue:43856", + "issue:41762|issue:42915", + "issue:43782|issue:44960", + "issue:45685|issue:45823", + "issue:44991|issue:45399", + "issue:43824|issue:43872", + "issue:43881|issue:43883", + "issue:42491|issue:44368", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:43828|issue:45823", + "issue:45715|issue:46315", + "issue:45081|issue:45399", + "issue:45020|issue:45356", + "issue:43742|issue:45399", + "issue:43404|issue:43950", + "issue:43827|issue:44509", + "issue:43257|issue:43866", + "issue:43872|issue:44291", + "issue:42915|issue:43872", + "issue:43792|issue:44220", + "issue:43257|issue:45305", + "issue:36331|issue:41093", + "issue:43940|issue:45313", + "issue:42947|issue:45923" + ], + "prepared_review_unit_hash": "e0f72825791ccfe2607c306a615d230d84b9eefdf17e2788697cb44a2ec849e9", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37209, + "estimated_input_tokens": 9303, + "estimated_eval_tokens": 18862 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous, with a few clear duplicate subgroups: SigLIP2 implementation mismatch, docs still referencing removed pipeline tasks, the `_is_hf_initialized`/`init_empty_weights` compatibility regression, and a Qwen3 ZeRO-3 weight-loading failure. No single issue cleanly represents the whole set.", + "confidence": 0.69, + "canonical_issue_reason": "The cluster is too mixed for one true global canonical; 43872 is the clearest concrete regression anchor because 44291 describes the same `_is_hf_initialized`/`init_empty_weights` failure.", + "canonical_pr_reason": null, + "best_issue_reason": "43872 is the strongest representative issue: precise, model-agnostic, and backed by an almost identical duplicate (44291).", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43054", + "right": "issue:43493", + "accept": true, + "reason": "Same SigLIP2 problem: HF implementation diverges from the original JAX version and yields degraded embeddings." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Unrelated: generic loading noise vs a specific model-loading failure." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Different models and different failure modes; only both are load errors." + }, + { + "left": "issue:43975", + "right": "issue:45237", + "accept": false, + "reason": "Completely different areas: tokenizer detokenization vs AMD GPU support." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Not the same bug; one is a generic noise issue, the other is a model-specific load failure." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "45020 is a broad remote_code regression umbrella, while 43643 is a specific missing-fields config bug; not the same concrete issue." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Different training problems: gradient checkpointing behavior vs MoE memory usage." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different model families and different ZeRO/FP8 failure paths." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different model-loading bugs with no clear shared code-path." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "Different device/backend issues; no shared underlying bug." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attn fallback gating; unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Import/export symbol issue vs `_is_hf_initialized` constructor incompatibility." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Different model classes and different attribute/load failures." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "One is a LoRA compatibility problem across branches; the other is a warning about tie_word_embeddings." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Broad remote_code regressions are not the same as a Gemma4 device-map forward mismatch." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "Tokenizer mapping edge case vs remote_code regression; different concrete bugs." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Different runtime errors: autocast dtype mismatch vs device placement mismatch." + }, + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer performance regressions, but they affect different conversion APIs and do not look like one bug." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer regex crash vs flash-attn fallback logic; unrelated." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Remote_code regressions and Kimi tokenizer codec issues are different failures." + }, + { + "left": "issue:43742", + "right": "issue:45399", + "accept": false, + "reason": "Model load key error vs flash-attn fallback gating; not the same issue." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "Different code paths: tied weights on Mistral3 vs non-persistent buffer corruption in from_pretrained()." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Same docs regression: v5 removed pipeline tasks, but the docs still reference them." + }, + { + "left": "issue:43257", + "right": "issue:43866", + "accept": false, + "reason": "Different problems: Qwen3 MoE loading/conversion vs a corrupted Ovis2 checkpoint." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Same `_is_hf_initialized` compatibility regression when loading with `init_empty_weights`." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "Different bugs: Qwen3 MoE FP8 failure vs bitsandbytes constructor incompatibility." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "The Whisper issue is too broad to \u092a\u0941\u0937\u094d\u091f\u093f this as the same `_torch_extract_fbank_features()` bug from the title alone." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Different DeepSpeed-related issues: weight conversion on load vs gradient averaging with GAS." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Completely unrelated training API TypeError vs tensor-mask shape IndexError." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": true, + "reason": "Same underlying Qwen3-family ZeRO-3 weight-loading failure; both describe missing weights under DeepSpeed ZeRO-3." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Related theme, but not the same concrete bug: one is PEFT LoRA checkpointing effectiveness, the other is a missing flag on a trust_remote_code variant." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The selected canonical anchor (43872) is supported by the near-duplicate 44291, and the soft-edge accepts that were marked true are plausible from the titles alone. No obvious overstatement or mergeability issue." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 5, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41093", + "issue:41553", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43645", + "issue:43646", + "issue:43824", + "issue:43828", + "issue:43854", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44479", + "issue:44488", + "issue:44530", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:44991", + "issue:45030", + "issue:45325", + "issue:45584", + "issue:45588", + "issue:45663", + "issue:45685", + "issue:45823", + "issue:45834", + "issue:45902" + ], + "soft_pairs": [ + "issue:43854|issue:44863", + "issue:41553|issue:44991", + "issue:44928|issue:45588", + "issue:41093|issue:44805", + "issue:44479|issue:44560", + "issue:44530|issue:44863", + "issue:41553|issue:44488", + "issue:24643|issue:30064", + "issue:45325|issue:45902", + "issue:43366|issue:43828", + "issue:45663|issue:45823", + "issue:43366|issue:43531", + "issue:44155|issue:45584", + "issue:43950|issue:45834", + "issue:44560|issue:45685", + "issue:43824|issue:43957", + "issue:43550|issue:44910", + "issue:43645|issue:43646", + "issue:42915|issue:43957", + "issue:43366|issue:44589", + "issue:44560|issue:44918", + "issue:42617|issue:43366", + "issue:36010|issue:42915", + "issue:44661|issue:45823", + "issue:44560|issue:44805", + "issue:43632|issue:45834", + "issue:38175|issue:43994", + "issue:44877|issue:45030", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:44661|issue:44805", + "issue:43317|issue:44756", + "issue:38175|issue:39692" + ], + "prepared_review_unit_hash": "9a42506daa19e7c38fe81d37b6448c8364c2d33893d497ac6a25ae30acea000c", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37729, + "estimated_input_tokens": 9433, + "estimated_eval_tokens": 19122 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous issue cluster, but there are a few likely duplicate subpairs: custom model init in Transformers v5, SigLIP2 bad outputs, mask-shape IndexError, and a Qwen video-input regression.", + "confidence": 0.67, + "canonical_issue_reason": "Issue 43646 is the broadest, clearest representative of the custom-model initialization regression; 43645 is the notebook-specific variant of the same underlying bug.", + "canonical_pr_reason": null, + "best_issue_reason": "43646 is the best single issue to anchor the cluster because it is the most general formulation of one of the strongest duplicate subclusters.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for unrelated model families; same broad symptom, but not the same bug." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-load problems, but they affect different models and likely different code paths." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Different failures: RLHF numerical explosion vs flash-attention AttributeError on sink-less models." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "Same underlying mask/index shape-mismatch IndexError; 44805 looks like a later generic report of the same bug." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": true, + "reason": "Both describe the same Qwen video-input regression in transformers 5.3.0, with 44560 as a concrete variant." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "Unrelated failures: Qwen3.5 PagedAttentionCache crash vs NemotronH checkpoint loading." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Separate tokenizer-loading issues for different models; no clear shared root cause." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "Completely different subsystems and errors: DeepSpeed training vs image processor segmentation maps." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "Different bugs in different model components: Qwen2.5-VL rope indexing vs Qwen3.5-MoE meta materialization." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "Unrelated: GGUF gpt-oss support vs Phi MoE autocast dtype mismatch." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4 issues, but one is an FSDP2 KeyError and the other is a device-map forward mismatch." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "Different problems in different model families: GGUF architecture support vs Qwen3-MoE sliding window behavior." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "Different audio/Whisper bugs with no shared code-path indicated." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Different regressions: non-persistent buffer corruption vs long OCR index error." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Video-input regression vs an MPS histogram integer support error; not the same bug." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Both are load-time failures, but one is a missing import and the other is meta-device loading behavior." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different attention/kernel failures: torch.compile+SDPA vs Flash Attention misreading position IDs." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same custom-model initialization regression; 43645 is just the Jupyter-notebook manifestation of 43646." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Different issues: FineGrainedFP8Config failure vs meta-device model loading." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "Unrelated: GGUF model architecture support vs missing Float8 storage type." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Both mention Qwen3.5/video-adjacent behavior, but one is video loading and the other is SFT embedding unpacking." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "Different topics entirely: 3d_parallel.py runtime vs GGUF gpt-oss support." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Different import-path regression vs FP8 configuration failure." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Different config/mapping logic: tokenizer mapping edge case vs Gemma4 device mismatch." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different errors and subsystems: video StopIteration vs mask/index shape mismatch." + }, + { + "left": "issue:43632", + "right": "issue:45030", + "accept": false, + "reason": "Different config/init regressions with no strong evidence of a shared root cause." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 model-output bug; the later issue is a broader description of the same incorrect behavior." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both involve stricter config validation, but they affect different model configs and failures." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Different bugs: DeepSpeed ZeRO3 empty-index error vs tokenizer mapping failure." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Different bugs with different triggers and code paths." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Unrelated: tokenizer mapping edge case vs generic mask/index shape mismatch." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Both are loading/memory related, but one is offload failure and the other is Strix Halo mmap OOM." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays appropriately conservative. The accepted soft pairs are plausible duplicate subclusters from the titles alone, and the rejected pairs avoid overclaiming despite superficial symptom overlap. The canonical/best-issue choice around 43646 vs 43645 is also well supported." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 6, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43257", + "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:44534", + "issue:44756", + "issue:44910", + "issue:44960", + "issue:44977", + "issue:44991", + "issue:45276", + "issue:45292", + "issue:45313", + "issue:45335", + "issue:45357", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45588", + "issue:45704", + "issue:45812", + "issue:46050" + ], + "soft_pairs": [ + "issue:43644|issue:44534", + "issue:42491|issue:43720", + "issue:41720|issue:42915", + "issue:44960|issue:44977", + "issue:45313|issue:45357", + "issue:44451|issue:44488", + "issue:42915|issue:44756", + "issue:42491|issue:45276", + "issue:43708|issue:44075", + "issue:43854|issue:43931", + "issue:43644|issue:43645", + "issue:44488|issue:44991", + "issue:39692|issue:43994", + "issue:43257|issue:43854", + "issue:45488|issue:45812", + "issue:43856|issue:43873", + "issue:43856|issue:44387", + "issue:42491|issue:43931", + "issue:44960|issue:46050", + "issue:45362|issue:45406", + "issue:42491|issue:45292", + "issue:44910|issue:45588", + "issue:44164|issue:44291", + "issue:45276|issue:45335", + "issue:43404|issue:43454", + "issue:41720|issue:44155", + "issue:43425|issue:43957", + "issue:39692|issue:44960", + "issue:44402|issue:44488", + "issue:42947|issue:45704", + "issue:36010|issue:39404", + "issue:43873|issue:44387" + ], + "prepared_review_unit_hash": "202ba87d9a872a938c251f60ee1f7701189946c941224bb516d9682d09e4c845", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36710, + "estimated_input_tokens": 9178, + "estimated_eval_tokens": 18612 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of distinct Transformers regression reports, with only a couple of clear duplicate/near-duplicate pairs around v5 model initialization and embedding-resize behavior.", + "confidence": 0.69, + "canonical_issue_reason": "Issue 43644 is the broadest and clearest regression report in the set: a Transformers 5.0.0 initialization bug affecting non-persistent buffers, which matches the general v5 regression theme better than the more model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43644 is the best single issue to represent the cluster because it is concrete, widely scoped, and closest to the repeated v5 loading/initialization failures seen across the nearby reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying bug: both report Transformers v5 filling non-persistent buffers with junk, and the titles are effectively the same regression." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Different failures: one is Qwen3 MoE LoRA incompatibility across branches, the other is BitNet packed-weight loading with accelerate." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Both are Qwen3-family, but the concrete failures differ: cudaErrorAssert on device mapping vs FineGrainedFP8Config loading." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Different models and different failure modes; one is GLM5, the other is Qwen3.5 flash-attention generation." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Both involve Qwen3.5, but one is ZeRO-3 weight loading and the other is incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but they target different models and do not clearly point to the same underlying bug." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Unrelated issues: Qwen3Moe FP8 loading failure vs a mmap/OOM problem on Strix Halo." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Different areas: LoRA model compatibility on branch changes vs resize_token_embeddings not propagating to embeddings." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint-resume step math and unused SGD args are unrelated training bugs." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different models and different load-time failures: GLM-4.7 unit test loading vs Qwen3-VL weight shape mismatch." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5 regressions, but one is junk-filled buffers and the other is custom-model/Jupyter initialization." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both mention tokenizer/loading regressions, but they concern different models and likely different root causes." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "One is a documentation example problem, the other is an inference-quality regression for SigLIP2; not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Different Qwen3/GLM loading problems with different symptoms and code paths." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizer families and model sets; too broad to treat as one bug." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Qwen3 MoE training memory use and quantization/offloading behavior are distinct issues." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Different memory regressions: MoE training memory usage vs int4 CUDA reserved-memory/OOM." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "Unrelated: Qwen3 MoE LoRA branch compatibility vs Qwen3-VL weight shape mismatch." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "Different model families and different lifecycle bugs; no clear shared code-path." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Qwen3.5 chat crash and Gemma4 processor attribute error are unrelated failures." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "LoRA/branch compatibility is unrelated to resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both involve flash attention, but the concrete crashes are different: packed position_ids vs s_aux=None attribute error." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different loading/save bugs: extra_state serialization vs init_empty_weights/_is_hf_initialized argument handling." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Same underlying resize_token_embeddings regression: embeddings/output heads are not updated for related Gemma/T5Gemma model variants." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both mention untied lm_head weights, but they are different model families and the evidence is not strong enough to treat them as the same concrete bug." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different subsystems: Qwen3 device-mapping crash vs AudioFlamingo3 batched inference token/embedding leakage." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version compatibility and meta-device model-loading breakage are distinct problems." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Different models and different kinds of failure; one is a doc example, the other is a general GLM5 issue." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Different model-loading complaints; one is tokenizer vocab mismatch, the other is a broader load failure for a different model." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing with PEFT LoRA is unrelated to a T5/apex RMSNorm memory leak." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin and Whisper pipeline return_language failure are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both are memory-related, but one is quantization offloading and the other is increased CUDA reserved memory under int4 quantization." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is grounded in the packet and stays conservative. The two accepted soft pairs are well-supported by near-identical titles and closely matching regression descriptions, while the rejects avoid overclaiming shared root cause across mostly model-specific failures." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 7, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:39404", + "issue:42947", + "issue:43257", + "issue:43335", + "issue:43381", + "issue:43425", + "issue:43611", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43828", + "issue:43854", + "issue:43906", + "issue:43957", + "issue:44155", + "issue:44164", + "issue:44265", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44898", + "issue:44912", + "issue:44928", + "issue:45084", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45701", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:45859", + "issue:46315" + ], + "soft_pairs": [ + "issue:45663|issue:45759", + "issue:44291|issue:44756", + "issue:45663|issue:45859", + "issue:44155|issue:45823", + "issue:44387|issue:44756", + "issue:44292|issue:44912", + "issue:43335|issue:45663", + "issue:43957|issue:44756", + "issue:42947|issue:43381", + "issue:45276|issue:45292", + "issue:43854|issue:45216", + "issue:44164|issue:44756", + "issue:43381|issue:44928", + "issue:44479|issue:44877", + "issue:45684|issue:45698", + "issue:45588|issue:45663", + "issue:43646|issue:43708", + "issue:45812|issue:46315", + "issue:43645|issue:44164", + "issue:45292|issue:45335", + "issue:43828|issue:43957", + "issue:44402|issue:44451", + "issue:43425|issue:43828", + "issue:43257|issue:44863", + "issue:44589|issue:44756", + "issue:39404|issue:45701", + "issue:44683|issue:45823", + "issue:43611|issue:43645", + "issue:43425|issue:45823", + "issue:43425|issue:43716", + "issue:44530|issue:45084", + "issue:43906|issue:44530", + "issue:39404|issue:44912", + "issue:44265|issue:44898" + ], + "prepared_review_unit_hash": "ad8c47ae41574a4ff2aeb1d99d67b01c1450c4b01e032829cb05197f4bf2ca53", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38046, + "estimated_input_tokens": 9512, + "estimated_eval_tokens": 19280 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly a grab bag of unrelated Transformers bugs. The only clear duplicate cluster is the resize_token_embeddings regression trio: issues 45276, 45292, and 45335 all describe the same output-embedding propagation bug, with 45292 as the broadest representative.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 45292 is the best canonical issue because it states the core bug generically (`resize_token_embeddings` not affecting `output_embeddings`) and covers the more specific model variants in 45276 and 45335.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45292 is the strongest global issue anchor: concise, generic, and directly matched by the two accepted duplicate edges.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different subsystems and failure modes: Gemma4 FSDP attention state vs Qwen multimodal config unwrapping." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "One is an init_empty_weights argument regression; the other is mmap/OOM on Strix Halo. Not the same bug." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Gemma4 FSDP KeyError and Qwen3.5 MoE CPU-offload rotary embedding incompatibility are unrelated." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "AudioFlamingo3 batched inference token leak is unrelated to Gemma4 device placement under auto device map." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Int4 quantization memory growth and Strix Halo mmap OOM are different problems." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Both involve quantized model loading, but the models and concrete failures differ; not enough evidence of one bug." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to Gemma4 FSDP attention state rebuilding." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading failures are a different class of issue than mmap/OOM on Strix Halo." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is ineffectiveness with PEFT LoRA and the other is an eval-mode restriction." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same underlying bug: `resize_token_embeddings` does not propagate to output embeddings; 45276 is a model-specific instance." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "GLM-4.7-Flash loading test failure is unrelated to Qwen3.5 save_pretrained checkpoint regression." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "Custom `extra_state` save/load handling is unrelated to Strix Halo mmap/OOM." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Eval-mode checkpointing restriction and RLHF NaN explosion from 3D position ids are different bugs." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Video-input regression for Qwen VL models is not the same as strict-config loading for granite_speech." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "One concerns read-only permissions during save_pretrained; the other is wrong custom-module resolution after save_pretrained." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention s_aux=None crash is unrelated to Gemma4 shared_kv_states rebuilding." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization in notebooks and resume_from_checkpoint max_steps calculation are unrelated." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer-related, but one is incorrect token IDs and the other is a performance regression." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Jupyter custom-model init breakage is unrelated to save/from_pretrained extra_state handling." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same resize_token_embeddings propagation bug; 45335 is a t5gemma-specific manifestation." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch on Phi-tiny-MoE is a different failure from meta-device loading issues." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab-size mismatch and ScandiBERT loading failure are not clearly the same defect." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch 2.10 incompatibility is a version support issue, not the same as an autocast dtype mismatch on a specific model." + }, + { + "left": "issue:43257", + "right": "issue:44863", + "accept": false, + "reason": "Qwen3 MOE deepspeed conversion and NemotronH checkpoint loading are unrelated model-loading bugs." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to disabling mmap on Strix Halo." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to tokenizer behavior changes." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 PLE device mismatch are different code paths." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both involve custom models, but one is `base_model_prefix` loading and the other is notebook initialization; not the same bug." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch version incompatibility is unrelated to Gemma4 PLE device mismatch." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch 2.10 incompatibility and Mistral-3 dtype mismatch are different issues." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "PagedAttentionCache group-type crash and non-template-node compilation error are unrelated." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "Isolated reproduction of another issue is not the same as the Qwen3.5 PagedAttentionCache crash." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper return_language regression is unrelated to MXFP4 quantization fallback." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure with torch_compilable_check is unrelated to Perceiver interpolation failure." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Summary is grounded in the packet, and the soft-edge verdicts are conservative. The only accepted cluster is the resize_token_embeddings trio (45276/45292/45335), which is well supported by the issue titles and similarity signal." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 8, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42915", + "issue:43299", + "issue:43421", + "issue:43425", + "issue:43452", + "issue:43541", + "issue:43582", + "issue:43596", + "issue:43611", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43716", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44737", + "issue:44756", + "issue:44805", + "issue:44811", + "issue:44918", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45584", + "issue:45759", + "issue:45859", + "issue:46050", + "issue:46326" + ], + "soft_pairs": [ + "issue:44479|issue:44805", + "issue:43906|issue:46326", + "issue:46050|issue:46326", + "issue:44589|issue:45084", + "issue:43901|issue:44509", + "issue:43425|issue:44918", + "issue:43299|issue:43931", + "issue:43994|issue:44079", + "issue:36032|issue:43452", + "issue:44164|issue:45759", + "issue:44671|issue:44991", + "issue:43596|issue:44683", + "issue:44991|issue:45081", + "issue:43856|issue:44756", + "issue:44164|issue:44479", + "issue:39404|issue:44671", + "issue:44737|issue:45859", + "issue:43582|issue:43632", + "issue:44488|issue:45081", + "issue:44661|issue:44683", + "issue:44488|issue:44987", + "issue:32090|issue:39404", + "issue:43611|issue:43644", + "issue:40990|issue:43421", + "issue:43716|issue:45084", + "issue:36032|issue:42915", + "issue:43643|issue:44530", + "issue:34567|issue:40444", + "issue:40444|issue:46326", + "issue:44811|issue:45584", + "issue:43541|issue:45084" + ], + "prepared_review_unit_hash": "4ab15ccd02b1d39fca4b2684553f4c76096db813caa4fffc5b49baf3f2125f91", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36997, + "estimated_input_tokens": 9250, + "estimated_eval_tokens": 18756 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly separate Transformers regressions across different models, tokenizers, and runtime paths; the soft pairs share theme words but not the same underlying bug. I\u2019d keep the broad v5 model-loading regression as the representative issue.", + "confidence": 0.67, + "canonical_issue_reason": "issue:43611 is the clearest broad, concrete regression report in the set: Transformers 5.0.0 breaks model loading for configs using `base_model_prefix`, making it a good representative of the loading-regression subgroup.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the most suitable overall representative because it is specific, actionable, and broadly applicable to the v5 loading-breakage theme without being overly model-specific.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Different failures: a v5 video-input regression for specific Qwen VL models vs a generic mask/index shape mismatch." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "Only one side references another issue; the titles do not establish the same concrete tokenizer bug." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized save/load corruption and a mistral_common continue_final_message bug are distinct code paths." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup failure vs template-node compilation error are unrelated runtime/compiler problems." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are documentation-related, but they concern different pipeline APIs and different removal/mismatch issues." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility and Qwen3.5 input-embedding unpacking with TRL are separate issues." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "One is a v5 loading regression for Qwen3VL MoE models; the other is a weight-shape mismatch on a specific Qwen3-VL checkpoint." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Siglip2 nonsensical outputs and `ModelOutput` key assignment with prior `None` are different bugs." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer special-token method conflict and gguf_file loading breakage are unrelated loading failures." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "`extra_state` save/load handling and `from_config` not unwrapping `text_config` are separate serialization/config issues." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT masked-LM correctness and tokenizer loading for EMBEDDIA/est-roberta are different problems." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "DeepSpeed ZeRO-3 init index error and compiled flex_attention failure on torch>=2.9 are unrelated." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Both involve tokenizers, but they target different models and different root causes." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3 MoE training memory inefficiency and Strix Halo mmap OOM avoidance are not the same defect." + }, + { + "left": "issue:44164", + "right": "issue:44479", + "accept": false, + "reason": "Serialization of `extra_state` is unrelated to the Qwen VL video-input regression." + }, + { + "left": "issue:39404", + "right": "issue:44671", + "accept": false, + "reason": "Whisper pipeline `return_language` regression and CamemBERT MLM prediction errors are different model-specific bugs." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "XLNet CPU placement bug and Qwen3_5 MoE CPU-offload incompatibility affect different code paths." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and `_is_hf_initialized` flag breakage are separate issues." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "A model-loading failure for `cjvt/sleng-bert` is not the same as the Mistral regex patch crash." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "An add-new-model-like mapping bug and a flex_attention compile failure are unrelated." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Both are loading failures, but for different models and different failure mechanisms." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer broadcast TypeError and Whisper `return_language` pipeline regression are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "Both are v5 regressions, but one is about `base_model_prefix` loading and the other about non-persistent buffers being initialized with junk." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "High perplexity on GPT-OSS and tokenizer backend post-processor updates are different concerns." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral-3 image dtype mismatch and template-node compilation errors are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer special-token conflict and Qwen3Moe FineGrainedFP8Config failure are separate model/tokenizer issues." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "Missing fields from `trust_remote_code=True` config loading and PagedAttentionCache linear_attention crashes are different bugs." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "TrainerState token counting and Qwen2.5-VL multi-image iterable-dataset training failure are unrelated." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Multi-image Qwen2.5-VL training failure and mistral_common `continue_final_message` behavior are not the same issue." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Both are Whisper-related, but one is batch_decode skipping special tokens and the other is empty-transcription generation after align_special_tokens." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 9, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36032", + "issue:42175", + "issue:42915", + "issue:43089", + "issue:43262", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43452", + "issue:43482", + "issue:43540", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43883", + "issue:43994", + "issue:44062", + "issue:44206", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44610", + "issue:44683", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45507", + "issue:45987", + "issue:46178", + "issue:46304", + "issue:46315", + "issue:46326" + ], + "soft_pairs": [ + "issue:44451|issue:44987", + "issue:42175|issue:46178", + "issue:44530|issue:46304", + "issue:43644|issue:43646", + "issue:44206|issue:44220", + "issue:43452|issue:43482", + "issue:44464|issue:45987", + "issue:43883|issue:44488", + "issue:36032|issue:43645", + "issue:43317|issue:43482", + "issue:45084|issue:45507", + "issue:44849|issue:45325", + "issue:44987|issue:44991", + "issue:43746|issue:46326", + "issue:43262|issue:43377", + "issue:43645|issue:44534", + "issue:43646|issue:44534", + "issue:43404|issue:44466", + "issue:43089|issue:45987", + "issue:44403|issue:45081", + "issue:43883|issue:44991", + "issue:42915|issue:45084", + "issue:43322|issue:43540", + "issue:43994|issue:44610", + "issue:43377|issue:43819", + "issue:44451|issue:45081", + "issue:44265|issue:44610", + "issue:44987|issue:45081", + "issue:44991|issue:46315", + "issue:42915|issue:44291", + "issue:45216|issue:45357", + "issue:44220|issue:44683", + "issue:36032|issue:43742", + "issue:44062|issue:44987" + ], + "prepared_review_unit_hash": "63da7724a872f51ec47ac8cd9052944ec3ae048df636b4544ac21ed5b348b496", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38036, + "estimated_input_tokens": 9509, + "estimated_eval_tokens": 19274 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous model-loading/regression reports. The clearest duplicate-like pair in this packet is the nearly identical non-persistent-buffer regression: issue:43644 and issue:44534. The 43644/43646 pairing is too title-only to merge confidently.", + "confidence": 0.79, + "canonical_issue_reason": "issue:43644 is the best anchor because it matches issue:44534 almost verbatim and sits in the clearest duplicate-like v5 buffer-regression subcluster; it is the most concrete representative available here.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest standalone issue to keep because it is specific, actionable, and closest to the near-identical 44534 report; it is a better cluster anchor than the broader custom-init titles.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are load failures, but they target different models and no shared concrete root cause is evident." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Backend-related, but one is about pip extras and the other about a missing backend mapping; different bugs." + }, + { + "left": "issue:44530", + "right": "issue:46304", + "accept": false, + "reason": "Different subsystems and symptoms: cache/group-type handling versus distributed build support." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "The titles suggest related v5 breakage, but with only titles available there is not enough evidence to say they are the same underlying defect." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio-related, but they point to different code paths and failure modes." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "Both involve loading, but one is a general API break and the other is a specific Qwen2.5-GGUF regression." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Chunked generation/compiled forward and seq-length return-type issues are distinct bugs." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing tied-weights metadata in Molmo and a model-specific load failure for sleng-bert are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method-name conflict and notebook custom-model initialization breakage are different regressions." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "GPU/CPU offload loading and Qwen2.5-GGUF loading are different failure paths." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Template compilation error and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen families and different bugs: hidden-state handling versus rope-index scaling." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Both are version-specific load failures, but on different models with no evidence of the same root cause." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "PEFT/local checkpoint loading and continue_final_message handling are separate issues." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and MIMI padding-mask batching are different bugs." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Notebook custom-model init and non-persistent-buffer junk are both v5 regressions, but not clearly the same concrete bug." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Broad custom-model initialization breakage and non-persistent-buffer corruption are related only at a high level, not enough to merge." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "lm_head tying in Mistral3 and tied-weight serialization differences are distinct failure modes." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation performance overhead and StaticCache seq-length type mismatch are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Logging/noise during loading and a Mistral regex patch crash are not the same issue." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Missing tied-weights metadata and tokenizer loading failure are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "FineGrainedFP8Config failure and template compilation errors are separate bugs." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Segfault on Llava Next and video-input ValueError in Qwen3OmniMoe are different multimodal bugs." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Nonsensical SigLIP2 outputs and OmDet-Turbo image-size mismatch are distinct model/processor issues." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Missing padding-mask support and DAC STE mismatch are unrelated audio-model defects." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "General model-load failure and Mistral tokenizer regex patch crash are not the same bug." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export failure and processor image-size mismatch are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different models and different failure mechanisms; no shared code-path evidence." + }, + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Tokenizer load failure and convert_tokens_to_ids performance regression are different issues." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FineGrainedFP8Config failure and init_empty_weights / unexpected _is_hf_initialized argument are separate initialization bugs." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 save_pretrained regressions, but one is incorrect checkpoint content and the other is incorrect visual encoder keys." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Audio fbank extraction and compiled flex_attention on torch>=2.9 are unrelated." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer special-token conflict and a MobileLLM loading key error are different bugs." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken special-argument duplication and a model loading failure for physical-intelligence/fast are unrelated." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The write-up is mostly grounded, but it is not conservative enough on the soft-edge accept for issue:43644 vs issue:43646. From the packet we only have titles, and \u2018fills non-persistent buffers with junk\u2019 versus \u2018breaks custom model initialization\u2019 is suggestive but not enough to assert the same underlying defect. The summary is also a bit too strong in calling this the only obvious duplicate subcluster, since issue:43644 and issue:44534 have nearly identical titles and look like the clearest duplicate pair in the packet. Overall: reasonable cluster impression, but one soft-edge verdict overstates the evidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 10, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43296", + "issue:43317", + "issue:43381", + "issue:43404", + "issue:43421", + "issue:43425", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43611", + "issue:43632", + "issue:43792", + "issue:43818", + "issue:43856", + "issue:44079", + "issue:44291", + "issue:44403", + "issue:44462", + "issue:44479", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44829", + "issue:44877", + "issue:44991", + "issue:45020", + "issue:45127", + "issue:45325", + "issue:45357", + "issue:45479", + "issue:45538", + "issue:45542", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:46209", + "issue:46326" + ], + "soft_pairs": [ + "issue:46209|issue:46326", + "issue:43632|issue:44756", + "issue:44479|issue:45325", + "issue:42915|issue:44589", + "issue:43404|issue:45127", + "issue:36032|issue:41720", + "issue:43531|issue:44877", + "issue:43611|issue:44534", + "issue:40444|issue:45325", + "issue:43296|issue:43531", + "issue:43425|issue:44403", + "issue:40990|issue:44811", + "issue:43792|issue:44589", + "issue:43632|issue:44534", + "issue:44079|issue:45663", + "issue:44403|issue:44991", + "issue:33666|issue:40444", + "issue:43421|issue:44568", + "issue:44683|issue:45020", + "issue:43381|issue:45923", + "issue:35707|issue:40444", + "issue:44291|issue:44589", + "issue:44829|issue:45479", + "issue:44462|issue:45812", + "issue:43482|issue:45538", + "issue:43317|issue:43856", + "issue:43482|issue:43525", + "issue:43818|issue:45357", + "issue:42175|issue:45542", + "issue:44079|issue:44534" + ], + "prepared_review_unit_hash": "f84c92c54d53776663ef07a06329c200ed41415d3eda4780cdace061f61f8c09", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36693, + "estimated_input_tokens": 9174, + "estimated_eval_tokens": 18604 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. The only strong duplicate-like pair is the AutoTokenizer/tokenizer.json regression; the rest are different model-, backend-, or feature-specific bugs with only loose thematic overlap.", + "confidence": 0.77, + "canonical_issue_reason": "issue:45812 is the clearest high-impact, broad tokenizer regression in the set: it affects all Granite models, has inbound references, and aligns with the strongest duplicate-like evidence around AutoTokenizer loading behavior.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45812 is the best overall issue to keep because it is specific enough to be actionable, broad enough to represent a real cluster theme, and appears to cover a silent v4\u2192v5 tokenization regression with wide impact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46209", + "right": "issue:46326", + "accept": false, + "reason": "Unrelated topics: FP8 inference on B200 vs tokenizer continue_final_message behavior." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "Different failure modes and code paths: _is_hf_initialized / v5 loading flag vs mmap OOM on Strix Halo." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen VL regressions, but the titles point to different bugs: generic video-input regression vs still-image temporal position_id scaling." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Both mention FP8, but one is a Qwen3Moe FineGrainedFP8Config failure and the other is a missing Float8 storage type error; not the same bug." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both involve tied embeddings/lm_head, but one is a model-definition tie bug and the other is a LoRA-merging collapse issue." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer attribute conflict vs device-map CUDA assert; no shared underlying bug." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior is unrelated to granite_speech config strictness." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Both are transformers v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization; different code paths." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Qwen2.5-VL multi-image dataset finetuning failure is not the same as the rope-index temporal position_ids bug." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and Qwen3-MoE sliding_window are unrelated." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch compatibility packaging issue vs noisy transformer loading messages; different problems." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "High perplexity on gpt-oss-20b is unrelated to Whisper processor.batch_decode skipping special tokens." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper runtime failure and missing Float8 storage are unrelated." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Both are v5 regressions, but the titles indicate distinct bugs: _is_hf_initialized handling vs junk-filled non-persistent buffers." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma-4 FSDP2 shared_kv_states rebuilding are different defects." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Generic loading noise vs a specific tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen VL training issues, but multi-GPU training and multi-image iterable-dataset failures are not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime special-token post-processor updates and add_special_tokens BOS/EOS behavior are related to tokenization, but not the same concrete failure." + }, + { + "left": "issue:44683", + "right": "issue:45020", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to remote_code model loading regressions." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing in eval mode is different from a missing supports_gradient_checkpointing flag." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image finetuning." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a different initialization-path bug than missing Float8 storage." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both concern degenerate classification training, but one is flash_attention_3-specific and the other is the num_labels=1 loss formulation." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": true, + "reason": "Both are AutoTokenizer/tokenizer.json regressions that affect token IDs; they plausibly share the same tokenizer-loading code path." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer max_length behavior are unrelated tokenizer issues." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "device_map=auto offload loading failure is unrelated to Qwen3 MoE training memory inefficiency." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "GGUF loading failure and Llama4Config missing pad_token_id are different model-loading problems." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Video-LLaVA tower weight/temporal-attention bug is unrelated to Qwen3.5 save_pretrained key mapping." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both concern TensorFlow packaging, but one is missing TF in the torch extra while the other is an undefined tf backend when only tensorboard is installed; different dependency issues." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key handling and non-persistent buffer initialization are unrelated defects." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The canonical issue choice is supported by the packet title/inbound refs, and the soft-edge decisions mostly stick to visible title-level evidence without overclaiming. The 44462\u219445812 acceptance is cautious enough to be plausible, while the rest of the rejects are appropriately conservative." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 11, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43377", + "issue:43454", + "issue:43482", + "issue:43531", + "issue:43582", + "issue:43606", + "issue:43611", + "issue:43632", + "issue:43644", + "issue:43646", + "issue:43701", + "issue:43746", + "issue:43749", + "issue:43792", + "issue:43819", + "issue:43873", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44466", + "issue:44479", + "issue:44568", + "issue:44877", + "issue:44898", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45381", + "issue:45491", + "issue:45584", + "issue:45701", + "issue:45800", + "issue:45823", + "issue:45834", + "issue:45920" + ], + "soft_pairs": [ + "issue:41553|issue:43582", + "issue:41720|issue:45081", + "issue:43531|issue:44291", + "issue:43377|issue:45491", + "issue:40444|issue:45381", + "issue:43644|issue:44079", + "issue:44265|issue:45072", + "issue:43749|issue:44079", + "issue:43611|issue:43646", + "issue:45325|issue:45381", + "issue:43632|issue:45072", + "issue:42175|issue:43296", + "issue:43792|issue:43957", + "issue:43994|issue:44898", + "issue:45325|issue:45834", + "issue:43454|issue:44466", + "issue:42915|issue:43482", + "issue:43701|issue:45325", + "issue:44479|issue:45823", + "issue:44291|issue:44877", + "issue:43873|issue:45701", + "issue:43606|issue:44265", + "issue:43262|issue:43746", + "issue:43296|issue:45800", + "issue:34634|issue:35707", + "issue:40990|issue:45584", + "issue:35707|issue:41720", + "issue:44568|issue:45920", + "issue:44186|issue:44265", + "issue:41720|issue:45823", + "issue:43606|issue:43819" + ], + "prepared_review_unit_hash": "8adeedbd6b245f0594d976a6dd595647daeae6dbe6af32304dc24bee7b5295f6", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37099, + "estimated_input_tokens": 9275, + "estimated_eval_tokens": 18806 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. A few pairs share broad themes like Transformers v5 model-loading regressions or Qwen2.5-VL position-id bugs, but none are clearly the same concrete bug/change, so I would not collapse any of the soft edges.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 43611 is the closest cluster center because several nearby items are Transformer v5 model-init/loading regressions, but the overall set is too diverse for a true duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "43611 is the cleanest representative of the only coherent subtheme in the set: v5 model-loading / initialization breakage.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Unrelated failure modes: tokenizer loading error message vs Apple Silicon warmup TypeError." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Different bugs in different code paths: auto device mapping CUDA assert vs tokenizer regex patch crash." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Both mention v5 regressions, but one is sliding-window behavior and the other is init_empty_weights/_is_hf_initialized loading." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Different model families and symptoms: batched-vs-single padding mask issue vs Gemma3 sliding-window NaNs." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen multimodal regressions, but one is training with multiple images per prompt and the other is video position IDs." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Serialization/tied-weights issue vs ModelOutput key assignment bug; not the same code-path." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "Export failure due to torch_compilable_check is unrelated to bf16 dtype mismatch inference." + }, + { + "left": "issue:43749", + "right": "issue:44079", + "accept": false, + "reason": "FSDP CPU RAM loading breakage is unrelated to ModelOutput key assignment." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Close theme, but still different bugs: base_model_prefix loading vs custom model initialization." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Same model family, but one is still-image temporal scaling and the other is video vision_position_ids; too specific to merge blindly." + }, + { + "left": "issue:43632", + "right": "issue:45072", + "accept": false, + "reason": "_is_hf_initialized flag breakage is unrelated to bf16 dtype mismatch in inference." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Different packaging/loading issues: missing TensorFlow backend in extras vs PaddleOCR-VL load failure in vLLM." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Whisper model runtime failure is not the same as meta-device model loading breakage." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Both are vision-model quality issues, but they involve different models and likely different preprocessing paths." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Both are multimodal, but one is Qwen2.5-VL rope/position IDs and the other is Kosmos2.5 OCR index bounds." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Tied-weight loading/generation bug vs inconsistent serialization of lm_head.weight; related area, not same defect." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Different failures: FineGrainedFP8Config on Qwen3Moe vs GGUF loading on Qwen2.5." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint key mismatch is unrelated to Qwen2.5-VL position-id computation." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Different model regressions: Qwen video input path vs Gemma4 PLE device mismatch." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "Unexpected _is_hf_initialized arg is a loading/init bug; strict granite_speech config is a separate config-validation issue." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offloading issue vs tokenizer behavior change across versions; no shared concrete bug." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch for Bark is unrelated to torch.export/exportability with torch_compilable_check." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio chat-template sampling-rate default is unrelated to PEFT checkpoint loading." + }, + { + "left": "issue:43296", + "right": "issue:45800", + "accept": false, + "reason": "Version compatibility reports, but for different stacks and failure modes." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "BarkProcessor voice_preset bug and progressive generation with inputs_embeds/past_key_values are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "Completely different models and symptoms: high perplexity vs empty-transcription generation failure." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation bug is unrelated to Qwen3 auto device mapping CUDA assert." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "One tokenizer-specific BOS/EOS regression vs a broader AutoTokenizer token-id regression; not the same concrete bug." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer padding/truncation crash is unrelated to torch.export failure on compilable-check models." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both involve device placement, but different models and different failure mechanisms." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "CPU offload device mismatch and DAC forward/from_latents mismatch are unrelated bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 12, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36331", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43450", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43708", + "issue:43720", + "issue:43749", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44008", + "issue:44186", + "issue:44387", + "issue:44466", + "issue:44488", + "issue:44534", + "issue:44610", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45325", + "issue:45357", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45923", + "issue:46050", + "issue:46291", + "issue:46305", + "issue:46315" + ], + "soft_pairs": [ + "issue:43454|issue:43550", + "issue:39692|issue:40444", + "issue:46305|issue:46315", + "issue:43881|issue:44488", + "issue:43377|issue:44610", + "issue:43257|issue:45923", + "issue:43454|issue:45072", + "issue:43572|issue:45759", + "issue:43317|issue:43873", + "issue:43377|issue:44898", + "issue:40990|issue:42947", + "issue:43322|issue:46050", + "issue:43454|issue:43720", + "issue:45325|issue:46315", + "issue:41720|issue:43322", + "issue:43749|issue:44534", + "issue:44466|issue:45357", + "issue:41720|issue:44977", + "issue:44991|issue:46291", + "issue:42617|issue:42915", + "issue:43526|issue:43883", + "issue:40990|issue:45701", + "issue:44387|issue:45704", + "issue:43881|issue:44991", + "issue:36331|issue:42617", + "issue:43577|issue:43708", + "issue:44186|issue:45072", + "issue:44008|issue:45823", + "issue:43404|issue:45357", + "issue:36331|issue:42915", + "issue:44877|issue:44991", + "issue:43450|issue:45701" + ], + "prepared_review_unit_hash": "3e0158722ad37f927444479efbbae2caebc7b30413d25887e515b15460e8191f", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37521, + "estimated_input_tokens": 9381, + "estimated_eval_tokens": 19018 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are unrelated model-specific regressions, tokenizer/loading issues, quantization/offload problems, or training quirks. The soft pairs share only superficial wording and should not be merged.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: the items span unrelated bugs across training, quantization, tokenization, loading, and model-specific behavior, with only superficial title overlap.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a strong global representative; if forced, the broadest regression-style report is the closest anchor, but it still does not match most items in the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "Different code paths and models: lm_head tying vs torch.compile/SDPA behavior." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Documentation example errors are unrelated to Qwen2.5-VL finetuning with multi-image iterable datasets." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Different subsystems: Moonshine sliding-window masking vs slow-tokenizer conversion performance." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Different failures: model loading for glm-4v-9b vs tokenizer loading for sleng-bert." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "Different bugs: MIMI batching/padding-mask inconsistency vs OmDet-Turbo processor/model resolution mismatch." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "Different models and mechanisms: Qwen3 MoE weight conversion vs missing gradient-checkpointing flag." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "Unrelated issues: tied lm_head weights vs bfloat16 dtype mismatches in inference." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Different configuration problems: missing pad_token_idx vs unwrapping text_config in composite multimodal configs." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both mention offloading/quantization, but the concrete failure modes are different enough to avoid deduping." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Different models and symptoms: MIMI batch masking vs Perceiver non-default resolution handling." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Perplexity regression is unrelated to LoRA gradient-checkpointing ineffectiveness." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Different issues: Llava Next loading segfault vs sinq quantized save/load breakage." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "Different code paths: lm_head tying vs BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:45325", + "right": "issue:46315", + "accept": false, + "reason": "Different bugs: Qwen2.5-VL rope/position_ids scaling vs tokenizer conversion speed regression." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Different model families and failure types: cudaErrorAssert on Qwen3 vs segfault loading Llava Next." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "Broken FSDP CPU RAM-efficient loading is unrelated to non-persistent buffers being filled with junk." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both involve saving, but one is tied lm_head serialization and the other is incorrect visual encoder keys." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Different Qwen3 failures: device-mapping cudaErrorAssert vs flash-attention generation issues." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading regression is unrelated to kernels v0.15 requiring version/revision." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "Different models and errors: 3d_parallel.py runtime failure vs FineGrainedFP8Config failure in Qwen3Moe." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Molmo tied-weights attribute error." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "High perplexity and tokenizer behavior changes are only loosely related; not the same underlying bug." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Different memory issues with different causes: int4 reserved-memory growth vs apex FusedRMSNorm leak." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Different models and failure modes: glm-4v-9b loading vs EMBEDDIA tokenizer loading." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer compute_loss API mismatch is unrelated to 3d_parallel.py execution failure." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "BLIP2 dtype propagation and resume_from_checkpoint step calculation are unrelated bugs." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer padding/truncation crash is unrelated to dtype mismatches in inference." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Different Gemma issues: variable name collision in forward vs device_map auto device mismatch." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Different model families and different save/load problems; not the same tied-weights bug." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "Unexpected compute_loss keyword argument is unrelated to Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Strict config parsing for granite_speech is unrelated to tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batch-shape bug is unrelated to tokenization changes across versions." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 13, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:29159", + "issue:30064", + "issue:33357", + "issue:34634", + "issue:35707", + "issue:36010", + "issue:36032", + "issue:36331", + "issue:41720", + "issue:42175", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43425", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43720", + "issue:43782", + "issue:43873", + "issue:43881", + "issue:44186", + "issue:44361", + "issue:44423", + "issue:44610", + "issue:44671", + "issue:44743", + "issue:44756", + "issue:44898", + "issue:44913", + "issue:45059", + "issue:45081", + "issue:45161", + "issue:45406", + "issue:45632", + "issue:45684", + "issue:45701", + "issue:45812", + "issue:46291" + ], + "soft_pairs": [ + "issue:43873|issue:44756", + "issue:43550|issue:43720", + "issue:42175|issue:46291", + "issue:41720|issue:42617", + "issue:44423|issue:45406", + "issue:43550|issue:43606", + "issue:44610|issue:44898", + "issue:44186|issue:44610", + "issue:45632|issue:45684", + "issue:43295|issue:45701", + "issue:42915|issue:43317", + "issue:36010|issue:42617", + "issue:43425|issue:46291", + "issue:41720|issue:45161", + "issue:44671|issue:44743", + "issue:43525|issue:46291", + "issue:45081|issue:45812", + "issue:43531|issue:43572", + "issue:43316|issue:44913", + "issue:43540|issue:43782", + "issue:42915|issue:43322", + "issue:44186|issue:44361", + "issue:43421|issue:45812", + "issue:36331|issue:41720", + "issue:43526|issue:43881", + "issue:29159|issue:36032", + "issue:43388|issue:45059", + "issue:33357|issue:35707", + "issue:30064|issue:34634" + ], + "prepared_review_unit_hash": "f400c1c1a6ca70c2441e7b258dcc37a022558599ddb12c4920ceaabd5901bd0b", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36348, + "estimated_input_tokens": 9087, + "estimated_eval_tokens": 18430 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items do not form a duplicate cluster: the soft pairs are all cross-subsystem or cross-model false positives with different failure modes. No soft edge should be merged.", + "confidence": 0.94, + "canonical_issue_reason": "Issue 45812 is the strongest representative: it describes a concrete, silent tokenization regression affecting all Granite models with clear user impact and good diagnostic value.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45812 is the best global issue candidate because it is specific, reproducible, and broadly impactful compared with the more niche or ambiguous reports here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Quantization/offload behavior vs mmap OOM workaround; different failure modes and likely different fixes." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "torch.compile + SDPA on Bamba is unrelated to accelerate loading and unpacking packed weights." + }, + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "Missing TensorFlow extra in pip install is a packaging issue, not the kernels version/revision import breakage." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and failing 3d_parallel.py are unrelated problems." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Continuous-batching multimodal crash differs from Gemma4Processor missing _tokenizer; separate code paths." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Bamba torch.compile/SDPA bug is not the same as Bark-small CPU offload device mismatch." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both involve image sizing, but they hit different models and different logic (processor output size vs interpolation on Perceiver)." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "Tokenizer NER padding/truncation crash is unrelated to an image processor producing the wrong resolution." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Cache-path collision for trust_remote_code is unrelated to permission propagation in save_pretrained." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "A regression around processor.tokenizer/image passing is not the same concrete bug as general version-dependent tokenization changes." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "FineGrainedFP8 Qwen3Moe failure and dequantized model offload loading failure are different quantization/offload bugs." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import failure is unrelated to a script runtime failure in 3d_parallel.py." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "Torch 2.10 incompatibility is a dependency/version issue, not the kernels import requirement change." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and GPT-OSS MoE-only-TP support are different model/runtime issues." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "CamemBERT masked LM regression and Qwen3.5 recurrent-state reset bug are unrelated model behaviors." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config is unrelated to the kernels package import constraint." + }, + { + "left": "issue:45081", + "right": "issue:45812", + "accept": false, + "reason": "Mistral regex patch/backend_tokenizer crash is a tokenizer backend attribute issue, not a Granite token-ID regression." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and StableLmConfig pad_token_idx missing are different config/model bugs." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy is unrelated to GPTNeoX rotary_pct not persisting on reload." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Video-input processing error in Qwen3OmniMoe is not the same as Qwen3VL weight_only loading failure." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "FineGrainedFP8 loading failure and Llava Next segfault are distinct model-loading bugs." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash and MLukeTokenizer task AttributeError are different tokenizer issues." + }, + { + "left": "issue:43421", + "right": "issue:45812", + "accept": false, + "reason": "Runtime post-processor update omission is a different tokenizer backend bug than Granite token-ID misgeneration." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "CustomTrainer compute_loss signature mismatch is unrelated to the Qwen3 device-mapping CUDA assert." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug has no overlap with glm-4v-9b loading failure." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Slow-vs-fast tokenizer inconsistency and T5Tokenizer add_special_tokens load conflict are different tokenizer bugs." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Metric gathering last-batch label dropping is unrelated to SAM3 text/bbox weird behavior." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "MacOS bus error in a CLIP model and progressive generation with inputs_embeds/past_key_values are unrelated." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation map processing and BarkProcessor voice_preset failure are different processor bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 14, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:34567", + "issue:34634", + "issue:35707", + "issue:39692", + "issue:41720", + "issue:42490", + "issue:43329", + "issue:43421", + "issue:43531", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:43927", + "issue:44062", + "issue:44077", + "issue:44368", + "issue:44451", + "issue:44492", + "issue:44509", + "issue:44514", + "issue:44530", + "issue:44545", + "issue:44589", + "issue:44610", + "issue:44625", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45357", + "issue:45381", + "issue:45406", + "issue:45464", + "issue:45479", + "issue:45588", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:46050", + "issue:46291" + ], + "soft_pairs": [ + "issue:41720|issue:43746", + "issue:43550|issue:43653", + "issue:43873|issue:45704", + "issue:43606|issue:43720", + "issue:45406|issue:45464", + "issue:44530|issue:46291", + "issue:43421|issue:45701", + "issue:34634|issue:41720", + "issue:43531|issue:45759", + "issue:44492|issue:44509", + "issue:43606|issue:43653", + "issue:42490|issue:43329", + "issue:34567|issue:35707", + "issue:44625|issue:45479", + "issue:44514|issue:44545", + "issue:44451|issue:44991", + "issue:45759|issue:45859", + "issue:44077|issue:44625", + "issue:43421|issue:43927", + "issue:43531|issue:44589", + "issue:44898|issue:45072", + "issue:45588|issue:45759", + "issue:44368|issue:45357", + "issue:44062|issue:44451", + "issue:44987|issue:45042", + "issue:43782|issue:45381", + "issue:44918|issue:44936", + "issue:39692|issue:44610", + "issue:43746|issue:46050" + ], + "prepared_review_unit_hash": "3ccd9e5587441cd95d0bf962d5b8b62ee9166f6055b032cd491e0c90bc23f58b", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "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; the only clear duplicate pair is the Qwen2_5_VLProcessor.apply_chat_template crash on batched input with padding=False (44514/44545).", + "confidence": 0.93, + "canonical_issue_reason": "issue:44514 is the best canonical for the duplicate pair because it was opened first and describes the same concrete crash as 44545 with the same repro conditions.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44514 is the strongest representative issue in the cluster: specific, reproducible, and the clearest duplicate target for 44545.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Different failures: Qwen3 auto device-mapping CUDA assert vs GraniteSpeech PEFT local checkpoint loading." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "Unrelated bugs in different areas: torch.compile/SDPA on Bamba vs BigBirdTokenizer special-token decode behavior." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Different code paths: quantization/offloading behavior vs T5 apex FusedRMSNorm memory leak." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "Different models and causes: CPU offload device mismatch vs BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Different processors and failure modes: missing _tokenizer in Gemma4Processor vs Qwen3.5 streaming chat/completions failure." + }, + { + "left": "issue:44530", + "right": "issue:46291", + "accept": false, + "reason": "Different subsystems: PagedAttentionCache group-type crash vs import-time kernel version/revision validation." + }, + { + "left": "issue:43421", + "right": "issue:45701", + "accept": false, + "reason": "Both mention tokenization, but one is runtime special-token/post-processor syncing and the other is version-dependent tokenization changes." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "No shared underlying bug: BarkProcessor voice_preset vs Qwen3 auto device-mapping CUDA assert." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Different Qwen3.5 issues: sliding_window behavior vs from_config not unwrapping text_config." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A typo in cache strategies is unrelated to removed pipeline-task docs." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Different bugs in different components: CPU offload mismatch vs tokenizer special-token registration/decode output." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Local-path trailing-slash import warning is unrelated to multimodal token counting video-branch undefined names." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "Trainer token counter not updating is unrelated to progressive generation with inputs_embeds and past_key_values." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Different classification bugs: num_labels propagation vs degenerate zero loss for num_labels=1." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same exact bug: Qwen2_5_VLProcessor.apply_chat_template crashes on batched input when padding=False." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer/load failures for different models and different root causes." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Different Qwen3.5 model-loading/runtime issues: multimodal config unwrapping vs CPU-offload incompatibility in rotary embedding." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "PatchTSMixer post-init policy is unrelated to Qwen3.5 num_labels propagation." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Runtime special-token/post-processor updates are unrelated to DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Sliding-window behavior for Qwen3-MoE is unrelated to Float8 storage lookup failures." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different image-model dtype/resolution issues: Perceiver non-default resolution vs SwitchTransformers/TimmWrapperModel bfloat16 mismatches." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "flash_attention s_aux=None crash is unrelated to Qwen multimodal config handling." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "Different Qwen3.5 LoRA/save_pretrained warnings vs incorrect visual encoder key persistence." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken keyword-argument error is unrelated to loading a specific model checkpoint." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Loading physical-intelligence/fast is unrelated to PIL backend processors requiring torchvision." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Different VL bugs: weight_only load error vs wrong vision_position_ids for video input." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "Unpacking Qwen3.5 input embeddings with TRL is unrelated to trainer.evaluate() failing after train()." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Docs/example errors are not the same bug as OmDet-Turbo processor/model resolution mismatch." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "PEFT checkpoint loading for GraniteSpeech is unrelated to sinq quantized model save/load corruption." + }, + { + "left": "issue:44936", + "right": "issue:44987", + "accept": false, + "reason": "Training/evaluation state issue is unrelated to loading physical-intelligence/fast." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: 44514/44545 are the same issue by title and chronology, and the rest of the soft-edge rejections stay conservative by citing distinct models, code paths, or failure modes. I don\u2019t see any overstatement that would force a reject." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 15, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:42915", + "issue:43295", + "issue:43322", + "issue:43381", + "issue:43388", + "issue:43421", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43618", + "issue:43645", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43746", + "issue:43819", + "issue:43844", + "issue:43873", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44488", + "issue:44661", + "issue:44671", + "issue:44843", + "issue:44898", + "issue:45072", + "issue:45081", + "issue:45183", + "issue:45356", + "issue:45479", + "issue:45663", + "issue:45685", + "issue:45704", + "issue:45834", + "issue:45859", + "issue:45907", + "issue:45923", + "issue:46032", + "issue:46294" + ], + "soft_pairs": [ + "issue:45704|issue:45923", + "issue:43381|issue:43844", + "issue:43421|issue:43653", + "issue:42886|issue:44843", + "issue:45663|issue:45923", + "issue:45704|issue:45859", + "issue:41720|issue:45704", + "issue:43531|issue:43645", + "issue:43994|issue:44671", + "issue:43421|issue:44186", + "issue:43746|issue:43873", + "issue:45081|issue:45356", + "issue:39692|issue:43873", + "issue:43540|issue:46294", + "issue:42915|issue:43746", + "issue:45356|issue:45907", + "issue:43596|issue:44220", + "issue:30064|issue:36010", + "issue:43720|issue:43819", + "issue:43653|issue:43720", + "issue:44220|issue:44661", + "issue:43540|issue:44008", + "issue:43550|issue:45072", + "issue:43295|issue:45834", + "issue:43618|issue:43697", + "issue:44062|issue:44488", + "issue:43550|issue:43819", + "issue:45183|issue:45479", + "issue:43388|issue:46032", + "issue:43322|issue:45685", + "issue:44186|issue:44898" + ], + "prepared_review_unit_hash": "d8228334165485221e6783549ea0cb20274c99432afd814cad7bcf195a65a8e1", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37488, + "estimated_input_tokens": 9372, + "estimated_eval_tokens": 19000 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a set of unrelated issue reports. Only one soft pair looks like the same underlying bug: offline tokenizer loading broken by `_patch_mistral_regex` / `model_info` behavior. The rest are at best broad subsystem overlaps, not true duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "issue:44843 is the strongest canonical issue because it names the concrete root cause (`model_info()` called unconditionally in `_patch_mistral_regex`) and directly explains the offline failure; issue:42886 is the broader symptom report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44843 is the best representative of the cluster\u2019s only clear duplicate-like bug: it is specific, root-cause oriented, and covers the offline tokenizer regression more precisely than the symptom-level report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45704", + "right": "issue:45923", + "accept": false, + "reason": "Different problems: T5 RMSNorm memory leak vs missing `supports_gradient_checkpointing` flag in a trust-remote-code model." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Both involve training/runtime stability, but one is eval-mode gradient checkpointing and the other is ZeRO-3 gradient explosion; not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Both concern tokenizer special-token behavior, but one is runtime post-processor updates and the other is BigBird mask-token registration / decode output." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": true, + "reason": "Same offline-tokenizer bug: both describe failure under `HF_HUB_OFFLINE=1`, and 44843 gives the specific root cause in `_patch_mistral_regex`/`model_info()`." + }, + { + "left": "issue:45663", + "right": "issue:45923", + "accept": false, + "reason": "Gemma4 FSDP2 shared-state KeyError is unrelated to the Nemotron gradient-checkpointing flag issue." + }, + { + "left": "issue:45704", + "right": "issue:45859", + "accept": false, + "reason": "Both mention performance/offload-ish areas, but one is an apex RMSNorm memory leak and the other is Qwen3 MoE CPU-offload incompatibility." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "CUDA assert in Qwen3 auto device mapping is unrelated to T5/apex RMSNorm memory leak." + }, + { + "left": "issue:43531", + "right": "issue:43645", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and Jupyter custom-model initialization are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different model failures: SigLIP2 AutoModel/pipeline output issue vs CamemBERT masked-LM prediction regression." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Both are tokenizer bugs, but they affect different code paths and symptoms; no evidence they are the same underlying defect." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Local checkpoint loading with PEFT adapters is not the same as quantization/offloading not working as expected." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both touch `fix_mistral_regex`, but one is a crash on loading Mistral tokenizer while the other is a Kimi-K2.5 codec/warning regression; not clearly one bug." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 docs example errors are unrelated to quantization/offloading behavior." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Both are video-input issues in Qwen models, but they fail in different layers (`process`/model input vs chat-template token-feature mismatch)." + }, + { + "left": "issue:42915", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3Moe FP8 config failure is unrelated to GraniteSpeech PEFT checkpoint loading." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression and CI `list_repo_files` revision forwarding are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed ZeRO-3/BertModel index error is unrelated to `_torch_extract_fbank_features()`." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Image processor segmentation-map handling is unrelated to the `GenerationMixin` import error." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "BitNet packed-weight loading and DAC `from_latents` vs forward mismatch are different model-specific bugs." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBird special-token decode bug is unrelated to BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio feature extraction issue and `add-new-model-like` failure in tokenizer mapping are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Video-input failure in Qwen3OmniMoe is unrelated to Gemma 3n audio-mask variable collision." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Different paths: torch.compile/SDPA on Bamba vs bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to Kosmos2.5 long-OCR index error." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIP attentions being unassigned is unrelated to RTDetrV2 output differences in v5." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "`AddedToken` keyword collision and failure to load `cjvt/sleng-bert` are different tokenizer-loading issues." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is unrelated to DAC `from_latents` mismatch." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Input validation error messaging and degenerate zero loss from `num_labels=1` are different bugs." + }, + { + "left": "issue:43388", + "right": "issue:46032", + "accept": false, + "reason": "`gather_for_metrics` last-batch label dropping is unrelated to Mamba2Mixer `use_cache` correctness." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Llava Next segmentation fault and MPS histogram-on-int error are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "LayoutLMv2 tokenizer NER/padding crash and Perceiver non-default-resolution failure are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall and appropriately conservative. The one accepted soft pair (42886/44843) is supported by the packet, and the remaining pairs are rejected for clearly different bugs rather than loose subsystem similarity. The canonical/best-issue choice is also reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 16, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:30064", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43335", + "issue:43421", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43643", + "issue:43644", + "issue:43653", + "issue:43716", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43818", + "issue:43819", + "issue:43825", + "issue:43950", + "issue:43994", + "issue:44008", + "issue:44186", + "issue:44442", + "issue:44448", + "issue:44479", + "issue:44485", + "issue:44509", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45381", + "issue:45584", + "issue:45834", + "issue:45987" + ], + "soft_pairs": [ + "issue:43335|issue:43756", + "issue:43716|issue:44008", + "issue:43531|issue:43643", + "issue:43653|issue:43819", + "issue:44843|issue:45081", + "issue:43950|issue:45987", + "issue:43643|issue:44877", + "issue:30064|issue:36331", + "issue:44448|issue:44671", + "issue:43540|issue:45381", + "issue:41720|issue:44991", + "issue:43421|issue:44811", + "issue:43577|issue:43818", + "issue:43782|issue:44977", + "issue:39692|issue:45834", + "issue:44821|issue:45084", + "issue:45584|issue:45834", + "issue:43644|issue:43749", + "issue:44186|issue:44442", + "issue:43606|issue:45072", + "issue:44479|issue:45834", + "issue:34634|issue:36010", + "issue:43577|issue:43582", + "issue:43825|issue:43994", + "issue:44805|issue:44936", + "issue:44485|issue:44509" + ], + "prepared_review_unit_hash": "93bba59b05dbd9c2b7f7b158c1669fd4d3b6858010fc18c06c746b1580faae51", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35697, + "estimated_input_tokens": 8925, + "estimated_eval_tokens": 18106 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. A few pairs share a subsystem or helper function, but none look like the same underlying bug/change closely enough to merge. The strongest recurring theme is non-persistent buffer/loading regressions, with issue:43950 as the clearest representative.", + "confidence": 0.69, + "canonical_issue_reason": "issue:43950 is the most specific and broadly useful statement of a core regression in the cluster: non-persistent buffer corruption during loading. It reads like the best representative of that loading-path bug family.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43950 is the best issue to anchor this cluster because it is concrete, regression-oriented, and broadly applicable to a core model-loading path rather than a narrow model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Both are model/configuration bugs, but they affect different models and different mechanics (sparse-layer setup vs RoPE-layer dropping). Not the same underlying issue." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Both are multimodal model bugs, but the failures are unrelated: dtype mismatch in image preprocessing vs a Gemma3n variable/attribute collision." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "One is a Qwen3-MoE sliding-window behavior issue; the other is a config-loading bug with trust_remote_code. Same area only at a very high level." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration and DAC latent/forward consistency are unrelated bugs with no shared code path implied." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both hit _patch_mistral_regex, but one is an offline model_info call and the other is a tokenizer-API AttributeError. Same helper, different root causes." + }, + { + "left": "issue:43950", + "right": "issue:45987", + "accept": false, + "reason": "Both mention loading/state handling regressions, but one is buffer corruption on load and the other is a StaticCache return-type contract bug." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "These are distinct config-loading issues: missing fields from AutoConfig with trust_remote_code versus strict config rejection for granite_speech." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "An image-segmentation processing failure and a Trainer compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Both are output regressions in v5, but they concern different models and different failure modes; no shared underlying bug is evident." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Both are video-input regressions in Qwen-family models, but they affect different model lines and likely different preprocessing/code paths." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Auto device mapping CUDA assertions and tokenizer-loading failures are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Special-token/post-processor update behavior and Whisper batch_decode skipping special tokens are different tokenizer-layer problems." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "A dtype-loading issue in BLIP2 is not the same bug as a Video-LLaVA temporal-attention/weight-sharing problem." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Both involve Qwen models, but one is a from_pretrained weight_only error and the other is flash-attention generation behavior. Different code paths." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "A SigLIP2 docs/example problem and a Kosmos2.5 OCR index error are not the same underlying issue." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading and Jinja/template compilation errors are completely separate." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty-transcription generation and Kosmos2.5 long-OCR indexing are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Both concern loading regressions, but one is non-persistent buffer corruption and the other is broken FSDP CPU RAM efficient loading. Too different to merge." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2 NER/padding crashes and AutoTokenizer failing to load a specific tokenizer are different tokenizer issues." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Device-mismatch under CPU offload for Bark is unrelated to dtype mismatches in SwitchTransformers/TimmWrapperModel inference." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "A broad Qwen video-input regression and a Kosmos2.5 OCR index error do not look like the same bug." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "A BarkProcessor voice_preset issue and a GenerationMixin import failure are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype loading and Apple Silicon caching_allocator_warmup TypeError are unrelated." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Pipeline deprecation/error-message behavior and SigLIP2 nonsensical outputs are distinct issues." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "A mask-shape IndexError and a trainer.evaluate() failure after training are not the same bug." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion and stale docs mentioning removed pipeline tasks are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 17, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:34634", + "issue:36331", + "issue:42175", + "issue:43262", + "issue:43329", + "issue:43526", + "issue:43582", + "issue:43650", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43782", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43906", + "issue:44060", + "issue:44291", + "issue:44368", + "issue:44554", + "issue:44561", + "issue:44589", + "issue:44610", + "issue:44671", + "issue:44857", + "issue:44912", + "issue:45003", + "issue:45020", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45372", + "issue:45406", + "issue:45538", + "issue:45600", + "issue:45701", + "issue:45987", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46326" + ], + "soft_pairs": [ + "issue:44561|issue:45020", + "issue:45987|issue:46192", + "issue:28218|issue:29159", + "issue:44671|issue:45701", + "issue:44610|issue:45072", + "issue:43720|issue:45072", + "issue:43782|issue:45325", + "issue:43825|issue:43827", + "issue:43749|issue:44291", + "issue:43906|issue:46209", + "issue:43582|issue:44857", + "issue:45081|issue:45406", + "issue:42175|issue:45081", + "issue:45372|issue:46326", + "issue:43262|issue:43819", + "issue:45003|issue:46285", + "issue:34634|issue:36331", + "issue:44060|issue:44368", + "issue:43653|issue:44912", + "issue:43526|issue:43697", + "issue:43742|issue:45020", + "issue:43650|issue:43906", + "issue:45084|issue:45538", + "issue:43526|issue:44554", + "issue:43329|issue:45600", + "issue:44589|issue:45042" + ], + "prepared_review_unit_hash": "26bb9c2a906d317100e840151c983dd40ddfe023533dcac5f5a13556dedb835c", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 26, + "serialized_chars": 35154, + "estimated_input_tokens": 8789, + "estimated_eval_tokens": 17834 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: it mixes unrelated tokenizer, model-loading, processor, cache, and CI issues. None of the soft pairs look like safe duplicates of the same underlying bug/change.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue is safe here; the items span unrelated failures across many subsystems, so picking one representative would be misleading.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue stands out as a global best representative because the cluster is not centered on one bug family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both involve remote_code breakage, but one is specifically about removing is_torch_fx_available while the other is a broad recent-version regression; different root causes and fixes." + }, + { + "left": "issue:45987", + "right": "issue:46192", + "accept": false, + "reason": "These are unrelated API/behavior bugs: StaticCache sequence length vs text_kwargs/return_tensors handling." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": false, + "reason": "Both touch tokenizer behavior, but one is about added-token spacing and the other is a broad slow/fast tokenizer inconsistency; too little evidence they are the same bug." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "CamemBERT masked-LM prediction regressions and a general tokenization change report are not the same concrete issue." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Completely different components and failure modes: image processor sizing vs bfloat16 dtype mismatch." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Packed-weight loading in BitNet and dtype mismatches in inference are separate code paths with different fixes." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Both are vision-language related, but one is a from_pretrained weight_only load failure and the other is a rope-index position_id bug." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "Both relate to v5 pipeline removals, but one is an error-message wording problem and the other is stale docs; not the same bug/change." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "Both concern loading flows, but FSDP CPU RAM efficient loading and init_empty_weights argument handling are distinct issues." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "An isolated reproduction of an older issue is unrelated to FP8 inference on B200." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Apple Silicon allocator warmup and CUDA float16 AMP crashes are unrelated hardware/backend bugs." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Both are AttributeErrors, but they arise in different processors and different loading paths; not the same underlying defect." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "TensorFlow extra-install packaging and Mistral tokenizer regex patching are unrelated." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Both mention mistral_common, but one is an import-time dependency break and the other is a chat-template behavior bug; different failures and fixes." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and DAC latent reconstruction mismatch are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "Unsafe sys.modules access and CLIP text-model structural inconsistency are different model utility issues." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "BarkProcessor voice preset and CustomTrainer compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention tied-weights warnings, but the affected models, triggers, and remedies differ; not enough to merge as one bug." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and MXFP4 quantization loading are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and RTDetrV2 output drift are different model/processor regressions." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "A key error on one specific model and broad remote_code breakage are not the same concrete issue." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "No meaningful overlap; one is a vague request, the other a reproduction of a separate issue." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Template compilation TypeError and CLIPTokenizer max-length handling are unrelated tokenizer/model config bugs." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "These concern different domains: image label reduction vs MPS attention correctness." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "A multimodal-token counting bug and stale auto_mappings references are unrelated CI/runtime problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 18, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:29159", + "issue:34567", + "issue:36010", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:42490", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43450", + "issue:43452", + "issue:43454", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43606", + "issue:43638", + "issue:43643", + "issue:43645", + "issue:43696", + "issue:43720", + "issue:43742", + "issue:43746", + "issue:43749", + "issue:43819", + "issue:43854", + "issue:43927", + "issue:43931", + "issue:44387", + "issue:44479", + "issue:44568", + "issue:44610", + "issue:44617", + "issue:44756", + "issue:44811", + "issue:44912", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45042", + "issue:45356", + "issue:45459", + "issue:45759", + "issue:45803", + "issue:45859" + ], + "soft_pairs": [ + "issue:43746|issue:44811", + "issue:43452|issue:44991", + "issue:43638|issue:44936", + "issue:44387|issue:44617", + "issue:43550|issue:44912", + "issue:43819|issue:43927", + "issue:43257|issue:43329", + "issue:43854|issue:44977", + "issue:36010|issue:43645", + "issue:43720|issue:44610", + "issue:39692|issue:44912", + "issue:38175|issue:44610", + "issue:36010|issue:36331", + "issue:43531|issue:45042", + "issue:43454|issue:43577", + "issue:45459|issue:45803", + "issue:43257|issue:43696", + "issue:42490|issue:43257", + "issue:43262|issue:44811", + "issue:43643|issue:45759", + "issue:43329|issue:44479", + "issue:43329|issue:43450", + "issue:43606|issue:45859", + "issue:34567|issue:40990", + "issue:43749|issue:44756", + "issue:43257|issue:43931", + "issue:44568|issue:45356", + "issue:29159|issue:34567", + "issue:42915|issue:43742" + ], + "prepared_review_unit_hash": "7ef9706be5904a0ee202ae0542ae0d7b6bdbab1aba0bbc0a5fc1c0d4280dc3b4", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36635, + "estimated_input_tokens": 9159, + "estimated_eval_tokens": 18574 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is heterogeneous: the soft pairs share broad subsystem vocabulary but not the same underlying bug or fixable PR. I would not merge any of the candidate pairs as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:43257 is the strongest standalone bug report here: it has a concrete Qwen3 MoE loading failure, a clear reproduction context, and the most actionable description among the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Different bugs and code paths: PEFT/local checkpoint loading vs Whisper batch decoding token handling." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "Both involve loading, but one is gguf_file path handling for Auto* classes and the other is a tokenizer regression for a specific model." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Different trainer/runtime failures: DeepSpeed ZeRO-3 index error vs trainer.evaluate() breaking after train()." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both are OOMs, but from different models and likely different memory paths; not the same bug." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Different features: torch.compile+SDPA on Bamba vs MXFP4 quantization fallback on gpt-oss." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "Unrelated model logic: DAC latent reconstruction mismatch vs DiaConfig token ID persistence." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Different failures: Qwen3 MoE weight conversion under accelerate/deepspeed vs undefined helpers in a video token counting branch." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "Different models and symptoms: GLM-4.7-Flash test loading issue vs Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "Both mention loading/imports, but one is a missing GenerationMixin import and the other is a notebook/custom-model initialization regression." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Distinct subsystems: BitNet packed-weight loading vs OmDet-Turbo processor image-size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Docs example errors for SigLIP2 are not the same as the gpt-oss MXFP4 quantization fallback bug." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Different models and symptoms: zero probabilities in SigLIP2 vs incorrect processor output size in OmDet-Turbo." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Importing GenerationMixin is unrelated to the compute_loss signature change with num_items_in_batch." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior is unrelated to PIL backend image processors requiring torchvision." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Both are model-loading/architecture issues, but one is lm_head weight tying and the other is dtype propagation for Blip2/QFormer." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve exception handling, but they are different functions and different error-masking bugs." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3 MoE weight conversion is unrelated to a plain CUDA OOM on gpt-oss-20b." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "Path-slash import warning and Qwen3 MoE deepspeed loading are different problems." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Both are audio-related, but one is chat-template sampling-rate defaulting and the other is batch_decode skip_special_tokens handling." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Different config-loading bugs: missing fields with trust_remote_code vs not unwrapping text_config for composite multimodal configs." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Both are video-related, but one is undefined helper usage in multimodal token counting and the other is a broader video input regression." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined video token helper references are not the same as batched video processor shape errors." + }, + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both mention CPU offload, but they are different models and different device-mismatch failures." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counting is unrelated to GPT-oss perplexity behavior." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "Different memory-related issues: FSDP CPU RAM efficient loading vs mmap on Strix Halo." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Both are Qwen3-family loading issues, but the failure modes are different: deepspeed conversion vs VL weight-shape mismatch." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Tokenizer regressions, but on different models and different root causes (special-token insertion vs codec handling/warning behavior)." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency is unrelated to TrainerState.num_input_tokens_seen not updating." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Different model/config loading failures with no shared underlying code path." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 19, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:24643", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:38175", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43257", + "issue:43317", + "issue:43452", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43824", + "issue:43927", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44560", + "issue:44568", + "issue:44737", + "issue:44821", + "issue:44849", + "issue:44898", + "issue:44912", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45137", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45488", + "issue:45538", + "issue:45588", + "issue:45701", + "issue:45907" + ], + "soft_pairs": [ + "issue:44291|issue:45042", + "issue:42915|issue:44821", + "issue:43452|issue:43582", + "issue:44849|issue:45701", + "issue:44987|issue:45538", + "issue:43927|issue:44912", + "issue:43697|issue:45357", + "issue:42915|issue:45538", + "issue:44568|issue:45488", + "issue:41720|issue:43540", + "issue:33357|issue:34567", + "issue:44479|issue:45381", + "issue:42175|issue:43824", + "issue:43317|issue:45538", + "issue:43540|issue:44479", + "issue:44062|issue:44991", + "issue:28282|issue:34689", + "issue:45137|issue:45405", + "issue:43540|issue:44560", + "issue:24643|issue:29159", + "issue:43742|issue:44291", + "issue:44987|issue:45907", + "issue:43577|issue:45072", + "issue:28282|issue:29127", + "issue:42915|issue:43749", + "issue:43257|issue:43756", + "issue:38175|issue:44898", + "issue:42175|issue:44488", + "issue:43582|issue:43782", + "issue:44737|issue:45588" + ], + "prepared_review_unit_hash": "fea73ee08d838d48913f2cc00a4a2629278fe8e9b632f6a7e4c58f18f649ffd6", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36747, + "estimated_input_tokens": 9187, + "estimated_eval_tokens": 18630 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are false positives: they share a broad subsystem (tokenizers, loading, video, dtype, quantization) but not the same concrete bug. The only strong duplicate-looking pair is the Qwen video-input regression pair.", + "confidence": 0.9, + "canonical_issue_reason": "issue:44479 is the broadest and most central report in the only plausible duplicate pair, and it directly covers the specific qwen2.5-vl video-position-id symptom described in issue:45381.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the best representative issue for this set because it captures the shared Qwen video-input regression theme more broadly than the more specific follow-up report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "Different bugs: init_empty_weights/_is_hf_initialized loading error vs image-processor torchvision dependency regression." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Different failures: FineGrainedFP8Config model loading vs AutoImageProcessor URL loading." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "Unrelated code paths: gguf_file tokenizer/model loading vs Apple Silicon warmup TypeError." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Different scope: Qwen3.5 hidden-states bug vs a broad tokenization behavior change report." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Different issues: model loading for physical-intelligence/fast vs CLIPTokenizer model_max_length handling." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "Different bugs: DiaConfig save/load token IDs vs MXFP4 quantization fallback." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "Different model behaviors: RTDetrV2 output mismatch vs Qwen3.5 save_pretrained key regression." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Different bugs and models: Qwen3Moe FP8 config failure vs CLIPTokenizer max length issue." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer regressions, but one is add_special_tokens behavior and the other is LlamaTokenizer pre-tokenizer override; not the same bug." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Different failures: auto device mapping cuda assert vs video-input ValueError in Qwen3OmniMoe." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "Completely different symptoms: Mac bus error on CLIP vs TrainerState token-count update bug." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "Same underlying Qwen video-input regression; 45381 is a specific qwen2.5-vl symptom within the broader regression reported in 44479." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "Different areas: DeepSpeed ZeRO3 deque error vs dependency-version bump for PEFT." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Both are loading-related, but one is a key error for MobileLLM and the other is an unexpected constructor argument in init_empty_weights." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Unrelated: a model-loading failure vs a repo-files helper not forwarding revision." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both concern dtype, but they affect different models and code paths; not one concrete bug." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "Different problems: missing PyTorch import detection vs a specific Llama Vision model-loading regression." + }, + { + "left": "issue:43749", + "right": "issue:45405", + "accept": false, + "reason": "Unrelated: FSDP CPU RAM loading breakage vs minimum PEFT version metadata issue." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both mention video inputs, but the models and failure modes differ; not enough evidence of one shared bug." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "Different domains: DeepSpeed training runtime error vs tokenizer inconsistency." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Different loading failures with different root causes; no shared code-path bug shown." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Same repo surface area, but distinct bugs and fixes." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Distinct dtype issues on different model families; not mergeable as one PR." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Different packaging/import issues: missing tensorflow backend vs missing Qwen2_5_VL class export." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "Different loading paths and different symptoms; no same underlying bug." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Missing PyTorch import vs a clarity/improved error-message request for LayoutLMv3." + }, + { + "left": "issue:41720", + "right": "issue:44560", + "accept": false, + "reason": "Different Qwen video/device failures with different stacks and symptoms." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "MOE weight conversion with accelerate/deepspeed is unrelated to Smollm3 RoPE layer dropping." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Both are vision-model issues, but one is zero probabilities in SigLIP2 and the other is non-default-resolution Perceiver classification failure." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Packaging/backend availability issue vs a model-loading failure; unrelated." + }, + { + "left": "issue:43582", + "right": "issue:43782", + "accept": false, + "reason": "Different bugs: Apple Silicon warmup TypeError vs Qwen3VL weight_only loading error." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "Different code paths: XLNet CPU tensor creation vs flash-attention sink handling." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded in the packet: the only accepted pair is the clearly related Qwen video-input regression/symptom pair, and the other verdicts stay conservative by rejecting only when titles point to different model families or code paths. No obvious evidence overstatement." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 20, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:24643", + "issue:29127", + "issue:34567", + "issue:36331", + "issue:41762", + "issue:43404", + "issue:43452", + "issue:43493", + "issue:43526", + "issue:43531", + "issue:43643", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:43994", + "issue:44062", + "issue:44291", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44514", + "issue:44589", + "issue:44625", + "issue:44792", + "issue:44821", + "issue:44877", + "issue:44898", + "issue:44913", + "issue:44936", + "issue:44964", + "issue:45005", + "issue:45042", + "issue:45081", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45375", + "issue:45381", + "issue:45405", + "issue:45479", + "issue:45600", + "issue:45920", + "issue:45987", + "issue:46050", + "issue:46121", + "issue:46202" + ], + "soft_pairs": [ + "issue:43493|issue:43994", + "issue:43404|issue:45987", + "issue:34567|issue:36331", + "issue:43526|issue:45479", + "issue:44291|issue:44821", + "issue:43756|issue:45325", + "issue:44062|issue:45081", + "issue:43906|issue:45381", + "issue:44913|issue:46121", + "issue:44792|issue:44964", + "issue:45042|issue:45081", + "issue:43452|issue:44488", + "issue:43742|issue:44589", + "issue:45357|issue:45920", + "issue:43531|issue:44821", + "issue:45405|issue:45600", + "issue:43643|issue:45375", + "issue:44625|issue:44877", + "issue:24643|issue:29127", + "issue:43697|issue:45216", + "issue:43720|issue:44898", + "issue:41762|issue:44936", + "issue:44493|issue:46050", + "issue:44514|issue:45381", + "issue:43906|issue:46202", + "issue:44466|issue:45005" + ], + "prepared_review_unit_hash": "a685fa590df622e5b75e2bba4a4a9c0d04bec551cf34a655ec936342c8184131", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35939, + "estimated_input_tokens": 8985, + "estimated_eval_tokens": 18226 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a grab-bag of transformer v5/model-loading regressions, but only two soft pairs look like true duplicates: SigLIP2 output mismatch and tied-weight serialization/regression issues. Most other pairs share a subsystem name only and should stay separate.", + "confidence": 0.77, + "canonical_issue_reason": "issue:44466 is the clearest representative of the cluster\u2019s strongest concrete theme: v5 tied-weight / `lm_head.weight` serialization behavior with a reproducible regression and broad model impact.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44466 is the best single issue to anchor this cluster because it is specific, regression-oriented, and matches one of the few genuinely duplicate-looking pairs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Both describe SigLIP2 producing wrong outputs versus the expected implementation; same underlying model-behavior bug." + }, + { + "left": "issue:43404", + "right": "issue:45987", + "accept": false, + "reason": "One is a Mistral3 tied-weight bug; the other is a TrainerState counter not updating. Different code paths and symptoms." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Trainer token-count tracking vs a custom `compute_loss` signature break. Related training area, but not the same bug." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BEiT image-label reduction bug versus sequence-classification zero-loss behavior. Different components and failure modes." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "`init_empty_weights` loading error versus loading `AutoImageProcessor` from a URL. Not the same underlying issue." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both involve RoPE/position handling, but they affect different models and different logic paths." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer `AddedToken` duplication error versus a Mistral regex patch crash. Different tokenizer bugs." + }, + { + "left": "issue:43906", + "right": "issue:45381", + "accept": false, + "reason": "The first is only an isolated reproduction of another issue; the second is a Qwen2.5-VL video position-id bug. Not clearly the same defect." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "Config reload defaulting and rope-validation list handling are separate configuration bugs." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Janus image-generation test failure versus Phi-4 multimodal load failure. Different model-specific regressions." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "Image-processor torchvision dependency regression versus tokenizer regex patch crash. Different subsystems and fixes." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "GGUF loading breakage versus a specific model that also fails to load. Same broad loading domain, but not the same concrete bug." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "MobileLLM load key error versus missing Float8 storage type. Distinct runtime failures." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "Incorrect save_pretrained visual encoder keys versus wrong token IDs from AutoTokenizer. Different serialization/tokenization bugs." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior versus AutoImageProcessor URL loading. No meaningful bug match." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "Unreleased PEFT version bump versus broken auto_mappings references. CI/package-version issue versus code-reference cleanup." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Generic remote-code config field loss versus a specific Qwen3.5MoEVision config field being dropped. Similar area, but not the same bug." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation bug versus granite_speech strict-config loading failure. Different config issues." + }, + { + "left": "issue:24643", + "right": "issue:29127", + "accept": false, + "reason": "DeepSpeed training runtime error versus LayoutLMv3 error-message clarity issue. Not the same bug." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "RTDetrV2 output regression versus Qwen3.5 save_pretrained checkpoint regression. Separate models and code paths." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "BitNet packed-weight loading bug versus Perceiver interpolation failure. Same release window, different defects." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Gemma3 ZeRO-3 load index error versus trainer.evaluate() failing after training. Different failure surfaces." + }, + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "Unexpected position-id keys in many models versus sinq quantization save/load breakage. Only loosely related by serialization." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template batching crash versus video position-id bug. Both Qwen2.5-VL, but not the same code-path problem." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "An isolated reproduction note versus a Gemma pipeline roundtrip issue. No clear duplicate relationship." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": true, + "reason": "Both are about tied weights / `lm_head.weight` serialization regressions in v5, and could plausibly be fixed by one change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays grounded in the packet, and the soft-edge verdicts are mostly conservative. The two accepted pairs are defensible from the titles alone (SigLIP2 output discrepancy; tied-weight / lm_head serialization regression), while the rest are appropriately rejected as different bug surfaces. No mergeability issue applies here because these are issue pairs, not PR pairs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 21, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28282", + "issue:29159", + "issue:30064", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:35141", + "issue:38175", + "issue:43295", + "issue:43452", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43819", + "issue:43927", + "issue:44361", + "issue:44442", + "issue:44464", + "issue:44479", + "issue:44484", + "issue:44514", + "issue:44568", + "issue:44625", + "issue:44704", + "issue:44743", + "issue:44811", + "issue:44843", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45290", + "issue:45356", + "issue:45372", + "issue:45440", + "issue:45491", + "issue:45542", + "issue:45701", + "issue:45704", + "issue:45823", + "issue:46002", + "issue:46170", + "issue:46178" + ], + "soft_pairs": [ + "issue:43452|issue:43577", + "issue:43688|issue:43749", + "issue:43452|issue:44843", + "issue:45372|issue:46170", + "issue:30064|issue:33666", + "issue:28282|issue:32090", + "issue:44936|issue:45823", + "issue:43819|issue:44811", + "issue:28282|issue:33357", + "issue:43295|issue:45356", + "issue:28282|issue:35141", + "issue:43653|issue:44361", + "issue:44484|issue:44704", + "issue:44361|issue:44442", + "issue:38175|issue:43295", + "issue:45542|issue:46178", + "issue:44464|issue:44743", + "issue:43653|issue:44568", + "issue:43577|issue:43697", + "issue:44514|issue:45290", + "issue:29159|issue:33357", + "issue:44843|issue:44991", + "issue:43646|issue:43749", + "issue:43688|issue:45704", + "issue:43531|issue:43742", + "issue:43688|issue:45440", + "issue:43653|issue:43927", + "issue:45491|issue:46002", + "issue:43688|issue:45701", + "issue:44464|issue:44977", + "issue:43550|issue:43927", + "issue:44479|issue:44625" + ], + "prepared_review_unit_hash": "4ccfd8d4285794da75d7e74380410aff3cc0d8d412f53c4dcfa75eb943e41e8b", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37414, + "estimated_input_tokens": 9354, + "estimated_eval_tokens": 18964 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly a grab bag of unrelated Transformers bugs. There is a loose tokenizer/processor regression theme, but none of the proposed soft pairs look like the same concrete defect, so all soft edges are rejected.", + "confidence": 0.7, + "canonical_issue_reason": "issue:45701 is the broadest and most central tokenizer-regression report in the set, and it already attracts other related reports via references.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45701 is the best representative issue for the cluster\u2019s loose tokenizer-change theme because it is broad, still open, and closer to an umbrella report than the narrower model-specific bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "Different subsystems: tokenizer/model loading with gguf_file vs Blip2 dtype propagation." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Different bugs: auxiliary loss normalization vs FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both involve loading, but one is gguf_file support and the other is offline model_info handling in tokenizer patching." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Unrelated failures: Gemma processor import error vs Windows ROCm torch.distributed dependency issue." + }, + { + "left": "issue:30064", + "right": "issue:33666", + "accept": false, + "reason": "Image processor void-segmentation failure is unrelated to Qwen2-VL multi-GPU training." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "PyTorch missing import error and trainer NoneType broadcast error are different code paths." + }, + { + "left": "issue:44936", + "right": "issue:45823", + "accept": false, + "reason": "Trainer evaluate-after-train issue is unrelated to Gemma4 device-map mismatch." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC latent conversion mismatch is unrelated to Whisper batch_decode special-token handling." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Missing PyTorch import and MacOS CLIP bus error are distinct problems." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but one is processor.tokenizer/image passing and the other is Kimi codec handling; not the same bug." + }, + { + "left": "issue:28282", + "right": "issue:35141", + "accept": false, + "reason": "ImportError for PyTorch and embedding reinitialization after resize are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "Different tokenizer bugs affecting different models and failure modes." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "save_pretrained shard size question is unrelated to AutoProcessor kwargs forwarding." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "MLukeTokenizer AttributeError and FastSpeech2ConformerTokenizer loading failure are not the same defect." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "SigLIP2 zero probabilities are unrelated to the processor.tokenizer regression." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Both touch TF backend detection, but one is missing TensorFlow with tensorboard installed and the other is an unmapped tensorflow_text backend." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Compiled chunked generation inconsistency and recurrent-state reset under cache are separate generation bugs." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both are special-token issues, but they affect different tokenizers and different behaviors." + }, + { + "left": "issue:43577", + "right": "issue:43697", + "accept": false, + "reason": "Blip2 dtype loading and RTDetrV2 output divergence are unrelated." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both are apply_chat_template crashes, but the triggers differ: batched padding=False vs tool-call messages with no content." + }, + { + "left": "issue:29159", + "right": "issue:33357", + "accept": false, + "reason": "Slow/fast tokenizer inconsistency is unrelated to the CLIP MacOS bus error." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Both involve tokenizer loading, but one is HF_HUB_OFFLINE/model_info patching and the other is a specific EMBEDDIA tokenizer regression." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Custom model initialization breakage and FSDP loading regression are different issues." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Aux-loss normalization and T5 apex RMSNorm memory leak are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and MobileLLM key error are different bugs." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Aux-loss normalization is unrelated to DeepseekV3MoE implementation divergence." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "BigBird special-token registration and DiaConfig custom token IDs are distinct tokenizer/config bugs." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Gemma3 NaNs from sliding-window attention are unrelated to classification head NaNs on checkout/init." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux-loss normalization does not match the broad tokenization-change report." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled chunked generation inconsistency and flash-attention generation failure are different code paths." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure is unrelated to DiaConfig token-ID loss." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Video-input regression in Qwen VL models is unrelated to Qwen3.5 num_labels propagation." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 22, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28218", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:34567", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43643", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43756", + "issue:43825", + "issue:43927", + "issue:43994", + "issue:44361", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44554", + "issue:44561", + "issue:44610", + "issue:44625", + "issue:44821", + "issue:44849", + "issue:44913", + "issue:44964", + "issue:44977", + "issue:45405", + "issue:45588", + "issue:45823", + "issue:46002", + "issue:46050", + "issue:46192", + "issue:46209" + ], + "soft_pairs": [ + "issue:41720|issue:44464", + "issue:43643|issue:44561", + "issue:43550|issue:44361", + "issue:44977|issue:46209", + "issue:32090|issue:34567", + "issue:43577|issue:45588", + "issue:43577|issue:44554", + "issue:32090|issue:33666", + "issue:44466|issue:46050", + "issue:43504|issue:43720", + "issue:33666|issue:36010", + "issue:43504|issue:43550", + "issue:43504|issue:44610", + "issue:43742|issue:44821", + "issue:45823|issue:46050", + "issue:43262|issue:43295", + "issue:43452|issue:44451", + "issue:43720|issue:43927", + "issue:43606|issue:43927", + "issue:43994|issue:46192", + "issue:44849|issue:46209", + "issue:33666|issue:36331", + "issue:43278|issue:43688", + "issue:43525|issue:43572", + "issue:44625|issue:46002", + "issue:43720|issue:44361", + "issue:43756|issue:44913", + "issue:43493|issue:43825", + "issue:28218|issue:33357", + "issue:39692|issue:43493", + "issue:44964|issue:45405" + ], + "prepared_review_unit_hash": "74135cd37031932f82489f28c6e35596862d5fd9d715a053e5997c298e4dbd92", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36990, + "estimated_input_tokens": 9248, + "estimated_eval_tokens": 18752 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose mix of unrelated Transformers bug reports around loading, config, training, and model-specific regressions. None of the soft pairs look like safe duplicates, so I would not merge any of them. If forced to pick one representative issue, the broadest cluster anchor is the v5 `trust_remote_code` regression.", + "confidence": 0.83, + "canonical_issue_reason": "`issue:44561` is the broadest, most central symptom in the set: a core v5 regression that breaks `trust_remote_code` models. It is more representative of the cluster than the narrower model-specific or single-path failures.", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:44561` is the best stand-in for this loose cluster because it names a concrete upstream regression with wider impact and clearer actionability than the other, more model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "Different failures: device-mapping CUDA assert vs compiled-forward chunked generation inconsistency. Same general area, not the same bug." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Both mention `trust_remote_code`, but one is missing config fields from `AutoConfig` while the other is a v5 `is_torch_fx_available` removal regression. Different code paths and fixes." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Unrelated: SDPA/torch.compile failure in Bamba vs tokenizer AttributeError in MLuke tasks." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/model issues: Qwen3.5 flash-attention generation bug vs FP8 inference failure on B200." + }, + { + "left": "issue:32090", + "right": "issue:34567", + "accept": false, + "reason": "Trainer broadcast TypeError and TrainerState token counter not updating are separate training bugs." + }, + { + "left": "issue:43577", + "right": "issue:45588", + "accept": false, + "reason": "Blip2 dtype mismatch on load is not the same as flash-attention crashing on `s_aux=None`." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Different subsystems: model dtype propagation vs an MPS attention correctness issue." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Generic multi-GPU/trainer error versus Qwen2-VL multi-GPU training request; not the same underlying defect." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Both are save/load related, but one is tied-weight serialization and the other is sinq quantized model persistence. Not one concrete bug." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Legacy field loading failure in BEiT is unrelated to BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training for a vision-language model is unrelated to a `GenerationMixin` import error." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Model-loading legacy field bug is not the same as a compile/SDPA runtime failure." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Different model-specific loading/preprocessing failures; no shared underlying code path." + }, + { + "left": "issue:43742", + "right": "issue:44821", + "accept": false, + "reason": "KeyError loading MobileLLM is unrelated to loading an AutoImageProcessor from URL." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Device-mapping forward mismatch and quantized save/load corruption are separate problems." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio `apply_chat_template` sampling-rate defaulting and processor/tokenizer regression are different issues." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Both are load failures, but for different artifacts and failure modes (`gguf_file` path vs loading ScandiBERT)." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Packed-weight loading in BitNet and DiaConfig losing token IDs after save/load are distinct persistence bugs." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch in bark-small is unrelated to DiaConfig save/load token loss." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 bad outputs and `return_tensors` being ignored are unrelated API bugs." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 hidden-state output bug is not the same as FP8 inference failure on B200." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Qwen2-VL multi-GPU training and `compute_loss` signature mismatch are different training issues." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "One is dtype drift between train/eval; the other is auxiliary-loss normalization. Same broad area, not one bug." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Both are missing-pad-token config regressions, but for different models and different fields (`pad_token_id` vs `pad_token_idx`)." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "`num_labels` propagation in Qwen3.5 is unrelated to NaNs in classification heads." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet loading/unpacking and MLukeTokenizer task AttributeError are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:44913", + "accept": false, + "reason": "Smollm3 RoPE-layer count bug and GPTNeoX rotary_pct reload issue both touch RoPE, but they are not the same defect." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy is unrelated to a pipeline error message about translation tasks." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer added-space behavior and MacOS bus error in CLIP are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Docs example mistakes are not the same as an HF-vs-JAX implementation discrepancy." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Model loading failure and an unreleased PEFT version bump are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 23, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:33357", + "issue:33666", + "issue:39692", + "issue:43257", + "issue:43278", + "issue:43295", + "issue:43299", + "issue:43334", + "issue:43493", + "issue:43504", + "issue:43540", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44112", + "issue:44291", + "issue:44361", + "issue:44464", + "issue:44466", + "issue:44488", + "issue:44493", + "issue:44568", + "issue:44589", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:44977", + "issue:45083", + "issue:45200", + "issue:45229", + "issue:45357", + "issue:45405", + "issue:45520", + "issue:45584", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:46050", + "issue:46285" + ], + "soft_pairs": [ + "issue:44849|issue:45357", + "issue:44568|issue:45812", + "issue:43540|issue:44977", + "issue:45200|issue:45229", + "issue:43493|issue:43844", + "issue:43278|issue:43844", + "issue:44488|issue:44843", + "issue:39692|issue:43295", + "issue:44493|issue:46285", + "issue:43756|issue:45083", + "issue:43646|issue:44466", + "issue:43257|issue:43299", + "issue:43334|issue:43572", + "issue:33357|issue:33666", + "issue:43334|issue:45759", + "issue:45229|issue:45823", + "issue:44589|issue:45520", + "issue:43927|issue:44361", + "issue:43824|issue:44589", + "issue:43504|issue:43606", + "issue:43278|issue:43701", + "issue:43572|issue:44291", + "issue:44464|issue:45584", + "issue:43575|issue:45405", + "issue:43504|issue:43653", + "issue:44792|issue:45405", + "issue:43606|issue:44112", + "issue:43994|issue:44493", + "issue:43873|issue:46050" + ], + "prepared_review_unit_hash": "bffba67cf4554cc523ffee0b853e2bf13c8d284b20b651fae0018bf8456374bd", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36546, + "estimated_input_tokens": 9137, + "estimated_eval_tokens": 18530 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the paired items share broad model-family or regression themes, but none appear to be the same concrete bug/change. I would not merge any soft edge pair.", + "confidence": 0.91, + "canonical_issue_reason": "No true duplicate center stands out; issue 43873 is the broadest currently open item with the most discussion, so it is the best representative if one must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the most general and most active issue in the set, making it the safest single representative for the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Both concern Qwen3.5, but one is about output_hidden_states and the other about save_pretrained visual encoder keys; different code paths and failures." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Tokenizer-related, but one is a special-tokens/BOS-EOS regression for mdeberta and the other is Granite token ID mapping; not the same defect." + }, + { + "left": "issue:43540", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen3-family multimodal behavior, but one is a video-processing ValueError and the other is a flash-attention generation issue." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "One is about missing mm_token_type_ids defaults for text-only fine-tuning, the other is a multi-GPU inference CUDA OOM; unrelated failures." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 is the only shared theme; one is an implementation discrepancy, the other is a DeepSpeed ZeRO-3 gradient increase bug." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype drift in eval is unrelated to the ZeRO-3 gradient anomaly, despite both involving training/evaluation stability." + }, + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "One is a model load failure for a specific checkpoint, the other is an offline-mode failure caused by an unconditional hub call." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Both are regression-style model issues, but one is a SigLIP2 doc example/model-processor mismatch and the other is a custom processor.tokenizer API break." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "Both mention unexpected keys/structure, but one is a position-id key warning across models and the other is a CLIPTextModel class-structure inconsistency." + }, + { + "left": "issue:43756", + "right": "issue:45083", + "accept": false, + "reason": "Smollm3 RoPE-layer loss and qwen3_omni_moe feature-length helper behavior are different implementation bugs." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "One breaks custom model initialization in v5, the other is inconsistent lm_head serialization; different code paths and symptoms." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen3 MoE loading, but one is accelerate+deepspeed conversion and the other is a transformers dev loading regression for Qwen3VL MoE models." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Both are missing-config-attribute bugs, but for different configs and different attribute names; not the same underlying issue." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "Completely different areas: MacOS bus error loading CLIP vs Qwen2-VL multi-GPU training." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Both concern multimodal Qwen configs, but one is a missing pad_token_id attribute and the other is unwrapping text_config in AutoModelForCausalLM.from_config." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4 multi-GPU/device issues, but one is OOM and the other is a device mismatch during forward." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Unrelated import/runtime failures: missing Float8 storage vs flash_attn KeyError on Python 3.13." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer/config issues, but one is save/load token ID loss and generation IndexError, while the other is an AttributeError in MLukeTokenizer tasks." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Different failures: missing Qwen2_5_VL import vs Float8 storage lookup/type error." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "Legacy-field loading bug in BeitForSemanticSegmentation is unrelated to bark-small CPU offload device mismatch." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift during eval and resume_from_checkpoint key mismatch are unrelated problems." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression and init_empty_weights unexpected argument error are different initialization bugs." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Chunked generation inconsistencies with compiled forward and Whisper empty-transcription failure are unrelated generation bugs." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "Qwen2 large-model TP OOM and an unsupported PEFT version bump have no shared defect." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "A pretrained-model legacy-field load failure is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and a PEFT version requirement issue are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "One is a real CPU-offload device mismatch in bark-small; the other is a stale CI test in GraniteSpeech." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 inference corruption and a generic position-id key warning are not the same bug." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Both mention quantization/offloading, but one is offloading not behaving as expected and the other is sinq save/load breakage; different code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 24, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:29159", + "issue:29942", + "issue:30333", + "issue:32090", + "issue:34567", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:42490", + "issue:43065", + "issue:43232", + "issue:43295", + "issue:43450", + "issue:43493", + "issue:43540", + "issue:43582", + "issue:43696", + "issue:43701", + "issue:43844", + "issue:43957", + "issue:44186", + "issue:44423", + "issue:44521", + "issue:44554", + "issue:44610", + "issue:44734", + "issue:44849", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45084", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45381", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45588", + "issue:45823", + "issue:46050", + "issue:46121", + "issue:46192", + "issue:46209", + "issue:46365", + "issue:46369" + ], + "soft_pairs": [ + "issue:43065|issue:46369", + "issue:45588|issue:46050", + "issue:44423|issue:44734", + "issue:29942|issue:30333", + "issue:42490|issue:43696", + "issue:45406|issue:46121", + "issue:43844|issue:44964", + "issue:29159|issue:32090", + "issue:28218|issue:34567", + "issue:43582|issue:44977", + "issue:44554|issue:45446", + "issue:45084|issue:45520", + "issue:15354|issue:38175", + "issue:43540|issue:45200", + "issue:44521|issue:45290", + "issue:45200|issue:45823", + "issue:45072|issue:45198", + "issue:37428|issue:43957", + "issue:43493|issue:43701", + "issue:44186|issue:46192", + "issue:43232|issue:44849", + "issue:45823|issue:46192", + "issue:45381|issue:46209", + "issue:43295|issue:43450", + "issue:44554|issue:46365", + "issue:43295|issue:44610", + "issue:39692|issue:46192" + ], + "prepared_review_unit_hash": "bff1ebee622686eafe14607aa67e20765bc2df790c3d9b3be43b92eac36e392d", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35839, + "estimated_input_tokens": 8960, + "estimated_eval_tokens": 18176 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set looks like a heterogeneous grab bag of unrelated issues; the soft-similarity pairs do not appear to be duplicate reports of the same underlying bug or change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Different Sam3 problems: one is a dummy Conv2d placeholder in Sam3PixelDecoder, the other is a memory overflow when using sam3-litetext." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "Unrelated failure modes: flash_attention.py crashes on s_aux=None, while sinq quantization breaks after save/load." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both hit serve crashes, but one is multimodal continuous-batching with a string/.to error and the other is KV-cache continuation tensor indexing." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Completely different subsystems: Flash Attention 2 test failures vs MLflow reporting not updating after a failed training job." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "No overlap beyond being runtime issues: local-path import warning vs GPT-oss-20b CUDA OOM." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Different code paths: Gemma4Processor missing _tokenizer in serve, versus JSON-loaded list type handling in rope validation." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "Different problems and models: ZeRO-3 gradient growth vs Phi-4 multimodal loading failure." + }, + { + "left": "issue:29159", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency is unrelated to the Trainer NoneType GPU broadcast error." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Both concern tokenizer/training behavior, but one is added-token spacing and the other is TrainerState token counting not updating." + }, + { + "left": "issue:43582", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated: Apple Silicon caching_allocator_warmup TypeError versus Qwen3.5 flash-attention generation bug." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "Different layers: MPS attention correctness with mismatched head dims vs a PyTorch version check for flex_attention imports." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Different root causes: template-node compilation error versus import_utils KeyError for flash_attn on Python 3.13." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "No meaningful match: torch.jit.script GeneratorExp export issue versus SigLIP2 zero-probability outputs." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Both are multimodal-related, but one is video input processing in Qwen3OmniMoe and the other is Gemma 4 token-type IDs for text-only fine-tuning." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Same API family but different bugs: all-zero assistant masks for multimodal inputs versus crash on tool-call assistant messages with empty content." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Different Gemma4 issues: defaulting mm_token_type_ids to zeros versus device-mismatch behavior with device_map=\"auto\"." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "Different model classes and symptoms: dtype mismatches in inference versus Wav2Vec2 save_pretrained/tokenization failure." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Both involve model loading/import paths, but one is a missing flash-attention symbol import and the other is meta-device loading breaking some models." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "Different concerns: SigLIP2 HF-vs-JAX discrepancy versus checkpoint resume key mismatch." + }, + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "Tokenizer-related but not the same bug: LayoutLMv2Tokenizer crashes on NER/batched padding, while return_tensors is ignored when text_kwargs is passed." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Both are generation-related, but they target different code paths: sync_gpus kwargs update versus output_hidden_states handling in Qwen3.5." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "Different areas: Gemma4 device placement with auto mapping versus return_tensors precedence behavior." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Completely different topics: Qwen2.5-VL video vision_position_ids versus FP8 inference failure on B200." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Both are processor issues, but one is processor.tokenizer access/images-to-tokenizer regression and the other is batched video processor shape." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "Unrelated: MPS attention correctness bug versus an import conflict around Trainer and datasets." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Different processor regressions: custom code accessing processor.tokenizer/passing images to tokenizer versus OmDet-Turbo processor output size mismatch." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 documentation/model-processor mismatch and quantization failure is unrelated to return_tensors being ignored with text_kwargs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 25, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28218", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43388", + "issue:43450", + "issue:43493", + "issue:43525", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43653", + "issue:43824", + "issue:43844", + "issue:43856", + "issue:43873", + "issue:43950", + "issue:43957", + "issue:43975", + "issue:44112", + "issue:44442", + "issue:44466", + "issue:44554", + "issue:44792", + "issue:44898", + "issue:44987", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45325", + "issue:45356", + "issue:45412", + "issue:45520", + "issue:45563", + "issue:46178", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46326" + ], + "soft_pairs": [ + "issue:43572|issue:44987", + "issue:44554|issue:45356", + "issue:43295|issue:43873", + "issue:43572|issue:43957", + "issue:43950|issue:46285", + "issue:45520|issue:46178", + "issue:44466|issue:45127", + "issue:43550|issue:43844", + "issue:37428|issue:43824", + "issue:44898|issue:45412", + "issue:45200|issue:45325", + "issue:43450|issue:43575", + "issue:43334|issue:43525", + "issue:44792|issue:45563", + "issue:41720|issue:46209", + "issue:42490|issue:43493", + "issue:43388|issue:43526", + "issue:42915|issue:43824", + "issue:43824|issue:43975", + "issue:28218|issue:35141", + "issue:43278|issue:43856", + "issue:38175|issue:43493", + "issue:43526|issue:46192", + "issue:45290|issue:46326", + "issue:43653|issue:44112", + "issue:43975|issue:45520", + "issue:44442|issue:45198", + "issue:43262|issue:43450", + "issue:38175|issue:45412" + ], + "prepared_review_unit_hash": "d7fdbf54992daa34ddf72dab138198064c2029b6b6cd54cd8bc63897a51c70c9", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36814, + "estimated_input_tokens": 9204, + "estimated_eval_tokens": 18664 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a grab bag of unrelated Transformers bug reports that only share broad surface similarity (tokenizers, model loading, configs, processors, regressions). I do not see any soft pair that clearly describes the same underlying bug or change, so no duplicate edges should be accepted.", + "confidence": 0.29, + "canonical_issue_reason": "issue:43957 is the broadest representative of the cluster\u2019s dominant theme: a general Transformers 5.x model-loading regression involving `torch.device(\"meta\")`, which is more central and reusable as a cluster anchor than the many narrower, model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43957 is the most generally useful issue to keep as the representative problem statement because it covers a cross-model loading failure rather than a single model or one-off symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "Both are loading-related, but one is a missing config field and the other is a different failure when loading a specific model; not the same bug." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "Different subsystems entirely: MPS attention correctness vs tokenizer regression in Kimi-K2.5." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "One is a processor/tokenizer regression, the other is quantization offloading behavior; no shared code-path bug." + }, + { + "left": "issue:43572", + "right": "issue:43957", + "accept": false, + "reason": "Both involve loading, but one is a StableLmConfig attribute issue and the other is a meta-device load regression affecting other models." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Non-persistent buffer corruption on load is unrelated to CLIPTextModel structural inconsistency." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "Different import/backend lookup failures in unrelated modules." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both mention tied weights, but one is serialization/device-dependent and the other is LoRA merge collapse after vocab extension; not the same concrete bug." + }, + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "A torch.compile+SDPA model failure and a DeepSpeed gradient issue are not the same problem." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different missing symbols and different modules." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Image resolution handling vs model memory release are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Different multimodal bugs: missing mm token type defaults vs incorrect RoPE scaling." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video processor shape handling and tensor-parallel OOM are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Same symptom class (missing `pad_token_id`), but different model families and likely separate config implementations." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "A failed image-generation test and a paged-generate warning are not the same bug." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/backend failures for different model classes." + }, + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "Local-path import warning and SigLIP2 implementation discrepancy are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Both involve labels, but one is metrics gathering and the other is image processor label reduction." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "FineGrainedFP8Config failure and missing model import are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Importing a class from transformers and incorrect detokenization are different bugs." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer spacing after added tokens and embedding reinitialization on resize are different code paths." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Training/eval dtype change is unrelated to Qwen3 MoE memory usage." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": false, + "reason": "Both mention SigLIP2, but one is zero probabilities and the other is a broader HF-vs-JAX discrepancy; not enough to treat as the same bug." + }, + { + "left": "issue:43526", + "right": "issue:46192", + "accept": false, + "reason": "Image label reduction and `return_tensors` being ignored are unrelated processor bugs." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both are chat-template related, but one is a tool-call crash and the other is `continue_final_message`; different failure modes." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Tokenizer special-token registration and a GraniteSpeech CI test issue are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Tokenizer detokenization regression and Python 3.13 flash-attn import failure are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Different tokenizers/models and different failure modes; not the same underlying issue." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and batched video-shape handling are unrelated processor bugs." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 probability behavior and RT-DETR memory release are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 26, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:32090", + "issue:37428", + "issue:39692", + "issue:43278", + "issue:43335", + "issue:43425", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43901", + "issue:43931", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44448", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44743", + "issue:44855", + "issue:44857", + "issue:45083", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45381", + "issue:45520", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45759", + "issue:45823", + "issue:45859", + "issue:46170", + "issue:46234", + "issue:46285" + ], + "soft_pairs": [ + "issue:45759|issue:46285", + "issue:45520|issue:46170", + "issue:44514|issue:44521", + "issue:44610|issue:44857", + "issue:44561|issue:46234", + "issue:43827|issue:43901", + "issue:44206|issue:45584", + "issue:43819|issue:44112", + "issue:43819|issue:45083", + "issue:44857|issue:45593", + "issue:43335|issue:43550", + "issue:43606|issue:45083", + "issue:37428|issue:43425", + "issue:29942|issue:32090", + "issue:44448|issue:44743", + "issue:45127|issue:45245", + "issue:45200|issue:45588", + "issue:43611|issue:43931", + "issue:43335|issue:43653", + "issue:44855|issue:45083", + "issue:43550|issue:44112", + "issue:43819|issue:44493", + "issue:44206|issue:45823", + "issue:45290|issue:45381", + "issue:37428|issue:44351", + "issue:43335|issue:43504", + "issue:29127|issue:32090", + "issue:45588|issue:45859", + "issue:43278|issue:43825", + "issue:39692|issue:43526" + ], + "prepared_review_unit_hash": "66fba942fccfd485e39d0179ab849448eb9ec397ad710c60c72b22656f059250", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36401, + "estimated_input_tokens": 9101, + "estimated_eval_tokens": 18458 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Heterogeneous cluster of mostly unrelated issues spanning model loading, chat templates, import compatibility, CI, and processor bugs; none of the soft pairs look like true duplicates or mergeable PR-style overlaps.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue: the cluster mixes unrelated regressions across chat templates, imports, configs, and CI, so no item clearly subsumes the rest.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong best issue for dedupe anchoring; the soft pairs are mostly superficial theme matches rather than the same underlying bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Different problems: composite multimodal config unwrapping vs CLIP text-model structure inconsistency." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "Both are import/runtime compatibility issues, but on different platforms and different missing dependencies." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Same area, but one is a batching/padding crash and the other is incorrect assistant masks for multimodal inputs." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "Processor input-size mismatch vs AMP float16 loss crash; different code paths and failure modes." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Both involve import_utils, but they concern different missing helpers and different breakages." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs-related, but one is about pipeline removals in v5 and the other about return_all_scores behavior." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Different bugs: unsupported feature-extractor arg vs Whisper empty transcription after special-token alignment." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent mismatch vs a stale CI device override test; not the same underlying defect." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "Unrelated code paths: DAC from_latents mismatch vs qwen3_omni_moe helper length behavior." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "Loss AMP crash vs auxiliary-loss logic when denoising is off; different symptoms and fixes." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "Config bug creating sparse layers at zero vs torch.compile/SDPA failure; distinct issues." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch vs helper output-length behavior; no shared underlying bug." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Missing internal import vs Torch 2.10 incompatibility; one is an API reference issue, the other a version break." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures and NoneType broadcast errors are unrelated failures." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Different model behavior bugs in different modules; not one fixable code-path problem." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA merge collapse and category-count limit errors are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "mm_token_type_ids defaulting issue vs flash_attention s_aux=None crash; different components." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Base-model-prefix loading regression vs Qwen3-VL weight-shape mismatch; separate model-loading problems." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Sparse-layer config bug vs tokenizer special-token registration bug; different subsystems." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Python 3.13 import parse error vs helper length behavior; no duplication." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "torch.compile SDPA failure and a stale CI device-override test are not the same bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different issues: DAC latent mismatch vs unexpected position-id keys in model loading." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Unsupported feature-extractor arg vs Gemma4 device-map mismatch; unrelated." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call crash vs wrong vision_position_ids for video input; distinct bugs." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Missing flash-support import vs missing HybridCache import; different symbols and breakages." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Sparse-layer creation bug vs BEiT legacy-field loading bug; unrelated." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "LayoutLMv3 error-message clarity issue vs Trainer NoneType broadcast crash; not duplicates." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Both mention offload/attention-adjacent code, but one is an AttributeError on s_aux and the other is CPU-offload incompatibility in a rotary embedding." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype regression vs pipeline task-support messaging; no common bug." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "Docs example errors vs BeitImageProcessorFast reduce_labels bug; different artifacts and fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 27, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:28218", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:41720", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43335", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43701", + "issue:43720", + "issue:43746", + "issue:43854", + "issue:43873", + "issue:43927", + "issue:44112", + "issue:44360", + "issue:44448", + "issue:44493", + "issue:44561", + "issue:44779", + "issue:44857", + "issue:44977", + "issue:44987", + "issue:45072", + "issue:45083", + "issue:45229", + "issue:45310", + "issue:45459", + "issue:45478", + "issue:45593", + "issue:45698", + "issue:45920", + "issue:46192", + "issue:46209", + "issue:46234" + ], + "soft_pairs": [ + "issue:45310|issue:45478", + "issue:45459|issue:46234", + "issue:43335|issue:45083", + "issue:39692|issue:44493", + "issue:43720|issue:44112", + "issue:44448|issue:44779", + "issue:43335|issue:44112", + "issue:43335|issue:45593", + "issue:42490|issue:43854", + "issue:34689|issue:35141", + "issue:43606|issue:44857", + "issue:44360|issue:44561", + "issue:45920|issue:46192", + "issue:38175|issue:44493", + "issue:44977|issue:45229", + "issue:44857|issue:45072", + "issue:44448|issue:44977", + "issue:28218|issue:32090", + "issue:43746|issue:46209", + "issue:40444|issue:43526", + "issue:43701|issue:45083", + "issue:32090|issue:33357", + "issue:29942|issue:35141", + "issue:43550|issue:45083", + "issue:43873|issue:45229", + "issue:43927|issue:45698", + "issue:42490|issue:43575", + "issue:44561|issue:44779", + "issue:10000|issue:15354", + "issue:43278|issue:43329", + "issue:41720|issue:45229", + "issue:37428|issue:44987" + ], + "prepared_review_unit_hash": "5fdbf31e0516f5ec74fdd918e21b708c2d8e0b133ea8f9a24f61c778dda471ef", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37339, + "estimated_input_tokens": 9335, + "estimated_eval_tokens": 18926 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The only clear duplicate pair is the Qwen3.5 MoE `from_pretrained` regression. All other soft pairs describe different models, subsystems, or failure modes and should stay separate.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45310 is the earlier, more detailed report of the same `transformers>=5.4.0` Qwen3.5 MoE `from_pretrained` failure; issue:45478 reads like a later duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the best representative issue here because it is the clearest and most established report of the only true duplicate pair in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same exact Qwen3.5 MoE `from_pretrained` regression on `transformers>=5.4.0`; later report duplicates the earlier one." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Unrelated bugs: tokenizer error handling with missing protobuf vs a missing import_utils symbol for Falcon Mamba." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "Different components and failures: SwitchTransformers sparse-layer config vs qwen3_omni_moe feature-length helper behavior." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 doc/quantization issues are not the same as the broad unexpected-position-id regression." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Different problems: packed-weight loading in BitNet vs a stale CI device-override test in GraniteSpeech." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Different regressions: Pegasus v4/v5 output mismatch vs DeepSeek tokenizer changes in v5." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "No shared underlying bug; config-layer creation and GraniteSpeech CI test failure are distinct." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "SwitchTransformers config bug and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "Local-path trailing-slash warning is unrelated to GLM-4.7-Flash loading failures." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Model-loading regression for Llama 3.2 Vision is unrelated to embedding reinitialization after resize." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch and float16 AMP crash are different code paths and symptoms." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA indexer activation question is unrelated to the removal of `is_torch_fx_available` breaking trust-remote-code models." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Wrong token IDs in AutoTokenizer is a different bug from `return_tensors` being ignored with explicit `text_kwargs`." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "Zero probabilities in SigLIP2 are not the same as unexpected position-id keys across models." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Flash-attention generation issue for Qwen3.5 is unrelated to Gemma4 multi-GPU CUDA OOM." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Different dtype/runtime issues in different models; not the same underlying bug." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Pegasus v4/v5 output drift and Qwen3.5 flash-attention generation failures are unrelated." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer spacing bug and Trainer `_gpu_broadcast_one` NoneType error are unrelated." + }, + { + "left": "issue:43746", + "right": "issue:46209", + "accept": false, + "reason": "PEFT local-checkpoint loading issue is unrelated to FP8 inference failure on B200." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Qwen2.5-VL iterable-dataset multi-image failure is unrelated to BEiT label reduction behavior." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "Checkpoint key mismatch and qwen3_omni_moe helper behavior are different issues." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer broadcast type error and MacOS bus error with CLIP are unrelated." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failures are unrelated to embedding dtype reinitialization on resize." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Bamba torch.compile SDPA failure is unrelated to qwen3_omni_moe output-length helper behavior." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization offloading issue and Gemma4 OOM are different failure modes." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Custom token ID loss on save/load is different from loading the wrong custom module after save_pretrained." + }, + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "Trailing-slash path warning is unrelated to tp-induced OOM when loading Qwen2-57B-A14B-Instruct." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Removed utility breaking trust_remote_code models is unrelated to DeepSeek tokenizer regression." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel issue and GeneratorExp torch.jit.script export failure are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype mismatch in evaluation is unrelated to undefined variables in multimodal token counting." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3 device-mapping assert on A800 is a different bug from Gemma4 multi-GPU OOM." + }, + { + "left": "issue:37428", + "right": "issue:44987", + "accept": false, + "reason": "Missing flash-attention import symbol and loading physical-intelligence/fast failure are not the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: issue:45310 and issue:45478 have identical titles and later/earlier timestamps, so treating them as the duplicate pair is reasonable. The other pairings are described conservatively as unrelated based on clearly different titles and failure modes. No overstatement stands out." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 28, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:25251", + "issue:29127", + "issue:29942", + "issue:30064", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:39692", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43278", + "issue:43381", + "issue:43412", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43873", + "issue:43976", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44361", + "issue:44448", + "issue:44466", + "issue:44479", + "issue:44568", + "issue:44743", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:45198", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45712", + "issue:45803", + "issue:45823", + "issue:46121", + "issue:46285" + ], + "soft_pairs": [ + "issue:44977|issue:45563", + "issue:44857|issue:45588", + "issue:44568|issue:45491", + "issue:45372|issue:45588", + "issue:43278|issue:43381", + "issue:43412|issue:45588", + "issue:43611|issue:46285", + "issue:45588|issue:45823", + "issue:45372|issue:45406", + "issue:44448|issue:44466", + "issue:43976|issue:45440", + "issue:45198|issue:45803", + "issue:25251|issue:32090", + "issue:45245|issue:45563", + "issue:43653|issue:44743", + "issue:43611|issue:46121", + "issue:32090|issue:33290", + "issue:44206|issue:44479", + "issue:45459|issue:45712", + "issue:29127|issue:34567", + "issue:33357|issue:39692", + "issue:44186|issue:45198", + "issue:42490|issue:43873", + "issue:43065|issue:45412", + "issue:44871|issue:44977", + "issue:43065|issue:43994", + "issue:44361|issue:45198", + "issue:42915|issue:42994", + "issue:45372|issue:45459", + "issue:29942|issue:30064", + "issue:43278|issue:43575" + ], + "prepared_review_unit_hash": "d14fdb644b9ed7763a8d7a0065a48535d320624ba6cb87908c23461d1a0876db", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37175, + "estimated_input_tokens": 9294, + "estimated_eval_tokens": 18844 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These are mostly distinct bug reports across unrelated components; the soft similarity pairs do not describe the same underlying issue, so none should be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single true canonical issue emerges because the set is not a duplicate cluster; if a representative is needed, issue:45372 is the clearest, most actionable report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45372 is the strongest standalone representative: it has a concrete failure, version scope, and enough context to reproduce, but it is not a duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Qwen3.5 flash-attn generation regression vs a stale warning in paged generate; different symptoms and code paths." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "LwDetrImageLoss AMP/CUDA crash vs flash_attention.py failing on s_aux=None; unrelated failures." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer special-token omission vs Gemma3 NaN embeddings on mixed-length GPU batches; distinct bugs." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma4 processor import failure vs flash_attention AttributeError; different failure points." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Embedding dtype drift in eval vs gradient-checkpointing disallowed in eval mode; related to eval but not the same bug." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export guard/erfinv issue vs flash_attention None handling; different paths and fixes." + }, + { + "left": "issue:43611", + "right": "issue:46285", + "accept": false, + "reason": "base_model_prefix loading regression vs CLIP text model structure inconsistency; separate model-loading issues." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "flash_attention None crash vs Gemma4 PLE device mismatch; not the same underlying problem." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Both involve Gemma4, but one is processor import loading and the other is transformers serve missing _tokenizer; distinct code paths." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Different outputs for Pegasus vs inconsistent lm_head serialization; both v5 regressions but not one bug." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility regression vs DeepseekV3MoE divergence; unrelated." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 save/tokenization failure vs a bare-except bug in Fuyu batching; different model/util code." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting bug vs Trainer _gpu_broadcast_one NoneType error; unrelated subsystems." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count runtime limit vs stale generate warning; different generate-related issues." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration vs Qwen3.5 recurrent state reset; unrelated." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "base_model_prefix loading regression vs rope-validation TypeError on JSON-loaded lists; different mechanisms." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer NoneType broadcast error vs deepspeed Adafactor OOM; unrelated training issues." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Feature-extractor center-arg crash vs video input regression for Qwen models; different components." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Tokenizer decode-error masking vs leftover dummy classes/check_repo cleanup; not the same bug." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error-message clarity vs TrainerState num_input_tokens_seen not updating; unrelated." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS CLIP bus error vs SigLIP2 docs example issues; different artifacts and failure modes." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash vs Wav2Vec2 save/tokenization failure; different tokenizer families." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "Local-path import warning vs quantization offloading behavior; not the same underlying issue." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy Conv2d placeholder cleanup vs RT-DETR memory release on delete; different bugs." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma3 eos_token_id mismatch vs Qwen3.5 flash-attention generation regression; unrelated model issues." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Sam3PixelDecoder dummy layer vs SigLIP2 nonsensical outputs; different code paths." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError vs Wav2Vec2 save/tokenization failure; unrelated." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Qwen3Moe FP8 config failure vs quantized model saving failure; both quantization-related but not the same defect." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma4 import regression vs protobuf decode-error masking; distinct tokenizer/import concerns." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures vs void segmentation-map processing failure; different issues." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Eval dtype change vs tp-load OOM on Qwen2-57B; unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 29, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:29127", + "issue:29942", + "issue:33357", + "issue:35141", + "issue:36010", + "issue:42490", + "issue:42915", + "issue:43066", + "issue:43262", + "issue:43334", + "issue:43381", + "issue:43421", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43575", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43931", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44279", + "issue:44387", + "issue:44442", + "issue:44462", + "issue:44464", + "issue:44625", + "issue:44743", + "issue:44811", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45198", + "issue:45245", + "issue:45412", + "issue:45440", + "issue:45491", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45834", + "issue:45859", + "issue:46369" + ], + "soft_pairs": [ + "issue:43066|issue:44462", + "issue:45412|issue:46369", + "issue:43994|issue:45412", + "issue:44265|issue:45198", + "issue:43756|issue:45440", + "issue:43381|issue:43701", + "issue:43421|issue:45491", + "issue:43653|issue:44442", + "issue:29127|issue:33357", + "issue:43824|issue:44279", + "issue:45440|issue:45593", + "issue:43994|issue:46369", + "issue:29942|issue:36010", + "issue:43526|issue:45834", + "issue:29127|issue:35141", + "issue:43526|issue:44186", + "issue:44625|issue:45584", + "issue:43761|issue:45859", + "issue:43575|issue:43701", + "issue:42915|issue:43931", + "issue:44442|issue:45072", + "issue:42490|issue:43381", + "issue:43575|issue:44387", + "issue:44977|issue:45245", + "issue:43334|issue:43531", + "issue:43526|issue:44857", + "issue:44464|issue:45491", + "issue:43262|issue:44206", + "issue:44743|issue:44811", + "issue:43452|issue:45588" + ], + "prepared_review_unit_hash": "282496423c60409b5b809ca68fc67a1709cd26b5276936cb84b7d8be2fd8dd97", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36644, + "estimated_input_tokens": 9161, + "estimated_eval_tokens": 18578 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: most items are distinct model-, tokenizer-, or runtime-specific bugs rather than one shared underlying defect. I found no soft pair that should be merged. The broadest representative issue is the AutoTokenizer repository-loading regression.", + "confidence": 0.66, + "canonical_issue_reason": "issue:44462 is the most general, reusable regression in the set: it affects AutoTokenizer loading from repository files and has downstream references, making it the best anchor for a tokenizer/loading subcluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44462 is the clearest cross-model loading bug among the candidates; it is broader and more central than the model-specific failures in the rest of the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Both involve tokenization behavior, but one is a decoder-type regression in v5 and the other is repository tokenizer loading; different code paths and failures." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "Both mention memory, but one is a deletion/GC leak in RT-DETR and the other is an overflow while running sam3-litetext; not the same bug." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "Unrelated issues: SigLIP inference quality vs RT-DETR memory cleanup." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failures." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Both are model-specific implementation mismatches, but they affect different architectures and different internal mechanisms." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch are separate training-time problems." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Runtime special-token/post-processor updates are unrelated to Gemma3 NaNs from sliding-window all-padding batches." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "Both are tokenizer-related, but one is BigBird special-token registration and the other is AutoTokenizer failing to load a specific tokenizer class." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 error messaging and a MacOS CLIP bus error are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "A missing Qwen2_5_VL import is not the same as a generic dependency issue." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepSeekV3 implementation divergence and D-FINE auxiliary-loss behavior are different model bugs." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "Different models and different failure modes: bad outputs vs memory overflow." + }, + { + "left": "issue:29942", + "right": "issue:36010", + "accept": false, + "reason": "Failing Flash Attention 2 tests are not the same as a GenerationMixin import error." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "Beit label reduction and Kosmos2.5 OCR index errors are distinct image-processing issues." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "An error-message clarity bug is unrelated to token-embedding reinitialization." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Both touch vision preprocessing, but one is label reduction in BeitImageProcessorFast and the other is LayoutLMv2 tokenization/padding crashes." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and Whisper empty-transcription generation are unrelated." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "Hidden-states forwarding in CLIPVisionModel is unrelated to Qwen3_5Moe CPU-offload incompatibility." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "A Qwen2 model-load OOM is not the same as a checkpoint key mismatch." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "FineGrainedFP8Config failure and Qwen3-VL weight-shape mismatch are separate Qwen model issues." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer loading and bfloat16 dtype mismatches are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "A local-path import warning is unrelated to gradient checkpointing behavior." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both are OOM-related, but one is a specific model load with tensor parallelism and the other is a global int4 quantization memory regression." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation and category-count limits are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id for Qwen3-VL and sliding-window behavior in Qwen3-MoE are different bugs." + }, + { + "left": "issue:43526", + "right": "issue:44857", + "accept": false, + "reason": "Beit label reduction and LwDetrImageLoss AMP crashes do not share the same code path." + }, + { + "left": "issue:44464", + "right": "issue:45491", + "accept": false, + "reason": "Compiled-forward generation inconsistency and Gemma3 NaNs from padding/windowing are different failure modes." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and a feature-extractor center-arg crash are distinct audio bugs." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen recurrent-state reset and Whisper batch_decode skip_special_tokens handling are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:45588", + "accept": false, + "reason": "GGUF loading failures and flash-attention sinkless-model crashes involve different subsystems and symptoms." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 30, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:33290", + "issue:34567", + "issue:34689", + "issue:35141", + "issue:36010", + "issue:36331", + "issue:42175", + "issue:42490", + "issue:42907", + "issue:43232", + "issue:43316", + "issue:43381", + "issue:43388", + "issue:43493", + "issue:43525", + "issue:43611", + "issue:43643", + "issue:43646", + "issue:43720", + "issue:43723", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44062", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44442", + "issue:44479", + "issue:44743", + "issue:44871", + "issue:44898", + "issue:44913", + "issue:44987", + "issue:45003", + "issue:45216", + "issue:45290", + "issue:45440", + "issue:45479", + "issue:45698", + "issue:45759", + "issue:45803", + "issue:45907", + "issue:46178", + "issue:46234" + ], + "soft_pairs": [ + "issue:43646|issue:44077", + "issue:33290|issue:34689", + "issue:44062|issue:45907", + "issue:33290|issue:34567", + "issue:33290|issue:35141", + "issue:43381|issue:43873", + "issue:45003|issue:45440", + "issue:44442|issue:44898", + "issue:43723|issue:45479", + "issue:43761|issue:44079", + "issue:42490|issue:44743", + "issue:43316|issue:43525", + "issue:43232|issue:43388", + "issue:34689|issue:43611", + "issue:42175|issue:46234", + "issue:43720|issue:44442", + "issue:43761|issue:44188", + "issue:43931|issue:45216", + "issue:34689|issue:36010", + "issue:44077|issue:44479", + "issue:43493|issue:43827", + "issue:33290|issue:36331", + "issue:44987|issue:46178", + "issue:44479|issue:45290", + "issue:44442|issue:45803", + "issue:42907|issue:45698", + "issue:43873|issue:44871", + "issue:43749|issue:43931", + "issue:43643|issue:44913", + "issue:43316|issue:45759" + ], + "prepared_review_unit_hash": "10638528ff09d95bb91ebdc8fb040ee9eab7a627cd796d174195f30cce3e01db", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36329, + "estimated_input_tokens": 9083, + "estimated_eval_tokens": 18422 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is heterogeneous: most items are unrelated Transformers regressions across loading, config, generation, tokenization, and docs. I did not find any soft pair that looks like the same underlying bug/change, so all soft edges should be rejected.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the best broad representative of the cluster because it concerns a general Transformers 5.0 custom initialization regression, but the items are not true duplicates of one another.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the most representative issue by breadth and centrality, but only as a loose theme anchor rather than a duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Both mention initialization/post_init behavior, but they describe different models and different bugs." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "OOM with Adafactor/DeepSpeed is unrelated to Llama 3.2 model loading." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizers AddedToken keyword collision is unrelated to list_repo_files revision forwarding." + }, + { + "left": "issue:33290", + "right": "issue:34567", + "accept": false, + "reason": "Optimizer memory usage and TrainerState token counting are unrelated." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "OOM in training is unrelated to embedding reinitialization during post_init." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient checkpointing in eval mode and quantization offloading are different failure modes." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "Unsafe sys.modules access and DeepSeekV3 divergence are different code paths." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer loading failure and Perceiver image classification resizing failure are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "AutoTokenizer loading in v5 and sequence-classification zero-loss behavior are unrelated bugs." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression and ModelOutput key assignment bug are not the same issue." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "Local-path slash warning and Qwen recurrent-state reset are unrelated." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both are config API issues, but for different models and different missing fields." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs handling and gather_for_metrics label truncation are unrelated." + }, + { + "left": "issue:34689", + "right": "issue:43611", + "accept": false, + "reason": "Llama 3.2 Vision loading failure and base_model_prefix loading regression are different bugs." + }, + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "Missing tensorflow backend in install extras and missing import_utils helper are unrelated packaging issues." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet packed-weight loading and tokenizer loading failure are unrelated." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "Vision model hidden_states regression and attention-kernel divergence under torch.compile are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch on load and Qwen3.5 save_pretrained regression are different bugs." + }, + { + "left": "issue:34689", + "right": "issue:36010", + "accept": false, + "reason": "Model loading breakage and GenerationMixin import error are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "PatchTSMixer post_init policy and Qwen video-input regression are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and docs still referencing pipeline() are unrelated." + }, + { + "left": "issue:33290", + "right": "issue:36331", + "accept": false, + "reason": "Adafactor/DeepSpeed OOM and custom Trainer compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "Physical-intelligence/fast loading failure and tensorflow_text backend mapping error are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45290", + "accept": false, + "reason": "Video-input regression and chat-template crash on tool-call messages are unrelated." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure and Fuyu bare-except error are different bugs." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Dequantized save failure and wrong custom-module loading after save_pretrained are both save/load themed but not the same underlying bug." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offloading behavior and Gemma-3 eos_token_id inconsistency are unrelated." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Qwen3-VL weight mismatch are different load-path failures." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "trust_remote_code missing fields and rotary_pct reverting on reload are both config-related but distinct bugs." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both involve config handling, but one is a Gemma3TextConfig API discrepancy and the other is unwrapping text_config for multimodal Qwen models." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 31, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:28282", + "issue:41950", + "issue:42617", + "issue:43065", + "issue:43295", + "issue:43316", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43452", + "issue:43475", + "issue:43540", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43646", + "issue:43650", + "issue:43688", + "issue:43701", + "issue:43716", + "issue:43742", + "issue:43761", + "issue:43825", + "issue:44077", + "issue:44162", + "issue:44190", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44792", + "issue:44855", + "issue:44871", + "issue:44877", + "issue:44913", + "issue:45137", + "issue:45584", + "issue:45803", + "issue:46050", + "issue:46170", + "issue:46202", + "issue:46365", + "issue:46369" + ], + "soft_pairs": [ + "issue:44190|issue:44496", + "issue:44855|issue:45137", + "issue:43650|issue:46202", + "issue:45803|issue:46365", + "issue:25251|issue:28282", + "issue:43295|issue:43825", + "issue:43550|issue:43701", + "issue:42617|issue:43575", + "issue:43606|issue:43701", + "issue:43065|issue:44617", + "issue:43575|issue:43742", + "issue:16998|issue:22355", + "issue:43540|issue:44514", + "issue:43316|issue:43716", + "issue:44871|issue:45584", + "issue:44610|issue:46369", + "issue:43295|issue:44871", + "issue:43316|issue:44877", + "issue:44162|issue:44792", + "issue:44077|issue:44534", + "issue:43404|issue:43441", + "issue:45803|issue:46170", + "issue:44871|issue:46050", + "issue:43388|issue:43688", + "issue:43761|issue:44479", + "issue:41950|issue:43475", + "issue:43646|issue:44913", + "issue:43452|issue:44496" + ], + "prepared_review_unit_hash": "ea791a0b5b9cc70dfad39efdde92bacd30eab91ed9dc61aa3221a60fec95c75d", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35819, + "estimated_input_tokens": 8955, + "estimated_eval_tokens": 18166 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a broad mix of unrelated Transformers issues; the soft-similarity links are mostly false positives across different models, environments, and failure modes.", + "confidence": 0.94, + "canonical_issue_reason": "No single issue is a good canonical representative: the items span unrelated bugs in pipelines, configs, model loading, runtime errors, and model-specific regressions.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue stands out as a strong global representative for deduplication; the cluster is too heterogeneous to map to one underlying bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Different problems: local dataset loading in a training script vs an unrecognized model/config error when loading a checkpoint." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Unrelated failures: Python 3.13 import/annotation parsing vs a DeepSpeed ZeRO3 deque underflow." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "No overlap in bug shape; one is a low-information post, the other is a Gemma pipeline roundtrip issue." + }, + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Different code paths: Fuyu tensor conversion swallowing exceptions vs Trainer import being affected by installed datasets." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "Pipeline output-shape bug vs missing PyTorch dependency on AutoModel import; not the same underlying issue." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "One is a processor/tokenizer regression, the other is an error-message/docstring issue about translation support." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and resume checkpoint key mismatch are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "A script execution problem is not the same as an OOM when loading a large Qwen model with tensor parallelism." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "Device-mismatch CPU offload bug vs checkpoint key mismatch; different failure modes and code paths." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Dummy Conv2d implementation issue vs SAM3 video OOM; not a duplicate." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Large-model OOM vs MobileLLM key loading error; unrelated symptoms and root causes." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "Model_max_length behavior question and missing transformers.onnx module are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError differs from a batched chat-template padding crash." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "API/config discrepancy in Gemma3TextConfig is unrelated to an image-preprocessor dtype mismatch." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma eos_token_id inconsistency and Whisper empty-transcription handling are separate model-specific bugs." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Processor size mismatch for OmDet-Turbo vs memory overflow in sam3-litetext; different root causes." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor/tokenizer regression and Gemma eos_token_id config mismatch do not share a code path." + }, + { + "left": "issue:43316", + "right": "issue:44877", + "accept": false, + "reason": "Gemma3TextConfig API mismatch is not the same as strict config rejection for granite_speech." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 breakage and a janus image-generation test failure are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "Optional post_init allowance vs junk-filled non-persistent buffers are distinct v5 issues." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Weight tying bug in Mistral3 is unrelated to Ministral FlashAttention failures." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "Bare except in Fuyu tensor conversion and Windows/ROCm torch.distributed import issues are unrelated." + }, + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "Gemma eos_token_id config mismatch is not the same as sinq quantized save/load breakage." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Metric gathering truncation bug and auxiliary-loss normalization bug affect different training internals." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are different model families and code paths." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification processor selecting image processors is unrelated to SAM3 output missing fpn_position_embeddings." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Custom model initialization breakage in v5 and GPTNeoX rotary_pct not persisting are separate config/init issues." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "gguf_file handling in from_pretrained is unrelated to an unrecognized model_type in config.json." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 32, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:41720", + "issue:42490", + "issue:43065", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43572", + "issue:43644", + "issue:43673", + "issue:43761", + "issue:43827", + "issue:43931", + "issue:43976", + "issue:43994", + "issue:44060", + "issue:44190", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44521", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44913", + "issue:44933", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45698", + "issue:46178", + "issue:46234", + "issue:46369" + ], + "soft_pairs": [ + "issue:43673|issue:44743", + "issue:45127|issue:45200", + "issue:32090|issue:35141", + "issue:43572|issue:46178", + "issue:43504|issue:45698", + "issue:44617|issue:45584", + "issue:44734|issue:45406", + "issue:45542|issue:46234", + "issue:45563|issue:45584", + "issue:43525|issue:43531", + "issue:44792|issue:44977", + "issue:43761|issue:44242", + "issue:43931|issue:44060", + "issue:45381|issue:46369", + "issue:44534|issue:44913", + "issue:44190|issue:44964", + "issue:44792|issue:45600", + "issue:44464|issue:45563", + "issue:35141|issue:42490", + "issue:44479|issue:44734", + "issue:43827|issue:43994", + "issue:44855|issue:45588", + "issue:41720|issue:45245", + "issue:44617|issue:46369", + "issue:43065|issue:44610", + "issue:44521|issue:45381", + "issue:44933|issue:45362", + "issue:43644|issue:44913", + "issue:43976|issue:44855" + ], + "prepared_review_unit_hash": "8b926d133c9d10ec5aafaa61768c9fc8c82f0d31b5ca7c54704548d8b8fb23e5", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36375, + "estimated_input_tokens": 9094, + "estimated_eval_tokens": 18444 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly unrelated issue reports spanning different models, backends, and failure modes. The only obvious duplicate family is the Transformers v5 non-persistent-buffer regression, with 44534 being the cleanest representative.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44534 is the clearest, most generic statement of the recurring v5 non-persistent-buffer bug and is a better canonical representative than the more version-specific wording in 43644.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44534 is the best single issue to keep because it most directly captures the underlying regression that appears to have a near-duplicate sibling (43644) and is phrased broadly enough to anchor that family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both mention cache/state behavior, but one is a missing GenerationMixin cache during chunked_prefill while the other is recurrent states resetting in modular_qwen3_5; different bugs and code paths." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge/tie_word_embeddings corruption and Gemma 4 mm_token_type_ids defaulting are unrelated model-loading/training issues." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Trainer NoneType GPU broadcast failure is unrelated to output embeddings being reinitialized after resizing token embeddings." + }, + { + "left": "issue:43572", + "right": "issue:46178", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig and an undefined tensorflow_text backend entry are unrelated configuration/import problems." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "A legacy field breaking Beit preset loading is not the same as from_pretrained selecting the wrong custom module after save_pretrained." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video CUDA OOM and Whisper empty-transcription generation failure are different models and different failure modes." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Wrong KV-cache tensor indexing in transformers serve is unrelated to Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "TensorBoard/TensorFlow backend setup error and missing is_falcon_mamba_ssm_available helper are unrelated import/system issues." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale paged-generate warning about num_return_sequences is unrelated to Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Llama4Config missing pad_token_id and Qwen3-MoE sliding_window behavior are different config bugs." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are different model-specific problems." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "CLIPVisionModel returning hidden_states=None and missing load-balancing loss are not the same bug." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Qwen3-VL weight shape mismatch and a wrong tied-weights warning in Qwen3-Next are distinct issues." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Qwen2.5-VL video vision_position_ids regression and Sam3-LiteText memory overflow are unrelated." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffers filled with junk and GPTNeoX rotary_pct reverting on reload are different config/state bugs." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading in no-trainer image classification and Phi-4 multimodal model loading errors are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "A Janus image-generation test failure is unrelated to auto_mappings referencing removed Sam3LiteText configs." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Chunked generation inconsistencies with compiled forward and a stale num_return_sequences warning do not describe the same problem." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Resizing token embeddings reinitializing the output embedding is unrelated to a local-path trailing-slash import warning." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "A v5.3 video-input regression in Qwen VL models and a transformers serve KV-cache indexing crash are different code paths." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Outdated docs referencing pipeline() and SigLIP2 producing nonsensical outputs with AutoModel/pipeline are not the same bug." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "A Python 3.13 IndentationError in DebertaV2Model is unrelated to a flash_attention crash on s_aux=None." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device mapping cuda asserts and the 2^24 category limit error are entirely different failures." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Both concern Sam3-family memory usage, but CUDA OOM in Sam3Video and memory overflow in sam3-litetext are different models and not the same concrete bug." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder and a processor/model input-size mismatch are different Sam3 issues." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "All-zero assistant masks for multimodal chat templates and wrong vision_position_ids for Qwen2.5-VL video inputs are different multimodal bugs." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "A nonexistent image_utils import and Qwen3.5 chat crashes are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "The non-persistent-buffer junk regression and GPTNeoX rotary_pct reload behavior are different bugs despite both touching model configuration/state." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "A Python-version compatibility regression in Transformers 5.1.0 is unrelated to the Python 3.13 DebertaV2Model parsing failure." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 33, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36246", + "issue:38617", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:43116", + "issue:43295", + "issue:43335", + "issue:43404", + "issue:43479", + "issue:43540", + "issue:43550", + "issue:43643", + "issue:43650", + "issue:44016", + "issue:44077", + "issue:44112", + "issue:44186", + "issue:44190", + "issue:44220", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44496", + "issue:44683", + "issue:44734", + "issue:44737", + "issue:44855", + "issue:44877", + "issue:44938", + "issue:45003", + "issue:45127", + "issue:45216", + "issue:45245", + "issue:45276", + "issue:45357", + "issue:45375", + "issue:45446", + "issue:45542", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46082", + "issue:46121", + "issue:46234" + ], + "soft_pairs": [ + "issue:43550|issue:44190", + "issue:44077|issue:44877", + "issue:45632|issue:46234", + "issue:43335|issue:44855", + "issue:45127|issue:45357", + "issue:43540|issue:44734", + "issue:44462|issue:45907", + "issue:44737|issue:45803", + "issue:43335|issue:44190", + "issue:44112|issue:45600", + "issue:43116|issue:44190", + "issue:46082|issue:46121", + "issue:44938|issue:45542", + "issue:45276|issue:45584", + "issue:43295|issue:45276", + "issue:45245|issue:45584", + "issue:36246|issue:41628", + "issue:43404|issue:45216", + "issue:43650|issue:44220", + "issue:42175|issue:44279", + "issue:44186|issue:44734", + "issue:44246|issue:45003", + "issue:41950|issue:44016", + "issue:36246|issue:38617", + "issue:44683|issue:45446", + "issue:43479|issue:45375", + "issue:43643|issue:44496" + ], + "prepared_review_unit_hash": "044ed1389f9f929d31f918eae4652653f934d9ce2fa362acac79f0ef1c0f9986", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35440, + "estimated_input_tokens": 8860, + "estimated_eval_tokens": 17976 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: most items are unrelated Transformers bugs that only share broad themes like imports, configs, or multimodal models. None of the soft pairs look like the same concrete bug/change, so all are rejected.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue stands out; the items span unrelated bugs across imports, configs, examples, CI, and model-specific regressions.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41628 is the best representative issue because it is a concrete, user-facing top-level import failure with a clear repro, but it is still not a good duplicate sink for most of the other items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Different failures and code paths: torch.compile with SDPA in Bamba vs local dataset loading in an image-classification example." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Both are config-related, but one is about an optional post_init field and the other about strict loading of granite_speech config; not the same bug." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "Different APIs and symptoms: cache-path collision for trust_remote_code local models vs a missing utility export in import_utils." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "Unrelated issues: SwitchTransformers sparse layer construction vs a Python 3.13 import/indentation parsing failure in DebertaV2Model." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Same model family, but different save/merge problems and likely different code paths; not one concrete underlying bug." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Both involve video/serve-ish contexts, but one is a processing error in Qwen3OmniMoe and the other is KV-cache continuation tensor indexing in transformers serve." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "Different functionality: AutoTokenizer ignoring tokenizer.json vs list_repo_files failing to forward revision." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "XLNet device placement bug vs Fuyu exception handling bug; no shared underlying defect." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "Different layers of the stack: model config construction vs example-script dataset loading." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "One is a stale CI device-override test, the other is auto_mappings referencing removed configs; same broad CI area only." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Classification example output bug vs image-classification example input-loading bug; not the same behavior or code path." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "Unrelated config validation problems: Llama head_dim rejection vs rope-validation list handling." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python version load failure vs backend-installation mismatch; different root causes." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Both are multimodal, but one is embedding resize propagation in gemma4 and the other is Whisper empty-transcription generation after align_special_tokens." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Regression in processor/tokenizer image passing vs gemma4 embedding resize behavior; distinct regressions." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Category cardinality limit error is unrelated to Whisper empty-transcription generation." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different symbols and modules; not the same missing-export bug." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Model-specific lm_head tying bug vs Qwen3.5 save_pretrained checkpoint regression; different mechanisms and symptoms." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "Placeholder-style issue title vs fbank feature extraction bug; no meaningful duplicate evidence." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "TensorFlow backend installation issue vs generic dependency issue; too vague and not the same defect." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "Tokenizer crash on NER padding/truncation is unrelated to KV-cache continuation indexing in serve." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import latency complaint vs unsafe sys.modules access; different problem class and code path." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Video-classification pipeline bug vs notebook syntax error; no shared underlying change." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Both are import failures, but one is a missing model class export and the other is a missing configuration utility symbol." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Compiled flex_attention on torch>=2.9 is unrelated to a version-check bug for AuxRequest import." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Both involve multimodal config fields, but different models and different missing defaults/strict-dropped fields." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "trust_remote_code field-loss issue vs unrecognized model/config issue; not the same load path failure." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 34, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:38175", + "issue:42907", + "issue:43065", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43441", + "issue:43526", + "issue:43531", + "issue:43577", + "issue:43650", + "issue:43673", + "issue:43824", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44112", + "issue:44220", + "issue:44242", + "issue:44279", + "issue:44336", + "issue:44368", + "issue:44496", + "issue:44561", + "issue:44734", + "issue:44933", + "issue:44964", + "issue:44987", + "issue:45042", + "issue:45103", + "issue:45200", + "issue:45216", + "issue:45335", + "issue:45357", + "issue:45372", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45600", + "issue:45632", + "issue:45759", + "issue:46123", + "issue:46209", + "issue:46234", + "issue:46369" + ], + "soft_pairs": [ + "issue:38175|issue:43065", + "issue:44964|issue:45600", + "issue:44016|issue:44220", + "issue:43854|issue:44964", + "issue:44112|issue:44964", + "issue:46123|issue:46369", + "issue:44561|issue:45632", + "issue:44734|issue:45632", + "issue:43976|issue:44987", + "issue:43531|issue:45600", + "issue:42907|issue:43441", + "issue:43526|issue:43901", + "issue:43295|issue:45335", + "issue:44496|issue:45042", + "issue:45372|issue:46234", + "issue:43441|issue:45588", + "issue:44016|issue:44062", + "issue:43299|issue:45216", + "issue:43673|issue:44336", + "issue:43901|issue:44496", + "issue:43381|issue:45593", + "issue:43824|issue:44933", + "issue:43976|issue:44279", + "issue:44496|issue:45759", + "issue:44075|issue:44368", + "issue:45335|issue:45584", + "issue:43650|issue:46209", + "issue:45103|issue:45372", + "issue:43577|issue:45216", + "issue:44242|issue:45200", + "issue:43299|issue:45357" + ], + "prepared_review_unit_hash": "72fc8e582fd784f2f9630ba1a607c33ad560e3ca7cd1cb434cebecf91f648a2d", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37030, + "estimated_input_tokens": 9258, + "estimated_eval_tokens": 18772 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: most items are distinct bugs spanning model loading, generation, processors, docs, CI, and utilities. I do not see any soft pair that clearly represents the same underlying change or defect, so no candidate merges are justified.", + "confidence": 0.87, + "canonical_issue_reason": "issue:43824 is the most representative single artifact here: it has the highest discussion/inbound-reference signal and is a concrete reproducible regression, while the rest of the cluster is too unrelated to treat as one duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43824 is the best stand-in for this cluster because it has the strongest activity signal and a clearly scoped user-facing failure; there is no meaningful PR candidate in the provided items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:43065", + "accept": false, + "reason": "Different models and different failure modes: Siglip probability output vs a dummy Conv2d in Sam3PixelDecoder." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "One is a Phi-4 multimodal loading regression; the other is an auto_mappings/removed-config CI break. Different bug sources." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and audio feature extraction issue are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different code paths; not the same bug." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "CI test flake in GraniteSpeech is unrelated to Phi-4 multimodal loading." + }, + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "MaskGenerationPipeline batching bug and sam3-litetext memory overflow are unrelated." + }, + { + "left": "issue:44561", + "right": "issue:45632", + "accept": false, + "reason": "One is a missing helper breaking trust_remote_code models; the other is a cache-path collision bug. Different root causes." + }, + { + "left": "issue:44734", + "right": "issue:45632", + "accept": false, + "reason": "KV-cache tensor indexing in transformers serve is unrelated to local-model cache path collisions." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python version compatibility issue and a model-loading failure for physical-intelligence/fast are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Sam3LiteText auto_mappings breakage do not share a code path." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Both mention Ministral/Devstral or FlashAttention, but one is save/dequantization and the other is a runtime attention failure; not the same defect." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug and pipeline docs mismatch are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Processor.tokenizer regression and resize_token_embeddings not updating decoder embeddings are different subsystems." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unrecognized model type in config and PIL image processor torchvision dependency regression are unrelated." + }, + { + "left": "issue:45372", + "right": "issue:46234", + "accept": false, + "reason": "Gemma 4 processor import failure and missing falcon mamba availability helper are different missing-symbol issues." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Both involve FlashAttention, but one is model-specific and the other is a generic s_aux=None crash; not enough to treat as one bug." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error and AddedToken special-argument conflict are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3VL MoE loading breakage and Qwen3.5 save_pretrained checkpoint corruption are separate regressions." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "GenerationMixin cache missing in chunked_prefill and ANSI code emission in loading_report are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Pipeline documentation issue and model config loading failure do not describe the same bug." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction and D-FINE auxiliary-loss omission are unrelated training behaviors." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Different missing imports from different modules; same broad symptom, but not the same underlying defect." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "Python support regression and generic dependency issue are not the same code-path problem." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized model type vs composite Qwen config unwrapping are different model-loading failures." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "Unused SGD args and a tie_word_embeddings warning are unrelated." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize bug and Whisper empty-transcription generation failure are different areas." + }, + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "Placeholder issue and FP8/B200 inference failure are unrelated." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotation crash and Gemma 4 processor import error are unrelated." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "BLIP2 dtype propagation bug and Qwen3.5 save_pretrained regression are different issues." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss omission and Gemma 4 text-only mm_token_type_ids defaulting are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5 regressions, but one is model loading and the other is incorrect save_pretrained visual encoder keys; not one bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 35, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:33290", + "issue:38175", + "issue:39692", + "issue:41950", + "issue:42994", + "issue:43295", + "issue:43299", + "issue:43317", + "issue:43441", + "issue:43450", + "issue:43479", + "issue:43525", + "issue:43650", + "issue:43653", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44279", + "issue:44291", + "issue:44297", + "issue:44351", + "issue:44462", + "issue:44466", + "issue:44521", + "issue:44617", + "issue:44623", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:45020", + "issue:45081", + "issue:45276", + "issue:45278", + "issue:45325", + "issue:45397", + "issue:45405", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:45907", + "issue:46132" + ], + "soft_pairs": [ + "issue:38175|issue:44617", + "issue:44521|issue:45325", + "issue:44279|issue:45405", + "issue:44351|issue:45278", + "issue:39692|issue:43901", + "issue:44779|issue:45812", + "issue:41950|issue:45020", + "issue:43295|issue:44779", + "issue:43479|issue:43525", + "issue:41950|issue:45823", + "issue:43299|issue:43854", + "issue:43976|issue:45405", + "issue:44683|issue:45542", + "issue:44462|issue:45081", + "issue:43650|issue:46132", + "issue:44297|issue:44462", + "issue:43450|issue:44279", + "issue:45081|issue:45907", + "issue:44466|issue:45276", + "issue:44016|issue:44291", + "issue:43441|issue:45834", + "issue:42994|issue:43317", + "issue:43653|issue:45593", + "issue:45600|issue:45712", + "issue:44623|issue:44749", + "issue:44623|issue:45397", + "issue:33290|issue:45656" + ], + "prepared_review_unit_hash": "0af117dd4f98706cf13871d92a938456a32389d607eea813bb708c345704eed9", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35097, + "estimated_input_tokens": 8775, + "estimated_eval_tokens": 17806 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All soft pairs are false positives: they share only broad topical overlap, while each item points to a different bug, model family, or subsystem. No duplicate merge is justified.", + "confidence": 0.97, + "canonical_issue_reason": "No exact canonical duplicate stands out in this heterogeneous set; if a representative issue is needed, issue 44779 is the clearest concrete regression report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44779 is a focused, reproducible tokenizer regression with clear impact and a well-defined failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:44617", + "accept": false, + "reason": "Different model and failure mode: SigLIP2 zero probabilities vs Sam3Video CUDA OOM." + }, + { + "left": "issue:44521", + "right": "issue:45325", + "accept": false, + "reason": "Both are multimodal, but one is assistant mask generation and the other is temporal position ID scaling; different code paths." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Dependency/versioning problems in different places: a transformers dependency issue vs a PEFT minimum-version bump." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "45278 is a broad import-breakage report; 44351 is a specific HybridCache import error. Too general to call the same bug." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Both mention docs, but they are unrelated docs issues: SigLIP2 example errors vs TextClassificationPipeline parameter docs." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths; not enough to merge as one bug." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification processor lookup and remote_code model breakage are unrelated subsystems." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Both involve tokenizers, but one is processor.tokenizer / multimodal input handling and the other is Deepseek tokenization correctness." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Different config bugs: default multimodal sub-config initialization vs missing pad_token_id on Llama4Config." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Video pipeline processor selection and Gemma4 device-mapping mismatch are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Distinct model-loading failures for different models and different symptoms." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python-version compatibility vs an unreleased PEFT minimum version are not the same issue." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Torch flex_attention compilation failure and TensorBoard/TensorFlow backend error are unrelated." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Both are tokenizer-related, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash; different paths." + }, + { + "left": "issue:43650", + "right": "issue:46132", + "accept": false, + "reason": "Unrelated: a vague data request vs a behavior change in AttentionInterface.register." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Tokenizer save metadata mismatch and AutoTokenizer ignoring tokenizer.json are different serialization/loading bugs." + }, + { + "left": "issue:43450", + "right": "issue:44279", + "accept": false, + "reason": "Video processor batch-shape bug vs dependency issue; no shared underlying defect." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regex crash and list_repo_files revision forwarding are unrelated." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both involve model weights/embeddings, but one is serialization inconsistency and the other is resize_token_embeddings behavior." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error and init_empty_weights argument handling are different failure modes." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Different models and errors: FlashAttention failure for Ministral-3 vs OCR length index error in Kosmos2.5." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Quantized saving failure and dequantized loading/offload failure are related only broadly to quantization, not the same bug." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "Tokenizer special-token registration and D-FINE auxiliary-loss behavior are unrelated." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both are repo-maintenance/CI style issues, but one is stale auto-mapping references and the other is leftover dummy classes." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Processor save_pretrained missing files and a slowdown regression after upgrade are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Different failures: processor serialization vs Gemma4 zero3 from_pretrained." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention deepspeed/optimizer, but one is OOM with Adafactor and the other is double optimizer stepping; not the same code-path problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 36, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:41628", + "issue:42175", + "issue:42491", + "issue:42886", + "issue:43262", + "issue:43299", + "issue:43441", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43643", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:43994", + "issue:44016", + "issue:44162", + "issue:44188", + "issue:44242", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44361", + "issue:44373", + "issue:44496", + "issue:44521", + "issue:44589", + "issue:44671", + "issue:44683", + "issue:44779", + "issue:45003", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45593", + "issue:45823", + "issue:46132" + ], + "soft_pairs": [ + "issue:44261|issue:44263", + "issue:44361|issue:45593", + "issue:44671|issue:44779", + "issue:44016|issue:45823", + "issue:42175|issue:43976", + "issue:43441|issue:44683", + "issue:43901|issue:43994", + "issue:43723|issue:44162", + "issue:43262|issue:44521", + "issue:43673|issue:43761", + "issue:43824|issue:43976", + "issue:44297|issue:45397", + "issue:43525|issue:43881", + "issue:43901|issue:44246", + "issue:43525|issue:43643", + "issue:43784|issue:45542", + "issue:44188|issue:44336", + "issue:41628|issue:43976", + "issue:44589|issue:45362", + "issue:45561|issue:45563", + "issue:44336|issue:44373", + "issue:42491|issue:43299", + "issue:44246|issue:44496", + "issue:44295|issue:45397", + "issue:44242|issue:45593", + "issue:43531|issue:45362", + "issue:43526|issue:45003", + "issue:42886|issue:46132" + ], + "prepared_review_unit_hash": "d3dec6a0acf8b6ed76691a572f359738311d507ec4691f3fd4264d6e8f9ed7a6", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35251, + "estimated_input_tokens": 8813, + "estimated_eval_tokens": 17882 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This looks like a loose bag of Transformers bug reports with only superficial lexical similarity. The soft pairs mostly share subsystem terms (tokenizers, Qwen, docs, configs, etc.) but describe different failures, so I do not see a real duplicate cluster.", + "confidence": 0.92, + "canonical_issue_reason": "No single issue is a good canonical anchor: the items span unrelated regressions in tokenization, model loading, generation, audio processors, attention kernels, docs, and utilities. The pairwise similarities are mostly keyword-level rather than same-bug duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick one representative issue, issue:43643 is the broadest regression-style report, but it still only represents a narrow config-loading problem and does not actually unify the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Different bugs: one is missing config.rms_norm_eps in MLA q_a_layernorm, the other is torch.split return values in GlmMoeDsaIndexer." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "Unrelated: MLukeTokenizer AttributeError on tasks versus a stale warning in paged generate()." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Different components and symptoms: CamemBERT masked LM predictions vs Deepseek tokenizer output regression." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error and Gemma4 PLE device mismatch are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Package dependency/backend install issue versus a Python-version compatibility problem." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Both involve attention/FlashAttention-ish failures, but they are different code paths and different failure modes." + }, + { + "left": "issue:43901", + "right": "issue:43994", + "accept": false, + "reason": "Docs mismatch for return_all_scores is not the same as incorrect model/pipeline outputs for SigLIP2." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading in v5 is separate from ESM2 model breakage." + }, + { + "left": "issue:43262", + "right": "issue:44521", + "accept": false, + "reason": "Both mention chat template/audio or multimodal processing, but the bugs are distinct and not the same underlying defect." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Cache missing during chunked_prefill is unrelated to CLIPVisionModel hidden_states being None." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Missing Qwen2.5-VL import and Python version incompatibility are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save_pretrained class mismatch is unrelated to gemma-4 zero3 from_pretrained loading." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Different models and different AttributeError/loader failures; no shared code-path evidence." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs inconsistency versus import-time slowness are unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Both are model/config loading errors, but for different models and different missing fields; not the same bug." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Importing sentence-transformers with latest transformers is unrelated to TensorFlow backend installation behavior." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Attention kernel divergence on torch.compile is unrelated to ANSI codes in loading_report." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Cannot import AutoImageProcessor has nothing to do with the Python 3.9/3.10 support issue." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage import error and Qwen3.5 chat crash are different failures." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "pytest-xdist race on captured_info.txt is unrelated to the stale num_return_sequences warning." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "ANSI output formatting issue and wrong docstring for position_ids are different problems." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both concern Qwen3 MoE family regressions, but one is LoRA compatibility across versions and the other is model loading breakage in v5; not the same underlying bug." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Import slowness is unrelated to unrecognized model/config model_type error." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Registering position_ids as a buffer causing read errors is unrelated to gemma-4 zero3 loading." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Load-balancing loss omission in MoE routing is unrelated to D-FINE auxiliary losses when denoising is off." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Sliding_window issue in Qwen3-MoE is unrelated to Qwen3.5-35B chat crashes." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline tokenizer cache loading failure is unrelated to AttentionInterface.register changing function behavior." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 37, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36246", + "issue:41950", + "issue:42757", + "issue:42886", + "issue:43232", + "issue:43257", + "issue:43262", + "issue:43352", + "issue:43475", + "issue:43596", + "issue:43673", + "issue:43901", + "issue:44162", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44351", + "issue:44466", + "issue:44485", + "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:45250", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45412", + "issue:45459", + "issue:45561", + "issue:45584", + "issue:45701", + "issue:45759", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46369" + ], + "soft_pairs": [ + "issue:41950|issue:44220", + "issue:44246|issue:44871", + "issue:44336|issue:44485", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:45003|issue:45561", + "issue:44779|issue:45292", + "issue:43475|issue:46369", + "issue:44242|issue:44857", + "issue:44295|issue:44749", + "issue:43352|issue:45250", + "issue:43673|issue:44561", + "issue:36246|issue:43596", + "issue:43901|issue:45003", + "issue:44162|issue:45200", + "issue:44496|issue:45003", + "issue:43257|issue:45137", + "issue:43262|issue:44246", + "issue:45375|issue:45759", + "issue:45292|issue:45584", + "issue:45412|issue:46123", + "issue:45292|issue:45701", + "issue:44877|issue:45375", + "issue:44987|issue:45459", + "issue:43232|issue:44336", + "issue:45987|issue:46082", + "issue:44995|issue:45200", + "issue:44466|issue:45335", + "issue:42886|issue:45229" + ], + "prepared_review_unit_hash": "c3d7ea87c0637b033a1af5ae4e7c735c27452b146ce6121f2d957e703a3277bf", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36376, + "estimated_input_tokens": 9094, + "estimated_eval_tokens": 18444 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is a heterogeneous mix of unrelated Transformers bug reports. None of the soft-edge pairs look like the same underlying defect, so no duplicate merges are justified. The most representative single issue is the high-activity Gemma4 CUDA OOM report.", + "confidence": 0.96, + "canonical_issue_reason": "issue:45229 is the strongest standalone issue here: it is detailed, high-activity, and describes a concrete reproducible multi-GPU inference failure with clear user impact.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45229 is the best representative issue overall because it has the clearest problem statement and the most discussion around a concrete, actionable bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Different subsystems and failure modes: video-classification processor lookup vs audio fbank feature extraction." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "One is an intermittent import slowdown; the other is a Gemma eos_token_id config mismatch." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "Unrelated bugs: terminal ANSI output handling vs GLM-5 RoPE implementation." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and different packages/modules." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Different missing exports: Qwen2_5_VLImageProcessor vs HybridCache." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "sys.modules access bug and pytest-xdist file race are unrelated issues." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer correctness regression is not the same as embedding resize not propagating." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "SAM3 attribute error and SAM3 memory overflow are distinct bugs." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Different models and symptoms: MoE load-balancing loss vs LwDetrImageLoss AMP crash." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "A position_ids buffer access error is unrelated to a performance regression after upgrade." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "A model-specific Flash Attention support limitation is not the same as a generic Flash Attention issue." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Generation cache/chunked_prefill bug and trust_remote_code import breakage are separate failures." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for a missing processor and a DeepSpeed/BertModel IndexError are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Documentation mismatch in a pipeline vs unsafe sys.modules access in modeling utils." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma4 mm_token_type_ids behavior are different model-specific bugs." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Model config loading failure and sys.modules access are not the same code-path problem." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed is distinct from a ZeRO3 deque underflow." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling-rate bug is unrelated to import performance." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Both involve composite Qwen configs, but one is a dropped config field and the other is missing text_config unwrapping." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize propagation and Whisper empty-transcription generation failure are different issues." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "Memory reclamation on deletion and batch-finalization logic in MaskGenerationPipeline are unrelated." + }, + { + "left": "issue:45292", + "right": "issue:45701", + "accept": false, + "reason": "A token embedding resize bug is not the same as a broad tokenization version-change report." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both mention strict config handling, but they describe different models and different failure mechanisms." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Model loading regression in v5.1.0 is unrelated to protobuf exception masking in tokenizers." + }, + { + "left": "issue:43232", + "right": "issue:44336", + "accept": false, + "reason": "Generation kwargs sync logic and ANSI logging output are unrelated." + }, + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "Cache-length return type and Llama head_dim validation are different API/validation bugs." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "Stale indexer cache crash on second forward pass is unrelated to mm_token_type_ids defaults." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Serialization inconsistency for lm_head.weight is not the same bug as resize_token_embeddings propagation." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are completely different failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 38, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:42994", + "issue:43065", + "issue:43232", + "issue:43408", + "issue:43475", + "issue:43482", + "issue:43540", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43673", + "issue:43701", + "issue:43827", + "issue:43931", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44188", + "issue:44297", + "issue:44361", + "issue:44368", + "issue:44479", + "issue:44483", + "issue:44623", + "issue:44734", + "issue:44749", + "issue:44869", + "issue:44871", + "issue:44877", + "issue:44977", + "issue:44995", + "issue:45137", + "issue:45230", + "issue:45356", + "issue:45381", + "issue:45397", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45656", + "issue:46123", + "issue:46132", + "issue:46170", + "issue:46202" + ], + "soft_pairs": [ + "issue:43065|issue:43408", + "issue:45397|issue:45656", + "issue:44479|issue:46170", + "issue:45137|issue:45600", + "issue:43475|issue:43540", + "issue:44060|issue:44077", + "issue:44483|issue:46202", + "issue:43653|issue:44162", + "issue:44368|issue:44977", + "issue:44995|issue:45588", + "issue:44368|issue:45563", + "issue:44188|issue:46202", + "issue:44877|issue:46170", + "issue:43931|issue:44368", + "issue:45584|issue:46170", + "issue:43673|issue:44734", + "issue:43065|issue:46123", + "issue:44297|issue:44623", + "issue:44361|issue:44749", + "issue:44869|issue:45356", + "issue:43646|issue:44162", + "issue:44623|issue:45656", + "issue:44749|issue:45656", + "issue:43232|issue:43673", + "issue:44871|issue:46132", + "issue:43701|issue:44749", + "issue:43673|issue:43827", + "issue:45230|issue:45397", + "issue:42994|issue:43482", + "issue:44038|issue:44623", + "issue:43475|issue:43643", + "issue:28282|issue:30990", + "issue:43475|issue:45381" + ], + "prepared_review_unit_hash": "b814494854e9abfd32fb8bc14e314f8a5f6c3d5f5b5566327202789107d1bdcc", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37459, + "estimated_input_tokens": 9365, + "estimated_eval_tokens": 18986 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated issue reports, with one clear duplicate-like pair around Qwen video input regression. The strongest anchor is the broad v5.3.0 Qwen video regression report.", + "confidence": 0.78, + "canonical_issue_reason": "issue:44479 is the best representative because it names a concrete v5.3.0 regression, identifies the affected Qwen video model families, and is broad enough to anchor the closest related report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44479 is the strongest global issue candidate: it is specific, high-signal, and covers the widest relevant surface among the listed issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both are SAM3-related, but one is a dummy Conv2d concern and the other is a model-type/config warning; different bug surfaces." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Different subsystems and symptoms: DeepSpeed zero3 from_pretrained vs optimizer stepping twice." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "Qwen video regression and Windows ROCm/torch.distributed import failure are unrelated." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "ZeRO3 deque error vs stale Sam3LiteText auto-mapping references; not the same underlying defect." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "Different model families and failures: missing SAM3 vision encoder output field vs Qwen3OmniMoe video input ValueError." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "A tied-weights warning and PatchTSMixer post_init policy change are distinct issues." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "API request rejection and Gemma pipeline roundtrip token failure do not share a concrete code path." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer special-token registration bug vs ESM2 model breakage are different problems." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Warning about tie_word_embeddings and flash-attention generation failure are not the same bug." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Stale indexer cache on second forward pass vs flash_attention s_aux=None crash are different code paths." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Both involve warnings, but one is about tied embeddings and the other about stale num_return_sequences validation." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "torch.compile branching divergence in attention kernels is unrelated to Gemma token roundtrip failure." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "Strict config loading for granite_speech and ROCm/Windows import dependency issues are unrelated." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Weight-shape mismatch during Qwen3-VL loading is not the same as a tied-embeddings warning." + }, + { + "left": "issue:45584", + "right": "issue:46170", + "accept": false, + "reason": "Whisper empty transcription failure is unrelated to the Windows ROCm import/dependency issue." + }, + { + "left": "issue:43673", + "right": "issue:44734", + "accept": false, + "reason": "Both are generation/cache-related, but one is a chunked_prefill cache gap and the other is a server-side tensor indexing crash; too different to merge." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 decoder dummy conv and MaskGenerationPipeline final-batch handling are different bugs." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Tokenizer_class mismatch and processor.save_pretrained missing files are both serialization-related but not the same concrete failure." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "MLukeTokenizer AttributeError and a Chinese report about data filtering slowdown are unrelated." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper trailing replacement-character crash and Kimi-K2.5 codec-handling regression are different decode bugs." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Custom model initialization breakage in v5 and ESM2 breakage are separate issues." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save_pretrained file emission and DeepSpeed double optimizer step are unrelated." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Training slowdown report and duplicate optimizer step report do not match." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Different generation internals: sync_gpus kwargs update vs missing cache during chunked_prefill." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "Gemma EOS-token config inconsistency and AttentionInterface.register behavior change are not the same bug." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is unrelated to data filtering slowdown." + }, + { + "left": "issue:43673", + "right": "issue:43827", + "accept": false, + "reason": "Generation cache bug and outdated pipeline docs are separate concerns; same theme only, not same issue." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A generic bug report cannot be treated as the same concrete issue as gemma-4 zero3 loading failure." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving failure and Qwen2.5-GGUF loading failure are different directions of the pipeline." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe v5.0 bug and processor save_pretrained missing files are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "Missing fpn_position_embeddings on a vision output object is not the same as missing fields in AutoConfig with trust_remote_code." + }, + { + "left": "issue:45381", + "right": "issue:45656", + "accept": false, + "reason": "Qwen2.5-VL video position regression and duplicate optimizer stepping are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": true, + "reason": "These describe the same Qwen video-input regression family in v5.3.0, with 45381 giving a specific qwen2.5-vl symptom already covered by 44479." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 39, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:30990", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42371", + "issue:42673", + "issue:42831", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43377", + "issue:43504", + "issue:43525", + "issue:43526", + "issue:43673", + "issue:43704", + "issue:43723", + "issue:43824", + "issue:43873", + "issue:43976", + "issue:44038", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44351", + "issue:44496", + "issue:44568", + "issue:44792", + "issue:44933", + "issue:44945", + "issue:44964", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45310", + "issue:45375", + "issue:45406", + "issue:45478", + "issue:45561", + "issue:45712", + "issue:45753", + "issue:46077", + "issue:46123", + "issue:46202", + "issue:46326" + ], + "soft_pairs": [ + "issue:45406|issue:46202", + "issue:42371|issue:45092", + "issue:41628|issue:44933", + "issue:45375|issue:45478", + "issue:42994|issue:43525", + "issue:43723|issue:46326", + "issue:45310|issue:45375", + "issue:45081|issue:45103", + "issue:43873|issue:44945", + "issue:46077|issue:46123", + "issue:42673|issue:43704", + "issue:43232|issue:44188", + "issue:41628|issue:44351", + "issue:44964|issue:45137", + "issue:44945|issue:45753", + "issue:44222|issue:45310", + "issue:42994|issue:43824", + "issue:44038|issue:45310", + "issue:36010|issue:41628", + "issue:43232|issue:44792", + "issue:43673|issue:44188", + "issue:42831|issue:44222", + "issue:44496|issue:45561", + "issue:42175|issue:45712", + "issue:43526|issue:45561", + "issue:43504|issue:45561", + "issue:43377|issue:44242", + "issue:42175|issue:44933", + "issue:42898|issue:44568", + "issue:43976|issue:44933", + "issue:43278|issue:45137", + "issue:30990|issue:35141" + ], + "prepared_review_unit_hash": "31387615e13846e05cb98bdf604eb67e5aae906e2ed0cd3de294e4506fdfd89b", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37183, + "estimated_input_tokens": 9296, + "estimated_eval_tokens": 18848 + }, + "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 bug reports; only one soft-edge pair is a clear duplicate. The best representative issue is a broad top-level import regression report.", + "confidence": 0.87, + "canonical_issue_reason": "issue:41628 is the most central and broadly applicable report here: a top-level `transformers` import failure with multiple nearby import-regression links.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41628 has the widest apparent scope among the candidates and best matches the cluster\u2019s repeated theme of API/import regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Different bugs: Gemma4Processor serving-time attribute error vs Gemma pipeline roundtrip tokenization failure." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance is unrelated to the InternVL2 v5 meta-initialization incompatibility." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Both are import-related, but one is a missing top-level class and the other is a nonexistent image_utils symbol; not the same bug." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Same model family, but one is a specific config-field-drop report and the other is only a generic from_pretrained failure; insufficiently identical." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model saving failure and missing pad_token_id in Llama4Config are different code paths." + }, + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer loading regression and continue_final_message behavior for mistral_common are different tokenizer bugs." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Related Qwen3.5 MoE area, but one is a generic load error and the other is a specific config-field omission." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Tokenizer regex crash and auto_docstring annotation crash are unrelated subsystems." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization/offloading behavior and pipeline-parallel incorrect output are unrelated runtime bugs." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Gemma4 attention-mask construction and partial-batch pipeline handling are different problems." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Same title and same Qwen3ForCausalLM VRAM leak symptom; these look like duplicate reports of the same bug." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Generation kwargs/sync_gpus issue and torch.compile attention-kernel branching issue are different generation-path bugs." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different exported symbols and likely different regressions." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel wrong output and Qwen3_5 infinite-loop-on-image are different failure modes." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained MoE and Qwen3.5 MoE from_pretrained error are not the same concrete bug." + }, + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Quantized model saving and a Qwen2.5-VL import error are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Qwen3-VL-Moe on transformers 5.0 and Qwen3.5 MoE load failure are different model/version issues." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but GenerationMixin-from-generation and AutoImageProcessor-top-level are separate regressions." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation state-update bug and janus image-generation test failure are unrelated." + }, + { + "left": "issue:43673", + "right": "issue:44188", + "accept": false, + "reason": "Chunked-prefill cache handling and compile-time attention-kernel divergence are different code paths." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "Both mention FP8, but one is an accuracy issue and the other is saving pretrained weights." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading failure and xdist file-race bug are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Backend dependency packaging and dummy-class cleanup are different maintenance issues." + }, + { + "left": "issue:43526", + "right": "issue:45561", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and xdist captured_info.txt race are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Legacy-field pretrained loading failure and xdist race are unrelated." + }, + { + "left": "issue:43377", + "right": "issue:44242", + "accept": false, + "reason": "MIMI padding-mask mismatch and MoE load-balancing loss gating are different model bugs." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging/backend selection and image_utils import exposure are unrelated." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer cleanup-space behavior change and BOS/EOS insertion regression are different tokenizer issues." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility and image_utils import exposure are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift in eval and ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence Transformers loading stall and embedding reinit after resize are different problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary stays within the packet evidence: it treats the cluster as mostly heterogeneous, highlights 41628 as a reasonable broad import-regression representative, and only accepts the clearly duplicated 42673/43704 pair. The other soft-edge rejections are conservative and grounded in the issue titles/descriptions." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 40, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:30333", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:36331", + "issue:42222", + "issue:42831", + "issue:42898", + "issue:43122", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:44016", + "issue:44060", + "issue:44075", + "issue:44079", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44448", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44933", + "issue:45071", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45200", + "issue:45245", + "issue:45310", + "issue:45356", + "issue:45405", + "issue:45656", + "issue:45712", + "issue:45812", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46231" + ], + "soft_pairs": [ + "issue:42222|issue:45200", + "issue:44623|issue:45310", + "issue:43479|issue:44869", + "issue:30333|issue:33290", + "issue:43479|issue:44079", + "issue:42898|issue:45812", + "issue:43784|issue:43824", + "issue:43723|issue:44568", + "issue:42898|issue:44779", + "issue:42898|issue:45356", + "issue:44734|issue:45103", + "issue:44016|issue:46231", + "issue:43874|issue:45092", + "issue:44060|issue:45125", + "issue:44295|issue:45656", + "issue:44075|issue:46231", + "issue:42898|issue:44448", + "issue:45405|issue:45712", + "issue:44448|issue:46132", + "issue:44222|issue:44297", + "issue:44933|issue:45712", + "issue:44297|issue:45656", + "issue:43479|issue:43644", + "issue:44222|issue:44623", + "issue:30333|issue:34689", + "issue:42831|issue:44623", + "issue:45245|issue:46132", + "issue:30333|issue:36331", + "issue:30333|issue:35141", + "issue:44190|issue:46082", + "issue:28282|issue:30333", + "issue:43122|issue:46209", + "issue:22355|issue:28282", + "issue:45071|issue:46082", + "issue:44297|issue:45310" + ], + "prepared_review_unit_hash": "07d1e10c132c981fadc4193f49f93dcef9db04732102a5d3bab4c3c737e6fa36", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38256, + "estimated_input_tokens": 9564, + "estimated_eval_tokens": 19384 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are a heterogeneous mix of unrelated Transformers issues; the soft links mostly reflect superficial similarity around v5 regressions, tokenization, save_pretrained, and model-loading, not the same underlying bug. I would not merge any of the soft pairs.", + "confidence": 0.97, + "canonical_issue_reason": "issue:42898 is the most centrally connected item in the soft graph and is the broadest representative of the recurring v5/tokenization-regression theme, though the cluster is too mixed to treat it as a true duplicate hub.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42898 is the best anchor among these issues because it has the most soft similarities and reflects the most repeated theme in the set, but it is still not a suitable duplicate target for the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "Different model families and bugs: vitpose model breakage vs Gemma 4 multimodal token type IDs." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Both involve model/processor loading, but one is missing save_pretrained files and the other is a Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Unrelated bugs: Phi4MultimodalConfig default initialization vs Whisper word-timestamp decode crash." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "Training-job MLflow reporting is unrelated to DeepSpeed/Adafactor OOM behavior." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Config default handling and ModelOutput key assignment are different code paths." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both are v5 tokenization regressions, but they affect different models and different failure modes." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Different import failures: sentence-transformers with latest transformers vs missing Qwen2.5-VL symbol." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer loading failure is not the same as add_special_tokens not emitting BOS/EOS." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they concern different tokenizers/models and likely different fixes." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Different tokenizer regressions for different model families; not the same underlying bug." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "Serve KV-cache indexing crash and auto-docstring AttributeError are unrelated." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Notebook syntax error and documentation parameter mismatch are not the same issue." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Different multimodal-loading problems: missing image-processor method vs remote-code checkpoint/meta-init incompatibility." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Both are Qwen-related, but one is a tied-weights warning bug and the other is a missing tensor-parallel plan." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Buffer read/registration bug is unrelated to optimizer step being called twice with DeepSpeed." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "Optimizer argument handling is unrelated to a docs-only parameter name mismatch." + }, + { + "left": "issue:42898", + "right": "issue:44448", + "accept": false, + "reason": "Different v5 regressions affecting different model/tokenizer behavior." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "PEFT version pinning and leftover dummy classes are unrelated maintenance issues." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Pegasus output regression and AttentionInterface.register behavior change are different code paths." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both touch save_pretrained, but one is FP8 MoE saving and the other is tokenizer metadata mismatch." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Nonexistent image_utils import and dummy_pt_objects leakage are different import/symbol problems." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer save_pretrained metadata bug is unrelated to DeepSpeed optimizer stepping." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config default initialization and non-persistent buffer junk are different regressions." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both concern save_pretrained, but one is FP8 MoE artifacts and the other is processor missing files." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "MLflow reporting on failed training and Llama 3.2 vision model loading are unrelated." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy problems and processor save_pretrained file omissions are distinct bugs." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-count runtime error and AttentionInterface.register behavior change are unrelated." + }, + { + "left": "issue:30333", + "right": "issue:36331", + "accept": false, + "reason": "Failed-job MLflow reporting and CustomTrainer.compute_loss signature change are different training issues." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "MLflow reporting and embedding reinitialization on resize are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Dataset loading in a script and LlamaConfig head_dim validation are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:30333", + "accept": false, + "reason": "Missing torch dependency import error is unrelated to MLflow job reporting." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenization behavior change and FP8 inference failure on B200 are unrelated." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Different import errors: transformers.onnx module missing vs AutoModel requiring PyTorch." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "PretrainedConfig type checking regression is not the same as LlamaConfig head_dim validation." + }, + { + "left": "issue:42898", + "right": "issue:46082", + "accept": false, + "reason": "Tokenizer behavior regression and LlamaConfig constructor validation are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer metadata save bug and Qwen3.5 MoE from_pretrained error are different problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 41, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:30990", + "issue:38175", + "issue:39401", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43232", + "issue:43475", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:43761", + "issue:43824", + "issue:44038", + "issue:44206", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44295", + "issue:44360", + "issue:44373", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44568", + "issue:44623", + "issue:44829", + "issue:44871", + "issue:44913", + "issue:44977", + "issue:44991", + "issue:45092", + "issue:45103", + "issue:45310", + "issue:45357", + "issue:45381", + "issue:45406", + "issue:45834", + "issue:46002", + "issue:46050", + "issue:46082", + "issue:46305", + "issue:46328" + ], + "soft_pairs": [ + "issue:43475|issue:46305", + "issue:41628|issue:42757", + "issue:44206|issue:45103", + "issue:43723|issue:44977", + "issue:45381|issue:46305", + "issue:43232|issue:45834", + "issue:44222|issue:44484", + "issue:44263|issue:45092", + "issue:44295|issue:44623", + "issue:44373|issue:44485", + "issue:43761|issue:46082", + "issue:44483|issue:45406", + "issue:43723|issue:44871", + "issue:44279|issue:46328", + "issue:43723|issue:44991", + "issue:42907|issue:46050", + "issue:44514|issue:45103", + "issue:45092|issue:45357", + "issue:44263|issue:44360", + "issue:30990|issue:38175", + "issue:43644|issue:46305", + "issue:44829|issue:46002", + "issue:42831|issue:45310", + "issue:44261|issue:45092", + "issue:43479|issue:44913", + "issue:41628|issue:43824", + "issue:42831|issue:44038", + "issue:44263|issue:44373", + "issue:39401|issue:44568", + "issue:42886|issue:42994" + ], + "prepared_review_unit_hash": "031e6ff4c54974c358fe7179482bc43e444acefe74d4fdcbcb0cf1fb00da6b13", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36436, + "estimated_input_tokens": 9109, + "estimated_eval_tokens": 18474 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft pair looks like a true duplicate. The set is mostly a grab-bag of unrelated Transformers issues; the closest matches are usually just shared keywords like tokenizer, save_pretrained, or config handling, not the same concrete bug.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 43723 is the broadest, most representative issue in this set: a tokenizer-loading regression in v5, which is the closest shared theme across the tokenizer-related items.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43723 is also the most actionable as a triage anchor: it is a concrete regression report with a clear failure mode and version context, while the others are either more model-specific or orthogonal.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "Different models and different bugs: missing attribute in SAM 3 video vs a streaming mask off-by-one." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but for unrelated symbols and subsystems; not the same underlying fix." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature extractor crash vs auto-docstring annotation crash; unrelated code paths." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading regression vs flash-attention generation failure; not the same bug." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Both mention position/masking semantics, but one is Qwen2.5-VL video positions and the other is Moonshine streaming window logic." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwargs update bug vs long OCR input index error; different failure modes." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 save_pretrained failure vs a question about default max_shard_size; not duplicates." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "Moe DSA torch.split issue vs remote-code meta-init incompatibility; different concrete problems." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Buffer registration/read error vs processor save_pretrained missing files; separate save/load paths." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Wrong docstring vs RoPE implementation bug; documentation and implementation are not the same issue." + }, + { + "left": "issue:43761", + "right": "issue:46082", + "accept": false, + "reason": "CLIP hidden_states regression vs Llama head_dim validation; unrelated model behaviors." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat/completions request rejection vs Gemma4Processor missing _tokenizer; different layers." + }, + { + "left": "issue:43723", + "right": "issue:44871", + "accept": false, + "reason": "Tokenizer load regression vs eos_token_id configuration mismatch; related area, but different bug." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "Generic dependency complaint vs feature extractor pad performance issue; no shared concrete defect." + }, + { + "left": "issue:43723", + "right": "issue:44991", + "accept": false, + "reason": "Both concern tokenizer loading, but for different models and failure causes." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both involve quantized-model persistence, but dequantized Ministrals/Devstrals saving and sinq save/load breakage are distinct." + }, + { + "left": "issue:44514", + "right": "issue:45103", + "accept": false, + "reason": "Batched chat template crash vs auto-docstring annotation crash; unrelated subsystems." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Meta-init incompatibility in old InternVL2 checkpoints vs incorrect visual encoder keys on save; different save/load bugs." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Same component family, but one is a torch.split return-value bug and the other is a missing ReLU; not the same underlying change." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Stuck loading sentence-transformers vs zero-probability outputs in SigLIP2; unrelated symptoms." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Non-persistent buffer junk in v5 vs streaming window off-by-one; no shared code-path." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Flash-attention training degeneration vs NaNs after checkout/init; both are training-quality issues but not the same bug." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FP8 accuracy issue vs Qwen3.5 MoE from_pretrained error; different parts of the stack." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "MLA normalization epsilon precision issue vs checkpoint meta-init incompatibility; unrelated." + }, + { + "left": "issue:43479", + "right": "issue:44913", + "accept": false, + "reason": "Default config initialization bug vs config value not persisting on reload; both are config-related but distinct." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Two import errors for different exported classes; too broad to be the same bug." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FineGrainedFP8 accuracy problem vs Qwen3-VL-Moe bug report; no concrete overlap." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "Indexer logic bug vs wrong docstring; one is code behavior, the other is documentation." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Tokenizer offset_mapping bug vs add_special_tokens/BOS-EOS regression; both tokenizer issues, but different behaviors and likely different fixes." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Offline tokenizer cache loading failure vs quantized model saving failure; separate persistence paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 42, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:22355", + "issue:28218", + "issue:30990", + "issue:33357", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:43066", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43576", + "issue:43646", + "issue:43756", + "issue:43784", + "issue:43937", + "issue:43976", + "issue:44060", + "issue:44117", + "issue:44188", + "issue:44261", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44368", + "issue:44373", + "issue:44464", + "issue:44488", + "issue:44521", + "issue:44671", + "issue:44704", + "issue:44743", + "issue:44945", + "issue:45278", + "issue:45290", + "issue:45593", + "issue:45600", + "issue:45753", + "issue:46002", + "issue:46082", + "issue:46132", + "issue:46305", + "issue:46326", + "issue:46369" + ], + "soft_pairs": [ + "issue:42886|issue:43066", + "issue:43408|issue:46369", + "issue:42757|issue:45278", + "issue:44368|issue:46305", + "issue:42831|issue:45753", + "issue:22355|issue:28218", + "issue:43232|issue:45593", + "issue:44464|issue:46082", + "issue:43756|issue:44060", + "issue:43576|issue:43784", + "issue:43646|issue:46002", + "issue:44671|issue:46132", + "issue:44117|issue:44704", + "issue:44295|issue:44297", + "issue:30990|issue:33357", + "issue:43122|issue:44521", + "issue:42371|issue:44263", + "issue:44188|issue:44373", + "issue:42907|issue:44945", + "issue:43122|issue:46326", + "issue:44261|issue:44373", + "issue:39401|issue:44488", + "issue:42907|issue:43937", + "issue:44704|issue:45290", + "issue:41628|issue:45278", + "issue:43976|issue:44945", + "issue:43232|issue:44743", + "issue:43784|issue:45600" + ], + "prepared_review_unit_hash": "12e25f49cf472004709dbf4b541e5c1634de3f5a970367285251d3489c6ea9b6", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35705, + "estimated_input_tokens": 8927, + "estimated_eval_tokens": 18110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous and mostly contains unrelated Transformers bugs/errors that only share generic library/version breakage language. There are no pull requests in the set. I would treat the cluster as low-quality for duplicate consolidation; if forced to pick an issue representative, the broad upgrade/import-breakage report is the closest umbrella, but it is still not a true duplicate anchor.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella-style report in the set (\u201cmany import errors after update\u201d), so it best matches the cluster\u2019s superficial similarity around API/import breakage. It is still not a strong duplicate canonical because the other items are about distinct subsystems and symptoms.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best available representative only because it is the most generic and can subsume several import-related complaints. It is not a good true canonical for the cluster\u2019s mixed contents.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading failure and a tokenizer decoder-type mismatch in v5 are different bugs in different code paths." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "A model-type warning for sam3_tracker is unrelated to a memory overflow in sam3-litetext." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "One is a specific missing import from huggingface_hub; the other is a broad upgrade regression with many import errors, not the same underlying bug." + }, + { + "left": "issue:44368", + "right": "issue:46305", + "accept": false, + "reason": "A tie_word_embeddings warning in Qwen3.5 LoRA fine-tuning is unrelated to MoonshineStreaming\u2019s sliding-window off-by-one mask bug." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FineGrainedFP8 accuracy degradation and a Qwen3_5 infinite loop on a specific image are distinct model/runtime failures." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "Missing transformers.onnx module import and extra tokenizer spacing after an added token are unrelated tokenizer/import issues." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation kwargs handling after sync_gpus and missing auxiliary losses when denoising is off affect different generation/training paths." + }, + { + "left": "issue:44464", + "right": "issue:46082", + "accept": false, + "reason": "Chunked generation inconsistency under compiled forward is not the same as LlamaConfig rejecting an explicit head_dim value." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Smollm3 RoPE-layer dropping and an incorrect tied-weights warning in Qwen3-Next are different architecture/config bugs." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "A broken `transformers env` command and a NameError when importing sentence-transformers are unrelated failure modes." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Custom model initialization breakage in Transformers 5.0 and NaNs in classification heads after checkout/init are separate problems." + }, + { + "left": "issue:44671", + "right": "issue:46132", + "accept": false, + "reason": "CamemBERT masked-LM prediction regression and AttentionInterface.register behavior change are different code-path issues." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "A tokenizer mapping assumption bug and AutoProcessor not forwarding kwargs to cached_file are unrelated API defects." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Reading position_ids after buffer registration and tokenizer_class metadata mismatch on save_pretrained are not the same bug." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Sentence Transformers hanging on load and a MacOS bus error with a CLIP model are distinct runtime failures." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Version-dependent tokenization differences and zero assistant masks for multimodal chat templates are different tokenizer/template issues." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration guidance and a torch.split return-value issue in GlmMoeDsaIndexer have no shared underlying bug." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "Torch.compile attention-kernel divergence and a wrong docstring are unrelated; one is a runtime correctness bug, the other is documentation." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized models and incorrect output under pipeline parallelism are different failures in different subsystems." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "Tokenization drift across versions is not the same as `continue_final_message` failing for mistral_common tokenizers." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "A missing rms_norm_eps causing precision error and a wrong docstring are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 offset_mapping errors and failure to load a specific model checkpoint are different tokenizer/model-loading bugs." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Dequantized model save failures and an invalid GenerationConfig for GLM-5 are separate issues." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "Passing kwargs to cached_file and apply_chat_template crashing on tool-call assistant messages involve different API paths." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Missing AutoImageProcessor import and broad import failures after upgrade are related only at a high level; they are not the same concrete bug." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python version incompatibility and pipeline-parallelism output corruption are unrelated regressions." + }, + { + "left": "issue:43232", + "right": "issue:44743", + "accept": false, + "reason": "Generation kwargs update logic and Qwen3.5 recurrent-state reset under cache are different generation bugs." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Importing sentence-transformers causing NameError and stale auto_mappings references to removed configs are unrelated import-time failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 43, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:22355", + "issue:28218", + "issue:29127", + "issue:29159", + "issue:30990", + "issue:36296", + "issue:38617", + "issue:39401", + "issue:42175", + "issue:42673", + "issue:42757", + "issue:42907", + "issue:43066", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43493", + "issue:43784", + "issue:43824", + "issue:43992", + "issue:44038", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44485", + "issue:44514", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44938", + "issue:44945", + "issue:44964", + "issue:45072", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45446", + "issue:45478", + "issue:45753", + "issue:46231" + ], + "soft_pairs": [ + "issue:42907|issue:44964", + "issue:39401|issue:44521", + "issue:43122|issue:44514", + "issue:43116|issue:45200", + "issue:44625|issue:44829", + "issue:43408|issue:43475", + "issue:44038|issue:44297", + "issue:44360|issue:44485", + "issue:44038|issue:44295", + "issue:42907|issue:43452", + "issue:28218|issue:30990", + "issue:43122|issue:44779", + "issue:44038|issue:44222", + "issue:45216|issue:45478", + "issue:36296|issue:43116", + "issue:43824|issue:44351", + "issue:43784|issue:45446", + "issue:22355|issue:29159", + "issue:42757|issue:44351", + "issue:43278|issue:44945", + "issue:42175|issue:44938", + "issue:43493|issue:45753", + "issue:42673|issue:45072", + "issue:43066|issue:44779", + "issue:22355|issue:29127", + "issue:43992|issue:44704", + "issue:44355|issue:46231", + "issue:38617|issue:45278" + ], + "prepared_review_unit_hash": "cc228e264968f8474e836bc2d407463a7ce9ba516fffe9476859413397d601ad", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35702, + "estimated_input_tokens": 8926, + "estimated_eval_tokens": 18108 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated Hugging Face Transformers issues; the soft pairs are mostly keyword-similarity false positives rather than true duplicates.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 43122 is the broadest tokenizer-regression report here and the closest thing to a reusable duplicate target, though the overall set is heterogeneous.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43122 is the best representative issue because it describes a user-facing tokenizer behavior regression across versions, which is broad enough to absorb the closest tokenizer-related complaints.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Different models and failure modes: dequantized save failure vs loading a multimodal checkpoint." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Both touch tokenizer/chat-template behavior, but one is wrong offset mapping and the other is assistant mask generation for multimodal inputs." + }, + { + "left": "issue:43122", + "right": "issue:44514", + "accept": false, + "reason": "Both are tokenizer-related, but one is general tokenization drift across versions and the other is a batched chat-template crash." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Different tasks and code paths: example script empty multi-label outputs vs mm_token_type_ids handling in Gemma 4." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Distinct bugs: config propagation for Qwen3.5 vs degenerate training with flash_attention_3." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Related to SAM 3 naming, but one is a model-type warning and the other is a missing attribute on an output object." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "One is a Qwen3-VL-Moe bug; the other is tokenizer save_pretrained writing the wrong tokenizer_class." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "Different components: DSA indexer activation choice vs GLM-5 RoPE implementation." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Unrelated symptoms: Qwen3-VL-Moe bug vs position_ids buffer read error." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Different failure paths: dequantized saving vs gguf_file breaking AutoTokenizer/AutoModel loading." + }, + { + "left": "issue:28218", + "right": "issue:30990", + "accept": false, + "reason": "Tokenizer spacing bug and Sentence Transformers loading hang are not the same underlying issue." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they affect different models and reproduce differently." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Same broad family name, but one is Qwen3-VL-Moe behavior and the other is FP8 save_pretrained for moe models." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 and version 5.4+, but one is save_pretrained corruption and the other is a from_pretrained load error; not enough evidence they're the same bug." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor parallel training bug vs example script empty results; different subsystems and symptoms." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they are for different missing symbols from different modules." + }, + { + "left": "issue:43784", + "right": "issue:45446", + "accept": false, + "reason": "Importing sentence-transformers breaks due to a missing nn symbol, while the other is a PyTorch version check bug in flex_attention." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "Unrelated issues: missing transformers.onnx module vs slow/fast tokenizer inconsistency." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different libraries and symbols, so they are not the same defect." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Different runtime problems: dtype shift between train/eval vs incorrect output under pipeline parallelism." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Backend packaging issue vs Python 3.14 load failure; too broad and not the same code path." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Both are Qwen3_5-related, but one is SigLIP2/JAX parity and the other is an image-triggered infinite loop." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "Different bugs in different areas: VRAM leak in Qwen3_5 threading vs bfloat16 inference dtype mismatch in unrelated models." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both mention tokenizer behavior in v5, but one is decoder-type assignment and the other is Deepseek tokenization output regression." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Completely different problem domains: missing onnx module vs LayoutLMv3 error-message clarity." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Separate from-pretrained/cached-file behaviors affecting different model families and different missing weights/kwargs." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled Python load errors are unrelated to the documentation mismatch in set_training." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "Both are import problems after upgrading, but the missing symbol and broad import errors are not the same 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": 1, + "component_count": 19, + "review_unit_index": 44, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:33453", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:43064", + "issue:43116", + "issue:43122", + "issue:43441", + "issue:43450", + "issue:43493", + "issue:43611", + "issue:43673", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43906", + "issue:43937", + "issue:43976", + "issue:43986", + "issue:44038", + "issue:44075", + "issue:44117", + "issue:44188", + "issue:44291", + "issue:44380", + "issue:44410", + "issue:44466", + "issue:44485", + "issue:44779", + "issue:44936", + "issue:44977", + "issue:45071", + "issue:45081", + "issue:45083", + "issue:45310", + "issue:45341", + "issue:45357", + "issue:45405", + "issue:45446", + "issue:45478", + "issue:45561", + "issue:45563", + "issue:45588", + "issue:46002", + "issue:46077", + "issue:46298", + "issue:46328" + ], + "soft_pairs": [ + "issue:43937|issue:44485", + "issue:44188|issue:46328", + "issue:43611|issue:46002", + "issue:43116|issue:43493", + "issue:42222|issue:43441", + "issue:44075|issue:44936", + "issue:42371|issue:43874", + "issue:45357|issue:45478", + "issue:44038|issue:45478", + "issue:43976|issue:45446", + "issue:43116|issue:43825", + "issue:43064|issue:43673", + "issue:45083|issue:45446", + "issue:43937|issue:45563", + "issue:33453|issue:39401", + "issue:44380|issue:46077", + "issue:39401|issue:44779", + "issue:44117|issue:45588", + "issue:42673|issue:43906", + "issue:43827|issue:45446", + "issue:43937|issue:44977", + "issue:43450|issue:43986", + "issue:43673|issue:46328", + "issue:44291|issue:45071", + "issue:44466|issue:46002", + "issue:45405|issue:45446", + "issue:43122|issue:46298", + "issue:44117|issue:45081", + "issue:45341|issue:45561", + "issue:44410|issue:45310", + "issue:45341|issue:46328" + ], + "prepared_review_unit_hash": "30bca90f019b743423e329fef998e0442698d08a8565936fc670805d747a8e8d", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36787, + "estimated_input_tokens": 9197, + "estimated_eval_tokens": 18650 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of Transformers v5 regression reports across unrelated subsystems; the soft edges are all superficial and should be rejected. The most representative issue is the broad tokenizer regression report, with no PRs present.", + "confidence": 0.72, + "canonical_issue_reason": "issue:33453 is the broadest, earliest tokenizer-regression report in the set and best serves as an umbrella representative for the tokenizer-related subset.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:33453 is also the best global representative because its title is generic enough to cover the tokenizer-loading regression theme and it has the strongest discussion signal among the tokenizer cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Both mention GLM-5, but one is a GenerationConfig validation failure and the other is a RoPE implementation discussion; different code paths and symptoms." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "torch.compile attention-kernel branching is unrelated to a pad() performance inefficiency in SequenceFeatureExtractor." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "Model loading with base_model_prefix and NaNs in classification heads are different bugs in different parts of the stack." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Empty multi-label results in an example script and SigLIP2 implementation discrepancies are unrelated issues." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "Vitpose model breakage and Ministral FlashAttention failure involve different models and failure modes." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "SGD arguments being ignored is distinct from trainer.evaluate() failing after training; they are separate trainer bugs." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "A TF32 settings API warning is unrelated to a missing get_number_of_image_patches method in a processor." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is save_pretrained key serialization and the other is a from_pretrained loading error; not the same bug." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "These target different model families/symptoms: Qwen3-VL-Moe behavior versus Qwen3.5 Moe loading failure." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version support mismatch and a flex_attention import version check are separate dependency issues." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example-script multi-label output failure is unrelated to a pipeline() error message about translation support." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption and missing GenerationMixin cache in chunked_prefill are different code paths." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "qwen3_omni_moe feature-length calculation and flex_attention PyTorch version gating are unrelated bugs." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Invalid GenerationConfig and a stale num_return_sequences warning are different generation issues." + }, + { + "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; not the same concrete bug." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "GPT2 attention scaling under SDPA/FlashAttention and Gemma4 bidirectional-mask construction are different attention bugs." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Tokenizer offset_mapping regressions for Qwen3 and incorrect Deepseek tokenization are similar in theme but distinct model/code-path failures." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES returning None is unrelated to a flash_attention crash on s_aux=None." + }, + { + "left": "issue:42673", + "right": "issue:43906", + "accept": false, + "reason": "A VRAM leak across dataloader threads is not the same as an isolated reproduction of an older unrelated issue." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() and a flex_attention import version check are unrelated documentation vs runtime issues." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Both are generation-related, but one is GenerationConfig validation and the other is flash-attention generation failure for a different model." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Batched video-processor shape bugs and missing torchvision crash handling are both video-related but not the same failure." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Missing cache in chunked_prefill and inefficient numpy-to-list conversion in pad() are unrelated issues." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Unexpected _is_hf_initialized handling and PretrainedConfig type-checking regressions are separate v5 breakages." + }, + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Inconsistent lm_head serialization and NaNs in classification heads are different bugs affecting different behaviors." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "A PEFT minimum-version bump and a flex_attention import version check are unrelated dependency problems." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Tokenizer behavior changes and a noisy logging message do not look like the same underlying issue." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer mapping None and a Mistral regex patch crashing on backend_tokenizer are both tokenizer-adjacent but distinct failures." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "A small testing_utils bug and a pytest-xdist race in captured_info.txt are separate testing-related issues." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing qwen3next projections and a Qwen3.5 Moe from_pretrained error are different model-implementation/loading problems." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "testing_utils cleanup bug and SequenceFeatureExtractor.pad inefficiency are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 45, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42175", + "issue:42222", + "issue:42371", + "issue:42490", + "issue:42617", + "issue:42673", + "issue:42831", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43408", + "issue:43421", + "issue:43577", + "issue:43582", + "issue:43673", + "issue:43742", + "issue:43874", + "issue:43937", + "issue:44075", + "issue:44220", + "issue:44261", + "issue:44263", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44496", + "issue:44829", + "issue:44877", + "issue:44933", + "issue:45071", + "issue:45216", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45362", + "issue:45464", + "issue:45478", + "issue:45542", + "issue:45656", + "issue:46178", + "issue:46231", + "issue:46328" + ], + "soft_pairs": [ + "issue:44829|issue:45216", + "issue:42222|issue:46178", + "issue:38617|issue:41628", + "issue:42617|issue:45341", + "issue:43408|issue:43742", + "issue:45362|issue:46328", + "issue:42673|issue:43582", + "issue:39401|issue:43421", + "issue:42490|issue:42673", + "issue:44877|issue:46231", + "issue:42371|issue:44261", + "issue:44484|issue:45310", + "issue:38617|issue:44220", + "issue:43937|issue:44496", + "issue:44933|issue:45278", + "issue:43066|issue:43122", + "issue:42222|issue:45542", + "issue:44410|issue:45478", + "issue:42371|issue:42831", + "issue:44483|issue:45464", + "issue:44075|issue:45656", + "issue:42175|issue:42222", + "issue:43012|issue:43577", + "issue:43097|issue:44493", + "issue:43673|issue:45341", + "issue:41628|issue:45071", + "issue:44263|issue:44485", + "issue:45071|issue:45542", + "issue:43012|issue:43408", + "issue:43874|issue:44261" + ], + "prepared_review_unit_hash": "fc394dbc5f902adabb1b49bb9ff66b3fc00e3e4f2900e644a271dc536314bd55", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36276, + "estimated_input_tokens": 9069, + "estimated_eval_tokens": 18394 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it mixes import regressions, tokenizer/decoder behavior changes, model-specific loading/runtime bugs, and backend/optimizer issues. None of the soft pairs look like strong duplicates of the same underlying bug/change, so I would not merge any of them.", + "confidence": 0.22, + "canonical_issue_reason": "Issue 45278 is the best umbrella representative because it most broadly captures the import-regression theme seen in several nearby reports, but the overall cluster is too mixed for a single true duplicate center.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the most suitable global representative: it is broad, user-facing, and closest to the repeated import-error complaints in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Both involve Qwen-related regressions, but one is a flash-attention training degradation and the other is a save_pretrained checkpoint regression; different code paths and symptoms." + }, + { + "left": "issue:42222", + "right": "issue:46178", + "accept": false, + "reason": "VitPose model breakage and a tensorflow_text backend-mapping error are unrelated failures." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols in different modules and do not point to the same bug." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "A failure to run 3d_parallel.py and a small testing_utils bug are not the same issue." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "Model-type warning for sam3_tracker vs a KeyError loading MobileLLM-125M are unrelated." + }, + { + "left": "issue:45362", + "right": "issue:46328", + "accept": false, + "reason": "Qwen chat crashes and SequenceFeatureExtractor.pad performance are different subsystems with no shared underlying defect." + }, + { + "left": "issue:42673", + "right": "issue:43582", + "accept": false, + "reason": "A VRAM leak in Qwen3ForCausalLM threads and an Apple Silicon TypeError in caching_allocator_warmup are unrelated." + }, + { + "left": "issue:39401", + "right": "issue:43421", + "accept": false, + "reason": "Both affect tokenization, but one is wrong offset_mapping and the other is runtime post-processor updates; not the same concrete bug." + }, + { + "left": "issue:42490", + "right": "issue:42673", + "accept": false, + "reason": "Local-path import warning and Qwen3 VRAM leak are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "Strict config loading and a docs parameter-name mismatch are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API warning and MLA rms_norm_eps accuracy issue do not share the same root cause." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "max_shard_size behavior and Qwen3.5 MoE from_pretrained errors are different issues." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for layer_type_validation and _torch_extract_fbank_features are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Invalid GenerationConfig and unrecognized model/config.json errors are separate loading problems." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "The first is a specific missing import from image_utils; the second is a broad upgrade-time import regression, so they are not the same bug." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both are tokenizer-related v5 regressions, but one is decoder-type selection and the other is changed tokenization output; too little evidence of one concrete shared defect." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose breakage and missing TensorFlow backend after installing tensorboard are unrelated." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Missing attn projections in qwen3next and Qwen3.5 MoE from_pretrained errors are different model-loading bugs." + }, + { + "left": "issue:42371", + "right": "issue:42831", + "accept": false, + "reason": "TF32 behavior control and FineGrainedFP8 accuracy issues are not the same problem." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Both mention chat/completions, but one is a general API acceptance failure and the other is a model-specific streaming inference failure." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "Both are optimizer-related, but one says SGD args are ignored and the other says optimizer.step is called twice under deepspeed; distinct bugs." + }, + { + "left": "issue:42175", + "right": "issue:42222", + "accept": false, + "reason": "Missing TensorFlow in a torch install and VitPose model breakage are different issues." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "A compile-time float32 warning for bf16 and a loading-time dtype mismatch in Blip2 are not the same underlying bug." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Removed tie_embeddings behavior and unexpected position_id keys are unrelated configuration/model-loading regressions." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "GenerationMixin cache handling and a testing_utils bug have no shared code path." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "ImportError for AutoImageProcessor and PreTrainedConfig type-checking breakage are unrelated regressions." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "A torch.split return-value issue in GlmMoeDsaIndexer and a RoPE implementation discussion are not the same concrete bug." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Config type checking and TensorFlow backend resolution are unrelated." + }, + { + "left": "issue:43012", + "right": "issue:43408", + "accept": false, + "reason": "These are both warnings, but they concern totally different models and mechanisms." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing get_number_of_image_patches in a GLM image processor and an MLA rms_norm_eps precision issue are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 46, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:38617", + "issue:42757", + "issue:42890", + "issue:43064", + "issue:43097", + "issue:43335", + "issue:43441", + "issue:43452", + "issue:43502", + "issue:43531", + "issue:43673", + "issue:43824", + "issue:43873", + "issue:43874", + "issue:43931", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44261", + "issue:44263", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44410", + "issue:44484", + "issue:44485", + "issue:44704", + "issue:44829", + "issue:44933", + "issue:44987", + "issue:45003", + "issue:45071", + "issue:45103", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45588", + "issue:45753", + "issue:46050", + "issue:46123", + "issue:46172", + "issue:46231", + "issue:46298", + "issue:46369" + ], + "soft_pairs": [ + "issue:44263|issue:46172", + "issue:44360|issue:46172", + "issue:44484|issue:45478", + "issue:43931|issue:44410", + "issue:43441|issue:44829", + "issue:43335|issue:45071", + "issue:43931|issue:44038", + "issue:43097|issue:43452", + "issue:45003|issue:45341", + "issue:43673|issue:46123", + "issue:43986|issue:45341", + "issue:44933|issue:46231", + "issue:45753|issue:46369", + "issue:44117|issue:44987", + "issue:43986|issue:44484", + "issue:43874|issue:44263", + "issue:45440|issue:46172", + "issue:43986|issue:45372", + "issue:43531|issue:45341", + "issue:44261|issue:44485", + "issue:43097|issue:46050", + "issue:43097|issue:45588", + "issue:43064|issue:44315", + "issue:38617|issue:42757", + "issue:44062|issue:44117", + "issue:44410|issue:45375", + "issue:44038|issue:44410", + "issue:42890|issue:46298", + "issue:43502|issue:43992", + "issue:38617|issue:44351", + "issue:43097|issue:43873", + "issue:44704|issue:45103", + "issue:43986|issue:44117", + "issue:43824|issue:43986" + ], + "prepared_review_unit_hash": "3743df9bdc065dd39a4ef37810a13af196b622c1d9da394faf4042a99a7c219b", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37751, + "estimated_input_tokens": 9438, + "estimated_eval_tokens": 19132 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: mostly unrelated Transformers regressions spanning imports, model loading, config/schema issues, training, attention, and utility bugs. None of the soft candidate pairs look like true duplicates or mergeable PR-equivalent changes, so all soft edges should be rejected.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43097 is the broadest and most impactful regression report in the set, with discussion and inbound references, but it still does not represent the rest of the unrelated issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "Different model families and different bugs: GlmMoeDsaIndexer split behavior vs DeepSeek v4 tensor transpose/cast logic." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "DSA indexer activation bug vs DeepSeek v4 comb transformation bug; no shared code path." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to Qwen3.5 MoE loading failures." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Both are Qwen-related, but one is a VL weight-shape mismatch and the other is missing projections in qwen3next; different defects." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Both mention flash attention, but one is a model-specific failure and the other is degenerate training behavior in classification." + }, + { + "left": "issue:43335", + "right": "issue:45071", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to PretrainedConfig type checking regressions." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Same broad model family, but different failures: loading shape mismatch vs version-specific Qwen3-VL-Moe breakage." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Embeddings tie-setting removal and gguf_file loading failure are unrelated from_pretrained/save-pretrained issues." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafely accessing sys.modules and a testing_utils bug are different code paths and symptoms." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Cache handling during chunked_prefill vs pipeline partial-batch is_last logic are unrelated generation/pipeline bugs." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "Processor loading without torchvision has nothing to do with the testing_utils issue." + }, + { + "left": "issue:44933", + "right": "issue:46231", + "accept": false, + "reason": "Missing import from image_utils is unrelated to a documentation mismatch in set_training()." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "Infinite-loop generation for a specific image is not the same as memory overflow in sam3-litetext." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Tokenizer mapping None handling and a specific model load failure are different from_pretrained problems, not the same bug." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Video-model processor crash without torchvision is unrelated to shard-size behavior." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Missing image-patch count method in an image processor is unrelated to torch.split return handling in a DSA indexer." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "DeepseekV3MoE parity mismatch and DeepSeek v4 tensor manipulation are distinct model implementation issues." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Different missing dependency/import problems on different processor-loading paths; not one underlying bug." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug is unrelated to a testing utility fix." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "RMS norm epsilon precision issue and GLM-5 RoPE implementation are separate model math bugs." + }, + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "Removed tie_embeddings behavior and quantized model save/load corruption are different persistence regressions." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "Tie-embedding removal is unrelated to flash_attention.py crashing on sink-less models." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "Trainer optimizer-state corruption under FSDP2/PEFT is unrelated to Liger Kernel application with model_init." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols in different modules and likely different fixes." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "AddedToken special-argument collision and tokenizer mapping None handling are unrelated tokenizer bugs." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Missing qwen3next projections and a missing Qwen3_5MoeVisionConfig field are separate configuration/model-definition problems." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Qwen3-VL-Moe version failure and qwen3next missing projections are not the same concrete code-path problem." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "Seed-related flaky test failure and noisy logging are completely unrelated." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "local_files_only networking violation and missing embed token weights are different from_pretrained bugs." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both are import-name errors, but they concern different APIs and likely different export paths; not one duplicate bug." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "A removed config behavior and quantization offloading failure are unrelated regressions." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and auto_docstring annotation parsing crash are separate utility issues." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Video processor dependency crash and tokenizer mapping None handling are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43986", + "accept": false, + "reason": "Different import/loading failures for different multimodal model paths; no shared underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 47, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36683", + "issue:38617", + "issue:42490", + "issue:42757", + "issue:42907", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "issue:43618", + "issue:43761", + "issue:43824", + "issue:43825", + "issue:43906", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44162", + "issue:44222", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44514", + "issue:44655", + "issue:44683", + "issue:44936", + "issue:44938", + "issue:45290", + "issue:45478", + "issue:45479", + "issue:45600", + "issue:45698", + "issue:45712", + "issue:45753", + "issue:46132", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365" + ], + "soft_pairs": [ + "issue:44938|issue:46178", + "issue:43066|issue:43906", + "issue:43208|issue:43550", + "issue:43992|issue:44484", + "issue:43012|issue:43550", + "issue:43825|issue:44655", + "issue:43012|issue:44368", + "issue:45712|issue:46298", + "issue:42907|issue:44655", + "issue:42907|issue:45753", + "issue:43352|issue:44380", + "issue:43097|issue:46298", + "issue:44162|issue:46172", + "issue:43352|issue:43976", + "issue:44380|issue:45479", + "issue:43064|issue:44380", + "issue:44380|issue:46132", + "issue:44222|issue:45478", + "issue:44483|issue:44514", + "issue:36683|issue:43582", + "issue:43208|issue:43526", + "issue:44936|issue:46365", + "issue:43012|issue:44683", + "issue:42490|issue:43012", + "issue:42757|issue:43824", + "issue:38617|issue:43502", + "issue:45600|issue:46298", + "issue:45698|issue:46172", + "issue:43618|issue:44380", + "issue:44038|issue:44230", + "issue:43618|issue:43761", + "issue:44483|issue:45290" + ], + "prepared_review_unit_hash": "bb0a4cd3f1508d60071b02ea82125dea177c94e2066fa8acc0b4256b2c09ea7c", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36431, + "estimated_input_tokens": 9108, + "estimated_eval_tokens": 18472 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are a heterogeneous set of unrelated bug reports that only share broad Transformers/HF vocabulary. None of the soft-edge pairs look like the same concrete underlying issue, so all should be rejected as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:43012 is the cleanest standalone bug report in the set: concrete symptoms, likely reproducible, and a focused code-path problem; however, it does not meaningfully subsume the other items.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Unrelated problems: Python 3.14 import failure vs an invalid backend name error in a backend mapping." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Different issues: tokenizer decoder type regression vs an isolated reproduction of another external bug." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "Separate model-specific failures in different code paths; not the same training/compile bug." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing weight loading vs save_pretrained shard-size behavior; no shared underlying defect." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A bfloat16 compile warning is not the same as a model compile failure under SDPA." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Pipeline task support messaging vs inability to save pipeline objects are different bugs." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Unrelated warning reports: PyTorch precision warning vs tie_word_embeddings config warning." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "Left is dummy-class cleanup/CI leakage; right is a noisy logging message. Not the same bug." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Saving dequantized models and saving pipeline objects involve different save paths and failures." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Different model families and symptoms; dequantized save failure is not an infinite-image-loop bug." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "FlashAttention support gating for Nemotron vs GPT2 attention scaling ignored under SDPA/FlashAttention." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "A removed API warning/compat issue is not the same as a noisy logging message." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 breakage and a DeepSeek tensor operation failure are distinct code paths and models." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Model attention backend support vs Python-version compatibility are unrelated." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "Attention backend scaling bug is unrelated to degenerate classification loss with num_labels=1." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "Distributed optimizer-state corruption is not the same as attention scaling being ignored." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "Attention scaling bug vs register() changing callback behavior are different issues." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "FP8 save_pretrained and from_pretrained errors in related models, but not the same concrete failure." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Chat completions API acceptance failure is different from apply_chat_template crashing on batched input." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Gemma3Config missing vocab_size and an Apple Silicon allocator TypeError are unrelated." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training bugs and BeitImageProcessorFast reduce_labels behavior are different subsystems." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "trainer.evaluate() failure after train is unrelated to importing Trainer conflicting with datasets." + }, + { + "left": "issue:43012", + "right": "issue:44683", + "accept": false, + "reason": "A compile-time precision warning is not the same as compiled flex_attention failing on torch>=2.9." + }, + { + "left": "issue:42490", + "right": "issue:43012", + "accept": false, + "reason": "Path-trailing-slash import warning vs bfloat16 compile warning are different concerns." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Different missing imports from different packages/symbols; not the same root cause." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "ImportError for layer_type_validation is unrelated to local_files_only still making API requests." + }, + { + "left": "issue:45600", + "right": "issue:46298", + "accept": false, + "reason": "Broken auto_mappings references are a CI/import bug, not a logging-noise issue." + }, + { + "left": "issue:45698", + "right": "issue:46172", + "accept": false, + "reason": "Wrong custom module loading after save_pretrained is unrelated to DeepSeek tensor conversion failure." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "Missing CLIPOutput attentions assignment and GPT2 attention scaling under SDPA are different regressions." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Qwen3-VL-Moe v5 incompatibility and fp8 inference support are different bugs/features." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "One is attentions not assigned; the other is hidden_states not returned. Similar area, different defect." + }, + { + "left": "issue:44483", + "right": "issue:45290", + "accept": false, + "reason": "Chat/completions request acceptance and tool-call chat template crashes are distinct paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 48, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:36010", + "issue:36683", + "issue:41669", + "issue:42175", + "issue:42757", + "issue:42994", + "issue:43010", + "issue:43064", + "issue:43066", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43525", + "issue:43611", + "issue:43618", + "issue:43756", + "issue:43867", + "issue:43901", + "issue:43931", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44075", + "issue:44077", + "issue:44242", + "issue:44246", + "issue:44355", + "issue:44380", + "issue:44485", + "issue:44492", + "issue:44625", + "issue:44655", + "issue:44855", + "issue:44938", + "issue:45071", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45310", + "issue:45375", + "issue:45584", + "issue:45593", + "issue:46077", + "issue:46082", + "issue:46202", + "issue:46209", + "issue:46365" + ], + "soft_pairs": [ + "issue:43756|issue:44485", + "issue:43299|issue:43867", + "issue:43066|issue:46209", + "issue:36683|issue:43475", + "issue:42994|issue:44655", + "issue:45276|issue:46077", + "issue:43901|issue:44655", + "issue:44077|issue:44380", + "issue:36010|issue:42757", + "issue:43352|issue:43937", + "issue:43611|issue:44655", + "issue:44625|issue:46077", + "issue:36683|issue:44077", + "issue:43867|issue:45216", + "issue:46077|issue:46082", + "issue:45584|issue:46365", + "issue:43976|issue:44938", + "issue:42175|issue:46365", + "issue:43352|issue:45593", + "issue:36683|issue:43525", + "issue:43618|issue:45216", + "issue:44077|issue:46365", + "issue:43352|issue:43931", + "issue:36683|issue:44242", + "issue:44075|issue:46365", + "issue:43992|issue:45310", + "issue:43064|issue:44492", + "issue:43010|issue:45278", + "issue:44355|issue:44855", + "issue:43867|issue:46202", + "issue:43867|issue:45375", + "issue:41669|issue:44246", + "issue:44938|issue:45071", + "issue:36683|issue:43064" + ], + "prepared_review_unit_hash": "4e4297c34b1df79ac716a1182eca2c0f05e6c5c93709bff7671ecfdfe48898ed", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 34, + "serialized_chars": 36819, + "estimated_input_tokens": 9205, + "estimated_eval_tokens": 18666 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous Transformers issues with only superficial similarity in a few titles. I reject all soft edges; none look like the same underlying bug/change. The broadest representative issue is the 4.57.0\u21925.5.0 import-error regression.", + "confidence": 0.28, + "canonical_issue_reason": "issue:45278 is the broadest umbrella here: it describes widespread import failures after upgrading to 5.5.0, which is the closest match to the cluster\u2019s general theme of breakage/regression reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best single issue to represent this set because it is the most general regression report and could plausibly subsume several import/load breakage reports, even though the cluster is not truly duplicate-like.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Different models and different bugs: Smollm3 RoPE layer count vs GLM-5 RoPE implementation discussion." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Both are loading-related, but one is Qwen3VL MoE loading and the other is a generic state_dict sorting failure; different failure modes." + }, + { + "left": "issue:43066", + "right": "issue:46209", + "accept": false, + "reason": "Wrong tokenizer decoder type and FP8 inference on B200 are unrelated code paths." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Gemma3 config missing vocab_size and SAM 3 Video missing fpn_position_embeddings are different model/config bugs." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized model saving and saving Pipeline objects are both save-related, but they are not the same underlying problem." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Same model family, but resize_token_embeddings and bidirectional attention masks are separate subsystems." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mentioning return_all_scores is a documentation mismatch; save_pretrained on Pipeline objects is a runtime serialization bug." + }, + { + "left": "issue:44077", + "right": "issue:44380", + "accept": false, + "reason": "PatchTSMixer post_init handling and GPT2 attention scaling under SDPA/FlashAttention are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols from different dependencies." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Flash Attention 2 support error for Nemotron and GLM-5 GenerationConfig validation are different problems." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Base_model_prefix loading regression is not the same as Pipeline save_pretrained serialization." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and Gemma4 bidirectional attention are unrelated issues." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Different models and unrelated attributes: Gemma3 vocab_size vs PatchTSMixer post_init." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Both mention loading/saving, but one is generic state_dict sorting and the other is a Qwen3.5 checkpoint regression." + }, + { + "left": "issue:46077", + "right": "issue:46082", + "accept": false, + "reason": "Both are Gemma4/Llama configs, but the actual bugs are different config semantics." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty-transcription generation failure is unrelated to importing Trainer being affected by datasets." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility complaints, but they are about different versions and likely different causes." + }, + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "TensorFlow backend packaging and Trainer import interference from datasets are not the same issue." + }, + { + "left": "issue:43352", + "right": "issue:45593", + "accept": false, + "reason": "Unsupported Flash Attention 2 and D-FINE auxiliary loss omission are unrelated model behaviors." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Different missing config attributes in different models." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIPOutput attentions regression and Qwen3.5 checkpoint regression are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "PatchTSMixer post_init and Trainer import shadowing by datasets are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:43931", + "accept": false, + "reason": "Nemotron Flash Attention support and Qwen3-VL weight-shape mismatch are different loading/runtime failures." + }, + { + "left": "issue:36683", + "right": "issue:44242", + "accept": false, + "reason": "Gemma3 config attribute bug and load-balancing loss omission are unrelated." + }, + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "SGD args not used is an optimizer bug, not an import/dependency issue." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both involve Qwen3.5 loading/saving, but one misses an encoder weight and the other is a from_pretrained error after 5.4.0; not the same bug." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "FSDP2/PEFT optimizer state corruption and a cache-strategy typo are unrelated." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "A cache update no_grad decorator bug is unrelated to widespread import failures after upgrading Transformers." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Compiled Python file errors and the Python 3.13 torch.jit decorator parsing bug overlap only loosely; not the same underlying issue." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "State_dict sorting load failure and Gemma pipeline roundtrip on \u2581 are different loading/serialization bugs." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Generic state_dict ordering failure and Qwen3_5MoeVisionConfig field dropping are different problems." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Import-time slowness from removing import * and intermittent import slowness are too broad to treat as the same concrete bug." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different compatibility issues." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3 config attribute error and Trainer FSDP2 optimizer-state corruption are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 49, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:28282", + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:36296", + "issue:42548", + "issue:42831", + "issue:42890", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43299", + "issue:43502", + "issue:43575", + "issue:43576", + "issue:43618", + "issue:43723", + "issue:43867", + "issue:43874", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44077", + "issue:44112", + "issue:44222", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44625", + "issue:44749", + "issue:44857", + "issue:44961", + "issue:44964", + "issue:45125", + "issue:45357", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45561", + "issue:45600", + "issue:45850", + "issue:46132", + "issue:46285", + "issue:46365" + ], + "soft_pairs": [ + "issue:42890|issue:45561", + "issue:43010|issue:45850", + "issue:44625|issue:46365", + "issue:43867|issue:45357", + "issue:43992|issue:45478", + "issue:43010|issue:43976", + "issue:36296|issue:43575", + "issue:43618|issue:44077", + "issue:28282|issue:33453", + "issue:43618|issue:46285", + "issue:43618|issue:44857", + "issue:44360|issue:45440", + "issue:43502|issue:44062", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:43723|issue:44556", + "issue:44387|issue:44556", + "issue:44222|issue:45125", + "issue:43010|issue:43065", + "issue:43576|issue:46132", + "issue:43502|issue:44485", + "issue:42548|issue:44351", + "issue:45125|issue:45478", + "issue:44355|issue:44749", + "issue:30333|issue:33453", + "issue:43010|issue:43232", + "issue:45125|issue:45375", + "issue:30990|issue:31795", + "issue:43502|issue:44222", + "issue:43299|issue:44556", + "issue:43502|issue:44961", + "issue:30333|issue:36296", + "issue:42831|issue:43874" + ], + "prepared_review_unit_hash": "ebdf204fc2aadd08b797ff98e3a46dd29b75f1a16a8fde653d604f578ec2c1a2", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37400, + "estimated_input_tokens": 9350, + "estimated_eval_tokens": 18956 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: import errors, loading regressions, tensor-parallel/FP8 issues, CI flakes, and docs problems. None of the soft-similarity pairs look like true duplicates or the same concrete fix path.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue cleanly represents the set; the items span unrelated bugs, regressions, docs, and CI/test flakes.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43502 is the most concrete and self-contained bug report in the set, but it is not a true duplicate hub for the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Both are test failures, but one is a missing set_seed in SAM HQ integration tests and the other is an xdist race in captured_info.txt; different bugs." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "Specific no_grad decoration request vs a broad complaint that 5.0 breaks many models; not the same code-path." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "Qwen3.5 num_labels propagation bug vs a Trainer import conflict tied to datasets; unrelated failures." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Model load failure on sorted state_dict vs incorrect visual encoder keys in save_pretrained; opposite directions and different causes." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "UMT5Encoder missing embed_tokens.weight vs Qwen3.5 MoE from_pretrained error; different model families and code paths." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "No_grad/cache update bug vs Python-version compatibility regression; unrelated." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Generic tensor-parallel training bug vs OOM while loading a specific large Qwen2 model; not the same defect." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Missing CLIPOutput attentions vs patchtsmixer post_init validation; different subsystems." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "PyTorch missing import error vs tokenizer loading regression; unrelated import/load issues." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Both mention CLIP, but one is an output field regression and the other is a model structure inconsistency; different fixes." + }, + { + "left": "issue:43618", + "right": "issue:44857", + "accept": false, + "reason": "CLIP output regression vs LwDetrImageLoss AMP crash; unrelated." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "DSA indexer ReLU discussion vs DeepseekV3MoE implementation divergence; different bugs." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "local_files_only still making API calls vs tokenizers.AddedToken keyword conflict; unrelated." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "SAM HQ test seeding issue vs Phi-4 multimodal load failure; different models and failure modes." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/test issues, but one is seeding in SAM HQ and the other is a stale device override in GraniteSpeech; not the same bug." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "SAM HQ flakiness vs auto_mappings referencing removed configs; unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Tokenizer loading regression vs checkpoint reload incompatibility after upgrade; both are loading-related but not the same concrete path." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "int4 quantization memory growth/OOM vs generic reload failure after version upgrade; different root causes." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained issue for MoE vs missing _tp_plan for tensor parallelism; distinct model/export paths." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache/layer no_grad decoration vs dummy Conv2d in Sam3PixelDecoder; unrelated." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "Broken transformers env command vs AttentionInterface.register behavior change; different APIs." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "local_files_only network leak vs GLM-5 RoPE implementation discussion; not the same issue." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Different top-level import errors for different symbols; likely separate export/regression issues." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Missing _tp_plan vs Qwen3.5 MoE from_pretrained error; related model line, but not the same concrete bug." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors vs a performance regression after upgrade; unrelated." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLFlow reporting on failed training jobs vs tokenizer loading regression; different functionality." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "no_grad on cache updates vs generation kwargs update after sync_gpus; different generation internals." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Tensor-parallel plan missing vs vision config field dropped by strict config handling; different bug classes." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Loading hang in Sentence Transformers vs documentation confusion in model.forward; not duplicates." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "local_files_only network issue vs FP8 MoE save_pretrained bug; unrelated." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Qwen3VL MoE loading regression vs generic checkpoint reload failure after upgrading from v4.57; too broad to be the same fix." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "API calls despite local_files_only vs unrelated 'racoon' issue; no overlap." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLFlow failure reporting vs tensor-parallel training bug; unrelated." + }, + { + "left": "issue:42831", + "right": "issue:43874", + "accept": false, + "reason": "FineGrainedFP8 accuracy issue vs GLM46V image processor missing method; different components and failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 50, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:41669", + "issue:42371", + "issue:42548", + "issue:42757", + "issue:43525", + "issue:43526", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43824", + "issue:43827", + "issue:43901", + "issue:44038", + "issue:44222", + "issue:44230", + "issue:44273", + "issue:44315", + "issue:44355", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44655", + "issue:44908", + "issue:44938", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45362", + "issue:45468", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46095", + "issue:46143", + "issue:46144", + "issue:46209" + ], + "soft_pairs": [ + "issue:44355|issue:45071", + "issue:43827|issue:44961", + "issue:44355|issue:44655", + "issue:44315|issue:44360", + "issue:33453|issue:34689", + "issue:42757|issue:43576", + "issue:33453|issue:36296", + "issue:43576|issue:45071", + "issue:43576|issue:43824", + "issue:31795|issue:38617", + "issue:44961|issue:45468", + "issue:43576|issue:44938", + "issue:44222|issue:44230", + "issue:46143|issue:46144", + "issue:41628|issue:42548", + "issue:44485|issue:44961", + "issue:44908|issue:44961", + "issue:44315|issue:46209", + "issue:36296|issue:43901", + "issue:36296|issue:43526", + "issue:43673|issue:44492", + "issue:45362|issue:45850", + "issue:44623|issue:45230", + "issue:44273|issue:45542", + "issue:44038|issue:45125", + "issue:44230|issue:46095", + "issue:44393|issue:45071", + "issue:44273|issue:45003", + "issue:43576|issue:44556", + "issue:44393|issue:45753", + "issue:43525|issue:45850", + "issue:41669|issue:43673", + "issue:44556|issue:45092", + "issue:43572|issue:45850", + "issue:44492|issue:45850", + "issue:42371|issue:44492" + ], + "prepared_review_unit_hash": "67ba8efbe2f62c696f93e2e54186fab1f20f201e3bfd471fcf6d72789f9a18bb", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 37610, + "estimated_input_tokens": 9403, + "estimated_eval_tokens": 19062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a mixed bag of Transformers v5 regressions, import/export breakages, and unrelated model-specific bugs. None of the soft pairs look like safe duplicate merges, so I reject them all.", + "confidence": 0.74, + "canonical_issue_reason": "issue:44556 is the clearest concrete regression in the set: reloading a v4.57 checkpoint after upgrading to v5.2/v5.3. It is specific, actionable, and representative of the recurring v5 loading-breakage theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44556 is the most actionable issue here because it names a concrete failure mode and affected versions, unlike the more generic or broad complaints.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Different bugs: compiled Python file loading errors vs a v5.4 PretrainedConfig/type-checking regression." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "One is a docs issue about removed `pipeline()` usage; the other is an unrelated placeholder issue." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Different failure modes: compiled `.py` errors vs `save_pretrained` on Pipeline objects." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger Kernel application via `model_init` is unrelated to the DSA indexer/ReLU discussion." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Both concern loading, but one is tokenizer loading and the other is Llama 3.2 Vision model loading; not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "`huggingface_hub` import failure is distinct from the broken `transformers env` command." + }, + { + "left": "issue:33453", + "right": "issue:36296", + "accept": false, + "reason": "Tokenizer loading regression and tensor-parallel training bug are different code paths and symptoms." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "CLI/environment command breakage is not the same as a `PretrainedConfig` typing regression." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Broken `transformers env` and missing `Qwen2_5_VLForConditionalGeneration` import are separate import/export issues." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion for `forward` arguments is unrelated to the `layer_type_validation` import error." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "No shared bug: a placeholder issue versus a Gemma-4 audio positional encoding bug." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "A broken CLI command is not the same as Python 3.14 load/import compatibility." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Both mention fp8/moe, but one is about `save_pretrained` and the other is about inference support; too different to merge." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Closely related RoFormer topics, but one is missing `**kwargs` passthrough while the other is attention-interface usage; not clearly the same concrete bug." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Missing `AutoImageProcessor` and missing `PreTrainedModel` are separate import surface regressions." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation bug is unrelated to the placeholder issue." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwarg handling has nothing to do with the placeholder issue." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "`model_init`/Liger Kernel behavior is unrelated to fp8 inference failure on B200." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug and pipeline docs mentioning `return_all_scores` are unrelated." + }, + { + "left": "issue:36296", + "right": "issue:43526", + "accept": false, + "reason": "Tensor-parallel training and BeitImageProcessorFast label reduction are different model/component bugs." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "A missing generation cache in chunked prefill is not the same as a cache-strategy typo." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "A specific chat crash in Qwen3.5-35B is not the same as the broad claim that v5 breaks too many models." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "`processor.save_pretrained(...)` missing files is concrete; `Bug report` is too vague and not the same issue." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy-loading behavior and a TensorBoard-without-TF backend error are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "A broad Qwen3-VL-Moe v5 bug report is not the same as missing `_tp_plan` for tensor parallelism." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 inference support for Qwen3.5/moe is unrelated to a DeepSeekV4 weight-conversion question." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "A multimodal bounding-box output bug is unrelated to `PretrainedConfig` type-checking breakage." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading problems are unrelated to unsafe `sys.modules` access in `modeling_utils`." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Broken `transformers env` and checkpoint reload failures after version upgrade are different bugs." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Bounding-box hallucination/errors and infinite looping on a specific image are different Qwen3-VL failures." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_id` in `Llama4Config` is a specific config bug, not the same as the generic v5 breakage complaint." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_idx` in `StableLmConfig` is a distinct config regression from the broad v5 complaint." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo in cache strategy text is not the same as a generic \"v5 breaks too many models\" report." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 behavior API settings and cache-strategy typo are unrelated issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 51, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:31515", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43298", + "issue:43352", + "issue:43408", + "issue:43502", + "issue:43519", + "issue:43576", + "issue:43606", + "issue:43704", + "issue:43827", + "issue:43976", + "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:44623", + "issue:44655", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:45071", + "issue:45230", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45478", + "issue:45850", + "issue:46018", + "issue:46095", + "issue:46172", + "issue:46218", + "issue:46328" + ], + "soft_pairs": [ + "issue:45278|issue:45850", + "issue:41669|issue:46218", + "issue:42890|issue:44908", + "issue:43408|issue:44315", + "issue:44246|issue:44273", + "issue:44297|issue:45230", + "issue:43352|issue:44393", + "issue:44704|issue:44908", + "issue:44295|issue:45230", + "issue:43576|issue:44908", + "issue:43976|issue:44393", + "issue:44908|issue:45071", + "issue:41669|issue:44351", + "issue:44492|issue:46328", + "issue:41669|issue:45341", + "issue:43976|issue:45850", + "issue:44393|issue:45478", + "issue:44077|issue:44861", + "issue:43976|issue:46218", + "issue:43827|issue:44908", + "issue:31795|issue:33453", + "issue:44315|issue:44829", + "issue:43576|issue:44273", + "issue:44655|issue:44861", + "issue:44360|issue:45468", + "issue:45230|issue:45310", + "issue:44273|issue:44908", + "issue:43576|issue:46218", + "issue:46018|issue:46172", + "issue:43519|issue:44485", + "issue:43519|issue:44908", + "issue:41669|issue:43704", + "issue:31795|issue:34689", + "issue:31515|issue:31795", + "issue:44485|issue:45468", + "issue:43704|issue:44485", + "issue:43298|issue:46095", + "issue:45850|issue:46218", + "issue:43502|issue:43519", + "issue:42371|issue:43704", + "issue:43519|issue:44623", + "issue:43519|issue:43704", + "issue:43606|issue:45468", + "issue:44263|issue:45468" + ], + "prepared_review_unit_hash": "4a35f86f71fd29ba987499e41e03be7c4d30964d0d596f9ec2ccba54c2928cbf", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 44, + "serialized_chars": 40175, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: loading/import/model-setup regressions, docs issues, and unrelated model-specific bugs. I don\u2019t see any soft pair that is clearly the same underlying defect or change, so all soft edges should be rejected.", + "confidence": 0.82, + "canonical_issue_reason": "issue:43502 is the best anchor because it is a concrete, user-facing loading/fetch bug with clear scope (`local_files_only=True` still triggering network access).", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43502 is the strongest representative of the cluster\u2019s loading/import regression theme while still being specific enough to triage against.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Both are broad release-breakage reports, but one is about import errors after upgrade and the other about many models breaking; not the same bug." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "One is about removing `import *` to speed imports; the other is a missing Qwen3.5 import symbol. Different fixes and code paths." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "A flaky integration test seed issue is unrelated to the inverse_sqrt scheduler ignoring kwargs." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Wrong model type warning for sam3 tracker vs Liger kernel not applied with `model_init`; unrelated." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import-time slowness and lazy loading failure are both performance/loading-adjacent, but not the same concrete defect." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer class mismatch in save_pretrained is unrelated to a generic bug report." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention 2.0 unsupported and Qwen3-VL bbox hallucination are different model-specific problems." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor kwargs propagation and scheduler kwargs handling are separate API bugs." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Reading `position_ids` after buffer registration is unrelated to the generic bug report." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "`transformers env` being broken and scheduler kwargs being ignored are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version compatibility and Qwen3-VL bbox output are unrelated." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwargs passthrough and `PretrainedConfig` type checking are different regressions." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import cleanup for startup performance is not the same as a missing `HybridCache` export." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "A cache-strategy typo and a `SequenceFeatureExtractor.pad` performance issue are unrelated." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import-style cleanup and a testing_utils bug do not share the same underlying problem." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "Python support matrix issues and broad model breakage after v5 are not the same defect." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Qwen3-VL bounding-box output issue and Qwen3.5 MoE `from_pretrained` failure are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "Optional `post_init` enforcement and `_get_tied_weight_keys` AttributeError are separate code paths." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python version compatibility and missing `Qwen3_5ForCausalLM` import are unrelated." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still referencing `pipeline()` and scheduler kwargs propagation are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion in `forward` and tokenizer loading regression are different issues." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger kernel not applied under `model_init` and degenerate training with flash_attention_3 are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env CLI and lazy loading behavior are different bugs." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Pipeline save_pretrained support and tied-weight-key crashes are unrelated serialization vs utility bugs." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer ReLU discussion and Gemma-4 audio positional encoding are unrelated model internals." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "A generic bug report is not the same as the specific Qwen3.5 MoE from_pretrained error." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy loading problems and scheduler kwargs handling are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:46218", + "accept": false, + "reason": "Env command breakage and missing Qwen3.5 import symbol are unrelated." + }, + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "DeepSeek-V4 gating logic and a weight conversion/dtype-transpose bug are different model-specific issues." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Qwen3VL timestamp calculation and GLM-5 RoPE implementation are separate model bugs." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak in multithreaded dataloading and RoPE implementation are unrelated." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Continuous batching support for audio models and a DeepSeek-V4 weight conversion question are not the same bug." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "Broad v5 breakage and a specific missing import are not the same underlying issue." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "Network requests despite `local_files_only=True` and Qwen3VL timestamp calculation are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 settings warning and VRAM leak in dataloader threads are unrelated." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp calculation bug and missing files in `processor.save_pretrained` are separate processor issues." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Qwen3VL timestamp bug and multithreaded VRAM leak are unrelated." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch and Gemma-4 audio positional encoding are different subsystems." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer `torch.split` behavior and Gemma-4 audio positional encoding are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 19, + "review_unit_index": 52, + "review_unit_count": 52, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-445", + "nodes": [ + "issue:30333", + "issue:31515", + "issue:33453", + "issue:36296", + "issue:37934", + "issue:38617", + "issue:42913", + "issue:42936", + "issue:43066", + "issue:43240", + "issue:43298", + "issue:43352", + "issue:43525", + "issue:43572", + "issue:43618", + "issue:43668", + "issue:43717", + "issue:43813", + "issue:44031", + "issue:44230", + "issue:44261", + "issue:44351", + "issue:44492", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44938", + "issue:44998", + "issue:45290", + "issue:45310", + "issue:45468", + "issue:45850", + "issue:46018", + "issue:46144", + "issue:46218" + ], + "soft_pairs": [ + "issue:44261|issue:45468", + "issue:44351|issue:46218", + "issue:43352|issue:46218", + "issue:42913|issue:43066", + "issue:44908|issue:44998", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44998|issue:45850", + "issue:44861|issue:45290", + "issue:44938|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:44230|issue:45310", + "issue:31515|issue:33453", + "issue:43298|issue:44230", + "issue:43717|issue:46144", + "issue:42936|issue:46018", + "issue:43668|issue:44031", + "issue:42936|issue:43240", + "issue:43240|issue:46144", + "issue:42936|issue:44031", + "issue:42936|issue:46144", + "issue:42936|issue:43813", + "issue:43240|issue:43668", + "issue:43668|issue:43813", + "issue:37934|issue:43668", + "issue:42936|issue:43668" + ], + "prepared_review_unit_hash": "0213c9cc2ca43b965682ed5aca99f53ff5e9d611e60faad1f79ef03663356c00", + "component_budget": { + "node_count": 445, + "item_count": 445, + "soft_pair_count": 1647, + "serialized_chars": 852382, + "estimated_input_tokens": 213096, + "estimated_eval_tokens": 426448 + }, + "budget": { + "node_count": 35, + "item_count": 35, + "soft_pair_count": 31, + "serialized_chars": 28702, + "estimated_input_tokens": 7176, + "estimated_eval_tokens": 14608 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: mostly unrelated model/config/tokenizer regressions plus a few broad v5 complaints. No soft duplicate pair is close enough to merge into one concrete fix.", + "confidence": 0.94, + "canonical_issue_reason": "Issue 45850 is the broadest umbrella-style report in the set (\u201cSince 5.0 version it breaks too many models\u201d), so it is the closest thing to a cluster anchor, even though it is too vague to be a strong duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44261 is the most concrete and actionable bug report: it names a specific model path, describes the precision regression, and has the strongest discussion signal among the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Different models and failure modes: MLA q_a_layernorm eps handling vs Gemma-4 audio positional encoding." + }, + { + "left": "issue:44351", + "right": "issue:46218", + "accept": false, + "reason": "Separate import failures for different symbols/models; not the same code-path bug." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash Attention support error for Nemotron vs Qwen3.5 import-module failure are unrelated." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both are tokenizer/v5-related, but one is a behavior regression and the other is a wrong decoder-type issue." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Inverse-sqrt scheduler kwargs bug is unrelated to the unrelated placeholder issue." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Cache typo report does not match the other issue\u2019s subject." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the other issue." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "One is a nonspecific complaint, the other is a broad v5 breakage report; not the same concrete bug." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Different failures in different helpers: tied weights vs chat template/tool-call crash." + }, + { + "left": "issue:44938", + "right": "issue:44998", + "accept": false, + "reason": "Python 3.14 import/load compatibility issue is unrelated to the other issue." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight-key handling." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "StableLm pad_token_idx regression is a different config issue." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Checkpoint loading slowdown vs missing layer_type_validation import are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 Qwen3.5 support request is unrelated to processor.save_pretrained missing files." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow reporting on failed training jobs is unrelated to checkpoint loading speed." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Checkpoint load slowdown and tensor-parallel training bug are different problems." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen3.5/MoE, but one is fp8 inference support and the other is a from_pretrained import error." + }, + { + "left": "issue:31515", + "right": "issue:33453", + "accept": false, + "reason": "Loading slowdown and tokenizer-loading regression are not the same code-path." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching audio-model support is unrelated to Qwen3.5 FP8 inference support." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba init_weights behavior and RoFormer attention interface use are different model implementations." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2former ignore_value handling is unrelated to DeepSeek-V4 shared-expert gating." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "Incorrect type hint and tokenizer regex warning are different classes of issues." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Model loss-parameter handling does not match a Mask2former ignore_value bug." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Loss kwargs propagation and RoFormer attention interface are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2former ignore_value bug and tokenizer regex warning are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "No shared underlying bug; one is Mask2former-specific, the other is RoFormer-specific." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "Mask2former ignore_value and a typo in quantization_operation are unrelated." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "ModernBERT type-hint issue and quantization typo are unrelated." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Llama4 RMS norm question is unrelated to ModernBERT norm_eps typing." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Mask2former ignore_value handling is unrelated to ModernBERT config typing." + } + ] + }, + "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": 19, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10519-3", + "nodes": [ + "issue:10519", + "issue:11005", + "issue:11307" + ], + "soft_pairs": [ + "issue:10519|issue:11307", + "issue:10519|issue:11005", + "issue:11005|issue:11307" + ], + "prepared_review_unit_hash": "0c6208b5ca0a2b9cf6e47ca2ecb97b81a684925d123f836f3502dfaceb88b478", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2878, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2878, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The three issues are not duplicates: one is about truncation strategy in tokenization, one about ReduceLROnPlateau-like scheduler behavior, and one about Wav2Vec2 word time offsets.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue fits this cluster because the issues are unrelated feature requests.", + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a good global representative; 10519 is the most self-contained, but it does not match the other two topics.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Different domains: truncation behavior vs Wav2Vec2 word offset extraction; no shared underlying bug/change." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "Unrelated requests: truncation-from-beginning option vs ReduceLROnPlateau-like functionality." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "Different features in different components: learning-rate scheduling vs speech model word timing offsets." + } + ] + }, + "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": 19, + "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": "b8cccd8b3bde5b1566cc710955767d5b2b81778a8496361a0f97bbae6df460ae", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2478, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These three issues do not appear to be duplicates of one another: one is about Tapas tokenization differences, one about a bad authentication-related download error, and one about failing to load bert-base-uncased. The only broad overlap is general model/package loading behavior, but the concrete problems differ.", + "confidence": 0.88, + "canonical_issue_reason": "Issue #16618 is the broadest and most representative of the cluster because it concerns a model-loading failure. The other issues are narrower and clearly different in scope: #15980 is specifically about private-model auth messaging, and #13244 is about tokenization behavior.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 is the best issue to keep as the cluster representative because it has the most generic user-facing symptom and closest overlap with the other model-loading complaint, while still being distinct from the tokenization and login-message issues.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Different underlying problem: Tapas tokenization mismatch vs. a bert-base-uncased model loading failure. They share only a broad NLP/model context." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "#15980 is about the error message when downloading a private model without login; #16618 is about loading a public model failing. Related area, but not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 4, + "component_count": 19, + "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": "6cce52829ae05b6f26bf483401a285a35819216e91a64c69b6d165d687651e95", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are both community contribution requests about adding architecture support, but they target different features: Flash Attention 2 vs GGUF. They are not duplicates and should not be clustered as the same underlying request.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different feature requests and code paths: one asks for Flash Attention 2 support across more architectures, the other asks for GGUF support across more architectures. They do not describe the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 5, + "component_count": 19, + "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-26413-2", + "nodes": [ + "issue:26413", + "issue:31501" + ], + "soft_pairs": [ + "issue:26413|issue:31501" + ], + "prepared_review_unit_hash": "0959fe5237863835eeb8f39f7f00d7d1f4d0bc07a90c576e8b49e30c1e88a628", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1593, + "estimated_input_tokens": 399, + "estimated_eval_tokens": 1054 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1593, + "estimated_input_tokens": 399, + "estimated_eval_tokens": 1054 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues both concern training startup failures, but only one describes a specific root-cause error. They do not appear to be confidently the same underlying bug.", + "confidence": 0.61, + "canonical_issue_reason": "issue #26413 is the better canonical issue because it names a concrete failure mode and error message: `resume_from_checkpoint` fails with `There seems to be not a single sample in your epoch_iterator`.", + "canonical_pr_reason": null, + "best_issue_reason": "issue #26413 is the best representative issue overall since it is specific, reproducible, and tied to an explicit code-path failure rather than a broad symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "Both involve training not starting, but #31501 is too broad and does not clearly point to the same checkpoint-resume/epoch_iterator bug. The overlap is only at the symptom level." + } + ] + }, + "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": 19, + "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-27088-8", + "nodes": [ + "issue:27088", + "issue:28936", + "issue:38435", + "issue:43834", + "issue:44365", + "issue:44370", + "issue:46222", + "issue:46226" + ], + "soft_pairs": [ + "issue:27088|issue:44370", + "issue:27088|issue:44365", + "issue:27088|issue:43834", + "issue:27088|issue:38435", + "issue:27088|issue:28936", + "issue:28936|issue:43834", + "issue:28936|issue:44365", + "issue:28936|issue:38435", + "issue:28936|issue:44370", + "issue:38435|issue:44370", + "issue:38435|issue:44365", + "issue:38435|issue:43834", + "issue:38435|issue:46222", + "issue:38435|issue:46226", + "issue:43834|issue:44370", + "issue:43834|issue:44365", + "issue:43834|issue:46222", + "issue:43834|issue:46226", + "issue:44365|issue:44370", + "issue:44365|issue:46222", + "issue:44365|issue:46226", + "issue:44370|issue:46222", + "issue:44370|issue:46226", + "issue:46222|issue:46226" + ], + "prepared_review_unit_hash": "8fc41a402ca28af86e515adaf8abd23ca00b53f78f5846de8e574296c7732b22", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13531, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13531, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly false-positive similarity across unrelated language translation requests. The only clearly supported duplicate subset is the exact placeholder-title trio 43834/44365/44370; the Thai-language pair 46222/46226 is not strong enough to fold into that group without body-level evidence.", + "confidence": 0.84, + "canonical_issue_reason": "43834 is the earliest issue in the exact placeholder-title trio, which is the only clearly supported duplicate set here.", + "canonical_pr_reason": null, + "best_issue_reason": "43834 is the cleanest representative of the exact placeholder-title trio and predates the other clearly matching copies.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Different language request vs generic placeholder issue; similarity is only superficial." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "Turkish docs translation request is distinct from the placeholder-template issue." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Different target language request; not the same underlying issue." + }, + { + "left": "issue:27088", + "right": "issue:38435", + "accept": false, + "reason": "Turkish and Romanian translation requests are separate issues." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Turkish and Spanish translation requests are not duplicates." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Spanish translation request is distinct from the generic placeholder issue." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "Different language-specific request; template similarity alone is not enough." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Spanish and Romanian translation requests are separate issues." + }, + { + "left": "issue:28936", + "right": "issue:44370", + "accept": false, + "reason": "Spanish request and generic placeholder issue are not the same change." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "Romanian translation request is distinct from the placeholder-template issue." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "Romanian translation request does not match the generic placeholder issue." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "Romanian-specific request is not the same as the placeholder-template issue." + }, + { + "left": "issue:38435", + "right": "issue:46222", + "accept": false, + "reason": "Romanian translation request and opaque Thai-text issue are not clearly the same underlying task." + }, + { + "left": "issue:38435", + "right": "issue:46226", + "accept": false, + "reason": "Romanian translation request and opaque Thai-text issue are not clearly the same underlying task." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": true, + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": true, + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "High soft similarity is not enough here; the title is different and there is no body evidence tying it to the placeholder trio." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "High soft similarity is not enough here; the title is different and there is no body evidence tying it to the placeholder trio." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": true, + "reason": "Exact same placeholder title and near-identical template body shape; clearly the same copied issue." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "Different title and no body-level evidence to treat it as the same placeholder issue." + }, + { + "left": "issue:46222", + "right": "issue:46226", + "accept": false, + "reason": "The Thai titles are identical, but without body content or explicit linkage the evidence is still too thin to assert the same underlying issue conservatively." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The writeup overstates the evidence for the duplicate subset. The packet clearly supports treating 43834/44365/44370 as near-identical placeholder-title issues, but it does not ground the claim that 46222 and 46226 are the same placeholder issue: their titles are completely different (Thai text), and only soft similarity is available. That makes those accept verdicts non-conservative. The summary should be narrowed to the explicitly matching placeholder-title trio unless there is stronger body-level evidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 19, + "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-36817-16", + "nodes": [ + "issue:36817", + "issue:36879", + "issue:37015", + "issue:41115", + "issue:41211", + "issue:42503", + "issue:42738", + "issue:43671", + "issue:43837", + "issue:43909", + "issue:44121", + "issue:44205", + "issue:45027", + "issue:45177", + "issue:45322", + "issue:46133" + ], + "soft_pairs": [ + "issue:36817|issue:41211", + "issue:36817|issue:43671", + "issue:36817|issue:36879", + "issue:36879|issue:41211", + "issue:36879|issue:43909", + "issue:36879|issue:44121", + "issue:36879|issue:43837", + "issue:36879|issue:43671", + "issue:37015|issue:43837", + "issue:37015|issue:44121", + "issue:37015|issue:43909", + "issue:36879|issue:37015", + "issue:37015|issue:42503", + "issue:41115|issue:43837", + "issue:41115|issue:44121", + "issue:41115|issue:42503", + "issue:41115|issue:45027", + "issue:41211|issue:43671", + "issue:42503|issue:43837", + "issue:42503|issue:43671", + "issue:42503|issue:42738", + "issue:42738|issue:45027", + "issue:43671|issue:43837", + "issue:43671|issue:45027", + "issue:43671|issue:45177", + "issue:43671|issue:44121", + "issue:43671|issue:43909", + "issue:43837|issue:43909", + "issue:43837|issue:45322", + "issue:43837|issue:44121", + "issue:43909|issue:44205", + "issue:44121|issue:45322", + "issue:43837|issue:45027", + "issue:37015|issue:45027", + "issue:43837|issue:45177", + "issue:45322|issue:46133" + ], + "prepared_review_unit_hash": "11d1bae104e9ff19dbb33363ce18cf8015696268ad9f1307d785e10e402dd4c5", + "component_budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 36, + "serialized_chars": 21658, + "estimated_input_tokens": 5415, + "estimated_eval_tokens": 11086 + }, + "budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 36, + "serialized_chars": 21658, + "estimated_input_tokens": 5415, + "estimated_eval_tokens": 11086 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly a set of distinct model-request issues that only share generic \"add/support model\" wording. The soft links are driven by superficial similarity in model names and request style, not the same underlying change, so none should be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "No single true canonical issue: these requests target different models, tasks, and code paths, so the cluster is not a duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43837 is the best representative hub for the cluster because it is one of the most connected support requests and sits near several of the other model-addition issues, but it still is not a duplicate of them.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBert config vs DEIMv2 are different model requests with unrelated architectures." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config and Qwen3-TTS are unrelated text vs audio model additions." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBert vs RF-DETR are different models and different tasks." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are distinct vision-model requests, not the same change." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR vs LFM2.5 Audio are different model families and modalities." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR and OpenAI Weight-Sparse Transformer are unrelated requests." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR and Qwen3-ASR are different models and code paths." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR and Qwen3-TTS are unrelated model-support requests." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT vs Qwen3-ASR are different model families and tasks." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT and the sparse transformer request describe different architectures." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT and LFM2.5 Audio are unrelated text vs audio requests." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR vs NeoBERT are distinct model additions." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are both BERT-like names but refer to different concrete model additions." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 and Qwen3-ASR are different models and different support paths." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 and the sparse transformer request are unrelated changes." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 and ModernVBERT are separate model requests, not the same issue." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 vs Voxtral TTS are different models and modalities." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated model-support requests." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are different model additions." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS are unrelated requests." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "One is a specific model request, the other is a broad RoPE/BERT-like architecture request; not the same underlying change." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like models with RoPE and Voxtral TTS are unrelated." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR are related in family but require different concrete support paths, so they are not the same change." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS vs Voxtral TTS are different model-support requests." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "TTS support and OCR2 are different modalities and code paths." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS and the sparse transformer request are unrelated." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS vs LFM2.5 Audio are different model additions." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio are different models and support implementations." + }, + { + "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": "ASR support and a sparse transformer request do not describe the same code change." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are different modalities and model families." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "Weight-Sparse Transformer and EUPE are separate research-model requests." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different audio-model support requests." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are unrelated model requests." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "ASR support and DeepSeek-OCR2 are different model tasks." + }, + { + "left": "issue:45322", + "right": "issue:46133", + "accept": false, + "reason": "EUPE and TIPSv2 are distinct vision-language model requests, not the same underlying feature." + } + ] + }, + "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": 19, + "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": "d6decbfccab1f1c064c6e6c051634d0cd25e40744a61e6793393b8ee10feaba1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: one is about block decoding, the other about DeepSeek v3.2 model support. They describe different problem domains with no explicit linkage or shared implementation target.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: the items address unrelated topics and should not be merged as duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "No best issue for a duplicate cluster; neither issue clearly subsumes the other.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different underlying problems: block decoding vs DeepSeek v3.2 support. No shared code path, filenames, or explicit references tying them together." + } + ] + }, + "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": 19, + "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": "28fce62e6d25aeb54d34302fd1dad5d1876f9bcf791c008a3a77da075e3c3ea3", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1582, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1582, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are related thematically (custom trainers and Trainer behavior) but they describe different problems: one is about `remove_unused_column` customizability, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": "No single canonical issue fits both; the titles point to separate Trainer concerns with different expected fixes.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue needs to represent the cluster, #43598 is the better umbrella because it is more general about Trainer customizability, but it still does not cover FLOPS estimation.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Only a broad topical overlap around custom trainers; the concrete bugs differ (`remove_unused_column` handling vs FLOPS estimation), 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": 10, + "component_count": 19, + "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": "7dc5f32f2631106dfaac72056f96a27d8951aa5fe396836c2df1f1f1bc38c46c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1611, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1611, + "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 in that both request AutoModel-family support for Qwen variants, but they target different models and different auto classes, so they are not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 43684 is the broader registration request and the more canonical representative of the cluster, but it does not subsume the Qwen3.5 sequence-classification support request.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43684 is the cleaner cluster representative because it asks for model registration in core auto classes and is phrased as a general support gap rather than a single downstream task.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying requests: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 support for AutoModelForSequenceClassification. Same broad subsystem, but not the same bug/change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 19, + "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": "ed750937ccaf09a913164f540dc2d4b1704531d87ad17ef8345c0cbff6ec38ef", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4267, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4267, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Three issues titled \"huggingface \" look like duplicates of the same report; \"Fett\" appears unrelated. I\u2019d use issue 43845 as the representative canonical issue among the duplicate set.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 43845 is the earliest of the three nearly identical \"huggingface \" issues and is the cleanest representative for that duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "43845 best represents the cluster because it matches the other duplicate issues\u2019 title/profile and is earliest among them.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different title and no evidence of the same underlying problem; similarity is too weak to treat as duplicates." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different title and no concrete shared bug/change signal beyond generic soft similarity." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different title and insufficient evidence that it reports the same issue." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same title, same general metadata profile, and created close together; these look like the same underlying issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same title and clearly overlapping report pattern; likely duplicate of the same issue." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same title and close temporal proximity strongly suggest duplicate reports of the same problem." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering is grounded in the packet: 43845/43846/43859 share the same title and have higher soft-similarity scores, while 43835 has a different title and much weaker similarity. The canonical choice of 43845 as the earliest issue in that duplicate set is also supported by the timestamps. The soft-edge verdicts are conservative overall." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 12, + "component_count": 19, + "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": "c1684091d65ce2b2542f988dedfa44f36fb5e952c84f126f3ea4c557878f06d7", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5791, + "estimated_input_tokens": 1448, + "estimated_eval_tokens": 3152 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are all low-signal, unrelated issue reports dominated by spammy or placeholder titles; there is no clear shared underlying bug or request.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 44418 is the most neutral, issue-like entry in an otherwise spam/placeholder cluster, so it is the least-bad representative.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44418 is the best global representative only because it looks like a real issue title; the others are more obviously spam or promotional noise.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both titles are just \"\"; identical placeholder text is not enough to show the same underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Both are placeholder spam issues, but there is no concrete bug, request, or target proving they are duplicates." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is spam placeholder text and the other is a generic documentation suggestion; they do not describe the same problem." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "\"\" and \"Check out \\\"Google LLC\\\"\" are clearly different spam-like posts, not the same issue." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "A documentation enhancement suggestion and a spam placeholder issue are not the same underlying request." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "The titles do not describe the same bug or change; one is a generic doc request, the other is spam." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Matching placeholder spam titles alone are insufficient to merge them as duplicates without shared concrete content." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "Two different spammy titles with no shared technical substance; no evidence of the same underlying issue." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "No shared bug, feature, or target; these are just separate low-content spam issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 19, + "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": "6ad466947253d50830dc473b87618cbd6394d5f573c80dd22ccbf65ed760aca2", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues have the same title and highly similar content, indicating they are duplicate reports of the same energy-reduction idea. Issue #45145 is the better canonical choice because it was created earlier and already has an inbound reference.", + "confidence": 0.99, + "canonical_issue_reason": "#45145 is the earlier report and has an inbound reference, so it is the best canonical issue for this duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "It is the earliest instance of the duplicated title/body and is therefore the most stable issue to keep as the canonical tracker.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": true, + "reason": "Same title, very similar body size/content, and no distinguishing metadata suggests they are duplicates of the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The pair is reasonably grounded as a duplicate: the issues have identical titles, very similar body sizes, and the packet flags soft similarity between them. The canonical choice of issue #45145 is conservative and supported by earlier creation time plus an inbound reference. No clear overstatement beyond the available evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 14, + "component_count": 19, + "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": 15, + "component_count": 19, + "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": "02f172400c17462b4a37e2d4b7da9c472dc4513e93d7c86137252a7cabe57b78", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1670, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1670, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are both 4-bit inference reports, but they describe different bugs: one is a Gemma4/audio-module torch.finfo() TypeError on uint8 weights, the other is a BitsAndBytesConfig skip-module regression that clears the default lm_head exclusion and triggers an AssertionError. Shared topic, different underlying defects.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Same broad quantized inference area, but different failure modes and code paths: Gemma4 audio-module uint8/tensor dtype handling vs BitsAndBytesConfig lm_head exclusion behavior. Not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 16, + "component_count": 19, + "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": "ba73de174d39eef1d6c0979a33a7e3338fc7f8205bc58c57c011b815ea5950a1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1476, + "estimated_input_tokens": 369, + "estimated_eval_tokens": 994 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are unrelated-looking title-only entries with different names and dates; there is not enough evidence that they describe the same underlying problem.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different person names and different dates; no shared bug report, code path, or explicit linkage indicating a duplicate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 19, + "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": "b0094e1332dbafad69640bb20c4d604ad6dc86b5fb0ec033297b9bea06720991", + "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": "These two issues are just unrelated personal-name/date entries with no project context or shared bug/feature. They do not look like duplicates and should not be clustered together 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:45764", + "right": "issue:45765", + "accept": false, + "reason": "Titles only contain different names and birth dates; there is no shared code-path, error, or issue description indicating the same underlying problem." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 18, + "component_count": 19, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45810-2", + "nodes": [ + "issue:45810", + "issue:46295" + ], + "soft_pairs": [ + "issue:45810|issue:46295" + ], + "prepared_review_unit_hash": "7e40d7f5d54309b2d72e201f6541349c3365589be4205a7ffa70e105b5ce404b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1541, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: one requests a new model class for token classification/value modeling, while the other asks for FP8 training support. They concern different features and different code paths.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue: these requests are about unrelated functionality, so neither can represent the other as a duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a good global best for the cluster because the cluster does not appear to represent one underlying bug or feature request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "Different feature requests in different areas: model/class support for token classification vs FP8 training support. They do not look like 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": 19, + "component_count": 19, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45869-5", + "nodes": [ + "issue:45869", + "issue:45870", + "issue:45871", + "issue:45872", + "issue:45873" + ], + "soft_pairs": [ + "issue:45869|issue:45870", + "issue:45869|issue:45871", + "issue:45870|issue:45871", + "issue:45870|issue:45872", + "issue:45870|issue:45873", + "issue:45871|issue:45872", + "issue:45871|issue:45873", + "issue:45872|issue:45873" + ], + "prepared_review_unit_hash": "1a5cd63eae185948bef63ce0743018df85b39709149f853919fbeeeeaf5e1dab", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5727, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5727, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two clear duplicate issue pairs emerge: 45870/45871 are exact duplicates, and 45872/45873 are exact duplicates. Issue 45869 is only loosely related by finance/data wording and is not the same underlying request.", + "confidence": 0.97, + "canonical_issue_reason": "Issue 45870 is the best canonical representative of the exact duplicate pair 45870/45871: it is slightly earlier and has the only discussion activity, while 45871 appears identical.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45870 is the strongest overall issue candidate among the set because it is part of an exact duplicate pair and is the earliest, most canonical wording. 45872 is equally a valid canonical for its own separate duplicate pair, but 45870 is the cleanest representative here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Different underlying requests: 45869 is a financial dataset offer, while 45870 is an example/tutorial about financial time series prediction with DataForge." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Same as above; the dataset bundle request is not the same issue as the DataForge prediction example." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "These are exact duplicates: same title, same body size, and no distinguishing evidence." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Both are finance-related, but one is a prediction example and the other is a scraping-service feature request; they are not the same concrete issue." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "The titles describe different requests with different product focus, so they should not be merged as duplicates." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "The example/tutorial issue and the scraping-service issue are distinct underlying requests." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Not the same bug or change; only broad financial/time-series similarity is present." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "These are exact duplicates: same title, same body size, and no distinguishing evidence." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: 45870/45871 and 45872/45873 have identical titles, matching body lengths, and the strongest similarity signal available, while 45869 is only loosely related by theme. The canonical choice for 45870 over 45871 is also supported by the earlier timestamp and discussion activity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-39895-16", + "nodes": [ + "pull_request:39895", + "pull_request:41855", + "pull_request:42668", + "pull_request:43251", + "pull_request:43254", + "pull_request:43424", + "pull_request:43823", + "pull_request:43995", + "pull_request:44827", + "pull_request:45170", + "pull_request:45213", + "pull_request:45655", + "pull_request:46048", + "pull_request:46267", + "pull_request:46344", + "pull_request:46367" + ], + "soft_pairs": [ + "pull_request:39895|pull_request:43424", + "pull_request:41855|pull_request:43424", + "pull_request:41855|pull_request:43823", + "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:46344", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:46267", + "pull_request:44827|pull_request:46367", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:44827|pull_request:45655", + "pull_request:46048|pull_request:46367", + "pull_request:44827|pull_request:46048", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "b396088bbe5aaccede7b265f84346086cedde74404fb8c2c6fb7e56e5740d6f1", + "component_budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 22, + "serialized_chars": 42942, + "estimated_input_tokens": 10736, + "estimated_eval_tokens": 21728 + }, + "budget": { + "node_count": 16, + "item_count": 16, + "soft_pair_count": 22, + "serialized_chars": 42942, + "estimated_input_tokens": 10736, + "estimated_eval_tokens": 21728 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Only one pair looks like the same underlying fix: the two `fixed_cross_entropy` PRs in `loss_utils.py`. The rest are unrelated feature/model additions, docs, or subsystem-specific changes and should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43254 is the clearest representative of the `fixed_cross_entropy` change: it updates the shared loss helper in the same file, adds `weight` and `label_smoothing`, and preserves `**kwargs` while matching the same bug fix as the other PR.", + "best_issue_reason": null, + "best_pr_reason": "PR 43254 is the best standalone representative of the cluster\u2019s only true duplicate pair because it cleanly captures the concrete `cross_entropy` API support change without extra issue-specific framing.", + "soft_edge_verdicts": [ + { + "left": "pull_request:39895", + "right": "pull_request:43424", + "accept": false, + "reason": "Unrelated PRs: one adds a new VideoPrism model and docs, the other adds Executorch export tests for dynamic shapes. Different code paths and different kinds of change." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Different targets: Mistral tokenizer helper methods vs Executorch dynamic-shape export tests. No shared underlying bug or change." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Tokenizer utility fix vs new MobileLLM model scaffolding. Same repo area only in the broad sense, not the same fix." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor robustness changes across existing models vs adding a new MobileLLM model. Different underlying work." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Loss-helper API fix vs new model implementation. No overlap in concrete bug or code path." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "`fixed_cross_entropy` kwargs support and Mistral4 test fixes are unrelated changes in different subsystems." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "Executorch export tests and Mistral4 test/model fixes do not address the same underlying issue." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Dynamic export testing vs adding MobileLLM. Different features and no shared bug." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "New MobileLLM model addition is unrelated to the config resolution-order fix in `modeling_utils.py`." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "New model scaffold vs Falcon output-collection refactor. Different model families and different change intent." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Falcon refactor and model output-capture work is unrelated to the model-creation skill PR." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon standardized output interface refactor and Mistral4 test fixes are separate model-specific changes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46344", + "accept": false, + "reason": "Mistral4 test/model changes vs VitPose `flip_back` graph-break fix are unrelated code paths." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test fixes and `layrnorm` renaming/conversion-mapping cleanup are different bugs." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46267", + "accept": false, + "reason": "Mistral4 test fixes vs serve chat parsing update are unrelated features/fixes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46367", + "accept": false, + "reason": "Mistral4 test/model work is unrelated to DeepGEMM availability and XPU enablement." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 fixes and the model-creation skill PR do not share a concrete bug or change." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "Adding MobileLLM and adding a model-creation skill are different artifacts with no shared underlying fix." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4-specific fixes and `cls.config` resolution-order logic in `modeling_utils.py` are unrelated." + }, + { + "left": "pull_request:46048", + "right": "pull_request:46367", + "accept": false, + "reason": "TDT loss-kernel integration and DeepGEMM/XPU enablement are different kernel integrations, not the same change." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46048", + "accept": false, + "reason": "Mistral4 test/model work is unrelated to the TDT loss-kernel integration PR." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46344", + "accept": false, + "reason": "Mistral4 test fixes and VitPose graph-break changes do not fix the same bug." + }, + { + "left": "pull_request:46048", + "right": "pull_request:46367", + "accept": false, + "reason": "Different integrations: TDT loss kernel vs DeepGEMM availability helper and XPU path. No shared concrete code-path issue." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Both PRs change `src/transformers/loss/loss_utils.py` in the same `fixed_cross_entropy` function to add `weight` and `label_smoothing` support for the same bug/fix. The differences are minor API-shape details, so they are duplicate implementations of the same underlying change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted pair, 43251/43254, is well supported by the shared file/function, the same issue target, and near-identical `fixed_cross_entropy` API changes. The other rejections also read as appropriately cautious and do not overclaim duplication." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 2, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-3", + "nodes": [ + "pull_request:40546", + "pull_request:42781", + "pull_request:46331" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781", + "pull_request:42781|pull_request:46331" + ], + "prepared_review_unit_hash": "ed65ea101bec4f5d64fc27686cbdf8f875d714b90d510df5fc43858298a8281c", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10380, + "estimated_input_tokens": 2595, + "estimated_eval_tokens": 5446 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10380, + "estimated_input_tokens": 2595, + "estimated_eval_tokens": 5446 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is not a duplicate set: one PR adds base VibeVoice, another adds VibeVoice Realtime as an extension, and the third is unrelated Parakeet-RNNT work that only overlaps on shared auto-registration plumbing. None of the soft edges should be merged.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 40546 is the best canonical representative because it is the original, focused VibeVoice implementation and the clearest anchor for the model family in this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 40546 is the best overall PR here: it is the most self-contained and central change, while the other PRs are either a larger derivative feature set (42781) or a different model family altogether (46331).", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Related model family, but not the same change: 40546 adds base VibeVoice, while 42781 adds VibeVoice Realtime plus acoustic-tokenizer and generation changes. They touch overlapping files, but the concrete code-path work is different and would not plausibly collapse into one duplicate PR." + }, + { + "left": "pull_request:42781", + "right": "pull_request:46331", + "accept": false, + "reason": "These are different models with only generic auto-mapping overlap. 42781 is VibeVoice Realtime, while 46331 is Parakeet-RNNT; they do not fix the same bug or implement the same 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": 3, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40755-2", + "nodes": [ + "pull_request:40755", + "pull_request:41224" + ], + "soft_pairs": [ + "pull_request:40755|pull_request:41224" + ], + "prepared_review_unit_hash": "70af54203d28a10e9288d263f7bc719f179190ae3d29663bdadc40048aa54d3f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests are unrelated: one adds covariate forecasting to TimesFM, the other adds DINOv3 image classification support. They touch different models, code paths, and user-facing features, so they should not be clustered as duplicates.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single PR is a good canonical representative because the items address different features in different model families and are not duplicates.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different underlying changes: TimesFM time-series covariate forecasting versus DINOv3 image classification support. They do not fix the same bug or implement the same change and would not plausibly be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "aead1308c167b0bae6ff11bf9358340baa5e76f816618383418b505a10a9cd01", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two DEIMv2 pull requests target the same model family, but they are not the same change: PR 41356 is an early/basic model + image processor add, while PR 44339 is a much broader follow-up that adds loss code, auto-mappings, conversion tooling, modular files, and more integration work.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44339 is the more complete and representative DEIMv2 implementation; PR 41356 looks like an earlier partial version of the same model bring-up.", + "best_issue_reason": null, + "best_pr_reason": "PR 44339 is the strongest single artifact for the cluster because it subsumes the basic model work and includes the broader integration needed for a full Transformers addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": false, + "reason": "Same model area, but not the same concrete change: 44339 expands the implementation substantially and adds new code paths beyond the basic PR 41356. They look like related iterations, not mergeable duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "0b14c91c651c93129d42b56fdb3db46cbf6418695142e43dd2a5b379be05508f", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10435, + "estimated_input_tokens": 2609, + "estimated_eval_tokens": 5474 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 10435, + "estimated_input_tokens": 2609, + "estimated_eval_tokens": 5474 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are three separate model-addition PRs for different models (DeepSeek OCR, Isaac, and Kimi2-6). They share boilerplate registration/docs files, but they do not appear to fix the same code-path problem or be mergeable into one PR.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No duplicate PR stands out here; if a single representative is needed, PR #45630 is the strongest cluster exemplar because it is the newest and most substantial model addition, with broad integration changes and high review activity.", + "best_issue_reason": null, + "best_pr_reason": "PR #45630 is the best representative of the cluster\u2019s general theme because it is a complete, heavily reviewed model-addition PR. The other two are similar in shape but target different models and different implementations.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41797", + "right": "pull_request:45186", + "accept": false, + "reason": "Different models and implementations: DeepSeek OCR vs Isaac. Shared registry/docs updates are boilerplate for adding a new model, not evidence of the same underlying change." + }, + { + "left": "pull_request:45186", + "right": "pull_request:45630", + "accept": false, + "reason": "Different models and code paths: Isaac vs Kimi2-6. The overlap is only in generic auto-registration and docs, so these 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": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42975-2", + "nodes": [ + "pull_request:42975", + "pull_request:42976" + ], + "soft_pairs": [ + "pull_request:42975|pull_request:42976" + ], + "prepared_review_unit_hash": "f74641d847753619d527c323eded87943ce2c15121a77509e414c9cb5f838e8c", + "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 GitHub Actions dependency bumps in workflow files, but they target different upgrades and different action families: Node 24 compatibility vs. docker/build-push-action v6. They are related in theme, not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the broader and more representative workflow-actions upgrade, spanning many workflows and matching the cluster theme more closely than the narrower Docker action bump.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the best single representative because it is the more comprehensive GitHub Actions compatibility update across the repository.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Different underlying changes: 42975 upgrades several GitHub Actions versions for Node 24 compatibility, while 42976 only bumps docker/build-push-action to v6 in Docker image workflows. They do not appear to fix the same concrete problem or be mergeable into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "19e7d47ee328d9c72fa152ef7c6cda7a3124b70c4d867d8220c55efbb5d49c66", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14253, + "estimated_input_tokens": 3564, + "estimated_eval_tokens": 7384 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 14253, + "estimated_input_tokens": 3564, + "estimated_eval_tokens": 7384 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All four pull requests are separate model-addition PRs, not duplicates. They share common Transformers scaffolding (docs, auto-mapping registrations, tests), but each targets a different model family and code path: ViT NEPA, Molmo, Molmo2, and CTSM.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43451 (Molmo2) is the broadest and most complete model-addition PR in this set, with the highest review activity and the most extensive model/processor/video support; it best represents the cluster as a whole if one PR must be chosen.", + "best_issue_reason": null, + "best_pr_reason": "PR 43451 is the strongest standalone PR to anchor on because it adds a distinct model family with multiple components and substantial implementation/test coverage, whereas the other PRs are also independent model additions but narrower or less developed.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo). Similarity is only in shared integration files and docs, not the same underlying change." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo2). Shared auto/docs boilerplate does not indicate the same bug or feature." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "Different model additions (ViT NEPA vs CTSM). They touch shared registry/docs files but implement unrelated models and code paths." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are related by naming but appear to be separate model families with different configs, processors, and test sets; not mergeable as one PR." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo and CTSM are unrelated model additions; overlap is limited to common auto-registration and documentation updates." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM are distinct model implementations with different filenames and functionality; shared framework files are insufficient for duplication." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43291-4", + "nodes": [ + "pull_request:43291", + "pull_request:43665", + "pull_request:44815", + "pull_request:46380" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43291|pull_request:46380", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:46380", + "pull_request:44815|pull_request:46380" + ], + "prepared_review_unit_hash": "d45176e2ae7de1d8a50a28979496b6dc471c1cf21d608734c3dd83e12f6a43a6", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 11142, + "estimated_input_tokens": 2786, + "estimated_eval_tokens": 5828 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 11142, + "estimated_input_tokens": 2786, + "estimated_eval_tokens": 5828 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are not duplicates: they target different code paths (Whisper tokenization, test-only edits, FP8 dequantization, and generation watermarking/config validation).", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No strong duplicate cluster exists; if one PR must be chosen as the reference, 46380 is the most substantive and self-contained production fix, with real runtime logic plus tests.", + "best_issue_reason": null, + "best_pr_reason": "46380 is the strongest standalone PR in the set because it addresses a concrete runtime validation bug and includes broad test coverage, while the others are either unrelated or look like test-focused noise.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Whisper tokenization/timestamp decoding vs. test-only edits in CLIP/ViT; no shared underlying bug or change." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper decode logic vs. FP8 dequantization/Mistral4; different subsystems and failure modes." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Whisper tokenization fix vs. generation watermarking/config validation; the shared Whisper filename is incidental, not the same bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Bogus test edits vs. FP8 dequant fix; unrelated." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "Test-only noise in vision tests vs. generation/watermarking runtime fix; no duplicate overlap." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "FP8 kernel padding/dequant path vs. generation config watermarking validation; distinct concrete issues." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-2", + "nodes": [ + "pull_request:43488", + "pull_request:43532" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:43532" + ], + "prepared_review_unit_hash": "3764f8271bab9f93a174563d4bab67fa29d491a1ced29c81e87f69a785e00f09", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4563, + "estimated_input_tokens": 1141, + "estimated_eval_tokens": 2538 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs only overlap on a shared Vit test file, but they address unrelated changes: one tweaks repo-bot/setup metadata, the other changes GitHub workflows and notification/testing plumbing. They should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the overlap is only a shared test file, while the underlying changes are different and not mergeable into one fix.", + "best_issue_reason": null, + "best_pr_reason": "Neither PR is a good cluster representative; if forced, 43532 is the broader change, but it still does not match 43488's repo-bot/version dependency tweak.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "Shared filename alone is not enough; 43488 is a setup/version/dependency and bot-format change, while 43532 is workflow/notification/test infrastructure work. Different code paths and different fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 10, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43612-2", + "nodes": [ + "pull_request:43612", + "pull_request:43613" + ], + "soft_pairs": [ + "pull_request:43612|pull_request:43613" + ], + "prepared_review_unit_hash": "6bad0104c754d12e3a97153940461097f5b891a18beae61442a0bde35cb84e71", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two related but distinct PRs add separate promptable segmentation pipelines. They overlap in shared docs/registration plumbing, but one implements concept segmentation and the other visual segmentation with different models/tests, so they are not duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43613 is the better representative of the cluster because it is the more comprehensive segmentation-pipeline change and touches a broader set of model integrations and tests.", + "best_issue_reason": null, + "best_pr_reason": "PR 43613 is the strongest representative of the set: it is a concrete feature PR with substantial implementation, docs, and test coverage for a specific pipeline.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Reject: they are adjacent features, not the same fix. 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation and integrates different underlying models/tests. Shared docs and pipeline registry changes are incidental." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "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": "f82a1394dda9bb7dd3b38871728f29759afc2d75cf21d41dd347f2cbe166116d", + "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": "Two separate Trainer extensibility PRs: one adds a hook for gradient-accumulation loss scaling, the other adds a hook for signature-column filtering. They share a file and theme but not the same underlying change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43663 is a clean, self-contained API extraction in `trainer.py` that introduces a reusable override point for column filtering and cleanly refactors existing logic.", + "best_issue_reason": null, + "best_pr_reason": "PR #43663 is the better representative because it is a focused trainer API change with a clear override surface and minimal surrounding scope.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Both touch `trainer.py` and aim to make custom trainers more extensible, but they address different code paths: loss scaling during gradient accumulation vs. dataset column filtering. They are not the same 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": 12, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43656-2", + "nodes": [ + "pull_request:43656", + "pull_request:43842" + ], + "soft_pairs": [ + "pull_request:43656|pull_request:43842" + ], + "prepared_review_unit_hash": "2c82f11e768e35c0d3892f8c25617a3e2a8a9886deed3038f0fd8e6170c26dd6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3663, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 3663, + "estimated_input_tokens": 916, + "estimated_eval_tokens": 2088 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs modify `src/transformers/cli/serve.py` and mention `TypeAdapter`, but they address different typing/import failure modes rather than the same concrete fix.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43656 is the cleaner representative of the cluster: it directly targets the CLI `TypeAdapter` NameError, is explicitly linked to issue 43576, and appears to fix the runtime annotation-evaluation problem in one focused change.", + "best_issue_reason": null, + "best_pr_reason": "PR 43656 is the best overall representative because its title and issue link match the shared symptom most directly, while the alternate PR is a narrower pydantic-availability fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "Same file and similar symptom, but different root causes: 43656 uses `from __future__ import annotations` to avoid eager annotation evaluation, while 43842 guards the `TypeAdapter` import behind `TYPE_CHECKING` for optional pydantic. They are not the same concrete 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": 13, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43779-2", + "nodes": [ + "pull_request:43779", + "pull_request:43816" + ], + "soft_pairs": [ + "pull_request:43779|pull_request:43816" + ], + "prepared_review_unit_hash": "e7c674adcc031aa8883109a5dfc678b4407fb20b10764302092e384d031ba83a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4362, + "estimated_input_tokens": 1091, + "estimated_eval_tokens": 2438 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests appear to implement the same SwanLab integration change: adding run id/resume support in `integration_utils.py` for issue #43698, with one variant also expanding the docs. They are strong duplicate PR candidates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43816 is the cleaner representative: it carries the same SwanLab id/resume support plus the accompanying documentation updates, making it the more complete version of the shared change.", + "best_issue_reason": null, + "best_pr_reason": "PR #43816 is the best global representative because it appears to include the full feature change and docs for SwanLab run id/resume support, while #43779 is the narrower implementation-focused variant.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43779", + "right": "pull_request:43816", + "accept": true, + "reason": "Same underlying SwanLab callback enhancement for id/resume support in the same file and same issue target; the doc differences do not change the core code-path change, so these are duplicate PRs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both PRs target issue #43698, touch the same file, and the previews indicate the same SwanLab id/resume support with #43816 adding docs. The duplicate/mergeable verdict is conservative enough given the shared target and overlapping code change." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44066|pull_request:44071", + "pull_request:44018|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44066|pull_request:44068", + "pull_request:44019|pull_request:44722", + "pull_request: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:44018|pull_request:44116", + "pull_request:44068|pull_request:44722", + "pull_request:44071|pull_request:44722", + "pull_request:44018|pull_request:44085", + "pull_request:44018|pull_request:44086", + "pull_request:44068|pull_request:44086", + "pull_request:44071|pull_request:44085", + "pull_request:44085|pull_request:44086", + "pull_request:44059|pull_request:44086", + "pull_request:44098|pull_request:44101", + "pull_request:44059|pull_request:44722", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:43997|pull_request:44073", + "pull_request:44071|pull_request:44086", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44059|pull_request:44071", + "pull_request:44073|pull_request:44161", + "pull_request:44056|pull_request:44101", + "pull_request:44018|pull_request:44059", + "pull_request:44059|pull_request:44068", + "pull_request:44101|pull_request:44161", + "pull_request:43997|pull_request:44017", + "pull_request:43997|pull_request:44056", + "pull_request:44071|pull_request:44076", + "pull_request:44074|pull_request:44101", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44018", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:43997|pull_request:44071", + "pull_request:44010|pull_request:44071", + "pull_request:44018|pull_request:44129", + "pull_request:44010|pull_request:44073", + "pull_request:43997|pull_request:44074", + "pull_request:44071|pull_request:44129", + "pull_request:44010|pull_request:44018", + "pull_request:43997|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44017", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154" + ], + "prepared_review_unit_hash": "de0b38e89a81f997d39964a77ebf027ba47d03f1fb770afe7321c02b2c0afb74", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84892, + "estimated_input_tokens": 21223, + "estimated_eval_tokens": 42702 + }, + "budget": { + "node_count": 27, + "item_count": 27, + "soft_pair_count": 72, + "serialized_chars": 75088, + "estimated_input_tokens": 18772, + "estimated_eval_tokens": 37800 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a broad issue-driven refactor campaign for standardized output tracing across many different models. Most pairs are not true duplicates because they touch different model families and different code paths; the only clear duplicate sets are the GPT-J/CodeGen PRs and the GPT-Neo PRs.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44066 is the best canonical representative for the GPT-J/CodeGen duplicate set: it is explicitly issue-linked, touches the same concrete output-tracing code paths in both files, and is the clearest umbrella refactor among the overlapping GPT-J variants.", + "best_issue_reason": null, + "best_pr_reason": "PR 44066 is the strongest single representative in the cluster because it covers the core standardized output-tracing change with explicit linkage to the tracking issue and the broadest direct overlap with the other true duplicates.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families and code paths; same high-level refactor theme is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different model families and code paths; only the output-tracing theme is shared." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both refactor GPT-J output tracing in modeling_gptj.py toward the same standardized decorator-based interface." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Different model families and implementations; same issue target alone does not make them duplicates." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "EfficientNet vs GPT-J/CodeGen are different concrete code paths despite similar refactor wording." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen and EfficientNet are unrelated model implementations." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTa are unrelated model code paths." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are different model families with no shared bug fix." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "Different model families and different output-handling implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Both are GPT-Neo output-tracing refactors in the same model file and appear to implement the same decorator-based capture change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are different model families and code paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen and GPT-Neo are separate implementations, not the same duplicate change." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet and GPT-J/CodeGen are unrelated model code paths." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are unrelated model implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are unrelated code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are different models with different output-tracing logic." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different model families." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert/XLM are unrelated model code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert/XLM are different model families." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are different model refactors." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen are unrelated implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model families." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 and MGP-STR are different model families." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert/XLM are unrelated model code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-2 and GPT-J/CodeGen are different model implementations." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are unrelated model families." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44722", + "accept": false, + "reason": "SegFormer and GPT-J/CodeGen are different code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are unrelated model implementations." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "SegFormer and VisualBERT are different model families." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated models." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "RegNet and VisualBERT are unrelated code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "CVT/FNet and SegFormer are different model families." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBERT and MPT are unrelated implementations." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBERT and Flaubert/XLM are different model code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-2 and MPT are unrelated models." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "VisualBERT and LongT5 are different model families." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44059", + "accept": false, + "reason": "GPT-Neo and GPT-2 are different models." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-2 and GPT-Neo are different model implementations." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are unrelated code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "MPT and MGP-STR are different model families." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBERT and VisualBERT are unrelated models." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are different code paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated model implementations." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are different model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are unrelated code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are unrelated models." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet and LongT5 are different implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44161", + "accept": false, + "reason": "GPT-Neo and LongT5 are unrelated model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44071", + "accept": false, + "reason": "RegNet and MPT are unrelated code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBERT and DeBERTa are different model families." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa and TextNet are unrelated implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa and VITS are different model families." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor, touching the same concrete files and code path." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both affect the GPT-J output-tracing path in modeling_gptj.py and look like the same underlying change." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are unrelated model code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and RegNet/ResNet are different model families." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated output-tracing changes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa and SpeechT5 are different model implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa and VITS are unrelated code paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and the duplicate verdicts stay conservative. The accepted soft edges are supported by shared files/issue targets and clearly overlapping GPT-J/CodeGen or GPT-Neo refactors, while the cross-family pairs are correctly rejected rather than over-merged." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 32, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44024", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44024|pull_request:44076", + "pull_request:44073|pull_request:44074", + "pull_request:44056|pull_request:44073", + "pull_request:44056|pull_request:44074", + "pull_request:43997|pull_request:44010", + "pull_request:44010|pull_request:44129" + ], + "prepared_review_unit_hash": "37d1d30e42fcded831f906349f7eb0e9e33d7c223dc729c17a8b1822ded386d9", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 78, + "serialized_chars": 84892, + "estimated_input_tokens": 21223, + "estimated_eval_tokens": 42702 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 6, + "serialized_chars": 17125, + "estimated_input_tokens": 4282, + "estimated_eval_tokens": 8820 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a set of related but model-specific refactors to standardized output tracing/capture; they are not duplicates of one underlying bug or one PR.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single PR is a true canonical duplicate here; each PR targets a different model and code path, even though they share the same refactor theme.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:44073 is the clearest representative of the shared refactor pattern, with an explicit standardized output-tracing title and direct capture_outputs/can_return_tuple changes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "Different models and code paths (FocalNet vs ImageGPT); same refactor theme, but not the same concrete fix." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Both are output-tracing refactors, but VisualBert and TextNet are separate implementations with different model-specific changes." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet and VisualBert are distinct model modules; these are parallel migrations, not one mergeable PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Different models and differing output-capture plumbing; not the same underlying bug or change." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated concrete code paths despite similar standardized-output refactoring." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "SqueezeBert vs SpeechT5 are separate model-specific output tracing changes; the shared theme is too broad for duplication." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 15, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43998-13", + "nodes": [ + "pull_request:43998", + "pull_request:43999", + "pull_request:44000", + "pull_request:44001", + "pull_request:44002", + "pull_request:44003", + "pull_request:44004", + "pull_request:44025", + "pull_request:44026", + "pull_request:44027", + "pull_request:44028", + "pull_request:44029", + "pull_request:44030" + ], + "soft_pairs": [ + "pull_request:43998|pull_request:44002", + "pull_request:43998|pull_request:44000", + "pull_request:43998|pull_request:44001", + "pull_request:43998|pull_request:44003", + "pull_request:43998|pull_request:44029", + "pull_request:43999|pull_request:44001", + "pull_request:43999|pull_request:44002", + "pull_request:43999|pull_request:44025", + "pull_request:43999|pull_request:44026", + "pull_request:43999|pull_request:44027", + "pull_request:44000|pull_request:44001", + "pull_request:44000|pull_request:44002", + "pull_request:44000|pull_request:44003", + "pull_request:44000|pull_request:44025", + "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", + "pull_request:44001|pull_request:44025", + "pull_request:44002|pull_request:44029", + "pull_request:44002|pull_request:44003", + "pull_request:44003|pull_request:44004", + "pull_request:43998|pull_request:44004", + "pull_request:44001|pull_request:44004", + "pull_request:44002|pull_request:44004", + "pull_request:44000|pull_request:44004", + "pull_request:44025|pull_request:44026", + "pull_request:44025|pull_request:44027", + "pull_request:44025|pull_request:44028", + "pull_request:44025|pull_request:44029", + "pull_request:44025|pull_request:44030", + "pull_request:44026|pull_request:44029", + "pull_request:44026|pull_request:44027", + "pull_request:44026|pull_request:44028", + "pull_request:44026|pull_request:44030", + "pull_request:44027|pull_request:44028", + "pull_request:44027|pull_request:44029", + "pull_request:44027|pull_request:44030", + "pull_request:44028|pull_request:44029", + "pull_request:44028|pull_request:44030", + "pull_request:44029|pull_request:44030" + ], + "prepared_review_unit_hash": "114d3cb6aa07f0f46d9d765b0f4d361ace01601de3680adcc0204059de8493b4", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 39, + "serialized_chars": 39787, + "estimated_input_tokens": 9947, + "estimated_eval_tokens": 20150 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 39, + "serialized_chars": 39787, + "estimated_input_tokens": 9947, + "estimated_eval_tokens": 20150 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are mostly model-specific refactors for the same output-tracing pattern, but they are not duplicates of one another: each PR targets a different model or code path, so they should stay separate.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single PR is a true duplicate anchor here; the closest representative of the shared theme is PR 44025 because it is a substantive output-tracing refactor spanning two related DepthAnything files.", + "best_issue_reason": null, + "best_pr_reason": "PR 44025 is the best representative of the cluster theme because it is a broader, concrete output-tracing change across closely related files, but it still is not a duplicate of the others.", + "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 theme, but not the same code-path or change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch unrelated models (`timm_backbone` vs `vision_text_dual_encoder`)." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Separate model implementations (`timm_backbone` vs `univnet`); no shared concrete bug or mergeable single PR." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "`timm_backbone` and `mamba/falcon_mamba` are distinct code paths; similarity is only at the refactor label level." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Different architectures (`timm_backbone` vs `rwkv`); not the same underlying change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "`mobilenet_v1` and `univnet` are unrelated models; same tracing refactor pattern, but not duplicate work." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model code paths (`mobilenet_v1` vs `upernet`); too broad to be the same change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "`mobilenet_v1` and `depth_anything` are unrelated implementations, even though both refactor output handling." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "Not the same code path: `mobilenet_v1` vs `vision_encoder_decoder`." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "Different subsystems (`mobilenet_v1` vs `speech_encoder_decoder`); similar refactor, not duplicate." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "`vision_text_dual_encoder` and `univnet` are unrelated models; same broad theme only." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "Different model families and code paths (`vision_text_dual_encoder` vs `upernet`)." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "`vision_text_dual_encoder` and `mamba/falcon_mamba` do not share a concrete bug or change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different models (`vision_text_dual_encoder` vs `depth_anything`); too generic to merge as one duplicate PR." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "`univnet` and `upernet` are separate model code paths; only the refactor style matches." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Distinct architectures (`univnet` vs `mamba/falcon_mamba`); not the same underlying fix." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "`univnet` and `depth_anything/prompt_depth_anything` are unrelated model implementations." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "`upernet` and `rwkv` are different model families; same refactor theme is insufficient for duplication." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "Separate model code paths (`upernet` vs `mamba/falcon_mamba`) with no shared concrete bug." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "`mamba/falcon_mamba` and `codegen` are unrelated internals; not plausibly one PR." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "`timm_backbone` and `codegen` affect totally different model paths; only the output-tracing refactor label is shared." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "`univnet` and `codegen` are unrelated changes, despite both touching output handling." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "`upernet` and `codegen` do not share the same concrete bug or implementation path." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "`vision_text_dual_encoder` and `codegen` are different models and different fix scopes." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Both are encoder-decoder wrappers, but they target different modalities (`depth_anything` vs `vision_encoder_decoder`) and separate code paths." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "`depth_anything` and `speech_encoder_decoder` are unrelated models; not the same underlying change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "Different model families (`depth_anything` vs `superpoint`); same refactor pattern only." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "`depth_anything` and `rwkv` are unrelated code paths." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Different models (`depth_anything` vs `dpr`); not a duplicate change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "Both are encoder-decoder wrappers, but vision and speech variants are distinct code paths and should not be merged as duplicates." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "`vision_encoder_decoder` and `rwkv` are unrelated; only the refactor style is similar." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "`vision_encoder_decoder` and `superpoint` are different model implementations." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "`vision_encoder_decoder` and `dpr` are separate code paths with no shared concrete bug." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "`speech_encoder_decoder` and `superpoint` are unrelated models; not the same change." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "`speech_encoder_decoder` and `rwkv` are different architectures and fixes." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "`speech_encoder_decoder` and `dpr` do not share a concrete code-path problem." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "`superpoint` and `rwkv` are unrelated; the overlap is only in refactor wording." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "`superpoint` and `dpr` are distinct models and should remain separate." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "`rwkv` and `dpr` are different model code paths; not a duplicate PR pair." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 16, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44662-2", + "nodes": [ + "pull_request:44662", + "pull_request:45401" + ], + "soft_pairs": [ + "pull_request:44662|pull_request:45401" + ], + "prepared_review_unit_hash": "9a678589e6d0f910bc0f690ea5baac79d8e70fcfcf6db24c3a233e5c0d44d455", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one PR adds PenguinVL, a vision-language model, while the other adds Voxtral TTS. The overlap is limited to shared Transformers plumbing and docs scaffolding.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: they implement different model families and different code paths, so they should stay separate.", + "best_issue_reason": null, + "best_pr_reason": "45401 is the cleaner representative of the cluster because it is more narrowly scoped, has an explicit linked issue, and focuses on one concrete TTS integration.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Reject: different underlying features and model code paths. PenguinVL is a multimodal vision-language model; Voxtral TTS is a text-to-speech model. Shared auto-registration and docs changes are not enough to indicate 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": 17, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-4", + "nodes": [ + "pull_request:44979", + "pull_request:45363", + "pull_request:45666", + "pull_request:46339" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363", + "pull_request:45363|pull_request:46339", + "pull_request:45666|pull_request:46339" + ], + "prepared_review_unit_hash": "d09e75066c01b900dbff85c338ad201f1035a4ee81515d0d95156208209b1334", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 9635, + "estimated_input_tokens": 2409, + "estimated_eval_tokens": 5074 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 9635, + "estimated_input_tokens": 2409, + "estimated_eval_tokens": 5074 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is centered on the n-to-1 kernel fusion / KernelConfig feature line. The later PRs (45363, 45666, 46339) look like iterations of the same concrete integration work, while 44979 is a separate lower-level module fusion API proposal.", + "confidence": 0.73, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46339 is the best representative because it appears to be the latest, most consolidated iteration of the same KernelConfig-based n-to-1 fusion work, with shared core-loading, hub-kernel, modeling, and test changes.", + "best_issue_reason": null, + "best_pr_reason": "PR 46339 most completely captures the feature family in this cluster and overlaps the most with the later KernelConfig fusion implementation work.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Related theme, but not the same concrete change: 44979 introduces a generic Module Fusion API, while 45363 implements KernelConfig-based kernel fusion integration and example plumbing." + }, + { + "left": "pull_request:45363", + "right": "pull_request:46339", + "accept": true, + "reason": "Both are the same n-to-1 KernelConfig kernel-fusion feature line, with overlapping integration files and very similar goal; 46339 looks like a refinement/simplification of the same change." + }, + { + "left": "pull_request:45666", + "right": "pull_request:46339", + "accept": true, + "reason": "These target the same concrete KernelConfig-based n-to-1 fusion path, share core files, and 46339 appears to be a later cleanup/extension of the same work." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary and soft-edge calls are grounded in the packet: 44979 is a separate Module Fusion API, while 45363/45666/46339 are clearly KernelConfig n-to-1 fusion iterations with overlapping filenames and consistent titles. The accept/false and accept/true pair verdicts look conservative rather than overstated." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45086-2", + "nodes": [ + "pull_request:45086", + "pull_request:45317" + ], + "soft_pairs": [ + "pull_request:45086|pull_request:45317" + ], + "prepared_review_unit_hash": "a431808eea965ba87d67b6f386afb5df32cf361647cf3fa45555d98e06bb97e2", + "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 are near-identical fixes for the same Mistral regex pre-tokenizer AttributeError in `src/transformers/tokenization_utils_tokenizers.py`, and both explicitly target issue #45081. They should be clustered together, with PR #45317 as the stronger canonical PR because it adds tests and has the clearer, more complete change set.", + "confidence": 0.97, + "canonical_issue_reason": "Issue #45081 is the shared tracker for both PRs; no separate issue artifact is present in the cluster.", + "canonical_pr_reason": "PR #45317 is the better canonical PR because it implements the same fix as #45086 and includes additional test coverage and a clearer title.", + "best_issue_reason": "If anchoring to an issue, #45081 is the only and shared issue target for the cluster.", + "best_pr_reason": "PR #45317 is the best representative PR: same underlying fix as #45086, plus tests and more explicit intent.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same code path, same file, same AttributeError fix, and both target the same issue; the diffs are effectively 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": 19, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-2", + "nodes": [ + "pull_request:45115", + "pull_request:45569" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569" + ], + "prepared_review_unit_hash": "53558b51ea2a824b1172f1829c2e248291ecd820fc36eb73cef1bd4c15e1d90b", + "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": "Adjacent NemotronH work, but the PRs are not duplicates: one is a refactor to align NemotronH with GraniteMoeHybrid and conversion mapping, while the other splits NemotronH into dedicated dense/sparse model variants with docs/tests.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: these are related NemotronH changes, but they target different concrete code paths and different deliverables.", + "best_issue_reason": null, + "best_pr_reason": "45569 is the more complete representative of the NemotronH family rework because it adds the dedicated dense/sparse implementations plus docs/tests, whereas 45115 is a narrower refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Different changes: 45115 refactors NemotronH inheritance/checkpoint conversion toward GraniteMoeHybrid; 45569 introduces separate NemotronHDense/NemotronHSparse packages and documentation. Shared files are 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": 20, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45332-2", + "nodes": [ + "pull_request:45332", + "pull_request:45333" + ], + "soft_pairs": [ + "pull_request:45332|pull_request:45333" + ], + "prepared_review_unit_hash": "b944dd7836e66b4cb8f80dc6055a178b8849b51ffcaf5b507f5dc22f160517ab", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5273, + "estimated_input_tokens": 1319, + "estimated_eval_tokens": 2894 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5273, + "estimated_input_tokens": 1319, + "estimated_eval_tokens": 2894 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are related but not duplicates: #45333 adds heterogeneous per-layer configuration plumbing, while #45332 builds broader heterogeneous model/runtime support on top of that, including cache/modeling changes. They share groundwork files, but they are different changesets and should not be merged as the same PR.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45332 is the broader, more central change: it covers heterogeneous model support end-to-end and incorporates the runtime/modeling pieces that would likely depend on the config groundwork.", + "best_issue_reason": null, + "best_pr_reason": "PR #45332 is the better representative for the cluster because it is the larger feature umbrella and touches the main code paths beyond configuration plumbing.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Related foundation and follow-on work, but not the same underlying change. #45333 is config-only per-layer support; #45332 extends into modeling/cache behavior and thus fixes a different, broader 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": 21, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45415-2", + "nodes": [ + "pull_request:45415", + "pull_request:45425" + ], + "soft_pairs": [ + "pull_request:45415|pull_request:45425" + ], + "prepared_review_unit_hash": "565c33728705fb6e261cd56348904739c5649317fefc521cdc497352bb76ac5a", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5514, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are typing-related, but they address different scopes: #45415 is a broad repo-wide type-checking sweep with config changes, while #45425 adds specific typing protocols/helpers for a narrower set of runtime code paths. They are not the same underlying change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45425 is the more concrete and self-contained typing change; #45415 is a much broader sweep across many files plus a pyproject typing-ignore update.", + "best_issue_reason": null, + "best_pr_reason": "#45425 is the better representative of the cluster because it targets specific typing gaps in implementation code, whereas #45415 is a general typing cleanup pass.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Different changes: #45415 is a wide type-checking pass and config adjustment, while #45425 adds specific typing protocols/helpers for modeling/PEFT-related code. Shared filenames alone do not indicate the same fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45513-2", + "nodes": [ + "pull_request:45513", + "pull_request:46084" + ], + "soft_pairs": [ + "pull_request:45513|pull_request:46084" + ], + "prepared_review_unit_hash": "6bc690e0ae5e6ed3c6cd11f5e2c58a583306787a5170d1ae8d21e6ccac59d282", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4858, + "estimated_input_tokens": 1215, + "estimated_eval_tokens": 2686 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4858, + "estimated_input_tokens": 1215, + "estimated_eval_tokens": 2686 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests are both cache-related, but they fix different model families and different code paths: Qwen3.5/OLMo hybrid linear-attention cached multi-token forward versus Mamba2 use_cache handling for seq_len > 1. They are not duplicates of the same underlying change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46084 is the stronger canonical PR: it has an explicit issue target (#46032), a narrowly defined Mamba2 cache bug, and a small focused diff that clearly addresses one concrete failure mode.", + "best_issue_reason": null, + "best_pr_reason": "PR #46084 is the best representative of the cluster because it is the most precisely scoped and best-documented fix; PR #45513 is a broader multi-model cache correction without an explicit target issue.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45513", + "right": "pull_request:46084", + "accept": false, + "reason": "Both involve cached forward passes, but they affect different architectures and different implementation paths. #45513 fixes multi-token cached forward in Qwen3.5/OLMo hybrid linear attention, while #46084 fixes Mamba2 use_cache when seq_len > 1. They are not the same bug 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": 23, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45548-2", + "nodes": [ + "pull_request:45548", + "pull_request:45662" + ], + "soft_pairs": [ + "pull_request:45548|pull_request:45662" + ], + "prepared_review_unit_hash": "503a6dacf26904bda8c58efabbe15bbbaf79ad8e50ad075153a8b473c47bd35b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs both touch expert-parallel loading paths, but they fix different bugs: #45548 is about making distributed_config survive serialization and exposing has_ep for EP + ZeRO-3 loading via accelerate, while #45662 fixes expert parameters being overwritten by rank-0 broadcast under EP + FSDP2. They are related but not the same change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR for a duplicate cluster: the PRs target different concrete failure modes and should stay separate; if forced, #45662 is the more specific bug fix.", + "best_issue_reason": null, + "best_pr_reason": "#45662 is the stronger representative because it fixes a concrete overwrite bug in the loading path with explicit code changes across the affected flow, whereas #45548 is a different serialization/loading compatibility fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Both concern expert-parallel loading, but the bugs and code paths differ: one preserves distributed_config / has_ep for ZeRO-3 loading, the other prevents rank-0 broadcast from overwriting experts in FSDP2." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45639-2", + "nodes": [ + "pull_request:45639", + "pull_request:45645" + ], + "soft_pairs": [ + "pull_request:45639|pull_request:45645" + ], + "prepared_review_unit_hash": "10f06d274aa97d82e877f3f8e8d87b08991763ba490b4024543e2a669b32ae6d", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4589, + "estimated_input_tokens": 1148, + "estimated_eval_tokens": 2552 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same xdist collision around `captured_info` debug artifacts in testing workflows. PR 45645 is the broader, more complete fix; PR 45639 is a narrower subset focused on the `testing_utils` path handling.", + "confidence": 0.83, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the better canonical PR because it covers the core `testing_utils` xdist-safe change plus the CI/workflow handling needed to preserve and aggregate the multiple `captured_info*.txt` outputs.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the best overall PR since it appears to fully address the collision problem end-to-end, while PR 45639 only fixes the artifact write path in `testing_utils`.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Same underlying bug: xdist workers colliding on `captured_info` debug output. PR 45645 is a superset that includes the same core fix and related CI log handling, so they plausibly belong to one merged 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": 25, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45670-2", + "nodes": [ + "pull_request:45670", + "pull_request:46320" + ], + "soft_pairs": [ + "pull_request:45670|pull_request:46320" + ], + "prepared_review_unit_hash": "50ab93664d2dbec8de4a421b401a3fc39388b3032421f732db5c22ec44d1d146", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4431, + "estimated_input_tokens": 1108, + "estimated_eval_tokens": 2472 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4431, + "estimated_input_tokens": 1108, + "estimated_eval_tokens": 2472 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated: one adds a model class registration entry for glmasr in AutoModel, while the other implements RecurrentGemma attn_implementation dispatch and related model/conversion/test updates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46320 is the more substantive, self-contained change and a better representative of this cluster; PR #45670 is just a narrow AutoModel mapping tweak.", + "best_issue_reason": null, + "best_pr_reason": "PR #46320 is the best PR candidate because it fixes a concrete code path in the model implementation and includes supporting conversion and test updates, unlike the small registry-only tweak in #45670.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45670", + "right": "pull_request:46320", + "accept": false, + "reason": "Different models and different fixes: glmasr AutoModel registration versus RecurrentGemma attention-dispatch support. They do not look like the same underlying change 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": 26, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45720-2", + "nodes": [ + "pull_request:45720", + "pull_request:46259" + ], + "soft_pairs": [ + "pull_request:45720|pull_request:46259" + ], + "prepared_review_unit_hash": "2b8a419a9051fc973d99f8d1da463126d8433e8ebe92da7b069513a3c6dabc91", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7189, + "estimated_input_tokens": 1798, + "estimated_eval_tokens": 3852 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7189, + "estimated_input_tokens": 1798, + "estimated_eval_tokens": 3852 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs are both linter/modeling-rule work, but they target different problems: TRF003 return-tuple handling in model forwards vs TRF018 dead do_* flag consumption in image/video processors. The shared model filenames are incidental overlap, not the same change.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR 46259 is the clearer representative because it introduces a new rule with accompanying documentation and a focused set of model fixes; however, it is not a duplicate of PR 45720.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45720", + "right": "pull_request:46259", + "accept": false, + "reason": "Different underlying bugs/rules: 45720 implements TRF003 for can_return_tuple/return_dict behavior in model forwards, while 46259 adds TRF018 for unused do_* flags in image/video processor overrides. Overlapping files do not make the code-path problem the same." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 27, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45737-2", + "nodes": [ + "pull_request:45737", + "pull_request:45738" + ], + "soft_pairs": [ + "pull_request:45737|pull_request:45738" + ], + "prepared_review_unit_hash": "cc2ade73a49ec6a8883a7a24c433988645588bcd049e8aad2bce56b1b24daa1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same MusicgenMelody cache initialization bug in `forward()`: replacing `EncoderDecoderCache(DynamicCache, DynamicCache)` with a single `DynamicCache`. PR #45737 is the more complete duplicate because it bundles the fix with a regression test, while #45738 is the fix-only variant.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45737 is the best canonical PR because it contains the same code fix as #45738 plus a regression test, making it the most complete representation of the bug fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45737 is the strongest overall candidate: it fixes the concrete cache bug and adds a targeted regression test, whereas #45738 only changes the implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45737", + "right": "pull_request:45738", + "accept": true, + "reason": "Same underlying code-path bug in `musicgen_melody` cache setup; the diffs match on the core fix and the test-only addition in #45737 could plausibly live in the same PR as #45738." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: both PRs make the same `musicgen_melody` cache initialization change, and #45737 only adds a regression test on top of that fix. The soft-edge accept is conservative enough because the diffs are mergeable into one PR for the same bugfix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 28, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45932-2", + "nodes": [ + "pull_request:45932", + "pull_request:46100" + ], + "soft_pairs": [ + "pull_request:45932|pull_request:46100" + ], + "prepared_review_unit_hash": "0ca5063a745ce6545de7620e7b4b9b99f50b9199e3c6e35b3d7979d18f511877", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4300, + "estimated_input_tokens": 1075, + "estimated_eval_tokens": 2406 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4300, + "estimated_input_tokens": 1075, + "estimated_eval_tokens": 2406 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are unrelated: one is a broad model/config/workflow change with many files touched, while the other is a tiny ViT test-only PR adding intentionally failing assertions. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45932 is the stronger canonical PR candidate because it contains the substantive code/config/workflow changes across multiple model files, whereas PR 46100 is only a test-only placeholder/migration change and does not represent the same underlying change.", + "best_issue_reason": null, + "best_pr_reason": "PR 45932 is the best representative of the cluster only in the sense that it is the more substantive artifact; PR 46100 is not suitable as a representative because it is a narrow test file edit with no overlap in bug/change intent.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45932", + "right": "pull_request:46100", + "accept": false, + "reason": "Different intent and code paths: 45932 touches multiple model implementations plus workflow/config files, while 46100 only edits a ViT test file with dummy failing assertions. They do not appear to fix the same underlying problem and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 29, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46073-3", + "nodes": [ + "pull_request:46073", + "pull_request:46154", + "pull_request:46196" + ], + "soft_pairs": [ + "pull_request:46073|pull_request:46154", + "pull_request:46073|pull_request:46196", + "pull_request:46154|pull_request:46196" + ], + "prepared_review_unit_hash": "e6e08d8e0c23b8757b62fc8391a1d211944fec65b3c0fdc5481da2dd031d5826", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 8082, + "estimated_input_tokens": 2021, + "estimated_eval_tokens": 4298 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 8082, + "estimated_input_tokens": 2021, + "estimated_eval_tokens": 4298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related only at a high level: all add SDPA dispatch support, but each PR targets different model families and code paths, so they are not duplicates and should not be merged together.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46196 is the best representative of the cluster because it covers the broadest set of model implementations and tests, making it the most general SDPA-dispatch change here.", + "best_issue_reason": null, + "best_pr_reason": "PR 46196 is the strongest standalone PR in this group due to its broader model coverage and more complete implementation surface compared with the single-model PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46073", + "right": "pull_request:46154", + "accept": false, + "reason": "Both are SDPA-dispatch PRs, but they fix different model-specific implementations (CTRL vs T5) and are not the same concrete code-path change." + }, + { + "left": "pull_request:46073", + "right": "pull_request:46196", + "accept": false, + "reason": "They share the same theme, but target different architectures and files; this is not one mergeable bugfix/change." + }, + { + "left": "pull_request:46154", + "right": "pull_request:46196", + "accept": false, + "reason": "T5-only SDPA dispatch and the wav2vec2/SeamlessM4T SDPA dispatch work are separate model-family changes, 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": 30, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46126-2", + "nodes": [ + "pull_request:46126", + "pull_request:46269" + ], + "soft_pairs": [ + "pull_request:46126|pull_request:46269" + ], + "prepared_review_unit_hash": "af4a40adc6c6fcac91eb0a09264ab175f03c5be38d3bdf2cb94d71f8f4404d37", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5571, + "estimated_input_tokens": 1393, + "estimated_eval_tokens": 3042 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5571, + "estimated_input_tokens": 1393, + "estimated_eval_tokens": 3042 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one PR is a focused EP/Trainer integration on top of DistributedConfig, while the other is a broad distributed-branch base that bundles many unrelated distributed changes (FSDP, tensor parallel, integrations, MoE, CI). The shared filenames reflect subsystem overlap, not the same concrete change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46126 is the more focused, concrete change: EP router masking plus Trainer/config integration on a specific distributed code path.", + "best_issue_reason": null, + "best_pr_reason": "PR #46126 is the best representative PR for this cluster because it is narrowly scoped and clearly describes the shared EP/distributed functionality, whereas #46269 is an umbrella branch base with many unrelated edits.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "Different underlying changes. #46126 targets EP router behavior and Trainer integration; #46269 is a broad distributed branch base touching many disparate areas, so they should not be merged 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": 31, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46152-2", + "nodes": [ + "pull_request:46152", + "pull_request:46207" + ], + "soft_pairs": [ + "pull_request:46152|pull_request:46207" + ], + "prepared_review_unit_hash": "d9addb867b6e4593c984b2011eddefbcd7dcb8b594b644f4250b94e5e9c8de6e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4276, + "estimated_input_tokens": 1069, + "estimated_eval_tokens": 2394 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4276, + "estimated_input_tokens": 1069, + "estimated_eval_tokens": 2394 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs replace file-open-based source access with safer source introspection, but they fix different code paths: one in import utilities, the other in attention-implementation detection. They are related in theme, not the same underlying change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46207 is the cleaner representative because it has a well-scoped, concrete fix in a single behavioral path (`_can_set_attn_implementation`) and includes tests, making it the more specific duplicate-triage anchor.", + "best_issue_reason": null, + "best_pr_reason": "PR 46207 is the best overall representative since it addresses a clearly defined bug pattern with an explicit code path and validation, whereas PR 46152 is a separate utility-level refactor/fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46152", + "right": "pull_request:46207", + "accept": false, + "reason": "Both avoid reading source via `open()`, but they target different functions and different failure modes; they are not the same concrete bug or mergeable as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 32, + "component_count": 32, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46337-4", + "nodes": [ + "pull_request:46337", + "pull_request:46362", + "pull_request:46366", + "pull_request:46368" + ], + "soft_pairs": [ + "pull_request:46337|pull_request:46368", + "pull_request:46337|pull_request:46362", + "pull_request:46362|pull_request:46368", + "pull_request:46362|pull_request:46366" + ], + "prepared_review_unit_hash": "cdb7b6c2670d0fad6ff6807c3516ee02d1961c8fecc557a6328f73cf03100406", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 8225, + "estimated_input_tokens": 2057, + "estimated_eval_tokens": 4370 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 4, + "serialized_chars": 8225, + "estimated_input_tokens": 2057, + "estimated_eval_tokens": 4370 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is mostly a set of small, model-specific test adjustments. Two PRs touch DeepSeek v4, but one skips quant-cache tests while the other extends a distributed test to XPU/CUDA; they do not fix the same concrete bug. The other PRs are unrelated model-specific skips in different files.", + "confidence": 0.83, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR stands out: the DeepSeek v4 PRs address different test failures, and the Cohere/Nemotron PRs are separate model-specific skips with no shared underlying change.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:46362 is the clearest standalone change: it directly skips a single invalid DeepSeek v4 quant-cache test with a specific incompatibility explanation. But it is not a good cluster representative because the rest of the items are about different models or different failure modes.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46337", + "right": "pull_request:46368", + "accept": false, + "reason": "Both are model-test skips, but for different architectures and different cache behavior; they are not the same underlying bug or fix." + }, + { + "left": "pull_request:46337", + "right": "pull_request:46362", + "accept": false, + "reason": "Cohere2 MoE offload/parallelism test adjustments are unrelated to DeepSeek v4 quant-cache incompatibility." + }, + { + "left": "pull_request:46362", + "right": "pull_request:46368", + "accept": false, + "reason": "Both skip quant-cache tests, but the incompatibilities are model-specific (DeepSeek v4 vs NemotronH) and not mergeable as one PR." + }, + { + "left": "pull_request:46362", + "right": "pull_request:46366", + "accept": false, + "reason": "DeepSeek v4 quant-cache skipping is unrelated to DeepSeek v4 distributed-worker backend selection for CUDA/XPU." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/manifest.json b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..46cc92b3d573bbb65f7b6b9868f63031919b05a1 --- /dev/null +++ b/snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260604t000026z", + "artifacts": { + "hybrid": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-06-04T00:08:55Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260604T000026Z", + "variant": "hybrid" +} diff --git a/snapshots/20260604T000026Z/manifest.json b/snapshots/20260604T000026Z/manifest.json index be2814df3fd229cd4d7c7663a3bdde62aa6232a4..4b4eb13234b484e7c52b302fc103e87449194565 100644 --- a/snapshots/20260604T000026Z/manifest.json +++ b/snapshots/20260604T000026Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 394 }, "extracted_at": "2026-06-04T00:00:26Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260604t000026z", + "runs": { + "hybrid-model-20260604t000026z": { + "analysis_id": "hybrid-model-20260604t000026z", + "artifacts": { + "hybrid": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260604T000026Z/analysis-runs/hybrid-model-20260604t000026z/manifest.json", + "model": null, + "published_at": "2026-06-04T00:08:55Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260604T000026Z", "watermark": {