diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 221a3f50ff277e2215fd10ef662606cbea6569f5..e29ed71f2ca652936cb6883bcc2c1c5ececab5bc 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": "20260623T060018Z", - "generated_at": "2026-06-23T06:11:43Z", + "snapshot_id": "20260623T120018Z", + "generated_at": "2026-06-23T12:12:33Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1326,7 +1326,7 @@ "cluster_id": "cluster-43240-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", "status": "open", - "confidence": 0.75, + "confidence": 0.55, "canonical_issue_number": 43240, "canonical_pr_number": 43251, "issue_numbers": [ @@ -1337,8 +1337,7 @@ 43254 ], "evidence_types": [ - "closing_reference", - "shared_issue_target" + "closing_reference" ], "pr_comparisons": [ { @@ -2019,69 +2018,198 @@ ] }, { - "cluster_id": "cluster-46562-2", + "cluster_id": "cluster-45332-2", "summary": "Cluster of 2 related pull requests linked by soft_similarity.", "status": "open", "confidence": 0.5, "canonical_issue_number": null, - "canonical_pr_number": 46562, + "canonical_pr_number": 45333, "issue_numbers": [], "pr_numbers": [ - 46562, - 46658 + 45332, + 45333 ], "evidence_types": [ "soft_similarity" ], "pr_comparisons": [ { - "left_pr_number": 46562, - "right_pr_number": 46658, - "code_similarity": 0.153, - "size_similarity": 0.387, - "file_overlap": 0.024, - "area_overlap": 0.0, - "patch_similarity": 0.457, + "left_pr_number": 45332, + "right_pr_number": 45333, + "code_similarity": 0.61, + "size_similarity": 0.344, + "file_overlap": 0.389, + "area_overlap": 0.988, + "patch_similarity": 0.525, "shared_filenames": [ - "src/transformers/models/auto/auto_mappings.py" + "docs/source/en/_toctree.yml", + "docs/source/en/heterogeneous_configurations.md", + "src/transformers/configuration_utils.py", + "src/transformers/integrations/heterogeneity/__init__.py", + "src/transformers/integrations/heterogeneity/configuration_utils.py", + "tests/heterogeneity/__init__.py", + "tests/heterogeneity/test_configuration_utils.py" ], "shared_file_areas": [ { - "filename": "src/transformers/models/auto/auto_mappings.py", + "filename": "docs/source/en/_toctree.yml", + "left_ranges": [ + [ + 16, + 23 + ] + ], + "right_ranges": [ + [ + 16, + 23 + ] + ] + }, + { + "filename": "docs/source/en/heterogeneous_configurations.md", + "left_ranges": [ + [ + 1, + 206 + ] + ], + "right_ranges": [ + [ + 1, + 206 + ] + ] + }, + { + "filename": "src/transformers/configuration_utils.py", "left_ranges": [ [ - 441, - 447 + 30, + 36 + ], + [ + 123, + 129 ], [ - 591, - 598 + 188, + 195 ], [ - 823, - 829 + 302, + 311 ], [ - 883, - 889 + 316, + 324 ], [ - 1082, - 1088 + 443, + 456 ], [ - 1205, - 1211 + 859, + 867 + ], + [ + 965, + 973 + ], + [ + 1026, + 1034 + ], + [ + 1076, + 1084 ] ], "right_ranges": [ [ - 362, - 368 + 30, + 36 + ], + [ + 123, + 129 + ], + [ + 188, + 195 + ], + [ + 302, + 311 + ], + [ + 316, + 324 + ], + [ + 443, + 456 + ], + [ + 859, + 867 + ], + [ + 965, + 973 + ], + [ + 1026, + 1034 ], [ - 795, - 801 + 1076, + 1084 + ] + ] + }, + { + "filename": "src/transformers/integrations/heterogeneity/__init__.py", + "left_ranges": [ + [ + 1, + 34 + ] + ], + "right_ranges": [ + [ + 1, + 21 + ] + ] + }, + { + "filename": "src/transformers/integrations/heterogeneity/configuration_utils.py", + "left_ranges": [ + [ + 1, + 389 + ] + ], + "right_ranges": [ + [ + 1, + 389 + ] + ] + }, + { + "filename": "tests/heterogeneity/test_configuration_utils.py", + "left_ranges": [ + [ + 1, + 323 + ] + ], + "right_ranges": [ + [ + 1, + 323 ] ] } @@ -2392,7 +2520,7 @@ 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." + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference evidence." }, { "cluster_id": "cluster-43698-3", @@ -2438,6 +2566,15 @@ "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-45332-2", + "canonical_pr_number": 45333, + "duplicate_pr_numbers": [ + 45332 + ], + "target_issue_number": null, + "reason": "PRs in cluster-45332-2 are treated as duplicates because they share soft_similarity evidence." + }, { "cluster_id": "cluster-45561-3", "canonical_pr_number": 45645, @@ -2473,15 +2610,6 @@ ], "target_issue_number": 46305, "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." - }, - { - "cluster_id": "cluster-46562-2", - "canonical_pr_number": 46562, - "duplicate_pr_numbers": [ - 46658 - ], - "target_issue_number": null, - "reason": "PRs in cluster-46562-2 are treated as duplicates because they share soft_similarity evidence." } ], "best_issue": { diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 412fd778ca5fab52069a2ba245af15901f21c4e8..670a6031d54db93624e29c7999483f43aec21c62 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": "20260623T060018Z", - "generated_at": "2026-06-23T06:11:43Z", + "snapshot_id": "20260623T120018Z", + "generated_at": "2026-06-23T12:12:33Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -16,7 +16,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:24643", "issue:30277", @@ -62,7 +62,6 @@ "issue:46015", "issue:46165", "issue:46294", - "issue:46304", "issue:46468", "issue:46515", "issue:46661" @@ -129,322 +128,339 @@ "issue:30277|issue:36331", "issue:34928|issue:36010", "issue:45237|issue:46015", - "issue:43039|issue:44805", - "issue:45800|issue:46304" + "issue:43039|issue:44805" ], - "prepared_review_unit_hash": "aea7d9545f14d1185ba67c0a923ba9718f712840bbd4270062ab4bccfdadd2ab", + "prepared_review_unit_hash": "7cbe1529059cde0fc240a6be98d67feeed44e5d7939bb7ed8bd66c735ab64e48", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 63, - "serialized_chars": 46598, - "estimated_input_tokens": 11650, - "estimated_eval_tokens": 23556 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 62, + "serialized_chars": 45724, + "estimated_input_tokens": 11431, + "estimated_eval_tokens": 23118 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, with only two high-confidence duplicate pairs: the BERT + DeepSpeed ZeRO-3 init index-error reports, and the Torch 2.4.1 compatibility reports. Most other pairs share only broad subsystem overlap or similar error shapes, not the same underlying bug.", + "summary": "This cluster is mostly noisy and cross-model/topic. I accept only one true duplicate family: the DeepSpeed ZeRO-3/Bert index-0 out-of-bounds reports. The Qwen3_5Moe rotary-embedding pair is related but distinct (CPU-offload incompatibility vs uninitialized memory after meta\u2192to_empty(cuda)), so it should not be merged.", "confidence": 0.74, - "canonical_issue_reason": "issue:43596 is the clearest concrete bug report and has a near-identical companion in 43638; both describe the same DeepSpeed ZeRO-3/BERT init path failing with an index-0-out-of-bounds error.", + "canonical_issue_reason": "issue:43596 is the clearest representative of the only confirmed duplicate family here: the ZeRO-3/Bert size-0/index-0 out-of-bounds failure shared with issue:43638.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best representative of the strongest duplicate group in this cluster; it is specific, reproducible, and clearly maps to a single underlying bug.", + "best_issue_reason": "issue:43596 is the best anchor for triage because it captures the core ZeRO-3/Bert failure mode and has the strongest true duplicate match in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different failures: missing config attribute in Qwen3.5 vs chat/completions streaming inference error." + "reason": "Different model/config bug vs streaming chat/completions failure; no shared code path." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Both involve backend/GPU issues, but the concrete bugs differ: AMD GPU failure for GPT-OSS-20B vs MPS histogram op not implemented for Int." + "reason": "Different backend/hardware failures: AMD GPU problem vs MPS histogram-on-Int error." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different model families and failure modes: Mistral-3 image-preprocessor dtype mismatch vs GPT-OSS running on AMD GPUs." + "reason": "Unrelated symptoms: Mistral-3 image dtype mismatch vs GPT-OSS failing on AMD GPUs." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Both are dtype/backend problems, but one is an autocast mismatch in Phi-tiny-MoE and the other is a GPT-OSS AMD runtime failure." + "reason": "Both are runtime dtype/backend issues, but on different models and paths." }, { "left": "issue:45237", "right": "issue:46468", "accept": false, - "reason": "Different issues: GPT-OSS-20B AMD GPU failure vs a broken Nemotron example." + "reason": "Generic GPU failure vs broken official example; not the same bug." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Not the same bug; GPT-OSS-20B on AMD GPUs and GraniteMoEHybrid calling an invalid method are unrelated code paths." + "reason": "Different models and failures: AMD GPU incompatibility vs invalid method call in GraniteMoEHybrid." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "DeepSpeed ZeRO-3 collective mismatch is unrelated to ModernBERT ONNX export errors." + "reason": "DeepSpeed collective mismatch is unrelated to ModernBERT ONNX export." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Both are model/export/runtime problems, but they involve different models and unrelated failures." + "reason": "ModernBERT ONNX export and Qwen3Moe FP8 failure are unrelated bugs." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "ModernBERT ONNX export and Qwen3 auto device-mapping cuda asserts are unrelated." + "reason": "Export-time ONNX issue vs Qwen3 auto device-mapping CUDA assert; different failures." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "Different topics: table-question-answering crash vs outdated RAG examples." + "reason": "Table QA crash and outdated RAG docs are unrelated." }, { "left": "issue:44509", "right": "issue:44512", "accept": false, - "reason": "Both are docs regressions in v5, but they refer to different removed documentation topics and do not look like the same concrete issue." + "reason": "Both are docs cleanup items, but they refer to different removed references." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "ONNX export error and GenerationMixin import error are distinct failures." + "reason": "ModernBERT ONNX export vs GenerationMixin import breakage are unrelated." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "TRl SFT embedding unpacking bug is unrelated to resuming training with local models and trust_remote_code." + "reason": "TRl SFT embedding unpacking failure vs resume-training with remote code are different code paths." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Mask-shape index error and Gemma4 PLE device mismatch are different forward-pass bugs." + "reason": "Mask-shape IndexError and Gemma4 device-map mismatch are not the same bug." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Both mention shape/indexing errors, but one is DeepSpeed ZeRO-3 BERT init and the other is a mask/tensor indexing mismatch." + "reason": "Both mention index/shape errors, but the failure modes and contexts differ too much." }, { "left": "issue:43638", "right": "issue:46165", "accept": false, - "reason": "No clear overlap between BERT ZeRO-3 init failure and resume-training/trust_remote_code behavior." + "reason": "ZeRO-3 Bert loading error vs remote-code resume issue are unrelated." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Different failures and code paths: ZeRO-3 BERT init vs a generic mask/tensor shape mismatch." + "reason": "ZeRO-3/Bert size-0 IndexError is different from a mask/tensor shape mismatch." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Mask-shape mismatch is unrelated to Qwen3.5 video token/feature count mismatch." + "reason": "Mask shape mismatch is unrelated to Qwen3.5 video token/feature mismatch." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "These are different model/runtime issues; only the error category is superficially similar." + "reason": "Different problems: generic mask index error vs GPT-OSS TP-only failure." }, { "left": "issue:46165", "right": "issue:46468", "accept": false, - "reason": "Resume-training with remote code is unrelated to the Nemotron example being broken." + "reason": "Resume-training with remote code is unrelated to a broken official Nemotron example." }, { "left": "issue:44560", "right": "issue:46165", "accept": false, - "reason": "Qwen3-vl video StopIteration and training resume failure are unrelated." + "reason": "Video StopIteration issue and remote-code training resume issue are unrelated." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "DeepSpeed ZeRO-3 BERT init error is unrelated to local-model resume training with trust_remote_code." + "reason": "ZeRO-3/Bert loading error is unrelated to remote-code resume training." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are unrelated bugs." + "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are unrelated." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "DeepSpeed collective mismatch in Jamba-v01 is not the same as the BERT ZeRO-3 index error." + "reason": "Different failures and models: Jamba collective mismatch vs Bert ZeRO-3 index error." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same underlying bug: DeepSpeed ZeRO-3 initialization with BERT models hits the same index-0-out-of-bounds failure." + "reason": "Same underlying ZeRO-3/Bert index-0 out-of-bounds failure; the titles differ mainly by model specificity." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Liger Kernel cross-entropy being called and GPT-OSS TP failure are different issues." + "reason": "Liger Kernel cross-entropy call path is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Different symptoms and code paths: detokenization regression vs LoRA merge/save/load mismatch." + "reason": "DeepSeek detokenization and LoRA merge/save mismatch are different bugs." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime error and LoRA merge/reload inconsistency are unrelated." + "reason": "Qwen-3 NVFP4 runtime error and LoRA merge/save inconsistency are unrelated." }, { "left": "issue:44841", "right": "issue:46661", "accept": false, - "reason": "Processor failure for Voxtral and remote-code config_class AttributeError are different problems." + "reason": "Processor failure for Voxtral is unrelated to remote-code model_class/config_class errors." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "Both involve training/DeepSpeed, but the concrete errors and failure modes are unrelated." + "reason": "DeepSpeed training weight-shape error and collective mismatch are different failures." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Activation-checkpointing/FSDP recomputation mismatch is not the same bug as a mask shape indexing error." + "reason": "FSDP activation-checkpointing size issue is unrelated to mask-shape IndexError." }, { "left": "issue:45859", "right": "issue:45902", "accept": false, - "reason": "Same class name appears, but one is CPU-offload incompatibility and the other is uninitialized memory after meta->to_empty materialization." + "reason": "Related Qwen3_5Moe rotary-embedding reports, but one is CPU-offload incompatibility and the other is uninitialized memory after meta\u2192to_empty(cuda); not the same failure mode." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "BERT ZeRO-3 init failure and GPT-OSS tensor-parallel failure are unrelated." + "reason": "Bert ZeRO-3 loading error is unrelated to GPT-OSS tensor-parallel failure." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "FSDP activation-checkpointing tensor-size recomputation and custom trainer loss signature mismatch are unrelated." + "reason": "FSDP activation-checkpointing size mismatch is unrelated to CustomTrainer compute_loss signature change." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different model/runtime layers: BERT ZeRO-3 init vs GPT-OSS tensor parallelism." + "reason": "ZeRO-3 Bert loading error is unrelated to GPT-OSS TP-only failure." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "Mistral-3 dtype mismatch and GraniteMoEHybrid invalid method call are unrelated." + "reason": "Mistral-3 image dtype mismatch and GraniteMoEHybrid invalid method are different issues." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime error and GPT-OSS AMD GPU failure are distinct model/back-end issues." + "reason": "Qwen-3 NVFP4 runtime error is unrelated to the AMD GPU failure for GPT-OSS." }, { "left": "issue:45874", "right": "issue:46294", "accept": false, - "reason": "Gemma4 reverse embedding lookup regression and Qwen3.5 video token mismatch are different issues." + "reason": "Gemma4 reverse embedding lookup and Qwen3.5 video token mismatch are unrelated." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Gemma3 config missing attribute for vLLM and Mistral tokenizer patch crashing on backend_tokenizer are unrelated." + "reason": "Gemma3 vLLM attribute error and Mistral tokenizer regex crash are unrelated." }, { "left": "issue:45823", "right": "issue:45874", "accept": false, - "reason": "Device-map forward mismatch and reverse embedding lookup regression are different Gemma4 bugs." + "reason": "Both are Gemma4 issues, but one is device-map mismatch and the other is expensive reverse embedding lookup." }, { "left": "issue:45800", "right": "issue:46515", - "accept": true, - "reason": "These describe the same underlying compatibility problem centered on torch 2.4.1; one is just a more specific wording of the same incompatibility." + "accept": false, + "reason": "Both mention torch 2.4.1 incompatibility, but the failure details are too broad to establish one bug." }, { "left": "issue:43425", "right": "issue:44292", "accept": false, - "reason": "Torch 2.10 incompatibility and Qwen-3-8B-NVFP4 runtime error are not the same issue." + "reason": "Torch version incompatibility and Qwen-3 NVFP4 runtime error are unrelated." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "FSDP checkpointing size recomputation and Qwen3 auto device-map cuda asserts are unrelated." + "reason": "FSDP activation-checkpointing size issue is unrelated to Qwen3 auto device-mapping CUDA assert." }, { "left": "issue:43828", - "right": "issue:45161", + "right": "issue:45685", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE and GPT-OSS only-TP failure are different bugs." + "reason": "Different model/backend paths: autocast dtype mismatch vs MPS histogram-on-Int error." }, { - "left": "issue:45800", - "right": "issue:46304", + "left": "issue:44918", + "right": "issue:45685", "accept": false, - "reason": "Torch 2.4.1 compatibility and USE_DISTRIBUTED=0 breakage are different version-incompatibility issues." + "reason": "TRl SFT embedding issue is unrelated to the MPS histogram error." }, { - "left": "issue:44509", - "right": "issue:44512", + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and trainer compute_loss signature change are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Activation-checkpointing/FSDP size mismatch is unrelated to GenerationMixin import failure." + }, + { + "left": "issue:45237", + "right": "issue:46015", + "accept": false, + "reason": "GPT-OSS GPU issue and outdated RAG examples are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44805", "accept": false, - "reason": "Both are v5 docs regressions, but one is about pipeline tasks and the other about the transformers run command; not the same concrete docs bug." + "reason": "Liger Kernel cross-entropy issue is unrelated to the mask/tensor shape IndexError." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The two accepted duplicate pairs are well supported by the issue titles and descriptions, and the rejected soft-edge pairs are explained in a cautious way without overclaiming sameness. No mergeability issue applies here since this is not a PR pair cluster." + "accept": false, + "feedback": "The writeup is mostly conservative, but it overstates at least one soft-edge pair: issue:45859 vs issue:45902 is treated as a true duplicate even though the packet only shows related Qwen3_5Moe rotary-embedding wording, not a clearly same failure mode. The two titles point to different concrete problems (CPU-offload incompatibility vs uninitialized memory after meta\u2192to_empty(cuda) materialization), and the packet marks the pair as non-deterministic. Because the summary promotes that pair as one of the only strong duplicate signals, it is not fully grounded and should be softened/rejected." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -460,7 +476,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:36010", "issue:39290", @@ -512,11 +528,12 @@ ], "soft_pairs": [ "issue:43866|issue:46402", + "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:44589|issue:46304", "issue:45237|issue:45901", "issue:43039|issue:44560", "issue:45823|issue:46294", @@ -552,280 +569,274 @@ "issue:45800|issue:45823", "issue:43296|issue:43366" ], - "prepared_review_unit_hash": "d6928c1b4e47bbf8cff6331a297f94a03fadbee51e6a54a2d4e019d13049c776", + "prepared_review_unit_hash": "3c5c01df1ba5f95e30263a259877d5f1a9ad7821cb3a53fd0ef4a3f0f3441b3f", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 40, - "serialized_chars": 38898, - "estimated_input_tokens": 9725, - "estimated_eval_tokens": 19706 + "soft_pair_count": 41, + "serialized_chars": 39201, + "estimated_input_tokens": 9801, + "estimated_eval_tokens": 19858 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous issue set with many superficial similarity matches around Transformers regressions. Only the AutoTokenizer token-ID regression pair looks like a true duplicate; the other soft pairs differ by model, failure mode, or code path.", - "confidence": 0.74, - "canonical_issue_reason": "No single canonical issue fits the whole set because the items span unrelated bugs. If one representative is needed, issue 45920 is the best broad v5 tokenizer-regression report and likely subsumes 45812.", + "summary": "Most pairs are clearly unrelated issue reports. The only strong duplicate candidate is the AutoTokenizer v5 regression pair, where 45920 is the broader, better canonical issue.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 45920 is the broadest and clearest report of the tokenizer-ID regression, covering multiple affected model families and subsuming the narrower Granite-specific report.", "canonical_pr_reason": null, - "best_issue_reason": "45920 is the strongest umbrella issue here: it covers multiple affected model families, has inbound references, and is the broader version of the one clear duplicate pair.", + "best_issue_reason": "45920 best represents the underlying bug because it states the shared failure mode across several models, making it the most useful canonical issue for triage.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43866", "right": "issue:46402", "accept": false, - "reason": "Different bugs: corrupted checkpoint vs missing text-encoder weights during model loading." + "reason": "Different model-loading failures in unrelated models; no shared bug path is evident." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both mention version incompatibility, but they describe different breakages and code paths." }, { "left": "issue:43296", "right": "issue:46304", "accept": false, - "reason": "Unrelated failure modes: PaddleOCR-VL/vLLM load issue vs USE_DISTRIBUTED=0 support regression." + "reason": "Unrelated failures: vLLM/PaddleOCR-VL loading vs distributed/PyTorch support." }, { "left": "issue:41762", "right": "issue:43596", "accept": false, - "reason": "Both are ZeRO-3 IndexErrors, but one is Gemma3 loading and the other is BertModel init; not enough to treat as the same concrete bug." + "reason": "Both are index errors under ZeRO-3, but the affected models and likely failure sites differ." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Different model-specific errors: missing Gemma3 sliding_window_pattern vs Qwen3 auto device-map cuda assert." + "reason": "Gemma3 config attribute error and Qwen3 device-map CUDA assert are unrelated." }, { - "left": "issue:44589", - "right": "issue:46304", + "left": "issue:43901", + "right": "issue:44512", "accept": false, - "reason": "Different subsystems: float8 storage lookup vs distributed build/runtime support." + "reason": "Both are docs complaints, but they concern different documentation regressions." }, { - "left": "issue:43901", - "right": "issue:44512", + "left": "issue:44589", + "right": "issue:46304", "accept": false, - "reason": "Both touch docs, but they are different documentation issues with different removed/changed behavior." + "reason": "Float8 storage lookup error is unrelated to the USE_DISTRIBUTED=0 support regression." }, { "left": "issue:45237", "right": "issue:45901", "accept": false, - "reason": "Different areas and failures: AMD GPU support vs table-question-answering runtime crash." + "reason": "Different subsystems and symptoms; no common underlying bug is apparent." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Liger kernel cross-entropy dispatch issue is unrelated to Qwen3-VL video StopIteration." + "reason": "Liger Kernel cross-entropy dispatch and Qwen3-VL video StopIteration are unrelated." }, { "left": "issue:45823", "right": "issue:46294", "accept": false, - "reason": "Both mention device/multimodal mismatches, but the failing code paths and symptoms are different." + "reason": "Device-mapping mismatch and video-token/template mismatch are different model-input bugs." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "Different problems: incorrect detokenization vs outdated RAG examples." + "reason": "Tokenizer detokenization bug vs outdated RAG docs; not the same issue." }, { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to Gemma4 PLE device placement." + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different problems." }, { "left": "issue:43866", "right": "issue:46772", "accept": false, - "reason": "Checkpoint corruption vs ModelOpt/NVFP4 compatibility are not the same bug." + "reason": "Corrupted checkpoint vs NVFP4 compatibility are distinct causes." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "NemotronH checkpoint loading problem is unrelated to documentation freshness." + "reason": "NemotronH checkpoint loading failure is unrelated to RAG documentation issues." }, { "left": "issue:45020", "right": "issue:46304", "accept": false, - "reason": "Both are version regressions, but one is remote_code breakage and the other is PyTorch build/support with USE_DISTRIBUTED=0." + "reason": "Remote-code model breakage and USE_DISTRIBUTED=0 breakage are separate regressions." }, { "left": "issue:45812", "right": "issue:45920", "accept": true, - "reason": "Same underlying tokenizer regression: AutoTokenizer returns wrong token IDs in v5, with 45920 being the broader multi-model report." + "reason": "Same underlying AutoTokenizer v5 regression causing wrong token IDs across multiple model families; 45920 is the broader report." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Different model-loading failures: corrupted checkpoint vs missing NemotronH checkpoint support." + "reason": "Different checkpoint/loading failures for different models." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "bitsandbytes Int8Params keyword-argument incompatibility is unrelated to the distributed support regression." + "reason": "bitsandbytes Int8Params constructor mismatch is unrelated to distributed support." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime error is unrelated to RAG documentation content." + "reason": "NVFP4 model runtime error vs documentation issue; no shared code path." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Different sliding_window issues in different model families and with different missing/error conditions." + "reason": "Both mention sliding-window/Qwen3, but they target different models and failures." }, { "left": "issue:41093", "right": "issue:43638", "accept": false, - "reason": "Both are IndexErrors, but one is a mask-shape mismatch and the other is a ZeRO-3/BertModel empty-tensor case." + "reason": "Both are shape/index errors, but one is a mask mismatch and the other is ZeRO-3/Bert initialization." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper cannot run and table QA crashing are unrelated." + "reason": "Whisper runtime failure and table-question-answering crash are unrelated." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Model load/test failure for GLM-4.7-Flash is not the same as a corrupted Ovis2 checkpoint." + "reason": "Model test loading failure and checkpoint corruption are not the same bug." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Different version compatibility bugs: Torch 2.10 compatibility vs pydantic PretrainedConfig field breakage." + "reason": "Torch-version incompatibility and pydantic PretrainedConfig regression are different breakages." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Wrong detokenization and table QA crash are unrelated bugs." + "reason": "Tokenizer detokenization bug and table QA crash do not match." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper load failure and PretrainedConfig pydantic regression do not share a code path." + "reason": "Whisper loading/runtime issue is unrelated to the pydantic model regression." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "ZeRO-3 BertModel IndexError is unrelated to torch/transformers version incompatibility." + "reason": "Different ZeRO-3/indexing issue versus a Torch/Transformers compatibility report." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Qwen3.5 chat crash and streaming inference API failure are different issues." - }, - { - "left": "issue:44292", - "right": "issue:45901", - "accept": false, - "reason": "NVFP4 model error and table QA crash are unrelated." - }, - { - "left": "issue:39290", - "right": "issue:45823", - "accept": false, - "reason": "Different model families and failure modes; both mention configuration/device handling only broadly." + "reason": "Same model family, but different API surfaces and failure modes; not enough to treat as one bug." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Streaming chat/completions failure is unrelated to MoE MPS histogram support." + "reason": "Chat/completions streaming failure and MPS histogram kernel error are unrelated." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "ImportError for GenerationMixin and Gemma3 sliding_window_pattern missing attribute are unrelated." + "reason": "ImportError for GenerationMixin and Gemma3 config attribute error are unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Both involve MoE-ish models, but the errors and code paths differ." + "reason": "Grouped_mm/dynamo tracing issue and MPS histogram Int error are different code paths." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Grouped_mm tracing failure and autocast dtype mismatch are not the same bug." + "reason": "Both are precision-related MoE errors, but the concrete failures and triggers differ." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Voxtral processor failure and template-node compilation error are unrelated." + "reason": "Processor failure and template-node compilation error are unrelated." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "Mistral3 lm_head tying bug is unrelated to Ovis2 checkpoint corruption." + "reason": "Weight-tying bug and checkpoint corruption are distinct issues." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are IndexErrors, but the mask-shape mismatch and ZeRO-3 empty-dimension load error are different failures." + "reason": "Different index-error signatures with different model/loading contexts." }, { "left": "issue:45399", "right": "issue:46661", "accept": false, - "reason": "Flash-attn2 fallback gating and remote-code config_class loading are different code paths." + "reason": "Flash-attn fallback logic and remote-code config_class failure are unrelated." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "StableLmConfig pad_token_idx missing is not the same as the PretrainedConfig pydantic-field regression." + "reason": "Specific StableLmConfig field regression vs broader pydantic PretrainedConfig regression." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch/transformers compatibility and Gemma4 device mismatch are unrelated." + "reason": "Torch/Transformers incompatibility and Gemma4 device mismatch are unrelated." }, { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "PaddleOCR-VL load failure and GGUF gpt-oss support are different feature requests/failures." + "reason": "PaddleOCR-VL loading in vLLM and GGUF gpt-oss support are different model-support requests." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it treats the set as heterogeneous, identifies the one clearly related tokenizer pair (45812/45920), and does not overclaim the rest. The canonical-issue rationale is also appropriately cautious given the unrelated items." + "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted duplicate pair, issue:45812 \u2194 issue:45920, is well supported by the shared AutoTokenizer v5 wrong-token-ID regression, and choosing 45920 as the broader canonical issue is reasonable." }, "evaluator_used": true, "retried": false, @@ -844,7 +855,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:39186", "issue:40990", @@ -907,8 +918,8 @@ "issue:43716|issue:46294", "issue:45020|issue:45800", "issue:41093|issue:43596", - "issue:44479|issue:46294", "issue:43054|issue:43994", + "issue:44479|issue:46294", "issue:45759|issue:45902", "issue:46640|issue:46661", "issue:43716|issue:44918", @@ -929,232 +940,226 @@ "issue:43975|issue:44292", "issue:44292|issue:45507" ], - "prepared_review_unit_hash": "e6d6ca6f7ff8847521644bfbd4562fbe6c8b06c935292a7d8ba1397a1ff7fdb8", + "prepared_review_unit_hash": "b617567ab8c6db251ac478ed507ff6481abdc2412e916db043131796b066fc99", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37142, - "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": "The cluster is mostly heterogeneous and contains many unrelated model-, tokenizer-, multimodal-, and config-specific bugs. The only strong duplicate-looking pair is the remote-code/custom-module loading cache-path issue (45632/45698).", - "confidence": 0.71, - "canonical_issue_reason": "Issue 45632 is the best root-cause candidate for the only clear duplicate subcluster: it describes the cache-path collision that plausibly leads to the wrong custom module being loaded in 45698.", + "summary": "Most pairs are not duplicates; the clear duplicate-looking groups are SigLIP2 output regression, Qwen3.5 video-input regression, remote-code cache/custom-module loading, and generic input-validation error-message bugs.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45698 is the best umbrella issue here: it describes the user-facing failure ('from_pretrained loads wrong custom module after save_pretrained'), while issue:45632 is the narrower cache-collision cause.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45632 is the strongest single representative of the only likely duplicate pair, and it is more root-cause-oriented than the symptom-level report in 45698.", + "best_issue_reason": "issue:45698 is the strongest canonical issue candidate among the set because it is broad, actively discussed, and cleanly subsumes the more specific leaf-directory cache collision report.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "Both are v5-era regressions, but one is int4 memory/OOM behavior and the other is tied-weight translation-model handling; different bugs." + "reason": "Both involve quantization/tied-weights themes, but they describe different concrete failures on different models/code paths." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Both mention shape/packing failures, but one is a mask-index mismatch and the other is a Qwen3.5 embedding/trainer failure; not the same code path." + "reason": "Both are Qwen-related runtime errors, but the mask-shape mismatch and the embedding-unpacking failure are different bugs." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Different MoE-related failures: one is a torch dynamo grouped_mm error, the other is a tensor-parallel support issue for GPT-OSS." + "reason": "Different failures: torch dynamo grouped_mm tracing vs tensor-parallel support on GPT-OSS MoE." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Different models and symptoms: autocast dtype mismatch versus streaming chat/completions failure." + "reason": "Different model families and symptoms: autocast dtype mismatch vs streaming chat/completions failure." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF architecture support for gpt-oss is unrelated to a Qwen3.5 streaming inference failure." + "reason": "gpt-oss GGUF architecture support is unrelated to the Qwen3.5 streaming inference bug." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "One is init_empty_weights/_is_hf_initialized argument handling; the other is a pydantic PretrainedConfig field breakage." + "reason": "One is an init_empty_weights/_is_hf_initialized constructor issue; the other is a pydantic config-field regression." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to the PretrainedConfig pydantic regression." + "reason": "Qwen3-MoE sliding-window handling is unrelated to the pydantic PretrainedConfig field breakage." }, { "left": "issue:43541", "right": "issue:44589", "accept": false, - "reason": "Grouped_mm tracing and missing Float8 storage are distinct runtime/serialization issues." + "reason": "Different concrete errors and code paths: grouped_mm tracing vs missing Float8 storage object." }, { "left": "issue:43716", "right": "issue:46294", "accept": false, - "reason": "Both are multimodal dtype/feature issues, but one is image-preprocessor dtype mismatch and the other is video token/feature mismatch." + "reason": "Both are multimodal input mismatches, but image preprocessor/model dtype mismatch is not the same bug as video-token/template mismatch." }, { "left": "issue:45020", "right": "issue:45800", "accept": false, - "reason": "Remote-code model loading regressions are too broad here; one is custom-module breakage, the other is a torch/transformers compatibility report." + "reason": "General remote_code breakage is too broad to merge with a torch/transformers version incompatibility report." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Different indexing failures in different subsystems; no shared concrete bug." - }, - { - "left": "issue:44479", - "right": "issue:46294", - "accept": false, - "reason": "Both involve video input, but they target different Qwen model families and describe different mismatch failures." + "reason": "Both are indexing/runtime errors, but the mask-length mismatch and deepspeed zero3 out-of-bounds issue are different." }, { "left": "issue:43054", "right": "issue:43994", - "accept": false, - "reason": "Both are SigLIP2 quality/sanity complaints, but one is comparative embedding quality and the other is general bad AutoModel/pipeline output; root cause is not clear enough." + "accept": true, + "reason": "Same SigLIP2 underlying regression: one reports degraded text embeddings, the other nonsensical AutoModel/pipeline outputs on the same model family." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": true, + "reason": "46294 looks like a specific reproduction of the broader Qwen3.5 video-input regression described in 44479." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Different multimodal config/materialization bugs: unwrapping text_config versus uninitialized inv_freq after to_empty(cuda)." + "reason": "Different multimodal/Qwen issues: config unwrapping from_config vs uninitialized rotary embedding memory after to_empty." }, { "left": "issue:46640", "right": "issue:46661", "accept": false, - "reason": "Flash Attention 2 integration and remote-code model_class/config_class lookup are unrelated." + "reason": "Flash Attention 2 behavior and remote-code config_class lookup are unrelated failures." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Different failure modes: multimodal dtype mismatch versus TRL SFT embedding unpacking." + "reason": "Different multimodal bugs: dtype mismatch in preprocessing vs TRL SFT unpacking failure." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP 2-D weight requirement and Qwen3.5 embedding unpacking are unrelated." + "reason": "FSDP 2-D weight error is unrelated to Qwen3.5 embedding unpacking." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "Perplexity regression on gpt-oss and a GLM5 issue do not appear to share a concrete bug." + "reason": "Perplexity quality regression and GLM5 support are not the same underlying bug." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Custom-model notebook initialization and flash-attn2 fallback gating are unrelated." + "reason": "Custom models in notebooks vs flash-attn2 fallback blocking are different issues." }, { "left": "issue:45632", "right": "issue:45698", "accept": true, - "reason": "Both point to the same underlying remote-code/custom-module loading bug: cache/path resolution appears to select the wrong module, causing incorrect from_pretrained behavior." + "reason": "Same underlying remote-code cache/custom-module bug; 45632 is the narrower leaf-directory collision cause, 45698 is the broader symptom." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Tied lm_head weights in a multimodal Mistral model is not the same as translation-model weight tying regressions in v5." + "reason": "Both mention tied weights, but they concern different model families and user flows; not enough evidence of the same concrete bug." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Fine-grained FP8 failure for Qwen3Moe is unrelated to flash-attn2 fallback selection." + "reason": "FineGrainedFP8Config failure and flash-attn2 fallback gating are unrelated." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash and tied-weights translation behavior are different code paths." + "reason": "PagedAttentionCache linear_attention crash on Qwen3.5 is a different code path from tied-weights regressions." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization offloading/memory behavior is not the same as tied-weight handling." + "reason": "Offloading/quantization behavior is broader and different from the specific tied-weights regression reported in 45005." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "A DSA indexer algorithm issue and a docs typo are unrelated." + "reason": "DSA indexer ReLU discussion and docs mentioning the removed run command are unrelated." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading is a model implementation bug, not a tied-weights translation regression." + "reason": "NemotronH checkpoint loading failure is not the same bug as tied weights in translation models." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "Cache-strategy typo and removed CLI docs are unrelated documentation issues." + "reason": "A typo in cache strategies is unrelated to stale docs about the removed run command." }, { "left": "issue:45183", "right": "issue:46463", - "accept": false, - "reason": "Both concern validation errors, but one is a broad input_ids/inputs_embeds XOR-message problem and the other is T5 decoder-input messaging; not clearly the same bug." + "accept": true, + "reason": "46463 appears to be a specific T5-family manifestation of the broader input-validation XOR error-message bug described in 45183." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "TRL embedding unpacking for Qwen3.5 and GraniteMoEHybrid calling an invalid method are different failures." + "reason": "Qwen3.5 embedding unpacking under TRL and GraniteMoEHybrid invalid-method calls are different failures." }, { "left": "issue:46489", "right": "issue:46491", "accept": false, - "reason": "Different tokenizer regressions affecting different model families and different symptoms." + "reason": "Both are tokenizer regressions, but they affect different models and different round-trip behavior." }, { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "DeepSeek-Coder detokenization and NVFP4 loading errors are unrelated." - }, - { - "left": "issue:44292", - "right": "issue:45507", - "accept": false, - "reason": "Qwen-3-8B-NVFP4 runtime failure and GraniteMoEHybrid invalid method are different bugs." + "reason": "DeepSeek detokenization and Qwen NVFP4 runtime failure are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: the cluster is clearly heterogeneous, and the 45632/45698 pair is the only one that reads as a plausible duplicate on titles alone. The soft-edge rejections are generally conservative and do not overclaim shared root cause." + "feedback": "Grounded overall: the canonical issue choice is well supported by the titles, and the accepted soft pairs are conservative duplicates-by-subsumption rather than overclaims. The rejected pairs also look appropriately distinct based on the packet." }, "evaluator_used": true, "retried": false, @@ -1173,7 +1178,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:24643", "issue:28530", @@ -1250,9 +1255,9 @@ "issue:41762|issue:42915", "issue:44991|issue:45399", "issue:43824|issue:43872", - "issue:45685|issue:45823", "issue:43881|issue:43883", "issue:43782|issue:44960", + "issue:45685|issue:45823", "issue:45020|issue:45823", "issue:45081|issue:45399", "issue:44661|issue:45020", @@ -1260,244 +1265,232 @@ "issue:43828|issue:45823", "issue:45020|issue:45356" ], - "prepared_review_unit_hash": "df9d52a457b677b686e0b39f4b1aeb40a4711d8036c787acaeb3b8b339ba8378", + "prepared_review_unit_hash": "e369832d8de57252858f7e6f005983745f14cec818765bba8534a131d1ef4dfc", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 38200, - "estimated_input_tokens": 9550, - "estimated_eval_tokens": 19356 + "serialized_chars": 38201, + "estimated_input_tokens": 9551, + "estimated_eval_tokens": 19358 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of Transformers v5-era loading/tokenization regressions, with only a few true duplicate pairs. The strongest duplicates are the non-persistent-buffer corruption reports and the tokenizer-output regression for DeepSeek-Coder/AutoTokenizer.", - "confidence": 0.61, - "canonical_issue_reason": "Issue 45020 is the broadest umbrella in the set and best captures the common v5 loading/remote_code regression theme that ties the cluster together, even though many members are narrower or unrelated sub-bugs.", + "summary": "This cluster is mostly a grab bag of unrelated Transformers issues, but there is a clear duplicate set around the v5 regression that corrupts non-persistent buffers during `from_pretrained()`/materialization. Most other pairs are only loosely related by model family or subsystem and should stay separate.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43950 is the best umbrella report for the non-persistent buffer corruption regression: it names the concrete API (`from_pretrained()`), the affected behavior (`register_buffer(persistent=False)`), and the version regression. It subsumes the near-identical reports in 43644 and 44534, plus the more specific manifestation in 45902.", "canonical_pr_reason": null, - "best_issue_reason": "45020 is the best representative issue overall because it is the most general and central to the dominant regression pattern in this cluster.", + "best_issue_reason": "43950 is the strongest representative issue because it is the most concrete and actionable description of the shared bug, while still broad enough to cover the duplicate phrasing variants around the same buffer-corruption regression.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Same underlying bug: transformers v5 corrupts non-persistent buffers / fills them with junk during from_pretrained loading." + "reason": "Same underlying v5 bug: non-persistent buffers getting filled/corrupted with junk." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Different problems: one is a SigLIP2 docs/example mismatch plus quantization failure, the other is a model quality complaint about text embeddings." + "reason": "One is a broken docs example; the other is a model-quality/embedding regression. Same model family, not the same bug." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Both describe the same non-persistent-buffer regression in from_pretrained on transformers 5.x." + "reason": "Near-identical report of the same non-persistent buffer corruption regression." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Unrelated failures: custom compute_loss API change vs DeepSpeed ZeRO-3 Gemma3 loading index error." + "reason": "Different failures: trainer API incompatibility vs DeepSpeed ZeRO-3 load-time index error." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are low-bit/loading TypeErrors, but they involve different code paths and different root causes." + "reason": "Both are compatibility errors, but they involve different storage/backends and different code paths." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Different bugs: an autocast dtype mismatch in Phi-tiny-MoE vs Qwen3.5 RLHF/position_ids causing SDPA fallback and NaNs." + "reason": "Different models and different symptoms: dtype mismatch under autocast vs gradient explosion from 3D position_ids/SDPA fallback." }, { "left": "issue:45920", "right": "issue:46489", - "accept": true, - "reason": "46489 looks like a specific remaining DeepSeek-Coder tokenizer regression covered by the broader AutoTokenizer wrong-token-id bug in 45920." + "accept": false, + "reason": "Both concern tokenizer regressions, but 45920 is a broad multi-model token-ID issue while 46489 is a specific DeepSeek-Coder detokenization problem tied to a prior fix gap." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "One is a SigLIP2 embedding quality issue; the other is a general tokenization change across versions." + "reason": "Comparing SigLIP2 embedding quality to SigLIP is not the same as a generic tokenizer/version-change report." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Both involve SigLIP2, but one reports zero probabilities while the other reports worse text embeddings; not the same concrete bug." + "reason": "Both are about SigLIP2, but one reports zero probabilities and the other reports worse text embeddings; not the same concrete bug." }, { "left": "issue:43950", "right": "issue:45902", - "accept": false, - "reason": "Related to buffer handling, but 45902 is a Qwen3_5Moe inv_freq materialization bug, not the generic non-persistent-buffer corruption in 43950." + "accept": true, + "reason": "Specific manifestation of the same non-persistent buffer/materialization corruption bug." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Different models and different failures: bad AutoTokenizer error for Voxtral vs missing all_tied_weights_keys on MolmoForCausalLM." + "reason": "Bad AutoTokenizer error message for Voxtral vs missing attribute on MolmoForCausalLM are unrelated failures." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Both touch Qwen3.5/MoE, but one is flash-attention 3D position_ids misinterpretation and the other is CPU offload incompatibility in rotary embedding." + "reason": "Flash-Attention illegal memory access from 3D position_ids is unrelated to CPU offload incompatibility in Qwen3_5MoeTextRotaryEmbedding." }, { "left": "issue:43054", "right": "issue:43493", "accept": false, - "reason": "SigLIP2-related, but one is a quality regression complaint and the other is an implementation discrepancy vs JAX; not clearly the same bug." + "reason": "Both mention SigLIP2, but one is an output-quality complaint and the other is a HF-vs-JAX implementation discrepancy; too broad to treat as the same bug." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Different regressions: ineffective gradient checkpointing with PEFT LoRA vs increased CUDA reserved memory under int4 quantization." + "reason": "Different issues: ineffective gradient checkpointing with LoRA vs increased reserved memory/OOM under int4 quantization." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Distinct model-specific bugs: Gemma4 FSDP2 shared_kv_states KeyError vs Qwen3_5Moe inv_freq uninitialized memory." + "reason": "Gemma4 FSDP2 KeyError is a different code-path problem from non-persistent buffer materialization corruption." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Different loading failures for different models; no shared code-path evidence." + "reason": "Different model-loading failures with no shared concrete code path." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "One is a logging/noise complaint; the other is a concrete model load failure for ScandiBERT." + "reason": "Generic loading noise vs a specific model load failure; not the same bug." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Different tokenizer/import issues with different symptoms and code paths." + "reason": "Tokenizer detokenization regression vs missing `flash_attn` import on Python 3.13 are unrelated." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "General loading-noise issue vs a specific model loading failure; not the same bug." + "reason": "Generic loading noise vs a specific tokenizer/model load failure for `cjvt/sleng-bert`." }, { "left": "issue:24643", "right": "issue:28530", "accept": false, - "reason": "Completely unrelated training/DeepSpeed error vs early-stopping metric warning." + "reason": "DeepSpeed training shape error and early-stopping metric handling are unrelated." }, { "left": "issue:42947", "right": "issue:43856", "accept": false, - "reason": "Different performance/regression reports: LoRA gradient checkpointing vs Qwen3 MoE memory usage." + "reason": "PEFT LoRA checkpointing behavior is not the same issue as Qwen3 MoE memory usage." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "43543 is a specific AutoConfig missing-fields bug; 45020 is a broader remote_code regression umbrella, but they are not clearly the same concrete issue." + "reason": "Both involve `remote_code`, but one is missing fields in `AutoConfig` while the other is a broader model-loading regression; not clearly the same bug." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Different model-loading failures with no shared underlying bug indicated." + "reason": "Different models and different failure modes: Gemma3 DeepSpeed load error vs Qwen3 MoE LoRA incompatibility across branches." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Tokenizer loading for a specific model vs flash-attn2 fallback gating; unrelated." + "reason": "Tokenizer loading failure for a specific model vs flash-attn2 fallback gating; unrelated." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Import error for a missing model class vs bitsandbytes storage-object TypeError; unrelated." - }, - { - "left": "issue:45685", - "right": "issue:45823", - "accept": false, - "reason": "MPS histogram kernel missing for Int vs Gemma4 PLE device mismatch; different problems." + "reason": "Importing a missing model class and a bitsandbytes constructor incompatibility are different problems." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different model-specific failures: glm-4v-9b loading vs Molmo tied-weights attribute missing." + "reason": "Different load-time failures on different models and code paths." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Different models and different failure modes; no concrete shared bug." + "reason": "Qwen3VL weight-only load error and a GLM5 issue are not the same bug." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram dtype limitation and Gemma4 device mismatch are unrelated." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Remote_code-related loading regressions are too broad; Gemma4 PLE device mismatch is a separate forward-pass issue." + "reason": "Broad `remote_code` breakage is not the same as a Gemma4 PLE device-placement bug." }, { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Tokenizer regex crash vs flash-attn2 fallback checks; unrelated." + "reason": "Tokenizer regex crash and flash-attn2 fallback checks are separate problems." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES edge-case bug vs broad remote_code regression; not the same concrete issue." - }, - { - "left": "issue:42491", - "right": "issue:44368", - "accept": false, - "reason": "Version-compatibility break for a qwen3_moe LoRA model vs a warning about tie_word_embeddings in Qwen3.5 LoRA training." - }, - { - "left": "issue:43828", - "right": "issue:45823", - "accept": false, - "reason": "Autocast dtype mismatch vs device_map auto device mismatch; different code paths." + "reason": "`TOKENIZER_MAPPING_NAMES` add-new-model-like failure is unrelated to the broader `remote_code` regression report." }, { - "left": "issue:45020", - "right": "issue:45356", + "left": "issue:44910", + "right": "issue:45859", "accept": false, - "reason": "Generic remote_code breakage vs a specific Kimi-K2.5 tokenizer regression; not the same bug." + "reason": "Flash-attn crash from packed-sequence misinterpretation is unrelated to CPU offload incompatibility." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is mostly grounded in the packet and stays fairly conservative about the many unrelated pairs. The accepted duplicate calls are defensible from the titles alone, especially the non-persistent-buffer regressions and the DeepSeek-Coder tokenizer regression. The canonical issue choice is arguable, but not overstated given the soft-link evidence around remote_code/loading regressions." + "feedback": "Grounded overall. The umbrella issue choice is supported by the packet, and the accepted pairs are conservative duplicates of the same non-persistent-buffer/materialization regression. The remaining rejections stay safely on the cautious side." }, "evaluator_used": true, "retried": false, @@ -1516,7 +1509,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:24643", "issue:30064", @@ -1571,8 +1564,8 @@ "issue:45715|issue:46315", "issue:45488|issue:46489", "issue:43827|issue:44509", - "issue:43404|issue:43950", "issue:45759|issue:46661", + "issue:43404|issue:43950", "issue:43872|issue:44291", "issue:42915|issue:43872", "issue:45020|issue:46661", @@ -1597,14 +1590,14 @@ "issue:43638|issue:46515", "issue:45663|issue:45823" ], - "prepared_review_unit_hash": "84f8767f83830e826ee89e474db0370ab1bf5d7e5a8c6aecb58bbf766409e51a", + "prepared_review_unit_hash": "bb2445bbd3eabef2ca35d498442d15b50d4a71fbf8f9a7fe83310bf6a64a19fb", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, @@ -1619,186 +1612,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, with only a handful of true duplicate-like pairs: docs cleanup for removed pipeline tasks, the `_is_hf_initialized`/bitsandbytes init-empty-weights TypeError, Whisper fbank extraction, the mask-shape IndexError, and one Qwen video regression. No PRs are present.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 41093 is the strongest canonical anchor: it has an almost exact duplicate in 44805 with the same mask/tensor shape mismatch error and a concrete, reproducible failure signature.", + "summary": "This set is mostly a mix of unrelated Transformers bug reports, with a few clear duplicate pairs around docs, `_is_hf_initialized` loading regressions, tokenizer regressions, and a video-input regression.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 45020 is the broadest umbrella here: a general recent-version/`remote_code` compatibility regression that best represents the cluster\u2019s recurring loader breakages.", "canonical_pr_reason": null, - "best_issue_reason": "41093 is the best global issue representative because it is the clearest true duplicate pair in the set and has the most specific error reproduction.", + "best_issue_reason": "45020 is the strongest anchor issue because it is the most general and can subsume several of the more specific model-loading failures.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Both are tokenizer performance regressions, but they hit different APIs (`convert_ids_to_tokens` vs `convert_tokens_to_ids`) and look like separate bugs." + "reason": "Both are tokenizer performance complaints, but they hit different APIs (`convert_ids_to_tokens` vs `convert_tokens_to_ids`) and don\u2019t look like the same underlying bug." }, { "left": "issue:45488", "right": "issue:46489", - "accept": false, - "reason": "Both involve tokenizer behavior on DeepSeek-family models, but the failures are different concrete code paths and not clearly one bug." + "accept": true, + "reason": "Both point to the same DeepSeek/Llama tokenizer regression in v5, where tokenizer handling changes produce incorrect tokenization/output for DeepSeek-family models." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same docs cleanup problem: both report stale references to pipeline tasks removed in v5." + "reason": "These are the same docs regression: v5 pipeline-task removals are still referenced in summarization/translation docs." }, { - "left": "issue:43404", - "right": "issue:43950", + "left": "issue:45759", + "right": "issue:46661", "accept": false, - "reason": "One is an lm_head weight-tying bug, the other is a from_pretrained non-persistent buffer regression; different code paths." + "reason": "Both involve model construction, but one is about unwrapping `text_config` for composite Qwen configs while the other is a remote-code `config_class` attribute error; different bugs." }, { - "left": "issue:45759", - "right": "issue:46661", + "left": "issue:43404", + "right": "issue:43950", "accept": false, - "reason": "Different auto-model instantiation failures: composite `text_config` handling versus remote-code `config_class` lookup." + "reason": "One is a tied-weight bug in a specific multimodal model, the other is a broad `from_pretrained` non-persistent-buffer regression; no shared code-path evidence." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same underlying `_is_hf_initialized` incompatibility causing TypeError during model materialization/loading." + "reason": "Both center on the same `_is_hf_initialized` argument regression during model initialization/loading, causing TypeErrors in related loading paths." }, { "left": "issue:42915", "right": "issue:43872", "accept": false, - "reason": "Different problems: FineGrainedFP8Config failure versus `_is_hf_initialized`/bitsandbytes TypeError." + "reason": "These are different failures: FineGrainedFP8Config on Qwen3Moe versus an `_is_hf_initialized` bitsandbytes/init bug." }, { "left": "issue:45020", "right": "issue:46661", "accept": false, - "reason": "`45020` is a broad umbrella about remote_code breakage; `46661` is one specific error, so they are not the same concrete bug." + "reason": "`45020` is a broad umbrella for remote_code breakage, but `46661` is a specific `config_class` attribute error; not enough evidence they are the same concrete bug." }, { "left": "issue:43792", "right": "issue:44220", - "accept": true, - "reason": "Likely the same Whisper audio-feature extraction bug; both point to `_torch_extract_fbank_features()`/fbank processing failures." + "accept": false, + "reason": "The Whisper model failure and the `_torch_extract_fbank_features()` issue are related only by audio processing; the titles don\u2019t show the same exact bug." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer loading issues, but for different models and different failure modes." + "reason": "Different tokenizer-load problems on different models; one is a bad error message, the other is a failed tokenizer load regression." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Both mention DeepSpeed, but one is weight loading and the other is gradient averaging; different bugs." + "reason": "Both mention DeepSpeed and Qwen, but one is about weight conversion/loading and the other about gradient averaging; distinct code paths." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Different trainer signature issue versus tensor-mask shape mismatch; no shared bug." + "reason": "Unexpected `compute_loss` kwarg and a mask/tensor shape mismatch are unrelated runtime failures." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Both mention gradient checkpointing, but one is PEFT/LoRA ineffectiveness and the other is a missing model flag." + "reason": "Gradient checkpointing not taking effect with PEFT LoRA is a different bug from a missing `supports_gradient_checkpointing` flag on a remote-code model." }, { "left": "issue:43940", "right": "issue:45313", "accept": false, - "reason": "Similar DeepSpeed ZeRO-3 weight-loading symptom, but different Qwen model families and distinct reported failures." + "reason": "Both are ZeRO-3 weight-loading failures in Qwen models, but they affect different model families with different missing-weight symptoms." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different model-loading failures for unrelated architectures." + "reason": "Different model-loading failures for different architectures; no shared concrete bug is evident." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Both concern tokenizer/model loading, but they target different models and errors." + "reason": "A tokenizer error message issue for Voxtral is unrelated to loading `cjvt/sleng-bert`." }, { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "Different failures: RLHF gradient explosion versus a flash-attention auxiliary-state crash." + "reason": "Catastrophic NaN/SDPA-fallback behavior in RLHF is unrelated to a flash-attention crash on sink-less models." }, { "left": "issue:41093", "right": "issue:44805", "accept": true, - "reason": "Near-exact duplicate: same mask shape mismatch error with the same underlying indexing bug." + "reason": "These are the same generic mask-vs-indexed-tensor shape mismatch error pattern, likely the same off-by-one or alignment bug." }, { "left": "issue:44479", "right": "issue:44560", "accept": true, - "reason": "Both describe the same v5.3.0 Qwen video-input regression affecting Qwen3/VL family models." + "reason": "Both describe the same v5.3.0 Qwen video-input regression family; `44560` looks like a specific manifestation of the broader video-input breakage in `44479`." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Unrelated failures: DeepSpeed training weight shape error versus void segmentation map preprocessing." + "reason": "Training with DeepSpeed and void-segmentation-map processing are unrelated issues." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Different model-specific failures: PagedAttentionCache linear_attention crash versus checkpoint loading." + "reason": "PagedAttentionCache/linear_attention and NemotronH checkpoint loading are unrelated." }, { "left": "issue:44863", "right": "issue:46402", "accept": false, - "reason": "Different models and different loading problems; not the same bug." + "reason": "Different models and different loading failures; no same bug signal." }, { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "Unrelated issues: GGUF architecture support versus autocast dtype mismatch." + "reason": "GGUF gpt-oss support and a Phi autocast dtype mismatch are unrelated." }, { "left": "issue:43596", "right": "issue:46515", "accept": false, - "reason": "DeepSpeed ZeRO-3 init bug versus torch version incompatibility; no shared concrete code path." + "reason": "Zero-3 BertModel index error and torch 2.4.1 incompatibility are not the same concrete issue." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "Different Qwen-related issues: GGUF support versus sliding_window behavior." + "reason": "GGUF gpt-oss support is unrelated to a Qwen3-MoE sliding_window bug." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Both involve Qwen3.5/VL-related internals, but one is rope index scaling and the other is uninitialized-memory materialization." + "reason": "These are different positional-encoding bugs in different Qwen variants; same general area, but not the same issue." }, { "left": "issue:43638", "right": "issue:46515", "accept": false, - "reason": "Different failures: ZeRO-3/BertModel index error versus a torch compatibility issue." + "reason": "A Zero-3/non-pretrained BertModel index error is unrelated to a torch-version incompatibility." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Different Gemma4 bugs: FSDP2 shared_kv_states rebuild versus device_map placement mismatch." + "reason": "Gemma4 FSDP2 shared_kv_states issues and device_map mismatches are different forward-pass bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the accepted soft pairs are phrased conservatively enough overall. The canonical anchor on 41093/44805 is well supported by the identical mask-shape mismatch title, and the other accepted pairs are plausibly duplicate-like based on the titles alone. No PR-pair issue applies here since there are no PRs in the packet." + "feedback": "The summary is broadly grounded in the packet, and the conservative false rejects look fine. The accepted soft pairs are mostly defensible from the titles alone: the docs issues, the `_is_hf_initialized` loading regressions, the mask-shape errors, and the Qwen video-input regression family all look related. The tokenizer pair is a little broader, but still stays within a plausible shared regression family rather than asserting an exact duplicate." }, "evaluator_used": true, "retried": false, @@ -1817,7 +1810,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:36010", "issue:38175", @@ -1878,8 +1871,8 @@ "issue:44560|issue:45685", "issue:43645|issue:43646", "issue:43550|issue:44910", - "issue:44560|issue:44918", "issue:44683|issue:46515", + "issue:44560|issue:44918", "issue:42617|issue:43366", "issue:36010|issue:42915", "issue:45812|issue:46489", @@ -1903,106 +1896,106 @@ "issue:44960|issue:44977", "issue:46459|issue:46512" ], - "prepared_review_unit_hash": "42c8607ad78dd7c31c3b021aa53c386839741f3d1c45e6b0d7b399cffc9503de", + "prepared_review_unit_hash": "ba5cdb299e2c788a40fce4dcb32acaedeca8855c0433a0d9bef468c65d8032f5", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37472, - "estimated_input_tokens": 9368, - "estimated_eval_tokens": 18992 + "serialized_chars": 37477, + "estimated_input_tokens": 9370, + "estimated_eval_tokens": 18996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a bag of unrelated Transformers regressions; only a few pairs look like true duplicates. The strongest duplicate anchor is the non-persistent-buffer v5 regression, with exact matches also for SigLIP2 output corruption and custom model initialization.", + "summary": "Mostly a heterogeneous issue set with a few clear duplicate pairs. The strongest duplicates are the custom-model initialization regression, the SigLIP2 bad-output reports, and the non-persistent-buffer corruption reports; most other soft pairs are different bugs.", "confidence": 0.74, - "canonical_issue_reason": "issue:43644 is the clearest anchor for the shared v5 non-persistent-buffer regression, and issue:44534 reads like the same bug with the same symptom.", + "canonical_issue_reason": "issue:43644 is the best representative issue: it describes a broad Transformers 5.0 regression with clear duplicate coverage from 44534, and it\u2019s more general than the more model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the broadest and most representative issue in the set for a concrete, actionable regression, and it has an exact duplicate in issue:44534.", + "best_issue_reason": "issue:43644 is the strongest global issue choice because it captures a concrete, broadly applicable v5 regression and has an obvious duplicate, making it a good cluster anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "Different bugs: GGUF/gpt-oss support vs a Float8 storage lookup error." + "reason": "Different problems: GGUF architecture support vs a Float8 storage lookup error." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "Different code paths: AudioFlamingo3 batch leakage vs Whisper empty-transcription handling." + "reason": "AudioFlamingo3 batching leak and Whisper empty-transcription failure are unrelated code paths." }, { "left": "issue:43716", "right": "issue:44560", "accept": false, - "reason": "Different models and failures; dtype mismatch in SigLIP2 vs Qwen3-vl StopIteration." + "reason": "Model/processor dtype mismatch is unrelated to a Qwen3-vl video StopIteration error." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Import error for a missing class vs a meta-device model-loading regression." + "reason": "Missing import for a specific class and meta-device loading failures are different issues." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Non-persistent buffer corruption vs a long-input OCR index error; not the same bug." + "reason": "Non-persistent-buffer corruption is unrelated to Kosmos2.5 long-OCR index errors." }, { "left": "issue:42915", "right": "issue:43957", "accept": false, - "reason": "FP8 MoE failure and meta-device loading are distinct problems." + "reason": "Qwen3Moe FP8 failure and meta-device loading breakage are different bugs." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Unrelated failures: Qwen3-vl video inference vs an MPS int-histogram backend issue." + "reason": "Video embedding StopIteration and an MPS Int histogram error are unrelated." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Both describe the same Transformers v5 custom model initialization breakage; one is just a more specific notebook report." + "reason": "Same underlying custom-model initialization regression in Transformers 5; one report is just notebook-specific." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Different attention/compile bugs in different models and failure modes." + "reason": "Different model and different failure mode: torch.compile/SDPA vs FlashAttention misreading 3D position_ids." }, { - "left": "issue:44560", - "right": "issue:44918", + "left": "issue:44683", + "right": "issue:46515", "accept": false, - "reason": "Different Qwen3.5 bugs: video StopIteration vs SFT embedding unpacking." + "reason": "Both mention torch versions, but they describe different incompatibilities and different failure paths." }, { - "left": "issue:44683", - "right": "issue:46515", + "left": "issue:44560", + "right": "issue:44918", "accept": false, - "reason": "Both are torch-compatibility issues, but they target different versions and different failures." + "reason": "Qwen3-vl video inference StopIteration and TRL SFT embedding unpacking are not the same bug." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "3d_parallel.py runtime issue is unrelated to GGUF gpt-oss support." + "reason": "3d_parallel.py runtime failure is unrelated to GGUF gpt-oss architecture support." }, { "left": "issue:36010", @@ -2014,109 +2007,121 @@ "left": "issue:45812", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer regressions, but the affected models and symptoms are different enough to keep separate." + "reason": "Both are tokenizer regressions, but for different model families and likely different fixes." }, { "left": "issue:45823", "right": "issue:46515", "accept": false, - "reason": "Device-mismatch in Gemma4 forward vs a generic torch 2.4.1 incompatibility are not the same bug." + "reason": "Gemma4 device-mapping forward mismatch is unrelated to a general torch 2.4.1 incompatibility report." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "Tokenizer-mapping model-addition bug vs Gemma4 device placement issue." + "reason": "Tokenizer-mapping constraint and Gemma4 PLE device mismatch are different problems." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Different symptoms and code paths; Qwen3.5 video error is not the mask-shape IndexError." + "reason": "Different failures: video StopIteration vs tensor/mask shape mismatch." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Same SigLIP2 model regression: zero probabilities / nonsensical outputs from AutoModel and pipeline." + "reason": "Same SigLIP2 base model regression with incorrect/zero outputs; 43994 is the broader symptom report." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "_is_hf_initialized regression and Kosmos2.5 OCR index error are unrelated." + "reason": "_is_hf_initialized flag breakage and Kosmos2.5 OCR indexing are unrelated." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Both involve config validation, but on different models and likely different root causes." + "reason": "Both involve config validation, but the affected models and concrete failures differ too much to be the same bug." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "Zero3 BertModel init failure is unrelated to the TOKENIZER_MAPPING_NAMES add-new-model bug." + "reason": "Deepspeed zero3 Bert indexing and add-new-model-like tokenizer mapping failure are different code paths." }, { "left": "issue:43596", "right": "issue:44661", "accept": false, - "reason": "Same as above; different failure mode and code path." + "reason": "Zero3 Bert init IndexError is unrelated to the tokenizer-mapping issue." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "Dequantized device-map loading failure vs mmap OOM on Strix Halo are separate issues." + "reason": "GPU+CPU offload loading failure and Strix Halo mmap OOM are different issues." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Qwen3_moe hf4.x/hf5.x incompatibility and BitNet packed-weight loading are different regressions." + "reason": "Qwen3_moe LoRA compatibility across versions and BitNet packed-weight loading are unrelated." }, { "left": "issue:43644", "right": "issue:44534", "accept": true, - "reason": "Exact same non-persistent-buffer junk regression in Transformers v5." + "reason": "Same non-persistent-buffer regression in Transformers 5; the titles describe the same underlying bug." }, { "left": "issue:38175", "right": "issue:39692", "accept": false, - "reason": "Same model family, but one is a model-output regression and the other is a documentation/example plus quantization issue." + "reason": "Both touch SigLIP2, but one is bad model output while the other is a docs example with multiple distinct errors." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Qwen3 auto-device-map cuda assert is not the same as the Qwen3Moe FP8 config failure." + "reason": "Qwen3 device-map cudaErrorAssert and Qwen3Moe FP8 failure are different problems." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Completely different bugs: MoE FP8 config vs mmap-related OOM." + "reason": "Qwen3Moe FP8 failure is unrelated to mmap-related OOM avoidance." }, { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "LoRA/qwen3_moe compatibility and Gemma4 token-embedding resize behavior are unrelated." + "reason": "Different models and different symptoms: LoRA incompatibility vs resize_token_embeddings not propagating." }, { "left": "issue:44451", "right": "issue:44488", - "accept": true, - "reason": "Both report the same current-version model-loading regression affecting closely related BERT models." + "accept": false, + "reason": "Both are model-load failures, but the models and the likely root cause are not specific enough to confirm a duplicate." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "GLM5 and Qwen3.5 flash-attention failures are separate model-specific issues." + }, + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "One is dtype being ignored during checkpoint loading; the other is config.dtype staying stale after from_config." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted duplicate/near-duplicate pairs are plausible from the titles and issue descriptions, and the non-persistent-buffer anchor on 43644/44534 is well supported." + "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted duplicate pairs are well supported by highly similar titles and descriptions, and the rejected pairs avoid obvious overreach. The canonical issue choice of 43644 is also defensible as the broad v5 non-persistent-buffer regression with a clear duplicate in 44534." }, "evaluator_used": true, "retried": false, @@ -2135,7 +2140,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:36010", "issue:39404", @@ -2206,26 +2211,26 @@ "issue:45276|issue:45335", "issue:44910|issue:45588", "issue:44402|issue:44488", - "issue:44079|issue:46829", "issue:43425|issue:43957", "issue:43873|issue:44387", + "issue:44079|issue:46829", "issue:41720|issue:44155", "issue:39692|issue:44960", - "issue:42947|issue:45704", "issue:45663|issue:45759", + "issue:42947|issue:45704", "issue:43404|issue:43454", "issue:44291|issue:44756", "issue:36010|issue:39404", "issue:44387|issue:44756" ], - "prepared_review_unit_hash": "6d96cf41482cf705968b8465e5ec0055f6f2011ea6b4fe5f755f3e879afa5cbd", + "prepared_review_unit_hash": "c0ed4cd6e9bd0f3d09618d975515d8c215818e3c8fda5aa411d2124250be2e1a", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, @@ -2240,199 +2245,202 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous and mostly contains unrelated Transformers bugs. I would not merge any of the soft pairs as duplicates. If forced to pick a cluster anchor, issue 43873 is the broadest and most centrally discussed runtime bug, but the overall duplicate confidence is low.", - "confidence": 0.39, - "canonical_issue_reason": "Issue 43873 is the broadest, most user-facing report in the set and has strong discussion/inbound activity around a core runtime behavior (quantization/offloading/memory), making it the best anchor despite the cluster being weakly related overall.", + "summary": "Mostly a heterogeneous set of Transformers issues across unrelated models/subsystems. I found one likely duplicate pair around `resize_token_embeddings`; the rest look like distinct bugs and should not be merged.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 45276 is the broader of the only clear duplicate pair: it covers `embed_tokens_per_layer` and `output_embeddings`, while 45335 is the narrower T5Gemma variant.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43873 is the most suitable representative because it is active, broadly scoped, and about a foundational loading/runtime failure mode rather than a narrow model-specific regression.", + "best_issue_reason": "45276 is the best representative issue for this cluster because it is the broadest, most general-looking ticket among the few genuinely overlapping reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45507", "right": "issue:46468", "accept": false, - "reason": "Different models and failures: GraniteMoEHybrid invalid method call vs a broken Nemotron example. No shared concrete code path." + "reason": "Different models and failure modes; one is a model-method bug, the other a broken example." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Both involve Qwen3.5, but one is ZeRO-3 language_model weight loading and the other is save_pretrained visual-encoder key regression. Not the same bug." + "reason": "Both involve Qwen3.5, but one is ZeRO-3 loading and the other is save_pretrained key serialization; different bugs." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are model/tokenizer loading complaints, but they affect different models and appear to stem from different tokenizer/model config issues." + "reason": "Both are tokenizer-loading complaints, but for different models and likely different root causes." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Both mention Transformers 5.0 regressions, but one is junk non-persistent buffers and the other is Jupyter custom-model initialization. Distinct problems." + "reason": "Both are Transformers 5 regressions, but one is non-persistent buffer initialization and the other is Jupyter custom model initialization." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both touch memory/efficiency, but one is Qwen3 MoE training memory use and the other is quantization offloading behavior. Different code paths." + "reason": "Different problems: Qwen3 MoE training memory usage vs quantization offloading behavior." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both are memory-related, but one is training memory usage and the other is int4 quantization causing extra CUDA reserved memory/OOM. Not the same underlying defect." + "reason": "Both mention memory/OOM, but one is MoE training and the other is int4 quantization reserved-memory growth." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Both mention SigLIP2, but one is a broken docs example with model/processor mismatch and the other is nonsensical inference results with AutoModel/pipeline. Different issues." + "reason": "Related SigLIP2 area, but one is a docs example with multiple errors and the other is incorrect outputs from AutoModel/pipeline." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "LoRA checkpoint compatibility on Qwen3 MoE is unrelated to a resize_token_embeddings propagation bug." + "reason": "Different models and different symptoms; one is LoRA/hf5 compatibility, the other is resize_token_embeddings output handling." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is a different loading path than the GLM-4.7-Flash unit-test load failure." + "reason": "Different Qwen3 MoE loading paths: accelerate+deepspeed conversion vs unit-test model loading failure." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Both are model load failures, but for different model families and different error modes (unit-test loading vs shape mismatch)." + "reason": "Both are loading errors, but for different models and different mismatch conditions." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint math and SGD argument handling are unrelated bugs." + "reason": "Trainer checkpoint-resume math is unrelated to SGD optimizer arguments." }, { "left": "issue:42491", "right": "issue:43931", "accept": false, - "reason": "LoRA qwen3_moe version skew is unrelated to a Qwen3-VL weight-shape mismatch." + "reason": "No shared bug: LoRA checkpoint compatibility vs Qwen3-VL shape mismatch on load." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "GLM5 is not the same problem as quantized model save/load breakage." + "reason": "Different generations of the same family, but one is a generic GLM5 issue and the other is a quantized save/load regression." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and different tokenizer behaviors; not one concrete bug." + "reason": "Both are tokenizer-related regressions, but one is a LlamaTokenizer pre-tokenizer override and the other is Granite AutoTokenizer ID mapping." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Extra-state save/load handling is distinct from init_empty_weights rejecting an unexpected _is_hf_initialized argument." + "reason": "Different save/load paths: extra_state handling vs init_empty_weights unexpected argument." }, { "left": "issue:45362", "right": "issue:45406", "accept": false, - "reason": "Different product paths: chat runtime crash for Qwen3.5-35B vs transformers serve crashing on Gemma4Processor." + "reason": "Both are Qwen3.5/Gemma4 adjacent, but one is chat crashes and the other is processor attribute access in serve." }, { "left": "issue:45276", "right": "issue:45335", - "accept": false, - "reason": "Both are resize_token_embeddings regressions, but they involve different model classes and different affected submodules; too model-specific to merge." + "accept": true, + "reason": "Same API regression class: `resize_token_embeddings` fails to propagate to model-specific embedding/output weights; these look like the same underlying bug across Gemma4 and T5Gemma." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Both involve flash-attention, but one is a Qwen3.5 packed-sequence/position_ids crash and the other is an s_aux=None AttributeError on sink-less models." + "reason": "Both are flash-attention crashes, but the root causes differ: Qwen3.5 packed sequence handling vs sink-less `s_aux=None` handling." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Tokenizer vocab-size mismatch and a specific model load failure are not the same bug." - }, - { - "left": "issue:44079", - "right": "issue:46829", - "accept": false, - "reason": "Both concern output object fields, but one is generic ModelOutput key assignment and the other is a VideoPrism classifier output shape/field bug." + "reason": "One is tokenizer vocab-size mismatch, the other is a separate model-loading failure." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch 2.10 compatibility is a broad version issue, while the meta-device bug is a specific model-loading path regression." + "reason": "Torch version incompatibility is unrelated to meta-device loading regressions." }, { "left": "issue:43873", "right": "issue:44387", "accept": false, - "reason": "Both are memory/OOM-related, but one is offloading with quantization and the other is increased reserved memory under int4 quantization. Different root causes." + "reason": "Quantization offloading behavior and int4 reserved-memory OOM are related broadly, but not the same concrete bug." + }, + { + "left": "issue:44079", + "right": "issue:46829", + "accept": false, + "reason": "ModelOutput key assignment bug is unrelated to VideoPrism output-field wiring." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert and AudioFlamingo3 batched inference leakage are unrelated." + "reason": "Different issues: Qwen3 auto device mapping crash vs AudioFlamingo3 batched inference leakage." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "SigLIP2 docs/example bugs are unrelated to GLM5 model-loading work." + "reason": "SigLIP2 docs/model mismatch is unrelated to a GLM5 issue." }, { - "left": "issue:42947", - "right": "issue:45704", + "left": "issue:45663", + "right": "issue:45759", "accept": false, - "reason": "Gradient checkpointing with PEFT LoRA is unrelated to the T5 apex.FusedRMSNorm memory leak." + "reason": "Different model families and different failures: FSDP2 attention state vs AutoModelForCausalLM config unwrapping." }, { - "left": "issue:45663", - "right": "issue:45759", + "left": "issue:42947", + "right": "issue:45704", "accept": false, - "reason": "Gemma-4 FSDP2 KeyError and Qwen composite config unwrapping are different model/infrastructure bugs." + "reason": "Different bugs: PEFT gradient checkpointing vs T5 using apex.FusedRMSNorm memory leak." }, { "left": "issue:43404", "right": "issue:43454", "accept": false, - "reason": "Both mention lm_head tying, but they affect different model families and likely different implementations; not enough to treat as one bug." + "reason": "Very similar symptom, but different model classes; not enough evidence they share one concrete code path." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "init_empty_weights loading failure is unrelated to disabling mmap on Strix Halo for OOM." + "reason": "init_empty_weights argument handling is unrelated to mmap OOM on Strix Halo." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "Importing GenerationMixin and Whisper pipeline return_language are unrelated issues." + "reason": "ImportError for GenerationMixin is unrelated to Whisper pipeline `return_language` behavior." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both are OOM-related, but the former is quantization memory reservation and the latter is mmap on Strix Halo. Different underlying problems." + "reason": "Different resource issues: int4 quantization reserved memory vs mmap-induced OOM on Strix Halo." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted 45276/45335 pair is plausibly the same resize_token_embeddings propagation bug across closely related model implementations, and the other soft-edge rejections do not overclaim shared root cause. No mergeability issue stands out." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -2447,7 +2455,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:39404", "issue:42796", @@ -2501,18 +2509,18 @@ "soft_pairs": [ "issue:44292|issue:44912", "issue:45663|issue:45859", - "issue:43957|issue:44756", "issue:42947|issue:43381", + "issue:43957|issue:44756", "issue:44155|issue:45823", "issue:43335|issue:45663", "issue:44863|issue:46468", "issue:45276|issue:45292", "issue:45685|issue:46468", "issue:44479|issue:44877", - "issue:44164|issue:44756", "issue:43381|issue:44928", - "issue:45588|issue:45663", + "issue:44164|issue:44756", "issue:45684|issue:45698", + "issue:45588|issue:45663", "issue:43854|issue:45216", "issue:42796|issue:46326", "issue:45292|issue:45335", @@ -2534,136 +2542,136 @@ "issue:44589|issue:45084", "issue:44479|issue:44805" ], - "prepared_review_unit_hash": "f1b2ac615e86ba54d409dd59f9f6d1a291e2375e124748d06dba441a591ef29d", + "prepared_review_unit_hash": "76afb22914413a2272209d49bc88157d153d4d8081d2880c74721d140eed64cf", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37875, - "estimated_input_tokens": 9469, - "estimated_eval_tokens": 19194 + "serialized_chars": 37872, + "estimated_input_tokens": 9468, + "estimated_eval_tokens": 19192 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous cluster of unrelated tokenizer, model-loading, training, and quantization regressions. The only clear duplicate-like subgroup is the resize_token_embeddings/output-embeddings reports; the rest look distinct.", - "confidence": 0.84, - "canonical_issue_reason": "issue:45292 is the broadest, most central report in the resize_token_embeddings subgroup: it describes the shared symptom of output embeddings not updating, which the more specific Gemma4/t5gemma reports appear to instantiate.", + "summary": "This cluster is mostly a grab bag of unrelated regressions across tokenizers, custom-model loading, quantization, device placement, and training behavior. The only clearly duplicate-looking edges are the resize_token_embeddings issues, which appear to be the same propagation bug in different model variants.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45698 is the broadest and most central custom-model loading regression in the set, with the most discussion and inbound references, and it overlaps with several nearby save/from_pretrained issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45292 is the best single issue to anchor the duplicate subgroup because it states the core API regression in generic terms rather than a model-specific variant.", + "best_issue_reason": "issue:45698 is the best anchor for triage because it describes a concrete, user-facing from_pretrained/save_pretrained failure and has the strongest evidence of ongoing impact.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Both are quantization-related, but one is an NVFP4 runtime error and the other is an MXFP4 load fallback; different failure modes and likely different fixes." + "reason": "Same general quantization/loading area, but the models and failure modes differ; not the same underlying bug." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Both involve model internals, but one is an FSDP2 training KeyError in Gemma4 attention and the other is a CPU-offload rotary embedding incompatibility in Qwen3_5Moe; not the same bug." + "reason": "Different models and different code paths: FSDP attention state rebuild vs CPU offload rotary embedding compatibility." }, { - "left": "issue:43957", - "right": "issue:44756", + "left": "issue:42947", + "right": "issue:43381", "accept": false, - "reason": "Meta-device loading breakage and Strix Halo mmap OOM are unrelated root causes and fixes." + "reason": "Both mention gradient checkpointing, but one is about ineffective checkpointing with LoRA and the other about eval-mode restrictions." }, { - "left": "issue:42947", - "right": "issue:43381", + "left": "issue:43957", + "right": "issue:44756", "accept": false, - "reason": "Both mention gradient checkpointing, but one is ineffective with PEFT LoRA while the other forbids eval mode; different concrete issues." + "reason": "Meta-device loading and mmap OOM are separate issues with different triggers and fixes." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "AudioFlamingo3 batched inference leakage and Gemma4 PLE device mismatch are unrelated model-specific bugs." + "reason": "AudioFlamingo3 batched inference leak and Gemma4 device mismatch are unrelated failures." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "Sparse-layer construction in SwitchTransformersConfig is unrelated to Gemma4 FSDP2 shared_kv_states rebuilding." + "reason": "Sparse-layer construction bug and Gemma4 FSDP attention KeyError are not the same bug." }, { "left": "issue:44863", "right": "issue:46468", "accept": false, - "reason": "NemotronH checkpoint loading and a broken official Nemotron example may be related in the same family, but the titles do not establish the same concrete defect." + "reason": "Both mention Nemotron, but one is checkpoint loading and the other is a broken example; no shared concrete failure." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Both report the same resize_token_embeddings regression: token embedding/output embedding tables are not updated as expected." + "reason": "Both describe the same resize_token_embeddings propagation bug affecting output embeddings; one is a Gemma4-specific manifestation." }, { "left": "issue:45685", "right": "issue:46468", "accept": false, - "reason": "MPS histogram-on-Int error in MoE code is unrelated to Nemotron example breakage." + "reason": "MPS histogram/int error and a broken Nemotron example are unrelated." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Video-input regression for Qwen VL models and strict-config Granite Speech loading are different problems." - }, - { - "left": "issue:44164", - "right": "issue:44756", - "accept": false, - "reason": "extra_state save/load handling and Strix Halo mmap OOM are unrelated." + "reason": "Video-input regression for Qwen VL models and strict granite_speech config loading are different issues." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction is unrelated to the RLHF NaN/SDPA fallback issue." + "reason": "Eval-mode gradient checkpointing restriction is unrelated to the RLHF NaN explosion from 3D position_ids/SDPA fallback." }, { - "left": "issue:45588", - "right": "issue:45663", + "left": "issue:44164", + "right": "issue:44756", "accept": false, - "reason": "flash_attention.py crash on s_aux=None is a distinct attention-path bug from Gemma4 FSDP2 KeyError." + "reason": "extra_state save/load handling and Strix Halo mmap OOM are separate problems." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Custom-model source-file permissions during save_pretrained and wrong custom-module loading after save_pretrained are separate bugs." + "reason": "Both involve custom model save/load, but one is about read-only file permissions and the other is about loading the wrong module." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash attention sink-less crash and Gemma4 FSDP KeyError are distinct attention-path bugs." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "GLM-4.7-Flash test loading failure and Qwen3.5 save_pretrained regression are different save/load issues." + "reason": "GLM-4.7-Flash loading failure and Qwen3.5 save_pretrained regression are not the same underlying issue." }, { "left": "issue:42796", "right": "issue:46326", "accept": false, - "reason": "Tokenizer BPE-marker handling and mistral_common continue_final_message behavior are different tokenizer paths." + "reason": "Tokenizer BPE-marker handling and continue_final_message behavior are different tokenizer bugs." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Both are the same resize_token_embeddings propagation bug, reported against different model variants and output embedding locations." + "reason": "Both are the same resize_token_embeddings bug: the embedding/output weights are not updated consistently across model variants." }, { "left": "issue:43828", @@ -2675,103 +2683,103 @@ "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom-model initialization breakage in notebooks and Trainer max_steps miscalculation on checkpoint resume are different subsystems." + "reason": "Custom model initialization in notebooks and Trainer resume max_steps miscalculation are separate issues." }, { "left": "issue:44402", "right": "issue:44451", "accept": false, - "reason": "Tokenizer vocab-size mismatch and ScandiBERT loading failure are not the same defect." + "reason": "Tokenizer vocab-size mismatch for one model and inability to load another model are too weakly connected to treat as the same bug." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Notebook custom model initialization and extra_state save/from_pretrained handling are distinct issues." + "reason": "Notebook custom model initialization and extra_state save/load handling do not look like one bug." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage lookup failure and mmap OOM on Strix Halo are unrelated." + "reason": "Missing Float8 storage and mmap OOM are unrelated runtime/environment failures." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch 2.10 incompatibility and autocast dtype mismatch are different compatibility failures." + "reason": "Torch 2.10 incompatibility is broader and not the same as the specific autocast dtype mismatch." }, { "left": "issue:45812", "right": "issue:46315", "accept": false, - "reason": "Granite tokenizer ID regression and slow convert_tokens_to_ids are different tokenizer problems." + "reason": "One is correctness (wrong token IDs), the other is performance (slow convert_tokens_to_ids); not the same bug." }, { "left": "issue:46315", "right": "issue:46491", "accept": false, - "reason": "Tokenizer performance regression and CodeLlama leading-space round-trip regression are unrelated." + "reason": "Tokenizer performance regression and CodeLlama leading-space round-trip regression are different issues." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper return_language pipeline regression and generic tokenization-version changes are not the same bug." + "reason": "Whisper pipeline return_language regression and tokenization changes are not the same underlying bug." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both involve custom models under Transformers 5.0.0, but one is base_model_prefix loading and the other is notebook initialization; not one concrete defect." + "reason": "Both are custom-model related, but base_model_prefix loading and notebook initialization failures are different failure modes." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "PagedAttentionCache linear_attention group-type crash and non-template-node compile error are unrelated." + "reason": "PagedAttentionCache linear_attention crash and non-template-node compilation error are unrelated." }, { "left": "issue:39404", "right": "issue:46665", "accept": false, - "reason": "Whisper pipeline return_language regression and from_pretrained-in-class inconsistency are different API behaviors." + "reason": "Whisper pipeline behavior and from_pretrained inside Python classes are not the same issue." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and Gemma4 PLE device mismatch are different bugs." + "reason": "Compiled flex_attention torch-version failure and Gemma4 device mismatch are separate code paths." }, { "left": "issue:43425", "right": "issue:45823", "accept": false, - "reason": "Torch version incompatibility and Gemma4 device mismatch are unrelated." + "reason": "Torch 2.10 incompatibility is not the same as a Gemma4 device-map forward mismatch." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch compatibility and Mistral-3 image preprocessor dtype mismatch are not the same issue." + "reason": "General torch-version incompatibility and a model/preprocessor dtype mismatch are unrelated." }, { "left": "issue:44589", "right": "issue:45084", "accept": false, - "reason": "Float8 storage lookup failure and template-node compile failure are distinct." + "reason": "Float8 storage lookup failure and compilation of non-template nodes are different problems." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Qwen VL video-input regression and mask/tensor shape IndexError are unrelated." + "reason": "Qwen video regression and mask/tensor shape IndexError do not appear to be the same bug." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it identifies one clearly supported duplicate-like subgroup around resize_token_embeddings/output_embeddings and treats the other pairs as distinct rather than over-merging them. The accepted soft-edge verdicts for issues 45276/45292 and 45292/45335 are consistent with the issue titles and evidence provided." + "feedback": "The summary is grounded in the packet and stays conservative overall. The anchor choice for issue:45698 is reasonable given its discussion activity and direct save/from_pretrained scope, and the only accepted soft-edge pairs are the resize_token_embeddings cases, which do look plausibly like the same underlying bug across model variants. The rejected pairs are generally justified as distinct failure modes." }, "evaluator_used": true, "retried": false, @@ -2790,7 +2798,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:32090", "issue:36032", @@ -2844,9 +2852,9 @@ "soft_pairs": [ "issue:43906|issue:44530", "issue:43901|issue:44509", - "issue:43425|issue:44918", "issue:43994|issue:44079", "issue:44671|issue:44991", + "issue:43425|issue:44918", "issue:44265|issue:44898", "issue:36032|issue:43452", "issue:39404|issue:44912", @@ -2866,228 +2874,228 @@ "issue:43716|issue:45084", "issue:36032|issue:42915", "issue:46285|issue:46402", - "issue:44737|issue:45859", "issue:32090|issue:39404", + "issue:44737|issue:45859", "issue:44661|issue:44683", "issue:43643|issue:44530", "issue:43541|issue:45084", - "issue:44451|issue:44987", - "issue:42915|issue:44530" + "issue:42915|issue:44530", + "issue:44451|issue:44987" ], - "prepared_review_unit_hash": "0c970c38fdf0e4561df43d48731fbc0ba50857d85da742ec09d01fe7f0ab043f", + "prepared_review_unit_hash": "8494029e0a0d359a7e5c3ff22d812a517dc2a7afa589fb4fce1f1c3fa3a3aac7", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37073, - "estimated_input_tokens": 9269, - "estimated_eval_tokens": 18794 + "serialized_chars": 37071, + "estimated_input_tokens": 9268, + "estimated_eval_tokens": 18792 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a very heterogeneous cluster of Transformers issues: model/tokenizer loading regressions, docs mismatches, quantization/training problems, and unrelated export/compile bugs. The soft edges are mostly semantic false positives; I did not find a convincing duplicate pair.", + "summary": "The items are heterogeneous: tokenizer/load regressions, model-shape/config bugs, compilation/tracing failures, docs issues, and resource/OOM reports. None of the soft pairs look like the same underlying bug or a PR-worthy merge target, so I reject all soft edges.", "confidence": 0.94, - "canonical_issue_reason": "issue:43611 is the broadest representative of the cluster\u2019s dominant theme: Transformers v5 loading regressions. It is still only an umbrella-like fit, not a true duplicate of most other items.", + "canonical_issue_reason": "No single issue cleanly represents this set; the entries span several unrelated regressions. If forced to pick a broad loading-regression umbrella, 43611 is the closest, but it does not subsume most of the others.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the best single representative because it captures the recurring model-loading breakage pattern seen in several nearby reports, even though the cluster overall is not a single bug.", + "best_issue_reason": "There isn\u2019t a strong global issue representative here; 43611 is the broadest loading-regression candidate, but it is still not a safe canonical for the whole collection.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43906", "right": "issue:44530", "accept": false, - "reason": "Different failures: isolated reproduction of an older issue vs Qwen3.5 PagedAttentionCache group-type crash." + "reason": "Different problems: a reproduction of a prior bug vs a Qwen3.5 paged-attention cache crash." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but they concern different pipeline docs changes and different regressions." - }, - { - "left": "issue:43425", - "right": "issue:44918", - "accept": false, - "reason": "Unrelated problems: Torch 2.10 compatibility vs TRL SFT trainer unpacking embeddings." + "reason": "Both are docs-related, but they cover different pipeline docs regressions and different removed/changed tasks." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "Different bugs: SigLIP2 producing bad outputs vs ModelOutput key assignment when a prior key is None." + "reason": "Different bug classes: bad model/pipeline outputs vs ModelOutput key assignment when a key was previously None." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "Both affect tokenization/model loading in v5, but they are different models and different failure modes." + "reason": "Different models and root causes: CamemBERT masked-LM regression vs tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility is not the same as a TRL trainer embedding-unpack failure." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export with torch_compilable_check is unrelated to Perceiver non-default-resolution image classification." + "reason": "Different code paths: torch.export/export with torch_compilable_check vs Perceiver interpolation image-classification failure." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "T5 tokenizer init conflict is unrelated to gguf_file handling in from_pretrained." + "reason": "T5 tokenizer method conflict is unrelated to gguf_file handling in AutoTokenizer/AutoModel loading." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback." + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback behavior." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "save/from_pretrained extra_state handling is a different serialization bug than text_config unwrapping." + "reason": "save/from_pretrained extra_state handling is a different issue from from_config text_config unwrapping." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "Quantized model save/load corruption is unrelated to continue_final_message tokenizer behavior." + "reason": "Quantized model save/load breakage is unrelated to mistral_common continue_final_message behavior." }, { "left": "issue:43906", "right": "issue:46326", "accept": false, - "reason": "No shared underlying bug; one is a reproduction of an older issue, the other is mistral_common final-message handling." + "reason": "A reproduction issue and a mistral_common tokenizer bug are not the same underlying defect." }, { "left": "issue:43299", "right": "issue:43931", "accept": false, - "reason": "Both involve Qwen3-VL loading, but one is a generic dev0 regression and the other is a specific weight-shape mismatch; not enough to treat as one bug." + "reason": "Both involve Qwen3-VL MoE loading, but one is a version regression and the other is a weight-shape mismatch; root causes are not clearly the same." }, { "left": "issue:44991", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer problems on different models; one is est-roberta loading, the other is Mistral regex patching." + "reason": "Different tokenizer failures on different model families; no shared concrete bug path." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Qwen3 MoE training memory usage is unrelated to disabling mmap on Strix Halo." + "reason": "Qwen3 MoE training memory inefficiency is not the same as disabling mmap on Strix Halo to avoid OOM." }, { "left": "issue:43596", "right": "issue:44683", "accept": false, - "reason": "Deepspeed ZeRO3/BertModel index error is unrelated to compiled flex_attention on newer torch." + "reason": "Zero3/BertModel indexing failure is unrelated to compiled flex_attention on newer torch." }, { "left": "issue:44164", "right": "issue:46821", "accept": false, - "reason": "extra_state serialization and TimesFM window_size AttributeError are different code paths." + "reason": "extra_state save/load handling does not match a TimesFM window_size AttributeError." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Different tokenizer loading bugs with different root causes and models." + "reason": "Loading sleng-bert vs mistral regex patching are different tokenizer bugs." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Both mention loading failures, but sleng-bert tokenizer loading and fast model loading on transformers>=5.1.0 are not the same defect." + "reason": "Different model loading failures with different symptoms and likely different fixes." }, { "left": "issue:43611", "right": "issue:43644", "accept": false, - "reason": "Both are v5 regressions, but one is base_model_prefix loading and the other is non-persistent buffer initialization." + "reason": "base_model_prefix loading breakage is not the same as non-persistent buffers being filled with junk." }, { "left": "issue:44462", "right": "issue:46489", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json is different from DeepSeek-Coder tokenizer output being incorrect after a prior fix gap." + "reason": "Ignoring tokenizer.json from repo is distinct from DeepSeek-Coder v1 producing incorrect output after a prior fix gap." }, { "left": "issue:42796", "right": "issue:43404", "accept": false, - "reason": "Ministral BPE-marker handling and Mistral3 lm_head tying are unrelated model-specific bugs." + "reason": "Ministral BPE-marker handling and Mistral3 lm_head weight tying are different model/tokenizer bugs." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Mistral image preprocessor dtype mismatch is unrelated to a template-node compilation error." + "reason": "Image preprocessor/model dtype mismatch is unrelated to a template-node compilation error." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "T5 tokenizer method conflict is unrelated to Qwen3 MoE FineGrainedFP8Config loading failure." + "reason": "T5 tokenizer method conflict and Qwen3Moe FineGrainedFP8Config failure are unrelated." }, { "left": "issue:46285", "right": "issue:46402", "accept": false, - "reason": "CLIP text-model structure drift is unrelated to Sam3 text encoder weights not loading." + "reason": "CLIP structure inconsistency is not the same as Sam3Model text encoder weights missing on load." }, { - "left": "issue:44737", - "right": "issue:45859", + "left": "issue:32090", + "right": "issue:39404", "accept": false, - "reason": "XLNet CPU device placement bug is unrelated to Qwen3_5Moe rotary embedding CPU offload compatibility." + "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper pipeline return_language behavior." }, { - "left": "issue:32090", - "right": "issue:39404", + "left": "issue:44737", + "right": "issue:45859", "accept": false, - "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper pipeline return_language regression." + "reason": "Both mention device/offload concerns, but they are different code paths and different failures." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "Tokenizer mapping/add-new-model-like failure is unrelated to flex_attention compilation on torch>=2.9." + "reason": "Model-addition mapping failure is unrelated to flex_attention compilation on torch>=2.9." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "trust_remote_code missing config fields and Qwen3.5 PagedAttentionCache crash are different bugs." + "reason": "trust_remote_code config field loss is not the same as a PagedAttentionCache linear_attention crash." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Mixtral grouped_mm tracing failure is unrelated to the template-node compilation error." + "reason": "Torch dynamo tracing of grouped_mm in Mixtral is different from a non-template-node compile error." }, { - "left": "issue:44451", - "right": "issue:44987", + "left": "issue:42915", + "right": "issue:44530", "accept": false, - "reason": "Loading vesteinn/ScandiBERT and loading physical-intelligence/fast fail for different reasons." + "reason": "Qwen3Moe FP8Config failure and Qwen3.5 linear_attention cache crash are different bugs." }, { - "left": "issue:42915", - "right": "issue:44530", + "left": "issue:44451", + "right": "issue:44987", "accept": false, - "reason": "Qwen3 MoE FineGrainedFP8Config failure is unrelated to Qwen3.5 linear_attention group-type crash." + "reason": "Different model-loading failures for different repositories and no shared concrete code path." } ] }, @@ -3109,7 +3117,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28530", "issue:34567", @@ -3118,7 +3126,6 @@ "issue:40990", "issue:42175", "issue:42915", - "issue:43089", "issue:43262", "issue:43317", "issue:43377", @@ -3136,6 +3143,7 @@ "issue:43819", "issue:43883", "issue:43994", + "issue:44062", "issue:44206", "issue:44220", "issue:44291", @@ -3166,9 +3174,9 @@ "issue:43644|issue:43646", "issue:34567|issue:40444", "issue:44811|issue:45584", + "issue:43883|issue:44488", "issue:42175|issue:46178", "issue:44206|issue:44220", - "issue:43883|issue:44488", "issue:40444|issue:46326", "issue:36032|issue:43645", "issue:43452|issue:43482", @@ -3176,8 +3184,8 @@ "issue:45084|issue:45507", "issue:44987|issue:44991", "issue:43262|issue:43377", - "issue:43883|issue:44991", "issue:43746|issue:46326", + "issue:43883|issue:44991", "issue:42915|issue:45084", "issue:44403|issue:45081", "issue:44464|issue:45987", @@ -3188,249 +3196,246 @@ "issue:43404|issue:44466", "issue:43994|issue:44610", "issue:44987|issue:45081", - "issue:42915|issue:44291", "issue:44451|issue:45081", + "issue:42915|issue:44291", "issue:28530|issue:34567", "issue:36032|issue:43742", "issue:42915|issue:44589", - "issue:43089|issue:45987" + "issue:44062|issue:44987" ], - "prepared_review_unit_hash": "3435688660bcb8bbf4c107363c41c9035b66a58855dcf722c709af37cf7b86f4", + "prepared_review_unit_hash": "561f43a490aa8fe7a3f3e78ddee583e7808deb40e6e0dc2253af76f08263e0e8", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 33, - "serialized_chars": 37689, - "estimated_input_tokens": 9423, - "estimated_eval_tokens": 19102 + "serialized_chars": 37705, + "estimated_input_tokens": 9427, + "estimated_eval_tokens": 19110 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous cluster of unrelated Transformers issues, with one small v5 custom-model-initialization/buffer-handling subcluster centered on issue 43646. Most soft pairs are different bugs or different model/code paths.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 43646 is the broadest and most representative report in the only clear duplicate-like mini-cluster: a v5 regression around custom model initialization and buffer handling.", + "summary": "This cluster is mostly heterogeneous. It contains several unrelated transformers v5 regressions, tokenizer/model-loading failures, and audio/vision edge cases, but the soft pairs do not look like true duplicates of the same underlying bug.", + "confidence": 0.34, + "canonical_issue_reason": "issue:43632 is the broadest and most central v5 regression here, centered on broken model initialization/state handling, so it is the best umbrella representative among the issues.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43646 is the best umbrella issue because it captures the shared initialization regression without narrowing to a notebook-only or buffer-only symptom.", + "best_issue_reason": "issue:43632 is the clearest representative of the cluster\u2019s main theme: transformers v5 initialization/state regressions, rather than a narrow model- or tokenizer-specific failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "Early stopping metric lookup and runtime tokenizer post-processor updates are unrelated bugs." + "reason": "Unrelated symptoms: perplexity on gpt-oss vs runtime special-token/post-processor mutation." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and `_is_hf_initialized` loading regression affect different code paths." + "reason": "Different failures: Apple Silicon allocator warmup TypeError vs `_is_hf_initialized` flag regression." }, { "left": "issue:43644", "right": "issue:43646", - "accept": true, - "reason": "Both point to the same v5 custom-model-initialization regression; 43644 is a concrete buffer symptom of the broader issue described in 43646." + "accept": false, + "reason": "Both are v5 init-related, but one is junk non-persistent buffers and the other is custom model notebook initialization; not the same bug." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Token-count tracking and multi-image iterable fine-tuning failure are unrelated." + "reason": "Trainer token accounting bug vs Qwen2.5-VL iterable dataset with multiple images; different subsystems." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper decode special-token handling and empty-transcription generation failure are different bugs." + "reason": "Both touch Whisper/tokenization, but one ignores `skip_special_tokens` and the other fails on empty transcription after `align_special_tokens`." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Different model-loading problems: missing `all_tied_weights_keys` vs failing to load a specific BERT model." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Package/backend inclusion and backend-mapping validation are separate issues." + "reason": "Packaging/backends mismatch for `.[torch]` vs an unsupported backend mapping entry for `tensorflow_text`." }, { "left": "issue:44206", "right": "issue:44220", "accept": false, - "reason": "They describe different audio feature-extraction failures with different triggers." - }, - { - "left": "issue:43883", - "right": "issue:44488", - "accept": false, - "reason": "Missing tied-weights metadata and a specific model-loading failure are not the same bug." + "reason": "Related audio area, but different code paths and errors: unsupported `center` arg vs `_torch_extract_fbank_features()` issue." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "Multi-image iterable finetuning and mistral_common final-message handling are unrelated." + "reason": "Different multimodal/tokenizer bugs: iterable dataset multi-image prompt handling vs `continue_final_message` on mistral_common tokenizers." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "T5 tokenizer method conflicts and custom-model initialization in notebooks are different issues." + "reason": "Tokenizer method conflict at load time vs custom model initialization in Jupyter; not the same underlying issue." }, { "left": "issue:43452", "right": "issue:43482", "accept": false, - "reason": "Both involve GGUF loading, but one is a general `gguf_file` API break and the other is a specific model/version failure; not clearly the same bug." + "reason": "Both involve GGUF loading, but one is a general `gguf_file` API break and the other is a specific Qwen2.5-GGUF v5 load failure." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Dequantized model offload loading and GGUF loading failure are distinct load paths." + "reason": "Dequantized device-map offload failure is a different path from Qwen2.5-GGUF loading failure." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "A template compilation error and a GraniteMoEHybrid invalid-method call are unrelated." + "reason": "Completely different errors: template compilation TypeError vs invalid method call in GraniteMoEHybrid." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Different models and likely different tokenizer/loading regressions." + "reason": "Both are load regressions in v5, but for different models and likely different tokenizer/model code paths." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Default sampling-rate handling and padding-mask batching mismatch are separate audio bugs." + "reason": "Audio processor sampling-rate defaulting and MIMI batch padding-mask behavior are different bugs." }, { - "left": "issue:43883", - "right": "issue:44991", + "left": "issue:43746", + "right": "issue:46326", "accept": false, - "reason": "A tied-weights attribute error and a tokenizer loading failure are unrelated." + "reason": "PEFT/local-checkpoint loading for GraniteSpeech is unrelated to mistral_common `continue_final_message` handling." }, { - "left": "issue:43746", - "right": "issue:46326", + "left": "issue:43883", + "right": "issue:44991", "accept": false, - "reason": "PEFT-adapter checkpoint loading and mistral_common message continuation are unrelated." + "reason": "Different failures: tied-weights attribute missing vs tokenizer loading for EMBEDDIA/est-roberta." }, { "left": "issue:42915", "right": "issue:45084", "accept": false, - "reason": "Qwen3Moe FP8 config failure and a template compilation error are different problems." + "reason": "FineGrainedFP8Config failure and non-template node compilation error are unrelated." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "Noise/logging on load and Mistral regex patch crashes are not the same bug." + "reason": "Generic load noise vs a specific Mistral tokenizer regex patch crash; not the same bug." }, { "left": "issue:44464", "right": "issue:45987", "accept": false, - "reason": "Chunked-generation inconsistency and cache length return-type mismatch are different generation regressions." + "reason": "Different generation/cache issues: chunked compiled-forward inconsistency vs `get_seq_length()` returning the wrong type." }, { "left": "issue:43646", "right": "issue:44534", - "accept": true, - "reason": "These are the same v5 custom-model-init / non-persistent-buffer corruption regression stated at different levels of specificity." + "accept": false, + "reason": "Custom model init failure and non-persistent buffer junk are different regressions, even if both are v5-related." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Notebook-specific custom-model initialization and buffer junk are related in theme but not clearly the same concrete bug." + "reason": "Notebook custom model initialization and non-persistent buffer serialization are distinct issues." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Qwen3.5 hidden-states behavior and Qwen2.5-VL rope-index scaling are different model-specific regressions." + "reason": "Qwen3.5 hidden-state output bug vs Qwen2.5-VL rope/position-id scaling bug; different models and paths." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "MIMI batching/padding-mask behavior and DAC latent conversion mismatch are different audio-model bugs." + "reason": "Batched-vs-single padding-mask mismatch in MIMI is unrelated to DAC from_latents / missing STE behavior." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both mention tied weights, but one is a model-specific tie bug and the other is device-dependent serialization behavior; not clearly the same change." + "reason": "One is missing weight tying in a model class; the other is inconsistent serialization of tied weights. Related theme, but not the same concrete bug." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Nonsensical model outputs and wrong processor input size are unrelated." + "reason": "Siglip2 nonsensical outputs and OmDet-Turbo 640px-vs-224px preprocessing mismatch are different issues." }, { "left": "issue:44987", "right": "issue:45081", "accept": false, - "reason": "A model load failure and a Mistral regex patch crash are different issues." + "reason": "Different tokenizer/model load failures: physical-intelligence/fast vs Mistral regex patch crash." }, { - "left": "issue:42915", - "right": "issue:44291", + "left": "issue:44451", + "right": "issue:45081", "accept": false, - "reason": "Fine-grained FP8 config failure and `_is_hf_initialized` init_empty_weights argument error are distinct regressions." + "reason": "Model loading failure for ScandiBERT is unrelated to Mistral tokenizer regex patching." }, { - "left": "issue:44451", - "right": "issue:45081", + "left": "issue:42915", + "right": "issue:44291", "accept": false, - "reason": "A model load failure for ScandiBERT and a tokenizer regex-patch crash are unrelated." + "reason": "FP8 config failure and `_is_hf_initialized` argument regression are separate v5 breakages." }, { "left": "issue:28530", "right": "issue:34567", "accept": false, - "reason": "Early stopping metric selection and token-count tracking are unrelated Trainer bugs." + "reason": "Early stopping/eval metric mismatch vs trainer token counter not updating; different trainer features." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "Tokenizer special-token conflicts and a key error while loading MobileLLM-125M are different loading issues." + "reason": "Tokenizer class method collision and MobileLLM key error are unrelated loading failures." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "FP8 config failure and missing Float8 storage are both quantization-adjacent, but not the same concrete bug." + "reason": "FP8 configuration failure and float8 storage lookup failure are different low-level regressions." }, { - "left": "issue:43089", - "right": "issue:45987", + "left": "issue:44062", + "right": "issue:44987", "accept": false, - "reason": "Generation performance overhead and cache length contract mismatch are different problems." + "reason": "AddedToken keyword collision vs loading failure for a specific model; not the same code-path." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays mostly conservative. The identified mini-cluster around 43644/43645/43646/44534 is plausible, and the other pair rejections are generally cautious rather than overstated. I don\u2019t see a clear precision error strong enough to reject." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -3445,7 +3450,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:33666", "issue:35707", @@ -3455,6 +3460,7 @@ "issue:41553", "issue:41720", "issue:42796", + "issue:43089", "issue:43296", "issue:43322", "issue:43381", @@ -3468,7 +3474,6 @@ "issue:43632", "issue:43746", "issue:43792", - "issue:44062", "issue:44079", "issue:44220", "issue:44265", @@ -3482,7 +3487,6 @@ "issue:44756", "issue:44811", "issue:44877", - "issue:44987", "issue:44991", "issue:45216", "issue:45325", @@ -3491,17 +3495,18 @@ "issue:45663", "issue:45812", "issue:45923", + "issue:45987", "issue:46315", "issue:46491", "issue:46710", "issue:46821" ], "soft_pairs": [ - "issue:43632|issue:46821", - "issue:44062|issue:44987", "issue:44265|issue:44610", - "issue:43322|issue:43540", + "issue:43089|issue:45987", + "issue:43632|issue:46821", "issue:44479|issue:45325", + "issue:43322|issue:43540", "issue:43531|issue:44877", "issue:36032|issue:41720", "issue:45216|issue:45357", @@ -3512,8 +3517,8 @@ "issue:43611|issue:44534", "issue:40444|issue:45325", "issue:45812|issue:46491", - "issue:43632|issue:44756", "issue:44403|issue:44991", + "issue:43632|issue:44756", "issue:43792|issue:44589", "issue:43381|issue:45923", "issue:42796|issue:43746", @@ -3521,228 +3526,231 @@ "issue:33666|issue:40444", "issue:40990|issue:44811", "issue:42796|issue:46710", - "issue:44079|issue:45663", "issue:43482|issue:43525", "issue:35707|issue:40444", - "issue:43482|issue:45538", + "issue:44079|issue:45663", "issue:41553|issue:43582", + "issue:43482|issue:45538", "issue:43632|issue:44534" ], - "prepared_review_unit_hash": "9c2d2f270fcc3adb33927c1fec462784ba09224529135c0b509139cf2f14d6df", + "prepared_review_unit_hash": "bdf952b5e2ebff877b7c2d689015adb079ed7267d6df85b535a484ae05f28ad9", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36497, - "estimated_input_tokens": 9125, - "estimated_eval_tokens": 18506 + "serialized_chars": 36501, + "estimated_input_tokens": 9126, + "estimated_eval_tokens": 18508 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a grab bag of unrelated issue reports across tokenizers, multimodal models, checkpointing, and version-compatibility regressions. I don\u2019t see a single canonical issue/PR, and none of the soft edges are convincing duplicates under a strict triage standard.", - "confidence": 0.96, - "canonical_issue_reason": "No clear canonical issue: the items span many unrelated bugs and regressions rather than one underlying problem.", + "summary": "These items are mostly unrelated issues. The only clear duplicate-like pair is the Qwen3.5 save_pretrained regression/follow-up; the rest share at most a broad subsystem or model family, which is not enough to merge.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45216 is the broader, earlier regression report for the Qwen3.5 save_pretrained bug, while issue:45357 looks like a narrower follow-up about one concrete manifestation (visual encoder keys).", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a strong representative because the cluster is heterogeneous; any pick would be arbitrary.", + "best_issue_reason": "issue:45216 is the strongest cluster anchor because it is the broad upstream regression report and the only issue with a clearly related follow-up in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43632", - "right": "issue:46821", + "left": "issue:44265", + "right": "issue:44610", "accept": false, - "reason": "Different regressions in different code paths: `_is_hf_initialized` loading vs TimesFM `window_size` attribute handling." + "reason": "Different bugs: torch.export/torch_compilable_check vs OmDet-Turbo processor image-size mismatch." }, { - "left": "issue:44062", - "right": "issue:44987", + "left": "issue:43089", + "right": "issue:45987", "accept": false, - "reason": "Both involve tokenizers, but one is an `AddedToken` kwarg conflict and the other is tokenizer loading for a specific model." + "reason": "Both touch generation/cache behavior, but one is a performance overhead report and the other is a cache-length contract bug." }, { - "left": "issue:44265", - "right": "issue:44610", + "left": "issue:43632", + "right": "issue:46821", "accept": false, - "reason": "`torch.export` failure for `torch_compilable_check` is unrelated to OmDet-Turbo processor image-size mismatch." + "reason": "Unrelated: a Transformers v5 init flag regression vs a TimesFM window_size AttributeError." }, { - "left": "issue:43322", - "right": "issue:43540", + "left": "issue:44479", + "right": "issue:45325", "accept": false, - "reason": "Both are multimodal model issues, but one is a Llava Next load crash and the other is a Qwen3OmniMoe video-processing `ValueError`." + "reason": "Both involve Qwen VL media handling, but one is a broad video regression and the other is a specific still-image position_id scaling bug." }, { - "left": "issue:44479", - "right": "issue:45325", + "left": "issue:43322", + "right": "issue:43540", "accept": false, - "reason": "Same broader Qwen-VL area, but video-input regression and still-image rope-index scaling are not the same concrete bug." + "reason": "Different models and failure modes: Llava Next loading segfault vs Qwen3OmniMoe video ValueError." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "`sliding_window` in Qwen3-MoE and strict Granite Speech config loading are unrelated problems." + "reason": "No same code path: Qwen3-MoE sliding_window behavior vs granite_speech config strictness." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer method-name conflict is unrelated to the Qwen3 auto device-mapping CUDA assert." + "reason": "Tokenizer method conflict vs Qwen3 auto device mapping CUDA assert; no shared bug." }, { "left": "issue:45216", "right": "issue:45357", - "accept": false, - "reason": "Both concern Qwen3.5 save_pretrained, but one is a broad checkpoint regression and the other is a specific lingering visual-encoder key issue." + "accept": true, + "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a narrower follow-up about incorrect visual encoder keys." }, { "left": "issue:44991", "right": "issue:46315", "accept": false, - "reason": "Tokenizer loading failure and `convert_tokens_to_ids` performance regression are different bugs." + "reason": "Both are tokenizer-related, but one is a load failure for a specific model and the other is a performance regression in convert_tokens_to_ids." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch 2.10 incompatibility is unrelated to the loading-noise issue." + "reason": "Different topics: Torch 2.10 compatibility vs noisy loading output." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL loading failure and Qwen3-MoE `sliding_window` behavior do not share the same code path." + "reason": "Different model families and bugs: PaddleOCR-VL loading in vLLM vs Qwen3-MoE sliding_window." }, { "left": "issue:44220", "right": "issue:44683", "accept": false, - "reason": "Fbank feature extraction and compiled flex_attention failures are separate subsystems." + "reason": "Unrelated audio feature extraction bug vs compiled flex_attention failure on newer torch." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "`base_model_prefix` loading breakage and non-persistent-buffer junk filling are different v5 regressions." + "reason": "Different regressions: base_model_prefix loading vs non-persistent buffer corruption." }, { "left": "issue:40444", "right": "issue:45325", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is not the same as Qwen2.5-VL still-image temporal position scaling." + "reason": "Both involve Qwen2.5-VL, but one is IterableDataset/multi-image finetuning and the other is rope/position_id scaling." }, { "left": "issue:45812", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer correctness issues, but Granite token IDs and CodeLlama space stripping are distinct bugs." + "reason": "Different tokenizers and symptoms: Granite token IDs regression vs CodeLlama leading-space round-trip issue." }, { - "left": "issue:43632", - "right": "issue:44756", + "left": "issue:44403", + "right": "issue:44991", "accept": false, - "reason": "`_is_hf_initialized` loading failure is unrelated to Strix Halo mmap OOM behavior." + "reason": "Noise during loading is not the same bug as EMBEDDIA/est-roberta tokenizer loading failure." }, { - "left": "issue:44403", - "right": "issue:44991", + "left": "issue:43632", + "right": "issue:44756", "accept": false, - "reason": "Loading-noise behavior and tokenizer-loading failure are not the same bug." + "reason": "Different problem domains: HF init flag regression vs Strix Halo mmap OOM workaround." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper model runtime failure is unrelated to the Float8 storage lookup error." + "reason": "Whisper model loading/runtime issue vs Float8 storage lookup error; no concrete shared code path." }, { "left": "issue:43381", "right": "issue:45923", "accept": false, - "reason": "Eval-mode gradient checkpointing restriction and missing `supports_gradient_checkpointing` flag are different issues." + "reason": "Related to gradient checkpointing in name only: eval-mode restriction vs missing supports_gradient_checkpointing flag." }, { "left": "issue:42796", "right": "issue:43746", "accept": false, - "reason": "Tokenizer BPE-marker handling and GraniteSpeech PEFT checkpoint loading are unrelated." + "reason": "Tokenizer BPE marker handling is unrelated to GraniteSpeech PEFT checkpoint loading." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Unexpected `_is_hf_initialized` init argument and missing Float8 storage object are separate regressions." + "reason": "init_empty_weights/_is_hf_initialized TypeError is a different failure than missing Float8 storage." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both mention Qwen VL training, but multi-GPU training and multi-image IterableDataset failure are different problems." + "reason": "Both are Qwen VL training issues, but one is multi-GPU training and the other is multi-image iterable dataset failure." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on GPT-OSS and Whisper `batch_decode` skipping specials are unrelated." + "reason": "Completely different subsystems: GPT-OSS perplexity vs Whisper batch_decode skip_special_tokens behavior." }, { "left": "issue:42796", "right": "issue:46710", "accept": false, - "reason": "Ministral tokenizer BPE markers and DeepSeek accuracy regression are not the same underlying bug." - }, - { - "left": "issue:44079", - "right": "issue:45663", - "accept": false, - "reason": "ModelOutput key assignment and Gemma-4 FSDP2 `shared_kv_states` rebuilding are unrelated." + "reason": "Tokenizer marker handling for Ministral is unrelated to DeepSeek accuracy regression." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Qwen2.5-GGUF loading and missing `pad_token_id` on Llama4Config are different compatibility errors." + "reason": "Different loading failures: Qwen2.5-GGUF with transformers v5 vs missing pad_token_id on Llama4Config." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with `inputs_embeds`/`past_key_values` is unrelated to multi-image finetuning failure." + "reason": "Different generation/training paths: progressive generation with inputs_embeds/past_key_values vs Qwen2.5-VL finetuning on multi-image datasets." }, { - "left": "issue:43482", - "right": "issue:45538", + "left": "issue:44079", + "right": "issue:45663", "accept": false, - "reason": "Qwen2.5-GGUF loading failure and CLIPTokenizer `model_max_length` default are unrelated tokenizer/model-loading issues." + "reason": "ModelOutput key assignment bug is unrelated to Gemma-4 FSDP2 shared_kv_states KeyError." }, { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "AutoTokenizer error messaging for Voxtral and AppleSilicon `caching_allocator_warmup` TypeError are unrelated." + "reason": "Bad AutoTokenizer error message vs Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Different tokenizer/model issues: Qwen2.5-GGUF loading vs CLIPTokenizer max length constant." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "`_is_hf_initialized` regression and junk-filled non-persistent buffers are separate v5 issues." + "reason": "Both are Transformers v5 regressions, but they affect different internals and symptoms." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted soft pair, issue:45216 and issue:45357, is plausibly a same-bug regression/follow-up based on their titles. The other rejections are generally well-justified by distinct models, subsystems, or failure modes. Minor caution: wording like \"upstream\" is a bit interpretive, but not enough to invalidate the cluster judgment." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -3757,7 +3765,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:40444", "issue:41720", @@ -3816,11 +3824,11 @@ "issue:41720|issue:45081", "issue:44079|issue:44534", "issue:44462|issue:45812", - "issue:44829|issue:45479", "issue:43296|issue:46821", + "issue:44829|issue:45479", "issue:43818|issue:46402", - "issue:43611|issue:43646", "issue:43377|issue:45491", + "issue:43611|issue:43646", "issue:42175|issue:43296", "issue:43644|issue:44079", "issue:43818|issue:45357", @@ -3833,220 +3841,220 @@ "issue:43454|issue:44466", "issue:42915|issue:43482", "issue:44291|issue:44877", - "issue:43873|issue:45701", "issue:43701|issue:45325", + "issue:43873|issue:45701", "issue:42796|issue:45381", - "issue:44479|issue:45823", - "issue:43606|issue:44265" + "issue:43606|issue:44265", + "issue:44479|issue:45823" ], - "prepared_review_unit_hash": "75433e0d57a4ae6022d6730c3c363edd72f808b7c41fd7ead32764f482990f1b", + "prepared_review_unit_hash": "0a31a62378b8194b8d57c1e7b8ec2dba3da1c5395e121af4673b4fbd0d8fb148", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36687, - "estimated_input_tokens": 9172, - "estimated_eval_tokens": 18600 + "serialized_chars": 36674, + "estimated_input_tokens": 9169, + "estimated_eval_tokens": 18594 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a heterogeneous set of unrelated Transformers bug reports across different models and subsystems (Qwen/VL, tokenizers, loading, offload, serialization). Only the two SigLIP2 reports look like the same underlying incorrect-output bug.", - "confidence": 0.46, - "canonical_issue_reason": "issue:43493 is the clearest root-cause-style report in the only convincing duplicate pair; it frames the SigLIP2 HF-vs-JAX discrepancy that likely explains the user-facing bad outputs in 43994.", + "summary": "Mostly a heterogeneous issue cluster with a few clear duplicate-style subpairs around Qwen visual position handling and SigLIP2 tokenizer/model behavior; the rest are different regressions in unrelated subsystems.", + "confidence": 0.66, + "canonical_issue_reason": "Issue 44479 is the broadest and most representative report of the Qwen2.5-VL/Qwen3* video-input regression family, capturing the same visual position-id bug line as the narrower Qwen visual-position issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43493 is the best representative issue here because it is concrete, diagnostic, and most likely to subsume the related SigLIP2 symptom report.", + "best_issue_reason": "Issue 44479 is the best single representative because it describes the regression most generally and across multiple affected Qwen model variants.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43317", "right": "issue:43856", "accept": false, - "reason": "Both mention Qwen3/MoE or offloading-related pain, but one is a device-mapping dequantization failure and the other is a training memory inefficiency; different bugs." + "reason": "Different problems: CPU/GPU offload loading failure vs MoE training memory usage." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to the init_empty_weights TypeError from the unexpected _is_hf_initialized argument." + "reason": "Different code paths: sliding_window/Qwen3-MoE behavior vs init_empty_weights/_is_hf_initialized loading error." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer-adjacent, but one is runtime post-processor refresh logic and the other is add_special_tokens/BOS-EOS insertion for a specific tokenizer." + "reason": "Both tokenizer-related, but one is runtime post-processor updates and the other is add_special_tokens/BOS-EOS handling; not the same bug." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Both involve TensorFlow packaging/backend setup, but they target different failure modes and install configurations." + "reason": "Both concern TF-related packaging, but the reported failures are different extras/dependency setups." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Different code paths: Qwen3 device-mapping CUDA assert vs Mistral regex patch crashing on tokenizer backend access." + "reason": "Unrelated failures: Qwen3 auto device-map CUDA assert vs Mistral regex patch AttributeError." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "ModelOutput key assignment and non-persistent buffer junk are separate Transformers v5 regressions with different symptoms and code paths." + "reason": "Different issues in different internals: ModelOutput key assignment vs non-persistent buffer initialization." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "Both are AutoTokenizer regressions, but one ignores repo tokenizer.json while the other produces wrong Granite token IDs; not the same bug." + "reason": "Tokenizer loading vs Granite token-ID regression; related area, but not clearly the same underlying bug." }, { - "left": "issue:44829", - "right": "issue:45479", + "left": "issue:43296", + "right": "issue:46821", "accept": false, - "reason": "Both describe degenerate training, but one is tied to flash_attention_3 and the other to problem_type=num_labels=1 classification behavior." + "reason": "Different models and failures: PaddleOCR-VL load in vLLM vs TimesFM window_size AttributeError." }, { - "left": "issue:43296", - "right": "issue:46821", + "left": "issue:44829", + "right": "issue:45479", "accept": false, - "reason": "Completely different models and failure modes: PaddleOCR-VL loading in vLLM vs TimesFM window_size AttributeError." + "reason": "Both cause degenerate training, but via distinct mechanisms and model families." }, { "left": "issue:43818", "right": "issue:46402", "accept": false, - "reason": "Different multimodal-model issues: Video-LLaVA tower/weights mismatch vs Sam3Model text encoder weights not loading." + "reason": "Different model bugs: Video-LLaVA tower/weight sharing vs Sam3Model text encoder loading." }, { - "left": "issue:43611", - "right": "issue:43646", + "left": "issue:43377", + "right": "issue:45491", "accept": false, - "reason": "Both are broad Transformers v5 loading regressions, but one is about base_model_prefix and the other custom model initialization; too different to merge." + "reason": "Both are batching/masking issues, but they affect different models and different attention/padding failure modes." }, { - "left": "issue:43377", - "right": "issue:45491", + "left": "issue:43611", + "right": "issue:43646", "accept": false, - "reason": "Both involve batched-input correctness, but MIMI missing padding-mask support is not the same as Gemma3 all-padding sliding-window NaNs." + "reason": "Both are Transformers v5 regressions, but one is about base_model_prefix loading and the other custom model initialization." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "TensorFlow backend packaging vs PaddleOCR-VL model loading are unrelated." + "reason": "Packaging/backend issue vs model-loading failure in vLLM; not the same bug." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Non-persistent buffer serialization and ModelOutput key assignment are distinct v5 regressions." + "reason": "Non-persistent buffer corruption vs ModelOutput key assignment bug." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Video-LLaVA architecture/weight-sharing bug is unrelated to Qwen3.5 save_pretrained key serialization." + "reason": "Different visual-model bugs: missing temporal attention in Video-LLaVA vs incorrect visual encoder keys in save_pretrained." }, { "left": "issue:43792", "right": "issue:43957", "accept": false, - "reason": "Whisper runtime failure and meta-device loading breakage are different bugs." + "reason": "Different loading failures: Whisper runtime issue vs torch.device('meta') loading breakage." }, { "left": "issue:44265", "right": "issue:45072", "accept": false, - "reason": "torch.export exportability and bfloat16 dtype mismatch are unrelated issues." + "reason": "Different problems: torch.export/torch_compilable_check vs bf16 dtype mismatch in inference." }, { "left": "issue:45325", "right": "issue:45381", - "accept": false, - "reason": "Both are Qwen2.5-VL position-id regressions, but one is still-image temporal scaling and the other is video-input vision_position_ids; not clearly the same underlying bug." + "accept": true, + "reason": "Same Qwen2.5-VL vision-position regression family in 5.3.0; both point to get_rope_index/position-id handling." }, { "left": "issue:43493", "right": "issue:43994", "accept": true, - "reason": "Same model family and same symptom class: SigLIP2 produces wrong/nonsensical outputs, with 43493 giving the diagnostic HF-vs-JAX discrepancy that likely explains 43994." + "reason": "Same SigLIP2 bug family: HF implementation disagrees with the original JAX behavior and yields bad outputs for the same model." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Qwen2.5-VL position-id logic is unrelated to Kosmos2.5 long-OCR index errors." + "reason": "Both are vision/ocr-adjacent, but one is Qwen2.5-VL position-id scaling and the other is a Kosmos2.5 long-input index error." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "IterableDataset multi-image finetuning failure is a different Qwen2.5-VL issue from video vision_position_ids." + "reason": "Both mention Qwen2.5-VL, but one is finetuning with multiple images per prompt and the other is a video-input position-id regression." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Both involve lm_head/tied-weights topics, but one is a missing tie causing garbage generation while the other is device-dependent serialization inconsistency." + "reason": "Related to lm_head/tied weights, but one is a weight-tying bug and the other is a serialization inconsistency." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Qwen3MoE FP8 config failure and Qwen2.5-GGUF loading in v5 are unrelated." + "reason": "Unrelated model-specific failures: Qwen3MoE FP8 config vs Qwen2.5-GGUF loading on v5." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "init_empty_weights argument handling and strict config loading are separate Transformers v5 compatibility problems." + "reason": "Different failures: init_empty_weights argument mismatch vs strict config rejecting granite_speech." }, { - "left": "issue:43873", - "right": "issue:45701", + "left": "issue:43701", + "right": "issue:45325", "accept": false, - "reason": "Quantization/offload behavior and tokenization-version changes are unrelated." + "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL position-id computation." }, { - "left": "issue:43701", - "right": "issue:45325", + "left": "issue:43873", + "right": "issue:45701", "accept": false, - "reason": "resume_from_checkpoint key mismatch is unrelated to Qwen2.5-VL rope/position-id bugs." + "reason": "Offloading/quantization issue vs tokenizer-version-dependent tokenization change." }, { "left": "issue:42796", "right": "issue:45381", "accept": false, - "reason": "Tokenizer BPE marker handling and Qwen2.5-VL video position ids are different bugs." + "reason": "Tokenizer BPE-marker handling is unrelated to Qwen2.5-VL vision position ids." }, { - "left": "issue:44479", - "right": "issue:45823", + "left": "issue:43606", + "right": "issue:44265", "accept": false, - "reason": "Qwen video-input regression and Gemma4 PLE device mismatch are unrelated model-specific failures." + "reason": "CPU-offload device mismatch vs torch.export/compilable-check export failure." }, { - "left": "issue:43606", - "right": "issue:44265", + "left": "issue:44479", + "right": "issue:45823", "accept": false, - "reason": "CPU-offload device mismatch and torch.export/torch_compilable_check export failure do not share the same code-path problem." + "reason": "Video-input regression in Qwen models vs Gemma4 PLE device-mapping mismatch; different symptoms and code paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet, and the soft-edge verdicts stay conservative overall. The only accepted pair, issue:43493 and issue:43994, is a plausible same-bug match based on the shared SigLIP2 incorrect-output symptom, while the other pairs are reasonably rejected as different failure modes." + "feedback": "The clustering summary is grounded in the packet and stays conservative. The two accepted soft pairs (Qwen position-id handling and SigLIP2 behavior) are supported by the issue titles, and the rejected pairs generally involve clearly different model families or failure modes. The choice of issue 44479 as the broad representative of the Qwen video/position-id regression family is reasonable from the titles provided." }, "evaluator_used": true, "retried": false, @@ -4065,7 +4073,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:34634", "issue:35707", @@ -4148,214 +4156,214 @@ "issue:43881|issue:44991", "issue:36331|issue:42617" ], - "prepared_review_unit_hash": "96a3f48ec315b569e6e76c574cf75b11e3c3c88caecb7feccb811e880b494d94", + "prepared_review_unit_hash": "d536694e56253d2139145f2b399c81fafb99f27718e9dc972441cb5be46f9f3a", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36799, - "estimated_input_tokens": 9200, - "estimated_eval_tokens": 18656 + "serialized_chars": 36810, + "estimated_input_tokens": 9203, + "estimated_eval_tokens": 18662 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues are only loosely related by broad themes like loading, offloading, tokenization, or generation, but they describe different models, code paths, and failure modes. I do not see any true duplicate pairs among the soft edges.", - "confidence": 0.95, - "canonical_issue_reason": "issue:43873 is the broadest and most central representative of the cluster\u2019s recurring loading/offloading/device-mismatch theme, with the strongest discussion/inbound activity. Even so, it is better treated as the umbrella issue than as a true duplicate match for the others.", + "summary": "The set is mostly heterogeneous issue reports across unrelated models and subsystems. I do not see a safe cluster-wide duplicate canonical; none of the soft-edge pairs look mergeable into one concrete fix.", + "confidence": 0.88, + "canonical_issue_reason": "No safe canonical issue for the whole set: the reports span unrelated bugs in tokenization, loading, device mapping, training, audio, and export.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 best captures the shared high-level theme of runtime offloading/loading regressions across the cluster, making it the most useful representative issue for triage.", + "best_issue_reason": "If forced to name a closest representative, issue 43873 is the broadest loading/offloading report, but it still does not represent the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "Both involve generation quality, but one is perplexity on gpt-oss and the other is Whisper empty-transcription handling after token alignment; different models and distinct bugs." + "reason": "Different models and bug surfaces: perplexity/eval behavior vs Whisper empty-transcription generation after token alignment." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio processor sampling-rate defaulting and GraniteSpeech PEFT local checkpoint loading are unrelated code paths and symptoms." + "reason": "Both touch audio, but one is a sampling-rate default bug and the other is a PEFT/local checkpoint loading failure." }, { "left": "issue:43828", "right": "issue:46821", "accept": false, - "reason": "Both mention dtype/argument issues in generation-like flows, but one is an autocast dtype mismatch in Phi MoE and the other is a TimesFM window_size AttributeError." + "reason": "Unrelated failures: autocast dtype mismatch in Phi-MoE vs an AttributeError for TimesFM window_size." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "glm-4v-9b loading failure and cjvt/sleng-bert tokenizer load failure are different model families and different failure mechanisms." + "reason": "Both are model-loading failures, but they involve different models and likely different root causes." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is a generation API bug; Qwen3 auto device mapping on A800 is a device-placement/runtime assert issue." + "reason": "Generation with inputs_embeds/past_key_values is unrelated to a device-map CUDA assert on Qwen3." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "BarkProcessor voice_preset and inputs_embeds/past_key_values are unrelated: audio processor preset handling vs generation cache/embed handling." + "reason": "Bark voice_preset handling and progressive generation with past key values are different code paths." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "AyaVision lm_head tying and Bamba torch.compile SDPA failures are different model implementation problems." + "reason": "Weight tying in AyaVisionConfig is unrelated to Bamba SDPA/torch.compile failures." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "CPU offload device mismatch for Bark-small and DAC.from_latents vs forward mismatch are separate bugs in different models." + "reason": "Both are audio-model issues, but one is CPU offload device mismatch and the other is DAC latent reconstruction mismatch." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 docs errors and Qwen2.5-VL iterable multi-image finetuning failure are documentation vs training/data-pipeline issues." + "reason": "Docs example errors for SigLIP2 do not match Qwen2.5-VL finetuning with multi-image iterable datasets." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crashes and torch.export.export with torch_compilable_check are unrelated tokenizer vs export/runtime issues." + "reason": "Tokenizer crash in LayoutLMv2 is not the same as torch.export failing on torch_compilable_check models." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Both are device_map-related, but Qwen3 auto-mapping cudaErrorAssert and Gemma4 PLE device mismatch are different models and different device bugs." + "reason": "Both involve device_map=auto, but the failures differ: CUDA assert on Qwen3 vs device mismatch during Gemma4 forward." }, { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "MIMI encoder padding-mask batching and OmDet-Turbo processor/model resolution mismatch are separate multimodal bugs." + "reason": "Missing padding-mask support in MIMI encoder is unrelated to OmDet-Turbo processor/model input-size mismatch." }, { "left": "issue:46305", "right": "issue:46315", "accept": false, - "reason": "MoonshineStreaming sliding window off-by-one and tokenizer convert_tokens_to_ids slowdown are unrelated streaming vs tokenizer performance issues." + "reason": "Streaming mask off-by-one and tokenizer convert_tokens_to_ids slowdown are different bugs." }, { "left": "issue:43257", "right": "issue:45923", "accept": false, - "reason": "Qwen3 MoE weights not converted during accelerate+deepspeed loading and Nemotron gradient-checkpointing flag missing are different loading vs config-flag bugs." + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed is unrelated to a missing supports_gradient_checkpointing flag." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "mdeberta add_special_tokens regression and AutoTokenizer wrong token IDs across multiple models are both tokenizer regressions but not the same underlying bug." + "reason": "Specific BOS/EOS handling in one tokenizer is not the same as broad AutoTokenizer ID regressions across many models." }, { "left": "issue:45823", "right": "issue:46821", "accept": false, - "reason": "Gemma4 PLE device mismatch during forward and TimesFM window_size AttributeError are unrelated model/runtime failures." + "reason": "Gemma4 device-mismatch forward failure and TimesFM window_size AttributeError are unrelated." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "AyaVision weight tying and dtype mismatches in SwitchTransformers/TimmWrapperModel are different failure classes." + "reason": "LM head weight tying and bfloat16 dtype mismatches in inference are different problems." }, { "left": "issue:42796", "right": "issue:46209", "accept": false, - "reason": "Ministral tokenizer BPE-marker handling and FP8 inference on B200 are unrelated tokenizer vs hardware-kernel issues." + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are unrelated." }, { "left": "issue:43317", "right": "issue:43873", "accept": false, - "reason": "Both concern offload/loading, but one is dequantized model loading with device_map=auto and the other is quantization/offloading behavior in general; different concrete code paths." + "reason": "They share offloading/quantization wording, but the repros and failure modes differ enough that they should not be merged." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "MIMI batching/padding-mask behavior and Perceiver non-default-resolution interpolation are both vision-model issues but not the same bug." + "reason": "Padding-mask batch consistency and Perceiver non-default resolution failures are different model/processor bugs." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "AyaVision lm_head tying and BitNet packed-weight unpacking during accelerate loading are distinct model-loading problems." + "reason": "AyaVision weight tying and BitNet packed-weight unpacking during accelerate loading are unrelated." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "High perplexity on a model benchmark and gradient checkpointing being ineffective with PEFT LoRA are unrelated training/evaluation issues." + "reason": "Perplexity on a benchmark and gradient-checkpointing behavior with PEFT LoRA are unrelated." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "3d_parallel.py runtime failure and Qwen3Moe FineGrainedFP8Config failure are different scripts/models and different failure modes." + "reason": "3d_parallel script execution and Qwen3MoE FP8 loading are different code paths." }, { "left": "issue:43749", "right": "issue:44534", "accept": false, - "reason": "FSDP CPU RAM efficient loading and non-persistent buffer junk in Transformers v5 are both loading-related but not the same bug." + "reason": "Both mention loading/regression, but FSDP CPU RAM efficient loading and junk non-persistent buffers are not the same defect." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "StableLm missing pad_token_idx and AutoModelForCausalLM not unwrapping text_config are separate config-handling regressions." + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to unwrapping text_config for composite Qwen multimodal configs." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug and Molmo missing all_tied_weights_keys are unrelated processor vs model attribute issues." + "reason": "BEiT image-processor label reduction and Molmo tied-weights attribute errors are unrelated." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next segfault on load and sinq quantized model save/load breakage are different model-loading failure modes." + "reason": "Llava Next load crashes and sinq save/load quantization breakage are different failures." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "Both can happen during loading, but Qwen3 device-map asserts and Llava Next segfaults are distinct model/runtime paths." + "reason": "Both are model-loading issues, but one is a device-map CUDA assert and the other is a segmentation fault in Llava Next loading." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "glm-4v-9b load failure and EMBEDDIA/est-roberta tokenizer loading regression are different model classes and symptoms." + "reason": "Different tokenizer/model-loading failures for different repositories; no shared concrete bug." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "CustomTrainer.compute_loss signature mismatch and 3d_parallel.py runtime failure are unrelated trainer API vs script execution bugs." + "reason": "Trainer compute_loss signature mismatch is unrelated to the 3d_parallel script failure." } ] }, @@ -4377,7 +4385,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:26413", "issue:28530", @@ -4423,7 +4431,6 @@ "issue:45701", "issue:45704", "issue:45823", - "issue:46291", "issue:46402", "issue:46665", "issue:46821" @@ -4432,18 +4439,18 @@ "issue:44387|issue:45704", "issue:41720|issue:44977", "issue:36331|issue:42915", - "issue:44877|issue:44991", "issue:44466|issue:45357", + "issue:44877|issue:44991", "issue:43450|issue:45701", "issue:40990|issue:45701", "issue:43065|issue:46402", - "issue:41720|issue:42617", "issue:43873|issue:44756", + "issue:41720|issue:42617", "issue:44186|issue:45072", "issue:26413|issue:28530", "issue:43550|issue:43720", - "issue:45685|issue:46821", "issue:43577|issue:43708", + "issue:45685|issue:46821", "issue:43295|issue:45701", "issue:43550|issue:43606", "issue:43646|issue:46665", @@ -4457,217 +4464,210 @@ "issue:45632|issue:45684", "issue:41720|issue:45161", "issue:43531|issue:43572", - "issue:28530|issue:33666", - "issue:42915|issue:46291" + "issue:28530|issue:33666" ], - "prepared_review_unit_hash": "fd3e2733fb1f276dff50d160896ec16fa2daa92d49cb76e21427af596891f755", + "prepared_review_unit_hash": "ec30d31d55ee7d0569890e6ab100da31cc2abba2312670390ca259e4f02c4306", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36794, - "estimated_input_tokens": 9199, - "estimated_eval_tokens": 18654 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35902, + "estimated_input_tokens": 8976, + "estimated_eval_tokens": 18208 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is broadly heterogeneous: it mixes unrelated trainer, tokenizer, multimodal, quantization, save/load, and serving bugs. None of the soft pairs look like the same underlying defect or a plausible single merged fix, so all soft edges are rejected. If a representative issue is needed, #43708 is the clearest, best-scoped regression report.", - "confidence": 0.97, - "canonical_issue_reason": "Issue #43708 is the strongest representative in the set because it is a concrete, reproducible Trainer regression with a precise failure mode and clear scope, even though it is not a duplicate of the other issues.", + "summary": "This cluster is highly heterogeneous: the soft pairs share only broad themes (same model family, same API area, or same symptom class) but not the same concrete bug. I would not merge any of the proposed soft edges.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45701 is the broadest, most hub-like issue here (generic version-regression/tokenization wording, multiple inbound references), so it is the best umbrella representative for the noisy cluster.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #43708 is the most actionable and well-specified issue here, with a direct repro path and a clearly stated bug in resume/checkpoint logic.", + "best_issue_reason": "Issue 45701 is the most representative issue by scope and connectivity, but only as a loose thematic hub rather than a true duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "Both mention memory/performance regressions, but one is int4 CUDA reserved-memory OOM and the other is a T5/apex RMSNorm leak; different root causes and models." + "reason": "Both mention memory pressure, but one is int4 quantization/OOM while the other is a T5 RMSNorm memory leak; different code paths and fixes." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Same broad Qwen3.5 area, but one is an auto device-map cuda assert and the other is a flash-attention generation failure; not the same bug path." + "reason": "Both are Qwen3-related, but one is a device-mapping cuda assert and the other is a flash-attention generation issue; not the same bug." }, { "left": "issue:36331", "right": "issue:42915", "accept": false, - "reason": "CustomTrainer compute_loss keyword mismatch is unrelated to Qwen3Moe failing with FineGrainedFP8Config." + "reason": "`compute_loss(num_items_in_batch)` is a trainer API compatibility issue; Qwen3MoE + FineGrainedFP8Config is a different model/config failure." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:44466", + "right": "issue:45357", "accept": false, - "reason": "Strict config loading for granite_speech and tokenizer loading for est-roberta are separate loading failures with different causes." + "reason": "Both concern saving/loading model weights, but one is tied `lm_head` serialization and the other is incorrect visual encoder keys; distinct failures." }, { - "left": "issue:44466", - "right": "issue:45357", + "left": "issue:44877", + "right": "issue:44991", "accept": false, - "reason": "Both involve model serialization, but one is tied lm_head.weight behavior and the other is incorrect visual encoder keys for Qwen3.5; different defects." + "reason": "Both involve loading/serialization regressions in v5, but one is strict config handling and the other is tokenizer loading for a specific model; not the same defect." }, { "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "Video processor batched-shape bug and tokenizer version-dependent behavior are unrelated." + "reason": "Video processor batch-shape handling is unrelated to the tokenization-version regression described in 45701." }, { "left": "issue:40990", "right": "issue:45701", "accept": false, - "reason": "High perplexity on gpt-oss-20b and tokenization changes across versions are not the same issue." + "reason": "High perplexity on gpt-oss is an evaluation/model-behavior problem, not the same underlying tokenization change bug." }, { "left": "issue:43065", "right": "issue:46402", "accept": false, - "reason": "Both touch Sam3, but one is a dummy Conv2d in the pixel decoder and the other is missing text-encoder weight loading; different code paths." + "reason": "Both are Sam3-related, but one is a dummy Conv2d concern and the other is missing text-encoder weight loading; different code paths." }, { - "left": "issue:41720", - "right": "issue:42617", + "left": "issue:43873", + "right": "issue:44756", "accept": false, - "reason": "Qwen3 auto device-map cuda assert and inability to run 3d_parallel.py are unrelated runtime failures." + "reason": "Quantization/offload behavior and Strix Halo mmap OOM are both memory-related but arise from different mechanisms and fixes." }, { - "left": "issue:43873", - "right": "issue:44756", + "left": "issue:41720", + "right": "issue:42617", "accept": false, - "reason": "Quantization offloading behavior and Strix Halo mmap OOM are separate memory-management problems." + "reason": "Qwen3 auto-device-map cuda assert and failure to run `3d_parallel.py` are not the same underlying issue." }, { "left": "issue:44186", "right": "issue:45072", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and dtype mismatches in SwitchTransformers/TimmWrapperModel are unrelated." + "reason": "Tokenizer crashes on NER/batched padding and bfloat16 dtype mismatches are unrelated preprocessing vs precision bugs." }, { "left": "issue:26413", "right": "issue:28530", "accept": false, - "reason": "Checkpoint resume epoch-iterator failure and early-stopping metric lookup are different Trainer control-flow bugs." + "reason": "Checkpoint resumption exhausting the epoch iterator and early-stopping metric lookup are separate trainer behaviors." }, { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba and BitNet weight-unpacking during accelerate loading are different model/backend issues." + "reason": "One is torch.compile + SDPA on Bamba, the other is packed-weight handling during accelerate loading; different failure modes." }, { - "left": "issue:45685", - "right": "issue:46821", + "left": "issue:43577", + "right": "issue:43708", "accept": false, - "reason": "MoE histogram MPS error and TimesFM window_size AttributeError are unrelated model bugs." + "reason": "BLIP2 dtype propagation and resume-from-checkpoint max_steps calculation are unrelated bugs." }, { - "left": "issue:43577", - "right": "issue:43708", + "left": "issue:45685", + "right": "issue:46821", "accept": false, - "reason": "Blip2 dtype staying float32 after load is unrelated to resume_from_checkpoint max_steps miscalculation." + "reason": "A MoE MPS histogram dtype error and a TimesFM window_size AttributeError are unrelated model-specific failures." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Processor.tokenizer regression for multimodal custom code and tokenization changes across versions are not the same defect." + "reason": "Processor/tokenizer API regression and a broader tokenization-version complaint are related only superficially, not the same concrete bug." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Bamba torch.compile/SDPA failure and suno/bark-small CPU offload device mismatch are different issues." + "reason": "Both touch device/runtime behavior, but one is compile+SDPA and the other is CPU-offload device mismatch; not mergeable as one fix." }, { "left": "issue:43646", "right": "issue:46665", "accept": false, - "reason": "Both are custom-code regressions after newer Transformers, but they affect different APIs and failure modes." + "reason": "Custom model initialization breakage and inconsistent `from_pretrained` behavior inside classes are separate API issues." }, { "left": "issue:44186", "right": "issue:44610", "accept": false, - "reason": "LayoutLMv2 tokenizer crash and OmDet-Turbo processor resolution mismatch are separate processor bugs." + "reason": "LayoutLMv2 tokenizer NER/padding crashes and OmDet-Turbo image-size mismatch are both processor issues but not the same bug." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "Qwen3Moe FP8 config failure and dequantized model offload loading failure do not share the same root cause." + "reason": "FineGrainedFP8Config failure and dequantized model loading with cpu offload are different quantization/offload problems." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Both involve image size handling, but OmDet-Turbo returns 640px while Perceiver fails on non-default resolution; different model logic." + "reason": "Both involve image sizing, but one is an OmDet processor/model expectation mismatch and the other is Perceiver interpolation behavior; distinct issues." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "lm_head tying in Mistral3 and Qwen3.5 visual-encoder save keys are distinct serialization/tie-weight problems." + "reason": "Tied `lm_head` weights in Mistral3 and incorrect visual encoder keys in Qwen3.5 save_pretrained are different serialization bugs." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma3n variable-name collision in forward and Gemma4 PLE device mismatch under auto mapping are different bugs." + "reason": "Gemma3n variable-name collision in forward and Gemma4 PLE device mismatch with `device_map=auto` are unrelated." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "GenerationMixin import error and 3d_parallel.py runtime failure are unrelated." + "reason": "GenerationMixin import regression and failure to run `3d_parallel.py` are unrelated issues." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both are serve/multimodal crashes, but they involve different processors and distinct missing-attribute errors." + "reason": "`continuous-batching` multimodal crash and Gemma4Processor missing `_tokenizer` are both serve-related but different code paths and fixes." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Trust-remote-code cache path collisions and read-only permission propagation in save_pretrained are unrelated filesystem issues." + "reason": "Trust-remote-code cache path collision and read-only permission propagation are unrelated save/load filesystem issues." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "Qwen3 auto device assert and GPT-OSS MoE tensor-parallel failure are separate parallelism problems." + "reason": "Both mention Qwen/GPT-OSS/MoE style model families, but one is an A800 device-map crash and the other is tensor-parallel support; different problems." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and missing pad_token_idx in StableLmConfig are unrelated config bugs." + "reason": "Sliding-window handling for Qwen3-MoE and missing `pad_token_idx` in StableLmConfig are separate config/model bugs." }, { "left": "issue:28530", "right": "issue:33666", "accept": false, - "reason": "Early stopping metric lookup and Qwen2-VL multi-GPU training are unrelated." - }, - { - "left": "issue:42915", - "right": "issue:46291", - "accept": false, - "reason": "Qwen3Moe FP8 config failure and transformers.pipelines import breaking on kernels version/revision are unrelated." + "reason": "Early-stopping metric lookup and Qwen2-VL multi-GPU training are not the same underlying issue." } ] }, @@ -4689,7 +4689,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28530", "issue:29159", @@ -4736,248 +4736,250 @@ "issue:45701", "issue:45704", "issue:45759", + "issue:46291", "issue:46665", "issue:46821" ], "soft_pairs": [ "issue:44671|issue:44743", + "issue:42915|issue:46291", "issue:36331|issue:41720", "issue:43526|issue:43881", "issue:44186|issue:44361", "issue:43425|issue:46821", "issue:43716|issue:46821", "issue:43316|issue:44913", - "issue:28530|issue:30064", "issue:42915|issue:43322", + "issue:28530|issue:30064", "issue:33357|issue:35707", "issue:43540|issue:43782", "issue:30064|issue:34634", "issue:43550|issue:43653", "issue:43873|issue:45704", + "issue:29159|issue:36032", "issue:44387|issue:45701", "issue:43606|issue:43720", - "issue:29159|issue:36032", "issue:44492|issue:44509", "issue:44534|issue:46665", "issue:43531|issue:45759", "issue:41720|issue:43746", "issue:42490|issue:43329", - "issue:43606|issue:43653", "issue:43388|issue:45059", + "issue:43606|issue:43653", "issue:45406|issue:45464", - "issue:43531|issue:46821", "issue:34567|issue:35707", + "issue:43531|issue:46821", "issue:34634|issue:41720", "issue:43531|issue:44589" ], - "prepared_review_unit_hash": "6f51e81ab92c984e8dc0cb2f97a97fe15b5c2416d017286f71d0a9008e634d77", + "prepared_review_unit_hash": "7bf1362ce6d44d9833b3ee15a705c132dcb681b606ca5b83a168aaed600c4cd0", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35715, - "estimated_input_tokens": 8929, - "estimated_eval_tokens": 18114 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36607, + "estimated_input_tokens": 9152, + "estimated_eval_tokens": 18560 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a duplicate set: the issues span unrelated areas (tokenizers, multimodal models, training APIs, quantization/offload, docs, and serving). All soft edges should be rejected.", - "confidence": 0.97, - "canonical_issue_reason": "No single canonical issue stands out because the items are heterogeneous and do not share one underlying bug or code path.", + "summary": "Highly heterogeneous cluster: it mixes unrelated tokenizer, config, multimodal, training, quantization, serving, and docs issues. None of the suggested soft pairs look like the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out; the set is too heterogeneous to have a meaningful duplicate anchor.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a good global representative; picking any one would be arbitrary because the cluster is not about one defect.", + "best_issue_reason": "No strong global best issue; even the broadest titles here describe different failure modes in different subsystems.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "Different model families and failures: CamemBERT masked-LM prediction bug vs Qwen3 modular recurrent-state reset logic." + "reason": "Different models and code paths: CamemBERT masked-LM predictions vs Qwen3.5 recurrent state reset with cache." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and pipelines import failure from kernels version/revision requirements are unrelated." }, { "left": "issue:36331", "right": "issue:41720", "accept": false, - "reason": "Training API signature mismatch in CustomTrainer vs Qwen3 device-mapping CUDA assert; unrelated code paths." + "reason": "Trainer compute_loss signature mismatch is unrelated to Qwen3 auto device mapping cudaErrorAssert." }, { "left": "issue:43526", "right": "issue:43881", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug vs glm-4v-9b loading failure; no shared underlying defect." + "reason": "BeitImageProcessorFast reduce_labels bug and glm-4v-9b loading failure are different subsystems and symptoms." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Both are tokenizer crashes, but on different tokenizers and causes (LayoutLMv2 NER/padding vs MLuke task AttributeError)." + "reason": "Both are tokenizer bugs, but LayoutLMv2 NER/padding crashes and MLukeAttributeError are different concrete defects." }, { "left": "issue:43425", "right": "issue:46821", "accept": false, - "reason": "Torch 2.10 compatibility issue vs TimesFM window_size AttributeError; different layers and symptoms." + "reason": "Torch 2.10 incompatibility is a version-support problem, not the TimesFM window_size AttributeError." }, { "left": "issue:43716", "right": "issue:46821", "accept": false, - "reason": "Multimodal dtype mismatch vs TimesFM argument handling; no common code path." + "reason": "Image preprocessor dtype mismatch and TimesFM window_size AttributeError are unrelated issues." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Gemma3TextConfig API discrepancy vs GPTNeoX rotary_pct reload regression; unrelated config bugs." + "reason": "Gemma3TextConfig API discrepancy and GPTNeoX rotary_pct reload regression involve different config fields and behaviors." }, { - "left": "issue:28530", - "right": "issue:30064", + "left": "issue:42915", + "right": "issue:43322", "accept": false, - "reason": "Early-stopping metric lookup failure vs image processor failing on void segmentation maps; different subsystems." + "reason": "Qwen3Moe fine-grained FP8 config failure and Llava Next segmentation fault are not the same bug." }, { - "left": "issue:42915", - "right": "issue:43322", + "left": "issue:28530", + "right": "issue:30064", "accept": false, - "reason": "Qwen3Moe FP8 config failure vs Llava Next segmentation fault; distinct model-loading problems." + "reason": "Early stopping/eval metric lookup has nothing to do with void segmentation map processing." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "MacOS bus error with CLIP model vs generation with inputs_embeds/past_key_values; unrelated inference bugs." + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values are different runtime problems." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Video-input processing error in Qwen3OmniMoe vs Qwen3VL weight_only load failure; different failure points." + "reason": "Video-input ValueError in Qwen3OmniMoe and Qwen3VL weight_only load failure are different code paths." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Vision segmentation-map processing bug vs BarkProcessor voice_preset issue; different processors and domains." + "reason": "Void segmentation map processing and BarkProcessor voice_preset handling are unrelated." }, { "left": "issue:43550", "right": "issue:43653", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba vs BigBirdTokenizer special-token registration; not the same bug." + "reason": "torch.compile SDPA failure in Bamba-9B-v2 is unrelated to BigBirdTokenizer special-token registration." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization/offloading behavior vs T5 apex RMSNorm memory leak; separate defects." + "reason": "Quantization/offloading behavior and T5 apex FusedRMSNorm memory leak are distinct bugs." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Slow/fast tokenizer inconsistency is different from the T5Tokenizer add_special_tokens naming conflict." }, { "left": "issue:44387", "right": "issue:45701", "accept": false, - "reason": "Int4 quantization memory growth vs tokenization changes across versions; different behavior and likely different fixes." + "reason": "Int4 quantization OOM and version-dependent tokenization changes are not the same underlying issue." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "Both CI-related, but one is CPU-offload device mismatch and the other is packed-weight unpacking in accelerate loading." - }, - { - "left": "issue:29159", - "right": "issue:36032", - "accept": false, - "reason": "Tokenizer inconsistency between slow and fast implementations vs T5Tokenizer load-time add_special_tokens method conflict." + "reason": "CPU offload device mismatch for Bark differs from BitNet packed-weights unpacking during accelerate loading." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "A typo in cache strategies vs stale docs mentioning removed pipeline tasks; different documentation problems." + "reason": "A typo in cache strategies is unrelated to stale docs mentioning removed pipeline tasks." }, { "left": "issue:44534", "right": "issue:46665", "accept": false, - "reason": "Non-persistent buffer initialization bug in v5 vs inconsistent from_pretrained behavior inside Python classes." + "reason": "Non-persistent buffers filled with junk and from_pretrained behavior inside Python classes are different loader semantics." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Qwen3-MoE sliding_window bug vs AutoModelForCausalLM.from_config not unwrapping text_config for composite multimodal configs." + "reason": "Qwen3-MoE sliding_window behavior and AutoModelForCausalLM config unwrapping are separate Qwen-related bugs." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Qwen3 auto device mapping CUDA assert vs GraniteSpeech PEFT local-checkpoint loading; unrelated model/runtime issues." + "reason": "Qwen3 auto device mapping cudaErrorAssert is unrelated to GraniteSpeech PEFT adapter loading." }, { "left": "issue:42490", "right": "issue:43329", "accept": false, - "reason": "Local-path trailing-slash import warning vs undefined variables in multimodal token counting video branch; different bugs." + "reason": "Trailing-slash local model import warning and undefined variables in multimodal token counting are different issues." }, { - "left": "issue:43606", - "right": "issue:43653", + "left": "issue:43388", + "right": "issue:45059", "accept": false, - "reason": "CPU-offload device mismatch vs BigBirdTokenizer mask-token registration; no shared underlying change." + "reason": "gather_for_metrics dropping tuple labels is unrelated to SAM3 PCS text/bounding-box behavior." }, { - "left": "issue:43388", - "right": "issue:45059", + "left": "issue:43606", + "right": "issue:43653", "accept": false, - "reason": "Metrics gathering truncation in the last batch vs SAM3 text/bounding-box behavior; different components." + "reason": "CPU offload device mismatch and mask token special-token registration are different bugs." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in transformers serve vs Qwen3.5 streaming chat/completions failure; separate serving paths." + "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to Qwen3.5 streaming chat/completions failure." }, { - "left": "issue:45701", - "right": "issue:45704", + "left": "issue:34567", + "right": "issue:35707", "accept": false, - "reason": "Tokenization behavior changed across versions vs T5 memory leak in apex RMSNorm; unrelated regressions." + "reason": "TrainerState token counters not updating and progressive generation cache behavior are unrelated." }, { "left": "issue:43531", "right": "issue:46821", "accept": false, - "reason": "Qwen3-MoE sliding_window bug vs TimesFM window_size AttributeError; only superficial parameter-name similarity." - }, - { - "left": "issue:34567", - "right": "issue:35707", - "accept": false, - "reason": "TrainerState num_input_tokens_seen not updating vs progressive generation with inputs_embeds/past_key_values; different training vs generation code." + "reason": "Both mention windowing, but Qwen3-MoE sliding_window and TimesFM window_size AttributeError are different models and bugs." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "BarkProcessor voice_preset bug vs Qwen3 auto device-mapping CUDA assert; unrelated model families." + "reason": "BarkProcessor voice_preset failure is unrelated to Qwen3 device-mapping cudaErrorAssert." }, { "left": "issue:43531", "right": "issue:44589", "accept": false, - "reason": "Qwen3-MoE sliding_window issue vs Float8 storage lookup error; different runtime failures and fixes." + "reason": "Qwen3-MoE sliding_window bug and missing Float8 storage object are unrelated." } ] }, @@ -4999,7 +5001,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:30064", "issue:36010", @@ -5018,6 +5020,7 @@ "issue:43994", "issue:44062", "issue:44077", + "issue:44186", "issue:44368", "issue:44451", "issue:44514", @@ -5040,15 +5043,13 @@ "issue:45759", "issue:45859", "issue:45923", - "issue:45987", "issue:46032", "issue:46050", "issue:46402", "issue:46463", "issue:46655", "issue:46665", - "issue:46693", - "issue:46829" + "issue:46693" ], "soft_pairs": [ "issue:44451|issue:44991", @@ -5056,10 +5057,10 @@ "issue:45357|issue:46402", "issue:44077|issue:44625", "issue:43644|issue:46665", - "issue:44625|issue:45479", "issue:44062|issue:44451", "issue:44898|issue:45072", "issue:44987|issue:45042", + "issue:44625|issue:45479", "issue:43381|issue:43844", "issue:39692|issue:44610", "issue:44368|issue:45357", @@ -5067,8 +5068,8 @@ "issue:41720|issue:45704", "issue:43746|issue:46050", "issue:43421|issue:43927", - "issue:43381|issue:45704", "issue:45704|issue:45923", + "issue:43381|issue:45704", "issue:44918|issue:44936", "issue:42886|issue:44843", "issue:44514|issue:44545", @@ -5079,217 +5080,223 @@ "issue:43994|issue:44671", "issue:43421|issue:43653", "issue:30064|issue:36010", - "issue:45987|issue:46829", + "issue:43421|issue:44186", "issue:43782|issue:45381" ], - "prepared_review_unit_hash": "9fbdb28675cf53ff22447cfb50b508d735b1f2086771fd0b767a62a78a6962f6", + "prepared_review_unit_hash": "83b52acba4548dd8206f45733b47441b136ddc09978c6a46d6633763d37e2711", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36985, - "estimated_input_tokens": 9247, - "estimated_eval_tokens": 18750 + "serialized_chars": 36365, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Only one soft pair looks like a true duplicate: the two Qwen2_5_VLProcessor batched-input crashes. The rest are similar only at a broad subsystem level (tokenizers, vision, loading, training, offline mode) and should stay separate.", - "confidence": 0.92, - "canonical_issue_reason": "Issue 44514 is the older, clearer report of the same Qwen2_5_VLProcessor.apply_chat_template crash seen in 44545, so it is the best canonical issue anchor.", + "summary": "Mostly heterogeneous Transformers issues with only two clear duplicate pairs: the offline tokenizer-cache regression and the Qwen2_5_VL batched padding crash. The rest are similar only at a broad subsystem level, not the same underlying bug.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43421 is the best cluster anchor among the issues: it is the broadest tokenizer-centric report and sits near several other special-token/tokenizer behavior reports, making it the most representative issue in this set.", "canonical_pr_reason": null, - "best_issue_reason": "44514 is the strongest representative issue in this cluster because it exactly matches its duplicate and states the shared failure mode cleanly.", + "best_issue_reason": "Issue 43421 is the strongest overall issue candidate because it captures a concrete recurring tokenizer state/post-processor problem and is broader than the more model-specific tokenizer reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model loading failures, but they involve different models and likely different regression paths." + "reason": "Both are tokenizer-loading failures, but they involve different models and likely different root causes." }, { "left": "issue:45759", "right": "issue:45859", "accept": false, - "reason": "Both are Qwen3.5-related, but one is a from_config/text_config unwrap bug and the other is a CPU-offload rotary embedding issue." + "reason": "Both mention Qwen 3.5/3.6, but one is a config unwrap issue and the other is CPU offload compatibility; not the same bug." }, { "left": "issue:45357", "right": "issue:46402", "accept": false, - "reason": "Both concern checkpoint loading, but they target different model families and different missing-weight symptoms." + "reason": "Different model families and different loading failures; no shared concrete code path." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "These are unrelated config bugs: optional post_init handling versus num_labels propagation." + "reason": "One is about post_init being optional, the other about num_labels propagation; unrelated behaviors." }, { "left": "issue:43644", "right": "issue:46665", "accept": false, - "reason": "One is about non-persistent buffer initialization; the other is about inconsistent from_pretrained behavior inside Python classes." - }, - { - "left": "issue:44625", - "right": "issue:45479", - "accept": false, - "reason": "Both touch classification config, but one is propagation of num_labels and the other is zero-loss behavior for single-label classification." + "reason": "Non-persistent buffer corruption and from_pretrained behavior in Python classes are unrelated." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "An AddedToken constructor TypeError and a model-loading failure for a specific repo are not the same concrete bug." + "reason": "A tokenizer AddedToken constructor error is not the same as a model-specific load failure." }, { "left": "issue:44898", "right": "issue:45072", "accept": false, - "reason": "Different code paths: Perceiver resolution/interpolation behavior versus dtype mismatches in bfloat16 inference." + "reason": "Different subsystems and failure modes: image resolution handling vs dtype mismatch." }, { "left": "issue:44987", "right": "issue:45042", "accept": false, - "reason": "One is a model loading regression, the other is an unnecessary torchvision dependency in PIL image processors." + "reason": "Loading a specific model and PIL backend torchvision requirements are separate bugs." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "One is config propagation, the other is loss calculation degeneracy; different code paths." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "These are both training-related, but eval-mode gradient checkpointing restrictions and ZeRO-3 gradient growth are distinct issues." + "reason": "Gradient checkpointing in eval mode and a ZeRO-3 gradient blow-up are distinct issues." }, { "left": "issue:39692", "right": "issue:44610", "accept": false, - "reason": "Both involve vision processors, but one is a docs/model mismatch plus quantization failure and the other is a processor input-size mismatch." + "reason": "Both are vision-related, but one is a doc/example mismatch and quantization failure while the other is a processor size mismatch." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "A tie_word_embeddings warning in fine-tuning is not the same as incorrect visual encoder keys being saved." + "reason": "A warning about tie_word_embeddings is unrelated to saving incorrect visual encoder keys." }, { "left": "issue:46655", "right": "issue:46693", "accept": false, - "reason": "Video padding calculation and FA2 long-context performance regression are different bugs." + "reason": "Incorrect video padding calculation and a FA2 performance regression are different problems." }, { "left": "issue:41720", "right": "issue:45704", "accept": false, - "reason": "Qwen3 auto device mapping cuda asserts and T5 apex RMSNorm memory leaks are unrelated." + "reason": "CUDA assert during Qwen3 device mapping is unrelated to a T5 Apex RMSNorm memory leak." }, { "left": "issue:43746", "right": "issue:46050", "accept": false, - "reason": "PEFT local checkpoint loading and sinq quantized save/load breakage are different persistence bugs." + "reason": "Local PEFT checkpoint loading and sinq save/load breakage are different serialization problems." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Runtime special-token post-processing and DiaConfig custom-token-ID loss after save/load are separate tokenizer/config paths." + "reason": "Both involve tokenizer metadata, but one is runtime special-token updates and the other is save/load losing custom token IDs." }, { - "left": "issue:43381", - "right": "issue:45704", + "left": "issue:45704", + "right": "issue:45923", "accept": false, - "reason": "Gradient checkpointing in eval mode and a T5 RMSNorm memory leak do not share the same underlying fix." + "reason": "A memory leak in T5 norm code and a missing gradient_checkpointing flag are not the same bug." }, { - "left": "issue:45704", - "right": "issue:45923", + "left": "issue:43381", + "right": "issue:45704", "accept": false, - "reason": "A memory leak in T5 is unrelated to a missing supports_gradient_checkpointing flag on Nemotron-3-Nano-Omni." + "reason": "Eval-mode gradient checkpointing behavior does not match a T5 Apex memory leak." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "Both mention trainer behavior, but unpacking embeddings in SFT and failing evaluate() after train() are different failures." + "reason": "Trainer embedding unpacking failure and evaluate-after-train failure are different trainer issues." }, { "left": "issue:42886", "right": "issue:44843", - "accept": false, - "reason": "Both relate to offline mode, but one is cache loading failure and the other is an unconditional model_info() call in a Mistral regex patch." + "accept": true, + "reason": "Both describe the same offline tokenizer-loading regression; 44843 identifies the unconditional model_info() call in _patch_mistral_regex as the root cause." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "These are the same crash report: Qwen2_5_VLProcessor.apply_chat_template fails on batched input when padding=False." + "reason": "These are the same Qwen2_5_VLProcessor.apply_chat_template crash on batched input with padding=False." + }, + { + "left": "issue:45759", + "right": "issue:45923", + "accept": false, + "reason": "Different Qwen model issues: from_config text_config unwrapping vs missing supports_gradient_checkpointing." }, { "left": "issue:46032", "right": "issue:46463", "accept": false, - "reason": "Mamba2Mixer use_cache correctness and an incorrect decoder-input validation error message are not the same bug." + "reason": "A silent wrong-result cache bug and an error-message bug are not the same underlying change." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Different problems: a SigLIP2 docs/example issue versus quantization offloading not behaving as expected." + "reason": "SigLIP2 doc/example issues and quantization offloading behavior are not the same bug." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "Local checkpoint loading for PEFT adapters is unrelated to quantization offloading behavior." + "reason": "PEFT checkpoint loading and quantized offloading misbehavior are separate problems." }, { "left": "issue:43994", "right": "issue:44671", "accept": false, - "reason": "They both affect model outputs, but the affected models and failure modes are different." + "reason": "Different models and failure modes: nonsensical pipeline outputs vs masked LM predictions." }, { "left": "issue:43421", "right": "issue:43653", "accept": false, - "reason": "Special-token post-processor updates and BigBirdTokenizer mask-token registration are distinct tokenizer bugs." + "reason": "Both concern tokenizer special tokens, but one is runtime post-processor updating and the other is a BigBirdTokenizer registration/decode bug." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "An image processor segmentation-map bug and a GenerationMixin import error are unrelated." + "reason": "ImageProcessor segmentation-map handling and a GenerationMixin ImportError are unrelated." }, { - "left": "issue:45987", - "right": "issue:46829", + "left": "issue:43421", + "right": "issue:44186", "accept": false, - "reason": "A return-type contract issue for StaticCache is unrelated to VideoPrism mislabelling hidden_states." + "reason": "Special-token post-processor updates and LayoutLMv2 NER/padding crashes are different tokenizer bugs." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Different Qwen multimodal bugs: a weight_only load error versus incorrect video vision_position_ids." + "reason": "Qwen3VL weight_only loading and Qwen2.5-VL video position IDs are separate multimodal issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft pair, issue:44514 vs issue:44545, is well-supported by matching titles and the same batched-input/padding=False crash. The rejected pairs are framed cautiously at the issue-title level without overstating duplicate evidence." + "feedback": "The summary is well grounded in the packet: it correctly characterizes the set as mostly heterogeneous and identifies the two strongest duplicate pairs (42886/44843 and 44514/44545). The remaining soft-edge rejections are conservative and do not overclaim shared root cause from the titles alone. I don\u2019t see an evidence-overstatement problem here." }, "evaluator_used": true, "retried": false, @@ -5308,7 +5315,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:30064", "issue:36331", @@ -5316,7 +5323,6 @@ "issue:41720", "issue:43295", "issue:43335", - "issue:43421", "issue:43531", "issue:43540", "issue:43550", @@ -5324,11 +5330,9 @@ "issue:43611", "issue:43618", "issue:43643", - "issue:43644", "issue:43653", "issue:43697", "issue:43720", - "issue:43749", "issue:43756", "issue:43819", "issue:43927", @@ -5355,13 +5359,17 @@ "issue:45381", "issue:45834", "issue:45907", + "issue:45987", "issue:46294", "issue:46665", - "issue:46693" + "issue:46693", + "issue:46829", + "issue:46832" ], "soft_pairs": [ - "issue:43421|issue:44186", + "issue:46829|issue:46832", "issue:45081|issue:45356", + "issue:45987|issue:46829", "issue:45356|issue:45907", "issue:43720|issue:43819", "issue:43653|issue:43720", @@ -5369,243 +5377,242 @@ "issue:43295|issue:45834", "issue:43550|issue:45072", "issue:43550|issue:43927", - "issue:43540|issue:46294", "issue:44062|issue:44488", + "issue:43540|issue:46294", "issue:45381|issue:46693", - "issue:43618|issue:43697", "issue:44220|issue:44661", "issue:43531|issue:43643", "issue:43611|issue:46665", + "issue:43618|issue:43697", "issue:43653|issue:43819", "issue:44186|issue:44898", "issue:30064|issue:36331", "issue:41720|issue:44991", - "issue:43643|issue:44877", "issue:43335|issue:43756", + "issue:43643|issue:44877", "issue:43550|issue:43819", "issue:43540|issue:44008", "issue:44821|issue:45084", "issue:44843|issue:45081", "issue:39692|issue:45834", - "issue:44186|issue:44442", - "issue:44448|issue:44671", "issue:44485|issue:44509", - "issue:43644|issue:43749" + "issue:44186|issue:44442", + "issue:44448|issue:44671" ], - "prepared_review_unit_hash": "7262b76498b9f437622e28382fa9343edc214bbd8dedd79e7d64677a0a587e7f", + "prepared_review_unit_hash": "a4dba96b6eb8c870385ade848c2ba953e93bc2fd5c4ca26a7369842b7c4e9c7d", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37162, - "estimated_input_tokens": 9291, - "estimated_eval_tokens": 18838 + "serialized_chars": 37253, + "estimated_input_tokens": 9314, + "estimated_eval_tokens": 18884 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose collection of distinct Transformers issues, mostly around v5-era loading, tokenizers, and multimodal/model-specific regressions. The high textual similarity appears to be theme-level, not duplicate-level; none of the candidate soft edges look like the same underlying bug.", - "confidence": 0.46, - "canonical_issue_reason": "issue:43611 is the most representative anchor because it is a broad v5 loading regression and sits near the center of the cluster\u2019s recurring model/config/tokenizer-loading complaints.", + "summary": "This is a loose topical neighborhood, not a true duplicate cluster: it mixes unrelated bugs across vision/video models, tokenizers, configs, and training/runtime paths. No soft edge is strong enough to merge, so all candidate links are rejected.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue is clearly justified because the cluster is heterogeneous; at most it loosely centers on tokenizer-regex regressions, but most items are unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the best single issue to represent the cluster since it is version-wide and general, whereas the others are narrower, model-specific failures.", + "best_issue_reason": "If forced to pick one representative issue, 45081 is the clearest concrete regression in the tokenizer-regex code path (_patch_mistral_regex crash with a reproducible failure).", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43421", - "right": "issue:44186", + "left": "issue:46829", + "right": "issue:46832", "accept": false, - "reason": "Runtime special-token/post-processor update bug vs LayoutLMv2 NER/padding crash; different tokenizer paths and models." + "reason": "Different model families and failures: VideoPrism output field mismatch vs RT-DETR feature-level shape error." }, { "left": "issue:45081", "right": "issue:45356", "accept": false, - "reason": "Same tokenizer area, but one is a Mistral regex backend crash and the other is a Kimi codec/regression warning issue." + "reason": "Both touch fix_mistral_regex, but one is a crash in _patch_mistral_regex and the other is a Kimi tokenizer regression with codec/warning issues; not clearly the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache seq length contract bug is unrelated to a video classification output field mix-up." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Tokenizer regression vs repo-file revision forwarding CI break; unrelated subsystems." + "reason": "Tokenizer regression and repo_files revision forwarding are unrelated changes." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "BitNet packed-weight loading issue vs DAC latent reconstruction mismatch; different models and code paths." + "reason": "Different subsystems and failures: accelerate loading of packed BitNet weights vs DAC from_latents/STE mismatch." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Tokenizer special-token registration bug vs BitNet accelerate-loading bug; not the same failure." + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight loading are unrelated tokenizer/model-loading bugs." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "BertModel/DeepSpeed Zero3 index error vs audio fbank feature extraction issue; unrelated." + "reason": "Deepspeed zero3/BertModel init error is unrelated to _torch_extract_fbank_features() behavior." }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "Processor/tokenizer regression in v4.57.5 vs Kosmos2.5 long-OCR input index error; both multimodal but distinct." + "reason": "Processor/tokenizer regression in v4.57.5 is unrelated to Kosmos2.5 long-OCR indexing." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "Bamba torch.compile SDPA failure vs SwitchTransformers/TimmWrapper dtype mismatch; different models and symptoms." + "reason": "Bamba torch.compile SDPA failure is unrelated to SwitchTransformers/TimmWrapperModel dtype mismatches." }, { "left": "issue:43550", "right": "issue:43927", "accept": false, - "reason": "Bamba compile/SDPA bug vs DiaConfig token-ID persistence bug; unrelated." - }, - { - "left": "issue:43540", - "right": "issue:46294", - "accept": false, - "reason": "Different Qwen video bugs with different models and failure points: video processing ValueError vs chat-template token mismatch." + "reason": "Torch.compile SDPA failure and DiaConfig save/load token-ID loss are different bugs." }, { "left": "issue:44062", "right": "issue:44488", "accept": false, - "reason": "AddedToken duplicate-arg TypeError vs tokenizer load failure for a specific model; different tokenizer code paths." + "reason": "AddedToken duplicate special-arg error is unrelated to loading a specific model/tokenizer checkpoint." }, { - "left": "issue:45381", - "right": "issue:46693", + "left": "issue:43540", + "right": "issue:46294", "accept": false, - "reason": "Video vision_position_ids bug vs FA2 long-context performance regression; not the same bug." + "reason": "Qwen3OmniMoe video-processing ValueError and Qwen3.5 video apply_chat_template token/feature mismatch are different code paths." }, { - "left": "issue:43618", - "right": "issue:43697", + "left": "issue:45381", + "right": "issue:46693", "accept": false, - "reason": "CLIPOutput attentions assignment regression vs RTDetrV2 output difference; different model behaviors." + "reason": "Qwen2.5-VL video position IDs and FA2 long-context performance regression are unrelated." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "Audio feature extraction bug vs add-new-model-like TOKENIZER_MAPPING_NAMES issue; unrelated." + "reason": "Audio fbank feature extraction issue and add-new-model-like TOKENIZER_MAPPING_NAMES failure are unrelated." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Qwen3-MoE sliding_window bug vs trust_remote_code AutoConfig missing fields; different code paths." + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to AutoConfig trust_remote_code missing fields." }, { "left": "issue:43611", "right": "issue:46665", "accept": false, - "reason": "Both are loading-related, but one is a broad v5 model-loading regression and the other is inconsistent from_pretrained behavior inside Python classes." + "reason": "base_model_prefix loading breakage and from_pretrained behavior inside Python classes are different loader regressions." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "Missing CLIPOutput attentions field and RTDetrV2 output drift are unrelated model-output regressions." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "BigBirdTokenizer special-token registration vs DAC latent/STE mismatch; no shared underlying bug." + "reason": "Tokenizer special-token registration and DAC forward/from_latents mismatch are unrelated." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "LayoutLMv2 tokenizer crash vs Perceiver image-classification resolution bug; different models and inputs." + "reason": "LayoutLMv2Tokenizer NER/padding crash and Perceiver non-default-res image classification failure are different bugs." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Image processor void segmentation map failure vs CustomTrainer compute_loss signature change; unrelated." + "reason": "Image Processor void segmentation map handling and CustomTrainer.compute_loss signature mismatch are unrelated." }, { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Qwen3 device-map cuda assert vs tokenizer load failure for est-roberta; different subsystems." + "reason": "Qwen3 auto device mapping CUDA assert and est-roberta tokenizer loading failure do not share a concrete code path." }, { - "left": "issue:43643", - "right": "issue:44877", + "left": "issue:43335", + "right": "issue:43756", "accept": false, - "reason": "Both concern config loading, but the specific bugs differ: missing fields with trust_remote_code vs strict granite_speech config rejection." + "reason": "SwitchTransformers sparse-layer creation and Smollm3 RoPE layer-dropping are different configuration/model-definition bugs." }, { - "left": "issue:43335", - "right": "issue:43756", + "left": "issue:43643", + "right": "issue:44877", "accept": false, - "reason": "SwitchTransformers sparse-layer construction bug vs Smollm3 RoPE-layer count bug; unrelated." + "reason": "AutoConfig trust_remote_code missing fields and granite_speech strict-config loading are distinct config-loading regressions." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Bamba SDPA/torch.compile failure vs DAC from_latents mismatch; no common code path." + "reason": "Bamba torch.compile SDPA failure and DAC latent conversion mismatch are unrelated model/runtime issues." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Both are multimodal, but one is Qwen video processing and the other is Gemma3n variable collision in forward()." + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma3n variable-name collision are unrelated." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "AutoImageProcessor URL loading issue vs template-node compilation TypeError; unrelated." + "reason": "AutoImageProcessor URL loading and TypeError compiling non-template nodes are unrelated." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Both touch Mistral regex patching, but one is offline-mode network access and the other is a backend_tokenizer AttributeError." + "reason": "Both involve _patch_mistral_regex, but one is offline-mode model_info() access and the other is a backend_tokenizer attribute crash; not clearly the same underlying bug." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "SigLIP2 documentation example errors vs Kosmos2.5 OCR long-input crash; unrelated." + "reason": "SigLIP2 docs/example errors and Kosmos2.5 OCR indexing are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion and removed pipeline-task docs are different change requests." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "LayoutLMv2 tokenizer crash vs AutoTokenizer failing to load FastSpeech2ConformerTokenizer; different models and failure modes." + "reason": "LayoutLMv2Tokenizer NER/padding crash and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are unrelated." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Pegasus v4/v5 output mismatch vs CamemBERT masked-LM regression; distinct model-specific bugs." - }, - { - "left": "issue:44485", - "right": "issue:44509", - "accept": false, - "reason": "GLM-5 RoPE implementation discussion vs docs mentioning removed pipeline tasks; different topics." - }, - { - "left": "issue:43644", - "right": "issue:43749", - "accept": false, - "reason": "Non-persistent buffer initialization bug vs FSDP CPU RAM efficient loading breakage; related to loading but not the same issue." + "reason": "Pegasus v4/v5 output differences and CamemBERT masked-LM prediction regression are different model-specific issues." } ] }, @@ -5627,7 +5634,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28218", "issue:29159", @@ -5640,6 +5647,7 @@ "issue:43577", "issue:43582", "issue:43606", + "issue:43644", "issue:43645", "issue:43716", "issue:43720", @@ -5655,7 +5663,6 @@ "issue:44479", "issue:44561", "issue:44610", - "issue:44671", "issue:44805", "issue:44811", "issue:44936", @@ -5668,7 +5675,6 @@ "issue:45479", "issue:45584", "issue:45685", - "issue:45701", "issue:45834", "issue:45987", "issue:46032", @@ -5679,9 +5685,10 @@ "issue:46665" ], "soft_pairs": [ - "issue:34634|issue:36010", - "issue:43322|issue:45685", + "issue:43644|issue:43749", "issue:45183|issue:45479", + "issue:43322|issue:45685", + "issue:34634|issue:36010", "issue:43606|issue:45072", "issue:45584|issue:45834", "issue:43716|issue:44008", @@ -5694,8 +5701,8 @@ "issue:43825|issue:43994", "issue:43782|issue:44977", "issue:46294|issue:46463", - "issue:44561|issue:45020", "issue:43421|issue:44811", + "issue:44561|issue:45020", "issue:44805|issue:44936", "issue:43540|issue:45381", "issue:44610|issue:45072", @@ -5704,206 +5711,208 @@ "issue:28218|issue:29159", "issue:43749|issue:44291", "issue:43645|issue:46665", - "issue:43577|issue:43582", - "issue:44671|issue:45701" + "issue:43577|issue:43582" ], - "prepared_review_unit_hash": "5569fd4788316576ed6c86a48be42e08bf1db039f8de547634b3ed33f81e9dd3", + "prepared_review_unit_hash": "25988cda0876a91002c474f381645b8c7fb1e0b49b42d5071e50cdff0a71b581", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 36205, - "estimated_input_tokens": 9052, - "estimated_eval_tokens": 18360 + "serialized_chars": 35657, + "estimated_input_tokens": 8915, + "estimated_eval_tokens": 18086 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: the soft links span unrelated tokenizer, multimodal, loading, dtype/device, and training/eval bugs. I do not see any pair that is clearly the same underlying defect, so none should be merged as duplicates.", - "confidence": 0.88, - "canonical_issue_reason": "issue:45020 is the broadest umbrella for recent-version breakages, but it is still only a loose fit and does not \u05d1\u05d0\u05de\u05ea unify the unrelated bugs in this cluster.", + "summary": "This cluster is mostly noisy and spans many unrelated regressions. The only strong duplicate-looking pair is the tokenizer added-token spacing vs. slow/fast tokenizer inconsistency report.", + "confidence": 0.61, + "canonical_issue_reason": "issue:45020 is the broadest umbrella report in the set for recent-version remote_code regressions, making it the best general anchor for this mixed cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the least-bad triage bucket for the v5/remote_code regression subset, but it is too generic to serve as a true canonical for the whole group.", + "best_issue_reason": "issue:45020 is the most general, high-level issue here; the rest are mostly narrow, model-specific regressions or unrelated bug reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:34634", - "right": "issue:36010", + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Different loading paths and failures: non-persistent buffer corruption vs. FSDP CPU RAM efficient loading." + }, + { + "left": "issue:45183", + "right": "issue:45479", "accept": false, - "reason": "Different bugs: Bark voice_preset handling vs a GenerationMixin import failure." + "reason": "One is input_ids/inputs_embeds validation messaging; the other is a classification-loss bug for num_labels=1." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "Unrelated failures in different code paths: Llava Next loading segfault vs MPS histogram on Int." + "reason": "Unrelated models and code paths: Llava Next load segfault vs. an MPS histogram/int support error." }, { - "left": "issue:45183", - "right": "issue:45479", + "left": "issue:34634", + "right": "issue:36010", "accept": false, - "reason": "One is input validation/error messaging; the other is a loss-function regression for single-label classification." + "reason": "BarkProcessor voice_preset behavior is unrelated to the GenerationMixin import failure." }, { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Both are device/dtype-adjacent, but on different models and different failure modes; not the same bug." + "reason": "Different bug classes: CPU offload device mismatch vs. bfloat16 dtype mismatch in other models." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty-transcription/alignment issue is unrelated to Kosmos2.5 long-ocr index errors." + "reason": "Whisper empty transcription handling is unrelated to Kosmos2.5 long OCR indexing." }, { "left": "issue:43716", "right": "issue:44008", "accept": false, - "reason": "Different multimodal model bugs: dtype mismatch in Mistral-3 preprocessing vs Gemma 3n attribute collision." + "reason": "Both are multimodal, but they describe different failures in different models and code paths." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "Qwen video regression and Kosmos2.5 OCR index error are separate model-specific issues." + "reason": "Qwen video regression and Kosmos2.5 OCR indexing are not the same underlying bug." }, { "left": "issue:43388", "right": "issue:46032", "accept": false, - "reason": "Training/eval metric gathering bug is unrelated to Mamba2 caching correctness." + "reason": "Gather-for-metrics label trimming is unrelated to Mamba2 cache misuse with seq_len > 1." }, { "left": "issue:43950", "right": "issue:45987", "accept": false, - "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to StaticCache sequence-length return type." + "reason": "Non-persistent buffer corruption and StaticCache return-type mismatch are different regressions." }, { "left": "issue:44991", "right": "issue:46291", "accept": false, - "reason": "Tokenizer loading failure and pipelines/kernel dependency import failure are different problems." + "reason": "Tokenizer loading failure and kernels version/revision import failure are distinct issues." }, { "left": "issue:43582", "right": "issue:46512", "accept": false, - "reason": "Apple Silicon TypeError in warmup is not the same as stale config.dtype after from_config." + "reason": "Apple Silicon warmup TypeError is unrelated to stale config dtype in from_config()." }, { "left": "issue:43577", "right": "issue:43818", "accept": false, - "reason": "Blip2 dtype propagation and Video-LLaVA temporal-attention/weight-sharing are unrelated." + "reason": "BLIP2 dtype mismatch and Video-LLaVA architecture/weight-sharing issues are not the same bug." }, { "left": "issue:43825", "right": "issue:43994", "accept": false, - "reason": "Error-message wording for removed translation pipelines is not the same as SigLIP2 bad outputs." + "reason": "Pipeline v5 error-message wording and SigLIP2 model output corruption are unrelated." }, { "left": "issue:43782", "right": "issue:44977", "accept": false, - "reason": "Qwen3VL weight_only loading and Qwen3.5 flash-attention generation are different code paths." + "reason": "from_pretrained weight_only loading error is unrelated to flash-attention generation problems." }, { "left": "issue:46294", "right": "issue:46463", "accept": false, - "reason": "Qwen video chat-template mismatch is unrelated to T5 decoder-input validation messaging." + "reason": "Qwen3.5 video template mismatch and T5 decoder-input validation messaging are different code paths." }, { - "left": "issue:44561", - "right": "issue:45020", + "left": "issue:43421", + "right": "issue:44811", "accept": false, - "reason": "Both mention v5/remote_code breakage, but one is a specific removed utility and the other is a broad umbrella report." + "reason": "Runtime post-processor updates and Whisper batch_decode skip_special_tokens handling are separate tokenizer/decoder issues." }, { - "left": "issue:43421", - "right": "issue:44811", + "left": "issue:44561", + "right": "issue:45020", "accept": false, - "reason": "TokenizersBackend runtime post-processor updates and Whisper batch_decode skip_special_tokens are different tokenizer behaviors." + "reason": "Both mention remote_code breakage, but one is specifically about is_torch_fx_available removal while the other is a broad umbrella report." }, { "left": "issue:44805", "right": "issue:44936", "accept": false, - "reason": "Mask/index shape mismatch and trainer.evaluate after training are not the same underlying bug." + "reason": "Mask shape IndexError and trainer.evaluate() after train() are not the same failure." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Qwen3Omni video-input ValueError and qwen2.5-vl vision_position_ids regression are separate video-input failures." + "reason": "Video input ValueError in Qwen3OmniMoe is unrelated to video position-id regression in Qwen2.5-VL." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "Processor image-size mismatch and bfloat16 dtype mismatch are distinct multimodal inference bugs." + "reason": "OmDet-Turbo image-size mismatch and bfloat16 dtype mismatches are different problems." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weights unpacking during accelerate loading is unrelated to bfloat16 dtype mismatches." + "reason": "Packed-weight loading failure in BitNet is unrelated to bfloat16 dtype mismatch bugs." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "Pipeline error-message regression and docs still referencing pipeline are related only broadly, not the same issue." + "reason": "A misleading runtime error and stale documentation references are related thematically but not the same artifact." }, { "left": "issue:28218", "right": "issue:29159", - "accept": false, - "reason": "Both are tokenizer-related, but one is added-token spacing and the other is a broader slow-vs-fast inconsistency report." + "accept": true, + "reason": "Both describe tokenizer behavior divergence around added tokens / spacing, with the slow-vs-fast inconsistency matching the extra-space symptom." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "FSDP CPU RAM efficient loading and init_empty_weights _is_hf_initialized argument errors are different loading regressions." + "reason": "FSDP CPU RAM loading and init_empty_weights/_is_hf_initialized argument handling are different initialization/loading bugs." }, { "left": "issue:43645", "right": "issue:46665", "accept": false, - "reason": "Custom-model initialization in notebooks and from_pretrained inside Python classes are not the same bug." + "reason": "Jupyter custom-model initialization failure is not the same as from_pretrained-in-class inconsistent behavior." }, { "left": "issue:43577", "right": "issue:43582", "accept": false, - "reason": "Blip2 dtype propagation and Apple Silicon caching allocator TypeError are unrelated." - }, - { - "left": "issue:44671", - "right": "issue:45701", - "accept": false, - "reason": "CamemBERT masked-LM prediction regression is not the same as a general tokenization-change report." + "reason": "BLIP2 dtype handling and Apple Silicon caching allocator warmup are unrelated." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge decisions stay conservative overall. The one accepted pair is plausible from the titles, and the rejections do not overclaim duplicate status." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -5918,7 +5927,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:34634", "issue:36010", @@ -5945,6 +5954,7 @@ "issue:44368", "issue:44554", "issue:44589", + "issue:44671", "issue:44857", "issue:44912", "issue:44936", @@ -5959,18 +5969,19 @@ "issue:45406", "issue:45538", "issue:45600", + "issue:45701", "issue:45987", "issue:46192", "issue:46209", "issue:46285", - "issue:46291", "issue:46326", "issue:46697", "issue:46728" ], "soft_pairs": [ - "issue:45987|issue:46192", + "issue:44671|issue:45701", "issue:42175|issue:45081", + "issue:45987|issue:46192", "issue:34634|issue:36331", "issue:45003|issue:46285", "issue:46697|issue:46728", @@ -5980,220 +5991,219 @@ "issue:43782|issue:45325", "issue:45081|issue:45406", "issue:44589|issue:45042", - "issue:43742|issue:45020", "issue:45372|issue:46326", + "issue:43742|issue:45020", "issue:43550|issue:44912", "issue:45084|issue:45538", - "issue:43526|issue:43697", "issue:43329|issue:45600", + "issue:43526|issue:43697", "issue:44060|issue:44368", - "issue:43526|issue:44554", "issue:43452|issue:44991", + "issue:43526|issue:44554", "issue:43257|issue:43329", "issue:43906|issue:46209", "issue:36010|issue:36331", "issue:43819|issue:43927", "issue:36010|issue:43645", "issue:43531|issue:45042", - "issue:43638|issue:44936", - "issue:42175|issue:46291" + "issue:43638|issue:44936" ], - "prepared_review_unit_hash": "83fb6e5518ca17451c691ee8f1b6f57a0f6182bec13773cc8ef88af3c97700fb", + "prepared_review_unit_hash": "a6cde3af40de2c8a143dbad6ca6ab94e5c6b605b2d4063bdb4b298e634706937", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 35924, - "estimated_input_tokens": 8981, - "estimated_eval_tokens": 18218 + "serialized_chars": 36422, + "estimated_input_tokens": 9106, + "estimated_eval_tokens": 18468 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: most items are unrelated Transformers regressions across different models, processors, trainers, and backends. None of the soft-edge pairs look like true duplicates under a strict same-bug/same-change standard.", - "confidence": 0.83, - "canonical_issue_reason": "issue:45020 is the broadest umbrella regression report in the set ('recent transformers versions break models using remote_code'), but the overall cluster is too mixed for a strong canonical duplicate anchor.", + "summary": "The set is mostly heterogeneous, and the soft edges are broad-topic false positives rather than true duplicates. I would not merge any of these pairs; no single issue/PR stands out as a valid canonical representative for the whole group.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the most representative of the cluster\u2019s general theme of version regressions breaking model loading, though it is still not a duplicate of the other issues.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45987", - "right": "issue:46192", + "left": "issue:44671", + "right": "issue:45701", "accept": false, - "reason": "Different code paths: cache sequence-length typing vs. text_kwargs/return_tensors handling." + "reason": "Different bugs: one is a CamemBERT masked-LM regression, the other is a general tokenization change across versions." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Unrelated failures: installation/backend packaging vs. tokenizer regex patching." + "reason": "Unrelated subsystems: pip backend dependency packaging vs a Mistral tokenizer regex patch crash." + }, + { + "left": "issue:45987", + "right": "issue:46192", + "accept": false, + "reason": "Different API surfaces: cache sequence length return type vs chat-template kwargs handling." }, { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "Different subsystems and symptoms: Bark voice presets vs. trainer loss signature changes." + "reason": "Unrelated models/code paths: Bark voice preset vs trainer compute_loss signature regression." }, { "left": "issue:45003", "right": "issue:46285", "accept": false, - "reason": "Both are modeling-layer issues, but one is sys.modules access and the other is CLIP class structure; not the same bug." + "reason": "Both touch model internals, but the bugs are different: sys.modules access vs CLIP text model structure inconsistency." }, { "left": "issue:46697", "right": "issue:46728", "accept": false, - "reason": "Both concern image patch sizing, but they affect different processors with different counting logic." + "reason": "Both are image-processor issues, but they affect different models and different counting/padding logic." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "Tokenizer special-token registration is unrelated to MXFP4 quantization fallback during model loading." + "reason": "Tokenizer special-token registration bug is unrelated to MXFP4 quantization loading fallback." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting is unrelated to DAC latent reconstruction/STE behavior." + "reason": "Audio chat-template sampling-rate default bug is unrelated to DAC latent/forward mismatch." }, { "left": "issue:43582", "right": "issue:44857", "accept": false, - "reason": "Different platforms and failures: Apple Silicon allocator warmup TypeError vs CUDA AMP image loss crash." + "reason": "Different platform and failure mode: Apple Silicon allocator TypeError vs CUDA AMP loss crash." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Distinct Qwen VL bugs: weight_only loading error vs rope index scaling for still-image positions." + "reason": "Both are Qwen VL-related, but one is a load-time weight_only error and the other is a rope/indexing bug." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Both are processor/tokenizer crashes, but on different models and different missing attributes/code paths." + "reason": "Similar processor/tokenizer loading failures, but different models and different missing attributes." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to torchvision requirement in PIL image processors." + "reason": "Completely unrelated: Float8 storage lookup error vs PIL backend torchvision dependency regression." }, { - "left": "issue:43742", - "right": "issue:45020", + "left": "issue:45372", + "right": "issue:46326", "accept": false, - "reason": "A generic key error loading one model is not enough to match the broader remote_code regression." + "reason": "Both involve mistral_common, but one is an import break in Gemma 4 loading and the other is a continue_final_message tokenizer bug." }, { - "left": "issue:45372", - "right": "issue:46326", + "left": "issue:43742", + "right": "issue:45020", "accept": false, - "reason": "Both involve mistral_common, but one is an import failure for Gemma 4 loading and the other is continue_final_message behavior." + "reason": "KeyError on a specific model is not the same underlying issue as remote_code breakage in newer versions." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "torch.compile+SDPA failure is a different bug from MXFP4 quantization fallback." + "reason": "Different concrete failures: torch.compile+SDPA on Bamba vs MXFP4 quantization fallback on git-oss-20b." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template compilation error and CLIP tokenizer max-length bug are unrelated." + "reason": "Different functionality entirely: template compilation error vs CLIPTokenizer max length constant." }, { - "left": "issue:43526", - "right": "issue:43697", + "left": "issue:43329", + "right": "issue:45600", "accept": false, - "reason": "Different model families and different defects: label reduction vs divergent outputs in RTDetrV2." + "reason": "Both affect model plumbing, but one is multimodal token counting and the other is stale auto-mapping references." }, { - "left": "issue:43329", - "right": "issue:45600", + "left": "issue:43526", + "right": "issue:43697", "accept": false, - "reason": "Video multimodal token counting bug is unrelated to removed auto-mapping config references." + "reason": "Vision-model issues, but not the same bug: label reduction in BEiT fast processor vs RTDetrV2 output drift." }, { "left": "issue:44060", "right": "issue:44368", "accept": false, - "reason": "Related theme around Qwen tied-weights warnings, but the concrete symptoms differ enough that they do not look like one fixable bug." + "reason": "Both mention Qwen tied-weights warnings, but one is an incorrect tying bug and the other is a user-facing warning in a different model/version." }, { - "left": "issue:43526", - "right": "issue:44554", + "left": "issue:43452", + "right": "issue:44991", "accept": false, - "reason": "Beit label reduction and MPS attention correctness are different code paths." + "reason": "Both are tokenizer loading failures, but they concern different tokenizer configs and different failure paths." }, { - "left": "issue:43452", - "right": "issue:44991", + "left": "issue:43526", + "right": "issue:44554", "accept": false, - "reason": "Tokenizer/model loading breakages, but for different models and different root causes." + "reason": "Unrelated vision processor label handling vs an MPS attention correctness issue." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Qwen3 MoE weight conversion and multimodal video token counting are unrelated." + "reason": "Different Qwen3-MoE problems: weight conversion/loading under accelerate+deepspeed vs undefined multimodal-token helper usage." }, { "left": "issue:43906", "right": "issue:46209", "accept": false, - "reason": "An isolated reproduction of an old issue is not the same as an FP8 inference failure on B200." + "reason": "No shared bug: isolated reproduction of an older issue vs FP8 inference failure on B200." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "GenerationMixin import error and CustomTrainer compute_loss signature mismatch are unrelated." + "reason": "Different regressions: missing GenerationMixin import vs trainer compute_loss signature change." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "DAC latent reconstruction mismatch is unrelated to DiaConfig custom token-ID persistence." + "reason": "Different model/config bugs: DAC latent reconstruction mismatch vs DiaConfig losing custom token IDs after save/load." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "Different breakages: GenerationMixin import path vs notebook custom-model initialization in v5." + "reason": "ImportError for GenerationMixin is unrelated to notebook custom-model initialization failures." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and PIL torchvision dependency regression are unrelated." + "reason": "Different code paths: Qwen3-MoE sliding-window logic vs PIL backend processor dependency handling." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "One is a zero3/non-pretrained Bert index error; the other is evaluate() failing after train()." - }, - { - "left": "issue:42175", - "right": "issue:46291", - "accept": false, - "reason": "Package installation backend selection is unrelated to kernels v0.15 version/revision requirements." + "reason": "Training-time Zero3/empty-tensor index crash is not the same as evaluate-after-train state issues." } ] }, @@ -6215,7 +6225,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28282", "issue:29159", @@ -6224,16 +6234,14 @@ "issue:38175", "issue:39692", "issue:40990", + "issue:42175", "issue:42490", "issue:42915", "issue:43257", "issue:43262", "issue:43329", "issue:43450", - "issue:43454", - "issue:43525", "issue:43540", - "issue:43577", "issue:43606", "issue:43643", "issue:43650", @@ -6254,6 +6262,7 @@ "issue:44756", "issue:44811", "issue:44821", + "issue:44857", "issue:44912", "issue:44977", "issue:45042", @@ -6264,234 +6273,235 @@ "issue:46291", "issue:46421", "issue:46424", - "issue:46701" + "issue:46701", + "issue:46832" ], "soft_pairs": [ + "issue:42175|issue:46291", "issue:43720|issue:44610", "issue:42490|issue:43257", "issue:39692|issue:44912", "issue:38175|issue:44610", "issue:43746|issue:44811", "issue:43643|issue:45759", - "issue:44387|issue:44617", "issue:43854|issue:44977", + "issue:44387|issue:44617", "issue:43257|issue:43696", "issue:43262|issue:44811", - "issue:43257|issue:43931", "issue:46421|issue:46424", - "issue:42915|issue:43742", + "issue:43257|issue:43931", "issue:43329|issue:44479", + "issue:42915|issue:43742", "issue:45459|issue:45803", "issue:43329|issue:43450", "issue:43749|issue:44756", - "issue:44291|issue:45042", "issue:43650|issue:43906", + "issue:44291|issue:45042", "issue:42915|issue:44821", - "issue:43606|issue:45859", "issue:34567|issue:40990", + "issue:43606|issue:45859", "issue:43540|issue:46701", + "issue:44857|issue:46832", "issue:43927|issue:44912", "issue:29159|issue:34567", - "issue:43525|issue:46291", - "issue:28282|issue:29870", - "issue:43454|issue:43577" + "issue:28282|issue:29870" ], - "prepared_review_unit_hash": "6d36a03a42de0e865803f871e0246af8513005f99e57b64863e0e875f315b446", + "prepared_review_unit_hash": "9b946c827a9e9da402f9d31288b3665049576879820e5614ca18dfe3dc468ad2", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36246, - "estimated_input_tokens": 9062, - "estimated_eval_tokens": 18380 + "serialized_chars": 36200, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft pair looks like the same underlying bug/change. Most are same broad area at best (loading, video, cache, tokenizer) but have different concrete failure modes, models, or code paths, so they should stay separate.", - "confidence": 0.9, - "canonical_issue_reason": null, + "summary": "This cluster is mostly heterogeneous: loading/configuration, quantization, multimodal/video, tokenizer, and training-state issues all mixed together. I don\u2019t see any soft pair that clearly describes the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 43749 is the broadest infrastructure/loading bug in the set and has the strongest cluster-central feel (high discussion and inbound references), so it is the least narrow anchor even though the cluster is not a true duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 is the strongest standalone candidate: it describes a broad, high-impact loading regression, has the most discussion/inbound references, and is better specified than the other unrelated issues in this set.", + "best_issue_reason": "Issue 43749 is the best overall representative of the cluster\u2019s loading/runtime-failure theme; it is broader and more central than the more model-specific or narrowly scoped reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "One is about missing TensorFlow backend support from a torch-only install; the other is an import failure caused by the kernels package version/revision requirement. Different dependency problems." + }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "Packed-weight loading under accelerate vs OmDet processor image-size mismatch are unrelated bugs." + "reason": "Accelerate loading with packed weights is unrelated to a processor/model input-size mismatch in OmDet-Turbo." }, { "left": "issue:42490", "right": "issue:43257", "accept": false, - "reason": "Local-path import warning is a path handling issue; Qwen3 MOE weight conversion under accelerate/deepspeed is a model-loading bug." + "reason": "A trailing-slash local-path import warning is not the same bug as Qwen3 MoE weights failing to convert under accelerate + deepspeed." }, { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "Doc-example errors for SigLIP2 do not match a GPT-OSS quantization fallback/loading regression." + "reason": "Documentation example errors and an MXFP4 quantization fallback bug are different problems in different code paths." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Zero probabilities in SigLIP2 is a model-output issue; OmDet-Turbo input resolution mismatch is a preprocessing bug." + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo\u2019s processor/output-size mismatch." }, { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "PEFT local checkpoint loading and Whisper batch_decode special-token handling are different code paths and failures." + "reason": "Local checkpoint loading with PEFT adapters is a different failure mode from Whisper batch_decode ignoring skip_special_tokens." }, { "left": "issue:43643", "right": "issue:45759", "accept": false, - "reason": "Missing fields from trust_remote_code AutoConfig is not the same as AutoModelForCausalLM.from_config failing to unwrap text_config." + "reason": "Both involve config loading, but one loses fields with trust_remote_code while the other fails to unwrap text_config for composite multimodal configs; not the same bug." }, { - "left": "issue:44387", - "right": "issue:44617", + "left": "issue:43854", + "right": "issue:44977", "accept": false, - "reason": "Int4 quantization memory regression is unrelated to Sam3Video OOM despite both mentioning memory." + "reason": "A unit-test model load failure for GLM-4.7 is not the same as a flash-attention generation regression in Qwen3.5." }, { - "left": "issue:43854", - "right": "issue:44977", + "left": "issue:44387", + "right": "issue:44617", "accept": false, - "reason": "GLM-4.7 loading-test failure and Qwen3.5 flash-attention generation bug are different model-specific issues." + "reason": "Int4 CUDA memory growth under Transformers 5.x is a different issue from Sam3Video\u2019s out-of-memory failure." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Missing weight conversion with accelerate/deepspeed is unrelated to GPT-oss-20b CUDA OOM." + "reason": "Qwen3 MOE weight-conversion failure during loading is unrelated to a plain CUDA OOM on GPT-oss-20b." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio chat-template sample-rate defaulting and Whisper batch_decode token skipping are separate processor behaviors." + "reason": "Defaulting audio chat templates to 16kHz is a different processor bug from Whisper batch_decode skipping special-token handling." }, { - "left": "issue:43257", - "right": "issue:43931", + "left": "issue:46421", + "right": "issue:46424", "accept": false, - "reason": "MOE weight conversion/loading under accelerate is not the same as Qwen3-VL weight-shape mismatch." + "reason": "Both are cache-related, but one is about lazy initialization causing recompilation and the other about max_cache_length only applying to Quantized Cache; not the same concrete bug." }, { - "left": "issue:46421", - "right": "issue:46424", + "left": "issue:43257", + "right": "issue:43931", "accept": false, - "reason": "Both are cache-related, but one is lazy-init/recompilation and the other is max_cache_length honoring; different defects." + "reason": "Accelerate/deepspeed weight conversion failure and Qwen3-VL weight-shape mismatch are distinct model-loading errors." }, { - "left": "issue:42915", - "right": "issue:43742", + "left": "issue:43329", + "right": "issue:44479", "accept": false, - "reason": "Qwen3Moe FP8 config failure and MobileLLM key error are distinct load-time errors." + "reason": "Both touch video regressions, but one is a specific undefined-helper bug in multimodal token counting while the other is a broader v5.3.0 generation regression across several Qwen video models." }, { - "left": "issue:43329", - "right": "issue:44479", + "left": "issue:42915", + "right": "issue:43742", "accept": false, - "reason": "Undefined helper variables in multimodal token counting is not the same bug as a broader v5.3.0 video regression." + "reason": "Qwen3Moe FineGrainedFP8Config failure and a MobileLLM key error are unrelated loading bugs." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Tokenizer error masking via protobuf handling and Fuyu's bare except are different exception-swallowing bugs." + "reason": "Both are exception-handling problems, but they affect different functions and different failure modes: protobuf decode errors vs bare except swallowing interrupts in Fuyu tensor conversion." }, { "left": "issue:43329", "right": "issue:43450", "accept": false, - "reason": "Undefined variables in the video token-count branch and batched video shape errors occur in different parts of the pipeline." + "reason": "Undefined variables in video token counting is not the same as batched video processor shape corruption." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "FSDP CPU RAM efficient loading breakage and Strix Halo mmap OOM are different memory-related loading problems." + "reason": "FSDP CPU RAM-efficient loading and disabling mmap on Strix Halo are different memory/loading issues with different root causes." }, { - "left": "issue:44291", - "right": "issue:45042", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized type error and PIL backend torchvision requirement are unrelated issues." + "reason": "One item is just a placeholder/nonsense title, the other is an isolated reproduction report; they are not the same bug." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:44291", + "right": "issue:45042", "accept": false, - "reason": "The first item is too vague to match the isolated reproduction of issue 38071; no concrete shared bug is evident." + "reason": "The unexpected _is_hf_initialized argument error is unrelated to PIL image processors incorrectly requiring torchvision." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Qwen3Moe FP8 loading failure and AutoImageProcessor-from-URL failure are unrelated." + "reason": "A Qwen3Moe FP8 configuration bug is unrelated to failing to load AutoImageProcessor from a URL." }, { - "left": "issue:43606", - "right": "issue:45859", + "left": "issue:34567", + "right": "issue:40990", "accept": false, - "reason": "Both mention CPU offload, but Bark device mismatch and Qwen3_5Moe rotary embedding are different model/code-path failures." + "reason": "TrainerState token counting and GPT-oss perplexity are different issues in different subsystems." }, { - "left": "issue:34567", - "right": "issue:40990", + "left": "issue:43606", + "right": "issue:45859", "accept": false, - "reason": "Trainer token accounting and GPT-oss perplexity are unrelated issues." + "reason": "Both mention CPU offload, but they involve different models and different code paths: device mismatch in bark-small versus rotary embedding incompatibility in Qwen3_5Moe." }, { "left": "issue:43540", "right": "issue:46701", "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError and DiffusionGemma image-text-to-text pipeline failure are different models and pipelines." + "reason": "Qwen3OmniMoe video input validation failure and DiffusionGemma pipeline failure are unrelated." + }, + { + "left": "issue:44857", + "right": "issue:46832", + "accept": false, + "reason": "A float16 AMP crash in LwDetrImageLoss is not the same as an RTDetr forward TypeError when feature-level counts do not match backbone outputs." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "DiaConfig save/load token-ID loss and GPT-OSS MXFP4 fallback are unrelated." + "reason": "DiaConfig save/load token-ID loss and MXFP4 quantization fallback are separate model/config bugs." }, { "left": "issue:29159", "right": "issue:34567", "accept": false, - "reason": "Tokenizer slow/fast inconsistency does not match TrainerState num_input_tokens_seen not updating." - }, - { - "left": "issue:43525", - "right": "issue:46291", - "accept": false, - "reason": "Llama4Config missing pad_token_id and pipelines import failing on kernels v0.15 are unrelated." + "reason": "Tokenizer slow/fast inconsistencies are unrelated to TrainerState\u2019s token counter not updating." }, { "left": "issue:28282", "right": "issue:29870", "accept": false, - "reason": "AutoModel missing PyTorch and Bart generate decoder_attention_mask UnboundLocalError are completely different problems." - }, - { - "left": "issue:43454", - "right": "issue:43577", - "accept": false, - "reason": "AyaVision weight tying and BLIP2 dtype propagation are separate model-loading bugs." + "reason": "Missing PyTorch during AutoModel import is a dependency issue, while the Bart generate crash is an unbound decoder_attention_mask bug." } ] }, @@ -6513,7 +6523,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:24643", "issue:27957", @@ -6523,7 +6533,6 @@ "issue:33357", "issue:34567", "issue:34689", - "issue:38175", "issue:41720", "issue:42175", "issue:42915", @@ -6531,6 +6540,8 @@ "issue:43404", "issue:43425", "issue:43452", + "issue:43454", + "issue:43525", "issue:43540", "issue:43577", "issue:43582", @@ -6546,7 +6557,6 @@ "issue:44568", "issue:44821", "issue:44849", - "issue:44898", "issue:44936", "issue:44987", "issue:44991", @@ -6565,6 +6575,8 @@ "issue:46821" ], "soft_pairs": [ + "issue:43525|issue:46291", + "issue:43454|issue:43577", "issue:44568|issue:45356", "issue:44987|issue:45538", "issue:28282|issue:34689", @@ -6575,10 +6587,10 @@ "issue:43404|issue:45127", "issue:42175|issue:43824", "issue:45137|issue:45405", - "issue:43742|issue:44291", "issue:28282|issue:29127", - "issue:44936|issue:46821", + "issue:43742|issue:44291", "issue:43317|issue:45538", + "issue:44936|issue:46821", "issue:44479|issue:46693", "issue:44062|issue:44991", "issue:44849|issue:45701", @@ -6589,221 +6601,214 @@ "issue:27957|issue:28282", "issue:24643|issue:29159", "issue:43540|issue:44560", - "issue:42175|issue:44488", "issue:44987|issue:45907", + "issue:42175|issue:44488", "issue:43540|issue:44479", "issue:44291|issue:44821", - "issue:38175|issue:44898", "issue:44568|issue:45488" ], - "prepared_review_unit_hash": "689b1bdecfe3b276734c6229c050362fc47ba05177aac6230158efbe46318fbe", + "prepared_review_unit_hash": "190711a3fb18f07370d5a33f50dc3a759722f0cf752b29dd7b003c4160bdcd6b", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36682, - "estimated_input_tokens": 9171, - "estimated_eval_tokens": 18598 + "soft_pair_count": 31, + "serialized_chars": 36968, + "estimated_input_tokens": 9242, + "estimated_eval_tokens": 18740 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose mix of Transformers regression reports, mostly around tokenization, model loading, and backend/device edge cases. The soft pairs look superficially similar but do not describe the same concrete bug, so I would reject all of them as duplicates. No pull requests are present in this cluster.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45488 is the best representative issue: it describes a concrete, high-impact tokenizer regression with a clear failure mode and active user impact, making it a stronger canonical anchor than the more generic or highly model-specific reports.", + "summary": "These items are a heterogeneous mix of unrelated bug reports; the soft pairs are mostly only loosely similar by subsystem or symptom, not the same underlying fix. I found no duplicate pair that is strong enough to merge, and there are no PRs in the set.", + "confidence": 0.86, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 45488 is the most suitable global issue candidate because it is specific, reproducible, and centrally about a real tokenizer behavior regression rather than a vague symptom or environment-specific failure.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Different failures: a missing `pad_token_id` config attribute versus a kernels package API/version requirement." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Both touch model loading internals, but one is about untied `lm_head` weights while the other is a dtype-casting bug in BLIP2." + }, { "left": "issue:44568", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but one is about add_special_tokens not adding BOS/EOS and the other is about codec handling / warning behavior in Kimi-K2.5; different bugs." + "reason": "Tokenizer-related, but different concrete regressions and model families; not the same bug path." }, { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Loading failure for physical-intelligence/fast is unrelated to CLIPTokenizer model_max_length behavior; different code paths and symptoms." + "reason": "A model load failure for one repo is not the same as CLIP tokenizer `model_max_length` behavior." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "PyTorch-missing ImportError is an environment dependency issue, while the Llama 3.2 Vision problem is a model-loading regression; not the same bug." + "reason": "One is missing PyTorch in the environment; the other is a specific model-loading regression in a newer Transformers release." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch 2.10 incompatibility is a version constraint issue, while the kernels import failure is about a missing revision/version argument; different root causes." + "reason": "Torch compatibility and kernels revision forwarding are unrelated dependency issues." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "FineGrainedFP8Config failure and CLIPTokenizer max_length bug are unrelated subsystems with different failure modes." + "reason": "FineGrainedFP8/Qwen3Moe loading is unrelated to CLIP tokenizer length settings." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "RTDetrV2 output divergence in v5 and incorrect visual encoder keys on save_pretrained are separate regressions affecting different model workflows." + "reason": "Different models and different bugs: output differences in RTDetrV2 versus incorrect save_pretrained keys in Qwen3.5." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "MacOS bus error in CLIP loading is a runtime crash, whereas num_input_tokens_seen not updating is a Trainer state bookkeeping issue." + "reason": "MacOS bus error during CLIP model load is unrelated to TrainerState token counting." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Untied lm_head weights in Mistral3 and LoRA merge collapse with extended vocab are different model-state bugs, not the same underlying issue." + "reason": "Both involve tied weights, but the concrete problems differ: missing tie in Mistral3 versus LoRA merge/extended-vocab collapse." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Missing TensorFlow backend in a torch install and missing Qwen2_5_VL import are unrelated packaging/import problems." + "reason": "Packaging/backend inclusion issue versus missing exported Qwen2_5_VL class import." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "DeepSpeed ZeRO3 deque IndexError and a PEFT version bump to an unreleased release are unrelated." - }, - { - "left": "issue:43742", - "right": "issue:44291", - "accept": false, - "reason": "KeyError loading MobileLLM-125M and the unexpected _is_hf_initialized argument during init_empty_weights are distinct loading-path regressions." + "reason": "DeepSpeed ZeRO3 deque error and a PEFT minimum-version bump are unrelated." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "Missing PyTorch dependency and unclear LayoutLMv3 box-validation error are different issues with different scopes." + "reason": "Environment import failure versus LayoutLMv3 error-message clarity." }, { - "left": "issue:44936", - "right": "issue:46821", + "left": "issue:43742", + "right": "issue:44291", "accept": false, - "reason": "Trainer evaluate-after-train failure and TimesFM window_size AttributeError are unrelated trainer/model issues." + "reason": "A key error while loading one model is not the same as the `_is_hf_initialized` argument regression." }, { "left": "issue:43317", "right": "issue:45538", "accept": false, - "reason": "device_map=auto offload/dequantized model loading and CLIPTokenizer model_max_length are different subsystems." + "reason": "Device-map/offload loading failure is unrelated to CLIP tokenizer max-length behavior." + }, + { + "left": "issue:44936", + "right": "issue:46821", + "accept": false, + "reason": "Trainer lifecycle bug versus a TimesFM `window_size` attribute error." }, { "left": "issue:44479", "right": "issue:46693", "accept": false, - "reason": "Video-input regression for several Qwen models and FA2 long-context performance regression are different problems: correctness vs performance." + "reason": "Both are regressions, but one is a video-input bug and the other is a long-context FA2 performance regression." }, { "left": "issue:44062", "right": "issue:44991", "accept": false, - "reason": "Duplicate AddedToken keyword argument and tokenizer loading failure for est-roberta are not the same code-path problem." + "reason": "Both are tokenizer load failures, but they appear to stem from different concrete code paths and model-specific issues." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "Qwen3.5 hidden-state bug and broad tokenization changes across versions are not the same concrete defect." + "reason": "One is a Qwen3.5 hidden-states bug; the other is a broader tokenization-version change." }, { "left": "issue:43577", "right": "issue:46512", "accept": false, - "reason": "BLIP2 dtype staying float32 and from_config leaving config.dtype stale are related to dtype handling but they are different states and failure symptoms." + "reason": "Model weight dtype mismatch and stale `config.dtype` are related themes but not the same underlying defect." }, { "left": "issue:42915", "right": "issue:43749", "accept": false, - "reason": "FineGrainedFP8Config failure and FSDP CPU RAM efficient loading breakage are separate distributed/loading regressions." + "reason": "Qwen3Moe FP8 loading and FSDP CPU RAM efficient loading are separate training/loading problems." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "gguf_file loading breakage and Apple Silicon caching_allocator_warmup TypeError are unrelated features." + "reason": "gguf/AutoModel loading behavior is unrelated to Apple Silicon allocator warmup type handling." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Qwen3 device-mapping cudaErrorAssert and Qwen3OmniMoe video ValueError are different failure modes and code paths." + "reason": "Different model families and failures: CUDA assert during device mapping versus video input processing error." }, { "left": "issue:27957", "right": "issue:28282", "accept": false, - "reason": "Flash Attention 2 issue on XLMRoberta and missing PyTorch dependency are unrelated." + "reason": "Flash Attention 2 on XLM-Roberta is unrelated to the PyTorch import error." }, { "left": "issue:24643", "right": "issue:29159", "accept": false, - "reason": "DeepSpeed weight shape runtime error and slow/fast tokenizer inconsistency are unrelated bugs." + "reason": "DeepSpeed weight-shape runtime error and tokenizer inconsistency are unrelated." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both involve video inputs, but one is a ValueError in Qwen3OmniMoe and the other is a StopIteration in Qwen3-vl-embedding; different model families and failure points." - }, - { - "left": "issue:42175", - "right": "issue:44488", - "accept": false, - "reason": "TensorFlow backend packaging and failure to load cjvt/sleng-bert are unrelated loading/package issues." + "reason": "Both mention video inputs, but the model families and failure modes differ enough that they do not look like one fix." }, { "left": "issue:44987", "right": "issue:45907", "accept": false, - "reason": "Model load failure for physical-intelligence/fast and list_repo_files revision forwarding are different bugs in different APIs." - }, - { - "left": "issue:43540", - "right": "issue:44479", - "accept": false, - "reason": "Qwen3OmniMoe video ValueError and v5.3.0 video-input regression across Qwen variants are related only by media input, but not the same concrete bug." - }, - { - "left": "issue:44291", - "right": "issue:44821", - "accept": false, - "reason": "init_empty_weights argument handling and AutoImageProcessor-from-URL failure are unrelated loader paths." + "reason": "Model loading failure versus Hugging Face Hub revision forwarding in `list_repo_files`." }, { - "left": "issue:38175", - "right": "issue:44898", + "left": "issue:42175", + "right": "issue:44488", "accept": false, - "reason": "Zero probabilities in SigLIP2 and Perceiver non-default resolution failure are different vision-model bugs." + "reason": "Backend install packaging issue versus a model-specific loading failure." }, { "left": "issue:44568", "right": "issue:45488", "accept": false, - "reason": "Both are tokenizer-related, but one is about add_special_tokens behavior in mdeberta and the other is a LlamaTokenizer override of tokenizer.json pre-tokenizer; different underlying bugs." + "reason": "Both involve tokenizer regressions, but one is about `add_special_tokens` behavior and the other about a tokenizer.json pre-tokenizer override." } ] }, @@ -6825,17 +6830,17 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:24643", "issue:29870", "issue:34567", "issue:36331", + "issue:38175", "issue:43257", "issue:43452", "issue:43526", "issue:43531", - "issue:43540", "issue:43577", "issue:43582", "issue:43697", @@ -6853,6 +6858,7 @@ "issue:44792", "issue:44821", "issue:44877", + "issue:44898", "issue:44964", "issue:45042", "issue:45072", @@ -6876,6 +6882,7 @@ "issue:46829" ], "soft_pairs": [ + "issue:38175|issue:44898", "issue:44479|issue:45381", "issue:44792|issue:44964", "issue:43257|issue:43756", @@ -6884,216 +6891,215 @@ "issue:43582|issue:46821", "issue:43742|issue:44589", "issue:45042|issue:45081", - "issue:44737|issue:46463", "issue:43716|issue:46701", + "issue:44737|issue:46463", "issue:43756|issue:46399", "issue:43756|issue:45325", - "issue:43531|issue:44821", "issue:44568|issue:46491", + "issue:43531|issue:44821", "issue:43906|issue:46468", "issue:45405|issue:45600", "issue:43577|issue:45072", "issue:45823|issue:46463", - "issue:43697|issue:46829", "issue:24643|issue:29870", "issue:43906|issue:46693", + "issue:43697|issue:46829", "issue:43452|issue:44488", "issue:44737|issue:45588", - "issue:44625|issue:44877", "issue:43526|issue:45479", - "issue:45357|issue:45920", - "issue:43540|issue:46693" + "issue:44625|issue:44877", + "issue:45357|issue:45920" ], - "prepared_review_unit_hash": "2ec7520fc80d8f86a5a05ca02eba6448f40edd088c35d126aa08f1e805e1407c", + "prepared_review_unit_hash": "7b45b69a4ccffd62ab991a9de362f7c821df4c2df03e3d5fdf01c8b68cc73169", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35549, - "estimated_input_tokens": 8888, - "estimated_eval_tokens": 18032 + "serialized_chars": 36162, + "estimated_input_tokens": 9041, + "estimated_eval_tokens": 18338 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a mixed set of unrelated issues across tokenization, vision/video, trainer behavior, config loading, dtype/device mismatches, and model-specific regressions. All soft pairs look like superficial title similarity rather than the same underlying bug, so no duplicate canonicalization is warranted.", - "confidence": 0.98, - "canonical_issue_reason": "No clear canonical issue: the reports span different models, subsystems, and failure modes, and none appears to subsume the others as the same bug.", + "summary": "This cluster is highly heterogeneous: most items are distinct model-specific regressions or unrelated loader/trainer bugs, so none of the soft edges are true duplicates. The best issue representative is the broad, high-impact tokenizer regression in issue 45920.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45920 is the broadest, most central regression here: it impacts AutoTokenizer across multiple model families, has clear reproduction, and reflects the kind of high-signal user-facing breakage the cluster otherwise lacks.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a strong global representative for deduping; the closest matches are still different code paths or model families.", + "best_issue_reason": "issue:45920 has the strongest combination of breadth, impact, and documentation quality among the issues in this noisy cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Both are image-model inference bugs, but they hit different architectures and different failure modes: SigLIP2 zero probabilities vs Perceiver non-default-resolution failure." + }, { "left": "issue:44479", "right": "issue:45381", "accept": false, - "reason": "Both involve Qwen VL/video wording, but one is a v5.3.0 video-input regression across several models and the other is a still-image position-id scaling bug in Qwen2.5-VL; different bug and fix path." + "reason": "Same general Qwen VL/video area, but not the same concrete bug: one is a broad v5.3.0 regression across several models, the other is a specific qwen2.5-vl vision_position_ids issue." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Janus image generation test failure and Phi-4-multimodal model loading failure are unrelated model-specific issues." + "reason": "Different model families and paths: Janus image generation test failure versus Phi-4 multimodal loading failure." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Qwen3 MoE deepspeed weight conversion and Smollm3 RoPE-layer dropping are entirely different problems." + "reason": "Unrelated bugs: Qwen3 MoE weight conversion with accelerate/deepspeed versus SmolLM3 RoPE-layer configuration mismatch." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "TrainerState token counting not updating is unrelated to a custom trainer signature breaking on a new keyword argument." + "reason": "Trainer token-count tracking not updating is unrelated to CustomTrainer.compute_loss() receiving an unexpected keyword argument." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "AddedToken duplicate-argument error and Mistral regex patch AttributeError involve different tokenizer failure points." + "reason": "Tokenizer AddedToken keyword handling and Mistral regex patching fail in different code paths with different root causes." }, { "left": "issue:43582", "right": "issue:46821", "accept": false, - "reason": "Apple Silicon warmup TypeError and TimesFM window_size AttributeError are different platform/model bugs." + "reason": "Apple Silicon allocator warmup TypeError and TimesFM window_size AttributeError are unrelated runtime issues." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "MobileLLM loading key error and Float8 storage lookup error are unrelated loading/runtime failures." + "reason": "Loading facebook/MobileLLM-125M key errors and Float8 storage lookup failures are distinct loader/runtime problems." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "PIL backend torchvision dependency regression and Mistral tokenizer regex crash are not the same issue." + "reason": "One is an image-processor torchvision dependency regression; the other is a tokenizer regex patch crash." }, { - "left": "issue:44737", - "right": "issue:46463", + "left": "issue:43716", + "right": "issue:46701", "accept": false, - "reason": "XLNet CPU arange regression and T5 decoder-input validation message are different code paths and not duplicates." + "reason": "Different symptoms and code paths: BLIP2 dtype mismatch versus DiffusionGemma image-text-to-text pipeline failure." }, { - "left": "issue:43716", - "right": "issue:46701", + "left": "issue:44737", + "right": "issue:46463", "accept": false, - "reason": "Mistral-3 image-preprocessor dtype mismatch and DiffusionGemma pipeline breakage are separate multimodal issues." + "reason": "XLNet CPU tensor creation bug and T5 decoder-input validation error message are not the same underlying issue." }, { "left": "issue:43756", "right": "issue:46399", "accept": false, - "reason": "Smollm3 RoPE-layer bug and Qwen3.5 kernel registration crash are unrelated model implementation problems." + "reason": "SmolLM3 RoPE-layer loss and Qwen3.5 kernel-registration crash are unrelated model implementation bugs." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Smollm3 RoPE-layer count bug and Qwen2.5-VL rope-index scaling bug both mention RoPE but are different bugs in different models." + "reason": "Both involve Qwen-family rotary/position behavior, but one is about SmolLM3 layer dropping and the other is a Qwen2.5-VL still-image position-id scaling bug." }, { - "left": "issue:43531", - "right": "issue:44821", + "left": "issue:44568", + "right": "issue:46491", "accept": false, - "reason": "Qwen3-MoE sliding_window issue and AutoImageProcessor-from-URL loading issue are unrelated." + "reason": "Both are tokenizer regressions, but one is BOS/EOS insertion behavior and the other is code-lama leading-space round-trip loss." }, { - "left": "issue:44568", - "right": "issue:46491", + "left": "issue:43531", + "right": "issue:44821", "accept": false, - "reason": "mdeberta add_special_tokens regression and CodeLlama space-preservation regression are distinct tokenizer behaviors." + "reason": "Qwen3-MoE sliding-window behavior and AutoImageProcessor URL loading are unrelated." }, { "left": "issue:43906", "right": "issue:46468", "accept": false, - "reason": "An isolated reproduction of an old issue and an official Nemotron example breakage are not the same bug." + "reason": "An isolated reproduction of an old issue and a broken Nemotron example are not the same concrete defect." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "PEFT version bump is a dependency/versioning issue; auto_mappings referencing removed configs is a separate CI breakage." + "reason": "PEFT version constraint problems are unrelated to stale auto_mappings references to removed configs." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Blip2 dtype propagation bug and SwitchTransformers/TimmWrapperModel bfloat16 mismatches are different model-specific dtype issues." + "reason": "Both mention dtype, but they affect different model families and different inference paths." }, { "left": "issue:45823", "right": "issue:46463", "accept": false, - "reason": "Gemma4 PLE device mismatch and T5 decoder-input validation messaging are unrelated." - }, - { - "left": "issue:43697", - "right": "issue:46829", - "accept": false, - "reason": "RTDetrV2 output changes in v5 and VideoPrism hidden_states field mismatch are different model-output regressions." + "reason": "Gemma4 device-map mismatch and T5 decoder validation messaging are unrelated bugs." }, { "left": "issue:24643", "right": "issue:29870", "accept": false, - "reason": "DeepSpeed weight-dimension runtime error and BART generate() decoder_attention_mask assignment bug are unrelated training/generation failures." + "reason": "DeepSpeed training weight-shape error and BART generate() unbound-local mask error are different failures." }, { "left": "issue:43906", "right": "issue:46693", "accept": false, - "reason": "An isolated reproduction of a prior issue and an FA2 long-context performance regression are not duplicates." + "reason": "The first is a reproduction of another issue; the second is a performance regression in long-context generation with FA2." + }, + { + "left": "issue:43697", + "right": "issue:46829", + "accept": false, + "reason": "RTDetrV2 output changes in v5 and VideoPrism hidden_states misassignment are distinct model-output bugs." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "gguf_file loader breakage and sleng-bert loading failure are different tokenizer/model-loading reports." + "reason": "Both are load-time problems, but one is a gguf_file integration breakage and the other is a specific model load failure; not the same root cause." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "XLNet positional encoding CPU regression and flash_attention sink-less-model AttributeError are unrelated." - }, - { - "left": "issue:44625", - "right": "issue:44877", - "accept": false, - "reason": "Qwen3.5 num_labels propagation and strict granite_speech config loading are different config bugs." + "reason": "XLNet CPU arange placement and flash_attention sink-less crash are unrelated implementation bugs." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BEiT Fast label reduction bug and single-label classification zero-loss bug are different classification/preprocessing issues." + "reason": "BEiT label reduction and sequence-classification zero-loss behavior are different issues, despite both touching classification preprocessing/training." }, { - "left": "issue:45357", - "right": "issue:45920", + "left": "issue:44625", + "right": "issue:44877", "accept": false, - "reason": "Qwen3.5 visual encoder save_pretrained regression and AutoTokenizer wrong token IDs regression affect different subsystems." + "reason": "Qwen3.5 num_labels propagation and strict-config loading failures are separate config-handling bugs." }, { - "left": "issue:43540", - "right": "issue:46693", + "left": "issue:45357", + "right": "issue:45920", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError and FA2 long-context performance regression are unrelated." + "reason": "Qwen3.5 save_pretrained visual-key corruption and AutoTokenizer wrong token IDs are different code paths and failure modes." } ] }, @@ -7115,7 +7121,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28282", "issue:29159", @@ -7128,8 +7134,8 @@ "issue:43452", "issue:43525", "issue:43531", + "issue:43540", "issue:43643", - "issue:43646", "issue:43653", "issue:43688", "issue:43697", @@ -7162,15 +7168,17 @@ "issue:46178", "issue:46463", "issue:46491", + "issue:46693", "issue:46701", "issue:46829" ], "soft_pairs": [ + "issue:43540|issue:46693", "issue:44913|issue:46121", "issue:28282|issue:32090", - "issue:43761|issue:46829", "issue:43697|issue:45216", "issue:43720|issue:44898", + "issue:43761|issue:46829", "issue:43643|issue:45375", "issue:45200|issue:46463", "issue:44466|issue:45005", @@ -7179,8 +7187,8 @@ "issue:44493|issue:46050", "issue:43653|issue:44361", "issue:43819|issue:44811", - "issue:43525|issue:46701", "issue:43688|issue:43749", + "issue:43525|issue:46701", "issue:38175|issue:43295", "issue:43452|issue:44843", "issue:43994|issue:44898", @@ -7189,200 +7197,199 @@ "issue:45542|issue:46178", "issue:42898|issue:46491", "issue:43295|issue:45356", - "issue:43646|issue:43749", "issue:29159|issue:33357", "issue:43653|issue:44568" ], - "prepared_review_unit_hash": "cebd70c0ad2ca2f2107f53e3fce787728565b55bdde83e66da49dd3f7c45fb5e", + "prepared_review_unit_hash": "8ca91472a8b5224c5f02bc28cdfebfbe04398250b1e9427ae695a6514435b4eb", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 35343, - "estimated_input_tokens": 8836, - "estimated_eval_tokens": 17928 + "serialized_chars": 35916, + "estimated_input_tokens": 8979, + "estimated_eval_tokens": 18214 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous and should not be merged wholesale. The only strong duplicate-looking pair is the tied-weights/save_pretrained regression in 44466 and 45005.", - "confidence": 0.88, - "canonical_issue_reason": "No single cluster-wide canonical issue is obvious because most items are unrelated. Among the true duplicate subcluster, 45005 is the best representative because it describes the broader tied-weights regression, while 44466 is a more specific device-dependent symptom.", + "summary": "Most items are unrelated one-off issue reports across different models/subsystems. The only soft pair that plausibly describes the same underlying bug is the v5 tied-weights serialization regression: issue 44466 and issue 45005.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45005 is the best representative because it is broader and more detailed: it describes the tied-weights regression in v5 across translation models, which subsumes the narrower lm_head.weight/device-specific serialization problem in issue 44466.", "canonical_pr_reason": null, - "best_issue_reason": "45005 is the best issue to keep as the representative for the tied-weights save_pretrained problem; it is broader and subsumes the specific lm_head.weight serialization behavior reported in 44466.", + "best_issue_reason": "Issue 45005 is the strongest overall issue candidate for the cluster because it most clearly captures the shared tied-weights serialization bug and appears to be the broader root report.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "Different problems: video input processing error in Qwen3OmniMoe vs a FA2 long-context performance regression." + }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "Both involve rope/config settings, but one is a config reload defaulting issue and the other is a TypeError in a rope validation helper; not the same bug." + "reason": "Both mention rope-related code, but one is config reload losing rotary_pct while the other is a TypeError in rope validation for JSON-loaded lists." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "Completely different failures: missing PyTorch import vs Trainer GPU broadcast TypeError." - }, - { - "left": "issue:43761", - "right": "issue:46829", - "accept": false, - "reason": "Both mention hidden_states, but they hit different model families and different output bugs; too broad to be the same underlying fix." + "reason": "Unrelated failures: missing PyTorch import for AutoModel vs a Trainer GPU broadcast TypeError on NoneType." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "RTDetrV2 inference mismatch and Qwen3.5 save_pretrained checkpoint corruption are unrelated." + "reason": "Different model/regression bugs: RTDetrV2 output drift vs Qwen3.5 save_pretrained checkpoint corruption." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "Packed-weight loading in BitNet and Perceiver position-encoding failures are different code paths and symptoms." + "reason": "Different code paths: BitNet packed-weight unpacking during accelerate loading vs Perceiver positional interpolation on non-default image size." + }, + { + "left": "issue:43761", + "right": "issue:46829", + "accept": false, + "reason": "Both are output-field issues in vision models, but they affect different models and different incorrect fields; not the same underlying bug." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "Both concern config field loss, but one is a generic trust_remote_code AutoConfig issue and the other is a specific @strict drop of one field; not enough to treat as the same bug." + "reason": "Both involve config fields, but one is trust_remote_code returning incomplete configs while the other is a specific Qwen vision config field dropped by strict validation." }, { "left": "issue:45200", "right": "issue:46463", "accept": false, - "reason": "Gemma 4 text-only fine-tuning mm_token_type_ids and T5 decoder-input validation error messaging are unrelated." + "reason": "Unrelated: Gemma 4 text-only token defaults vs an incorrect decoder-input validation message for T5." }, { "left": "issue:44466", "right": "issue:45005", "accept": true, - "reason": "Both report the same tied-weights/save_pretrained regression in v5; 44466 is a concrete lm_head.weight/device symptom of the broader issue described in 45005." + "reason": "Both describe the same tied-weights/save_pretrained regression in v5, centered on lm_head.weight serialization and translation models." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "DeepSpeed ZeRO-3 Gemma loading IndexError and trainer.evaluate after train are unrelated trainer/loading failures." + "reason": "Different failures: DeepSpeed ZeRO-3 loading IndexError for Gemma3 vs trainer.evaluate() failing after training." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "Both are tokenizer loading errors, but they involve different tokenizers and different root causes." + "reason": "Both are tokenizer-loading problems, but they involve different tokenizer classes and distinct failure modes." }, { "left": "issue:44493", "right": "issue:46050", "accept": false, - "reason": "Unexpected position-id key warnings are not the same as quantized save_pretrained/from_pretrained breakage." + "reason": "Both mention unexpected keys/quantization, but one is a generic position-id key issue and the other is sinq quantization save/load corruption." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "BigBird special-token decode failure and MLukeTokenizer task AttributeError are unrelated tokenizer bugs." + "reason": "Different issues: BigBird mask token registration/empty decode vs MLukeTokenizer AttributeError on tasks." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "DAC latent reconstruction mismatch and Whisper batch_decode skip_special_tokens handling are different subsystems." + "reason": "Unrelated: DAC latent reconstruction mismatch vs Whisper processor.batch_decode ignoring skip_special_tokens." }, { - "left": "issue:43525", - "right": "issue:46701", + "left": "issue:43688", + "right": "issue:43749", "accept": false, - "reason": "Missing pad_token_id in Llama4Config and DiffusionGemma image-text pipeline failure are unrelated." + "reason": "Different subsystems: auxiliary-loss normalization in MoE models vs broken FSDP CPU RAM efficient loading." }, { - "left": "issue:43688", - "right": "issue:43749", + "left": "issue:43525", + "right": "issue:46701", "accept": false, - "reason": "Auxiliary-loss normalization and FSDP CPU RAM efficient loading are different bugs." + "reason": "Different model families and failures: Llama4Config missing pad_token_id vs DiffusionGemma image-text pipeline breakage." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Siglip2 zero probabilities and custom processor.tokenizer regression are unrelated." + "reason": "Both are regressions, but one is SigLIP2 zero-probabilities output quality and the other is CLIPVisionModel hidden_states handling." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "gguf_file loading breakage and offline model_info() call in mistral regex patch are different from one another." + "reason": "Both involve AutoTokenizer.from_pretrained, but one is gguf_file breakage for tokenizer/model loading and the other is an offline-mode mistral regex hub call." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "Siglip2 nonsensical outputs and Perceiver non-default-resolution failure are unrelated model-specific regressions." + "reason": "Different vision-model bugs: nonsensical SigLIP2 outputs vs Perceiver resize/interpolation failure." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "Gemma 4 processor import failure and Windows ROCm torch.distributed dependency issue are unrelated." + "reason": "Unrelated dependency/import problems: Gemma 4 processor loading needs ReasoningEffort vs Windows ROCm torch.distributed import/type issues." }, { "left": "issue:43531", "right": "issue:43742", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and MobileLLM-125M key error are different loading/model config problems." + "reason": "Different model-specific bugs: Qwen3-MoE sliding_window behavior vs a key error loading MobileLLM-125M." }, { "left": "issue:45542", "right": "issue:46178", "accept": false, - "reason": "Both touch backend detection, but one is tensorboard-without-TensorFlow and the other is an unmapped tensorflow_text backend; not the same bug." + "reason": "Both involve backend detection, but one is tensorboard-without-TensorFlow backend handling and the other is a missing tensorflow_text backend mapping entry." }, { "left": "issue:42898", "right": "issue:46491", "accept": false, - "reason": "Both are tokenizer round-trip/cleanup regressions, but 46491 is CodeLlama-specific leading-space loss and 42898 is a broader v5 cleanup behavior change; not clearly the same fix." + "reason": "Related only at a broad tokenizer level; one is a global clean_up_tokenization_spaces semantic change, the other is a CodeLlama leading-space round-trip regression." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "CLIPVisionModel hidden_states regression and Kimi-K2.5 tokenizer codec/regression warning are unrelated." - }, - { - "left": "issue:43646", - "right": "issue:43749", - "accept": false, - "reason": "Custom model initialization breakage and FSDP CPU RAM loading breakage are unrelated." + "reason": "Different regressions: CLIPVisionModel hidden_states output vs Kimi-K2.5 tokenizer codec handling and mistral regex warning behavior." }, { "left": "issue:29159", "right": "issue:33357", "accept": false, - "reason": "Slow vs fast tokenizer inconsistency and MacOS bus error on CLIP are unrelated." + "reason": "Unrelated: slow-vs-fast tokenizer inconsistency vs a MacOS bus error with a community CLIP model." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "Both involve special tokens, but BigBird mask-token decode and mdeberta-v3 BOS/EOS insertion are different tokenizer behaviors." + "reason": "Both are special-token/tokenizer bugs, but they affect different tokenizers and different behaviors; not the same underlying issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: it correctly treats the cluster as mostly heterogeneous, and the only accepted duplicate pair (44466/45005) is supported by the issue titles and framed conservatively. The other soft-edge rejections are appropriately cautious and do not overclaim sameness." + "feedback": "The cluster summary is conservative and grounded in the packet. The only clearly related soft pair is 44466/45005, and the rest of the pairwise rejections are appropriately cautious given the issue titles. No overstatement stands out." }, "evaluator_used": true, "retried": false, @@ -7401,27 +7408,27 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:32090", "issue:33666", "issue:36010", "issue:36331", - "issue:39692", "issue:41720", "issue:43262", "issue:43278", "issue:43295", "issue:43452", - "issue:43493", "issue:43504", "issue:43550", "issue:43577", "issue:43606", "issue:43643", + "issue:43646", "issue:43653", "issue:43688", "issue:43720", + "issue:43749", "issue:43756", "issue:43782", "issue:43927", @@ -7453,6 +7460,7 @@ "issue:46821" ], "soft_pairs": [ + "issue:43646|issue:43749", "issue:44464|issue:44743", "issue:43452|issue:43577", "issue:43756|issue:46402", @@ -7465,8 +7473,8 @@ "issue:44484|issue:44704", "issue:45491|issue:46002", "issue:43550|issue:44361", - "issue:33666|issue:36010", "issue:43688|issue:45440", + "issue:33666|issue:36010", "issue:44734|issue:46612", "issue:32090|issue:33666", "issue:44464|issue:44977", @@ -7475,238 +7483,236 @@ "issue:44977|issue:46693", "issue:43504|issue:43550", "issue:43504|issue:43720", - "issue:43504|issue:44610", "issue:43577|issue:44554", "issue:43643|issue:44561", "issue:43720|issue:43927", "issue:33666|issue:36331", + "issue:43504|issue:44610", "issue:43606|issue:43927", "issue:43278|issue:43688", "issue:43782|issue:46693", - "issue:43262|issue:43295", - "issue:39692|issue:43493", - "issue:43720|issue:44361" + "issue:43262|issue:43295" ], - "prepared_review_unit_hash": "2ddd0c213ab6da9bb2c713a2c20db3177c115764a6226c39eb5e108b5bfdc98d", + "prepared_review_unit_hash": "ffd7ee935324fc01dfea8518bfa027c9d84294fe8f1f84b84591f72a6e05cfe3", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37785, - "estimated_input_tokens": 9447, - "estimated_eval_tokens": 19150 + "soft_pair_count": 32, + "serialized_chars": 37395, + "estimated_input_tokens": 9349, + "estimated_eval_tokens": 18954 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set; the items share surface-level similarities around Transformers loading/generation/training, but they describe many different bugs across unrelated models and code paths.", - "confidence": 0.96, - "canonical_issue_reason": "No canonical issue: the cluster is heterogeneous, with no single underlying bug or fix that unifies these reports.", + "summary": "This cluster is mostly a grab-bag of unrelated regressions, loading bugs, and model-specific issues. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "If a single representative is needed, issue 43646 is the broadest cross-cutting regression report in the set, but the cluster is too heterogeneous to justify a true canonical duplicate target.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong cluster representative for dedupe triage; the overlaps are mostly broad subsystem/topic similarity rather than the same defect.", + "best_issue_reason": "Issue 43646 is the best overall representative because it describes a wide Transformers 5.0 breakage affecting custom model initialization, which is broader than the mostly model-specific reports here.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Different failure domains: custom model initialization vs FSDP CPU-RAM-efficient loading." + }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Both concern generation/cache behavior, but one is about compiled forward inconsistency and the other about recurrent states resetting in modular_qwen3_5; different bugs and code paths." + "reason": "Both involve generation/caching, but one is chunked generation with compiled forward and the other is recurrent-state reset in modular Qwen3.5." }, { "left": "issue:43452", "right": "issue:43577", "accept": false, - "reason": "Different failures: gguf_file loading for tokenizer/model creation vs Blip2 dtype propagation during model load." + "reason": "Tokenizer/model loading via gguf_file is unrelated to Blip2 dtype propagation." }, { "left": "issue:43756", "right": "issue:46402", "accept": false, - "reason": "One is a Smollm3 RoPE-layer configuration mismatch; the other is Sam3 text encoder weights not loading. Not the same bug." + "reason": "Different models and bugs: RoPE layer dropping vs missing text encoder weights on load." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "Aux-loss normalization and a T5 apex RMSNorm memory leak are unrelated issues." + "reason": "Aux-loss normalization bug is unrelated to T5 using apex RMSNorm and leaking memory." }, { "left": "issue:43653", "right": "issue:43927", "accept": false, - "reason": "Both affect token-related behavior, but one is BigBird mask-token registration and the other is DiaConfig losing custom token IDs after save/load." + "reason": "Tokenizer special-token registration issue is not the same as config token IDs being lost on save/load." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "Aux-loss normalization is unrelated to tokenization changes across versions." + "reason": "Aux-loss normalization is unrelated to tokenizer behavior changes across versions." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Both touch tokenizer loading, but one is an unconditional hub lookup breaking offline mode; the other is a specific tokenizer-load regression for EMBEDDIA/est-roberta." + "reason": "Offline hub call in tokenizer patching is different from a model-specific tokenizer load failure." }, { "left": "issue:44821", "right": "issue:46821", "accept": false, - "reason": "Different domains entirely: AutoImageProcessor URL loading vs TimesFM window_size AttributeError." + "reason": "Image processor URL loading and TimesFM window_size AttributeError are unrelated." }, { "left": "issue:44479", "right": "issue:44625", "accept": false, - "reason": "One is a multimodal video-input regression across Qwen VL models; the other is a Qwen3.5 config propagation issue for num_labels." + "reason": "Video-input regression in Qwen variants is different from num_labels propagation in Qwen3.5 config." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "max_shard_size defaulting is unrelated to AutoProcessor failing to pass kwargs through to cached_file." + "reason": "save_pretrained shard-size question is unrelated to AutoProcessor kwargs forwarding." }, { "left": "issue:45491", "right": "issue:46002", "accept": false, - "reason": "Both mention NaNs, but the model/components and trigger are different: Gemma3 sliding-window GPU batching vs classification heads after checkout/init." + "reason": "Both mention NaNs, but one is Gemma3 sliding-window attention and the other is classification-head init/checkpointing." }, { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "torch.compile/SDPA on Bamba-9B-v2 is unrelated to MLukeTokenizer task-time AttributeError." + "reason": "torch.compile/SDPA failure is unrelated to MLukeTokenizer AttributeError." }, { - "left": "issue:33666", - "right": "issue:36010", + "left": "issue:43688", + "right": "issue:45440", "accept": false, - "reason": "Multi-GPU Qwen2-VL training and a GenerationMixin import error are separate issues." + "reason": "Aux-loss normalization and DeepseekV3MoE implementation divergence are different problems." }, { - "left": "issue:43688", - "right": "issue:45440", + "left": "issue:33666", + "right": "issue:36010", "accept": false, - "reason": "Aux-loss normalization is not the same as DeepseekV3MoE diverging from the remote implementation." + "reason": "Multi-GPU Qwen2-VL training and GenerationMixin import failure are unrelated." }, { "left": "issue:44734", "right": "issue:46612", "accept": false, - "reason": "Both involve cache behavior, but one is serving KV-cache continuation indexing and the other is beam-search cache reorder support for certain models." + "reason": "Both involve generation internals, but the bugs are different: KV-cache indexing crash vs skipped beam-search cache reorder." }, { "left": "issue:32090", "right": "issue:33666", "accept": false, - "reason": "Both mention training, but one is a Trainer _gpu_broadcast_one NoneType error and the other is general Qwen2-VL multi-GPU training." + "reason": "Generic Trainer broadcast error is not the same as Qwen2-VL multi-GPU training behavior." }, { "left": "issue:44464", "right": "issue:44977", "accept": false, - "reason": "Different regressions: compiled-forward generation inconsistency vs Qwen3.5 flash-attention generation failure." + "reason": "Compiled-forward chunked generation inconsistency is distinct from Qwen3.5 flash-attention generation failure." }, { "left": "issue:44466", "right": "issue:46050", "accept": false, - "reason": "Both concern save/load behavior, but one is device-dependent tied-weight serialization and the other is sinq quantized-model persistence." + "reason": "Device-dependent lm_head serialization is not the same as sinq quantization save/load breakage." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "Qwen3 auto device-mapping cudaErrorAssert is unrelated to chunked generation inconsistency with compiled forward." + "reason": "cudaErrorAssert on auto device mapping is unrelated to compiled-forward generation drift." }, { "left": "issue:44977", "right": "issue:46693", "accept": false, - "reason": "FA2 is a shared theme, but one is a Qwen3.5 correctness regression and the other is a broader long-context performance regression." + "reason": "Correctness failure in Qwen3.5 flash-attention is not the same as a performance regression in FA2 long-context generation." }, { "left": "issue:43504", "right": "issue:43550", "accept": false, - "reason": "Beit legacy-field loading and Bamba torch.compile/SDPA are unrelated." + "reason": "Legacy-field model loading bug is unrelated to Bamba torch.compile SDPA failure." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Legacy-field preset loading and BitNet packed-weight loading during accelerate are different bugs." - }, - { - "left": "issue:43504", - "right": "issue:44610", - "accept": false, - "reason": "Beit pretrained preset loading is unrelated to OmDet-Turbo\u2019s processor/model image-size mismatch." + "reason": "Legacy-field load failure is different from packed-weights handling during accelerate loading." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Blip2 dtype propagation and an MPS attention correctness issue are different code paths." + "reason": "Blip2 dtype propagation and MPS attention correctness are unrelated." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Both mention trust_remote_code, but one is missing fields in AutoConfig output and the other is a v5 removal of is_torch_fx_available breaking remote models." + "reason": "Missing fields from trust_remote_code config loading is different from removal of is_torch_fx_available breaking remote-code models." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "BitNet packed-weight loading and DiaConfig token-ID loss are unrelated." + "reason": "Accelerate packed-weight loading and DiaConfig token-ID persistence are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Qwen2-VL multi-GPU training and compute_loss signature mismatch are different trainer issues." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Beit preset loading bug is unrelated to OmDet-Turbo processor/model input-size mismatch." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch in bark-small is unrelated to DiaConfig losing token IDs." + "reason": "CPU-offload device mismatch does not match DiaConfig token-ID loss." }, { "left": "issue:43278", "right": "issue:43688", "accept": false, - "reason": "Embedding dtype drift in eval is unrelated to auxiliary-loss normalization." + "reason": "Embedding dtype drift between train/eval is unrelated to auxiliary-loss normalization." }, { "left": "issue:43782", "right": "issue:46693", "accept": false, - "reason": "Qwen3VL weight_only loading error and a long-context FA2 performance regression are different failures." + "reason": "weight_only loading error for Qwen3VL is unrelated to FA2 performance regression." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and processor.tokenizer/image-passing regression are separate processor issues." - }, - { - "left": "issue:39692", - "right": "issue:43493", - "accept": false, - "reason": "Both are about SigLIP2, but one is a broken documentation example and the other is a core HF-vs-JAX implementation discrepancy." - }, - { - "left": "issue:43720", - "right": "issue:44361", - "accept": false, - "reason": "BitNet accelerate-loading failure is unrelated to MLukeTokenizer task AttributeError." + "reason": "Audio sampling-rate default bug and processor.tokenizer regression are different processor issues." } ] }, @@ -7728,7 +7734,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28218", "issue:33357", @@ -7739,13 +7745,11 @@ "issue:43295", "issue:43452", "issue:43493", - "issue:43504", "issue:43525", "issue:43572", - "issue:43575", - "issue:43606", "issue:43646", "issue:43701", + "issue:43720", "issue:43824", "issue:43825", "issue:43844", @@ -7765,10 +7769,10 @@ "issue:44849", "issue:44964", "issue:44977", + "issue:45072", "issue:45290", "issue:45357", "issue:45405", - "issue:45520", "issue:45584", "issue:45812", "issue:45823", @@ -7777,13 +7781,17 @@ "issue:46209", "issue:46365", "issue:46489", - "issue:46710" + "issue:46710", + "issue:46821", + "issue:46832" ], "soft_pairs": [ - "issue:43525|issue:43572", - "issue:44554|issue:46365", + "issue:39692|issue:43493", + "issue:43720|issue:44361", "issue:44514|issue:45290", "issue:44964|issue:45405", + "issue:44554|issue:46365", + "issue:43525|issue:43572", "issue:43493|issue:43825", "issue:44977|issue:46209", "issue:44625|issue:45584", @@ -7792,233 +7800,232 @@ "issue:28218|issue:33357", "issue:43994|issue:46192", "issue:42796|issue:44977", - "issue:43278|issue:43844", "issue:44849|issue:46209", + "issue:43278|issue:43844", "issue:39692|issue:43295", "issue:43493|issue:43844", "issue:33357|issue:33666", "issue:43646|issue:44466", "issue:46209|issue:46710", "issue:44568|issue:46489", - "issue:43278|issue:43701", "issue:44849|issue:45357", - "issue:43927|issue:44361", + "issue:43278|issue:43701", "issue:44792|issue:45405", + "issue:43927|issue:44361", + "issue:45072|issue:46832", "issue:43824|issue:44589", + "issue:46821|issue:46832", "issue:44568|issue:45812", - "issue:44488|issue:44843", - "issue:43575|issue:45405", - "issue:44589|issue:45520", - "issue:43504|issue:43606" + "issue:44488|issue:44843" ], - "prepared_review_unit_hash": "6597dfc45150af600b7f6b6e702e2d5c8a7c06367fa93d651a0533d848af79cd", + "prepared_review_unit_hash": "bfdc4d8c5d4a412964245fdae1e741a36a53b7e2e54ecdc47b5177dd9e56224b", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36722, - "estimated_input_tokens": 9181, - "estimated_eval_tokens": 18618 + "soft_pair_count": 31, + "serialized_chars": 37067, + "estimated_input_tokens": 9267, + "estimated_eval_tokens": 18790 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very broad and mostly contains distinct model/config/tokenizer regressions; none of the soft pairs look like true duplicates.", - "confidence": 0.87, - "canonical_issue_reason": "issue:45812 is the strongest representative issue: it is a broad tokenizer regression affecting multiple Granite models and has the clearest cross-cutting impact.", + "summary": "This cluster is heterogeneous: it contains several unrelated tokenizer/config/load regressions, mostly around Transformers v5+, but none of the proposed soft pairs look like the same underlying bug. I would use the Granite tokenizer regression as the closest representative, but with low confidence.", + "confidence": 0.66, + "canonical_issue_reason": "Issue 45812 is the most concrete and central tokenizer regression here: a silent v4\u2192v5 change causing wrong token IDs across Granite models, which matches the tokenizer-focused theme better than the broader or model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 is the best global issue candidate for the same reason: it is the most general, high-signal tokenizer bug among the set.", + "best_issue_reason": "45812 is the best anchor for this cluster because it is broad, user-visible, and directly about tokenizer output correctness rather than a one-off model-specific failure.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43525", - "right": "issue:43572", + "left": "issue:39692", + "right": "issue:43493", "accept": false, - "reason": "Both are missing-pad-token config regressions, but they affect different model configs and fields (`pad_token_id` vs `pad_token_idx`), so they are not the same bug." + "reason": "Different problems: documentation/example errors and quantization failure vs an implementation discrepancy with JAX." }, { - "left": "issue:44554", - "right": "issue:46365", + "left": "issue:43720", + "right": "issue:44361", "accept": false, - "reason": "Completely unrelated: one is an MPS attention correctness issue, the other is a Trainer import failure tied to `datasets`." + "reason": "Unrelated subsystems: BitNet accelerate-loading packed weights vs an MLukeTokenizer AttributeError." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Both involve `apply_chat_template`, but the failing inputs and code paths differ (`padding=False` batched VL input vs assistant tool-call messages with no content)." + "reason": "Both involve apply_chat_template crashes, but the triggers and code paths differ (batched padding=False vs tool-call assistant messages without content)." }, { "left": "issue:44964", "right": "issue:45405", "accept": false, - "reason": "One is a model-loading failure for Phi-4 multimodal; the other is a dependency/version bump problem for `MIN_PEFT_VERSION`." + "reason": "Model load failure vs dependency version bump; not the same underlying bug." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "MPS attention correctness issue and Trainer import failure are unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Same general class of missing config attributes after v5, but different models, attributes, and failure modes." }, { "left": "issue:43493", "right": "issue:43825", "accept": false, - "reason": "SigLIP2 implementation mismatch and pipeline translation-task error are unrelated issues in different subsystems." + "reason": "SigLIP2 implementation discrepancy and a pipeline error-message regression are unrelated." }, { "left": "issue:44977", "right": "issue:46209", "accept": false, - "reason": "Different failures: Qwen3.5 flash-attention generation bug versus FP8 inference failure on B200 hardware." + "reason": "Different hardware/precision failures: flash-attention generation vs FP8 inference on B200." }, { "left": "issue:44625", "right": "issue:45584", "accept": false, - "reason": "Different model-specific bugs: Qwen3.5 `num_labels` propagation versus Whisper empty-transcription handling after special-token alignment." + "reason": "Different features and symptoms: num_labels propagation vs Whisper empty transcription generation failure." }, { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "Both are loading-related, but one is a `gguf_file` regression in `from_pretrained`, while the other is a specific model load failure for `cjvt/sleng-bert`." + "reason": "Both are load-time failures, but they affect different models and likely different root causes." }, { "left": "issue:45823", "right": "issue:46050", "accept": false, - "reason": "Different root causes: device placement mismatch in Gemma4 versus quantized model save/load breakage in `sinq`." + "reason": "Different bugs: device-map mismatch in Gemma4 forward vs sinq save/load quantization breakage." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer spacing behavior and a MacOS bus error in CLIP are unrelated." + "reason": "Tokenizer spacing bug and a MacOS bus error in CLIP are not the same issue." }, { "left": "issue:43994", "right": "issue:46192", "accept": false, - "reason": "One is a SigLIP2 inference-quality regression; the other is an argument-merging bug where `return_tensors` is ignored." + "reason": "Wrong SigLIP2 outputs and silently ignored return_tensors are unrelated." }, { "left": "issue:42796", "right": "issue:44977", "accept": false, - "reason": "A tokenizer BPE-marker handling bug and a Qwen3.5 flash-attention generation bug are not the same underlying defect." + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation failure are different bugs." }, { - "left": "issue:43278", - "right": "issue:43844", + "left": "issue:44849", + "right": "issue:46209", "accept": false, - "reason": "One concerns dtype drift between training and evaluation; the other is a gradient-growth issue with DeepSpeed ZeRO-3." + "reason": "Qwen3.5 hidden-state output bug is unrelated to FP8 inference on B200." }, { - "left": "issue:44849", - "right": "issue:46209", + "left": "issue:43278", + "right": "issue:43844", "accept": false, - "reason": "Different Qwen3.5 regressions: hidden-states behavior versus FP8 inference on B200." + "reason": "Evaluate-time dtype change and ZeRO-3 gradient blow-up are not the same root cause." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Documentation example errors and a custom code regression around `processor.tokenizer`/image passing are different problems." + "reason": "Different regressions: SigLIP2 doc example errors vs processor.tokenizer/custom image-tokenizer access breakage." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 implementation mismatch is unrelated to the DeepSpeed gradient regression." + "reason": "Different model/runtime issues: SigLIP2 discrepancy vs ZeRO-3 gradient anomaly." }, { "left": "issue:33357", "right": "issue:33666", "accept": false, - "reason": "MacOS bus error on a CLIP model and Qwen2-VL multi-GPU training are different issues." + "reason": "MacOS bus error in CLIP and Qwen2-VL multi-GPU training are unrelated." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization breakage and inconsistent `lm_head.weight` serialization are separate v5 regressions." + "reason": "Generic custom model initialization breakage vs lm_head.weight serialization inconsistency are distinct problems." }, { "left": "issue:46209", "right": "issue:46710", "accept": false, - "reason": "Different failures: FP8 inference on B200 versus an accuracy regression after upgrading Transformers." + "reason": "FP8 inference failure and an accuracy regression after upgrade are not the same bug." }, { "left": "issue:44568", "right": "issue:46489", "accept": false, - "reason": "Both are tokenizer regressions, but the concrete failures differ: missing special tokens for one tokenizer vs incorrect DeepSeek-Coder v1 token IDs/BPE handling." - }, - { - "left": "issue:43278", - "right": "issue:43701", - "accept": false, - "reason": "Embedding dtype drift and resume-from-checkpoint key mismatch are unrelated training bugs." + "reason": "Different tokenizer regressions affecting different models and behaviors; not a shared code-path bug." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Qwen3.5 hidden-states output bug is unrelated to incorrect visual encoder key serialization." + "reason": "Hidden-states bug and visual-encoder key serialization bug are unrelated." }, { - "left": "issue:43927", - "right": "issue:44361", + "left": "issue:43278", + "right": "issue:43701", "accept": false, - "reason": "DiaConfig custom token IDs being dropped is unrelated to MLukeTokenizer task-time AttributeError." + "reason": "Dtype mismatch in eval vs resume_from_checkpoint key mismatch are unrelated." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure is unrelated to the PEFT version bump issue." - }, - { - "left": "issue:43824", - "right": "issue:44589", - "accept": false, - "reason": "ImportError for a missing Qwen2.5-VL class and a Float8 storage type error are unrelated." + "reason": "Janus image-generation test failure and a PEFT version bump are unrelated." }, { - "left": "issue:44568", - "right": "issue:45812", + "left": "issue:43927", + "right": "issue:44361", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and behaviors; one is special-token insertion, the other is Granite token-ID generation." + "reason": "DiaConfig token-id persistence bug and MLukeTokenizer AttributeError are different failures." }, { - "left": "issue:44488", - "right": "issue:44843", + "left": "issue:45072", + "right": "issue:46832", "accept": false, - "reason": "A model loading failure and an offline-mode regression from an unconditional hub call are different bugs." + "reason": "BF16 dtype mismatch issues are unrelated to RT-DETR feature-levels shape/type error." }, { - "left": "issue:43575", - "right": "issue:45405", + "left": "issue:43824", + "right": "issue:44589", "accept": false, - "reason": "Tensor-parallel OOM on Qwen2 is unrelated to the PEFT version requirement issue." + "reason": "ImportError for a model class and missing Float8 storage type are unrelated." }, { - "left": "issue:44589", - "right": "issue:45520", + "left": "issue:46821", + "right": "issue:46832", "accept": false, - "reason": "Different low-level runtime failures: missing Float8 storage type versus `flash_attn` import key error on Python 3.13." + "reason": "TimesFM window_size AttributeError and RT-DETR forward TypeError are distinct bugs." }, { - "left": "issue:43504", - "right": "issue:43606", + "left": "issue:44568", + "right": "issue:45812", "accept": false, - "reason": "Beit pretrained loading with a legacy field and Bark CPU-offload device mismatch are unrelated." + "reason": "Both are tokenizer regressions, but they affect different tokenizers and failure modes; not the same underlying bug." } ] }, @@ -8040,7 +8047,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28218", "issue:29159", @@ -8049,12 +8056,12 @@ "issue:32090", "issue:34567", "issue:42490", - "issue:43065", "issue:43257", "issue:43299", "issue:43334", "issue:43504", "issue:43572", + "issue:43575", "issue:43606", "issue:43653", "issue:43696", @@ -8066,7 +8073,7 @@ "issue:44291", "issue:44464", "issue:44493", - "issue:44554", + "issue:44589", "issue:44869", "issue:44964", "issue:44977", @@ -8074,7 +8081,7 @@ "issue:45200", "issue:45229", "issue:45245", - "issue:45446", + "issue:45405", "issue:45479", "issue:45520", "issue:45584", @@ -8083,19 +8090,23 @@ "issue:45823", "issue:46050", "issue:46285", - "issue:46369", "issue:46424", + "issue:46468", "issue:46701", "issue:46710", "issue:46752", - "issue:46829" + "issue:46829", + "issue:46832" ], "soft_pairs": [ + "issue:44589|issue:45520", + "issue:43575|issue:45405", "issue:43257|issue:43299", + "issue:43504|issue:43606", "issue:43606|issue:44112", "issue:43334|issue:45759", - "issue:43334|issue:43572", "issue:45245|issue:46710", + "issue:43334|issue:43572", "issue:43572|issue:44291", "issue:44493|issue:46285", "issue:44464|issue:46424", @@ -8103,9 +8114,10 @@ "issue:43937|issue:46701", "issue:44464|issue:45584", "issue:45200|issue:45229", + "issue:46468|issue:46832", + "issue:43844|issue:44964", "issue:44869|issue:46752", "issue:45229|issue:45823", - "issue:43844|issue:44964", "issue:43873|issue:46050", "issue:29942|issue:30333", "issue:43994|issue:44493", @@ -8113,202 +8125,212 @@ "issue:29159|issue:32090", "issue:42490|issue:43696", "issue:44977|issue:46701", - "issue:45479|issue:46829", "issue:45084|issue:45520", "issue:28218|issue:34567", - "issue:43065|issue:46369", - "issue:44554|issue:45446" + "issue:45479|issue:46829" ], - "prepared_review_unit_hash": "64ebe77f6ca2b8c9a64942bb1eab851543b94c316469a601aadc9d8bfdf629bf", + "prepared_review_unit_hash": "9168df70f907084be56dd0c4d501da7185fb4c32fa6041ce973c2f6b4e2262a8", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 27, - "serialized_chars": 35244, - "estimated_input_tokens": 8811, - "estimated_eval_tokens": 17878 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36410, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is highly heterogeneous: tokenizer spacing, model-loading/config regressions, training/runtime errors, CI failures, and unrelated multimodal issues. None of the proposed soft pairs look like the same underlying bug or change, so I would not merge any of them as duplicates.", - "confidence": 0.86, - "canonical_issue_reason": "issue:43873 has the strongest signal in this batch (high discussion/inbound refs and an open, broadly relevant bug report), so it is the best representative issue if one must be chosen, though it is not a duplicate of the others.", + "summary": "No soft edge looks like a true duplicate; the cluster is a broad false-positive mix of tokenizer, loading/config, quantization, and model-specific regression reports. I\u2019d treat it as heterogeneous rather than one underlying bug cluster.", + "confidence": 0.87, + "canonical_issue_reason": "If a representative issue is needed, #44291 is the broadest loader-regression report in the set (generic model loading failure under transformers >= 5.0), but this cluster is too heterogeneous to treat as a clean duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 is the most active and broadly applicable issue here; it is the best triage anchor among otherwise unrelated reports.", + "best_issue_reason": "#44291 is the best overall issue representative because it captures a concrete, broadly applicable loading regression; still, it is not a true canonical anchor for the whole mixed cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Different failures: Float8 storage lookup vs flash_attn import_utils on Python 3.13; unrelated code paths." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "OOM on Qwen2 tensor-parallel load vs a PEFT version bump; not the same bug." + }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Both involve Qwen3/MoE loading, but one is about missing conversion with accelerate+deepspeed and the other about a transformers-dev load regression for Qwen3VL; different failure modes and code paths." + "reason": "Both are Qwen-family loading regressions, but one is accelerate/deepspeed weight conversion and the other is Qwen3VL MOE config loading; different concrete paths." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field load failure vs Bark CPU-offload device mismatch; unrelated." }, { "left": "issue:43606", "right": "issue:44112", "accept": false, - "reason": "Both mention CI/device mismatch, but one is a bark-small CPU-offload failure and the other is a stale GraniteSpeech test; not the same bug." + "reason": "One is a real model/offload mismatch, the other a stale GraniteSpeech CI test; same wording, different underlying issue." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Both touch Qwen3VL multimodal config handling, but one is an AttributeError for missing pad_token_id and the other is AutoModelForCausalLM.from_config not unwrapping text_config; related area, different issue." + "reason": "Both touch multimodal Qwen config handling, but pad_token_id missing and text_config unwrapping are distinct loader bugs." }, { - "left": "issue:43334", - "right": "issue:43572", + "left": "issue:45245", + "right": "issue:46710", "accept": false, - "reason": "Different models and different missing fields: Qwen3VL pad_token_id vs StableLmConfig pad_token_idx." + "reason": "Runtime category-limit error vs accuracy regression after upgrade; unrelated." }, { - "left": "issue:45245", - "right": "issue:46710", + "left": "issue:43334", + "right": "issue:43572", "accept": false, - "reason": "RuntimeError about too many categories is unrelated to an accuracy regression after upgrading Transformers." + "reason": "Both mention missing token fields after 5.0, but they affect different configs/models and different attributes." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "One is a config-field regression, the other is an init_empty_weights constructor argument incompatibility; no shared underlying bug." + "reason": "Both are 5.0-era loading regressions, but StableLm config metadata and init_empty_weights arg handling are different code paths." }, { "left": "issue:44493", "right": "issue:46285", "accept": false, - "reason": "Unexpected position-id key warnings are unrelated to CLIP text model structure changes." + "reason": "Unexpected position-id key vs CLIP text-model structure change; unrelated." }, { "left": "issue:44464", "right": "issue:46424", "accept": false, - "reason": "Chunked generation with compiled forward and cache max length handling are different code paths and symptoms." + "reason": "Compiled-forward chunked generation inconsistency vs cache max-length semantics; not the same bug." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "A legacy-field pretrained load bug is unrelated to BigBirdTokenizer mask-token registration and empty decode output." + "reason": "BEiT legacy preset loading vs BigBirdTokenizer special-token registration; unrelated." }, { "left": "issue:43937", "right": "issue:46701", "accept": false, - "reason": "A GenerationConfig validation error is not the same as an image-text-to-text pipeline failure." + "reason": "GenerationConfig validation failure vs image-text pipeline output issue; different problems." }, { "left": "issue:44464", "right": "issue:45584", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency and Whisper empty-transcription failure after align_special_tokens are distinct issues." + "reason": "Compiled-forward generation inconsistency vs Whisper empty-transcription failure; unrelated." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Defaulting mm_token_type_ids to zeros is unrelated to Gemma4 multi-GPU CUDA OOM." + "reason": "Same model family, but one is input defaulting for fine-tuning and the other is multi-GPU inference OOM." + }, + { + "left": "issue:46468", + "right": "issue:46832", + "accept": false, + "reason": "Nemotron example breakage vs RT-DETR forward type error; unrelated." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient behavior vs Phi-4 multimodal loading failure; unrelated." }, { "left": "issue:44869", "right": "issue:46752", "accept": false, - "reason": "Both are decode/chat-template crashes, but one is Whisper word-timestamp trailing replacement-character handling and the other is apply_chat_template on an empty list." + "reason": "Both are input-edge-case crashes, but one is Whisper timestamp decoding and the other is empty chat-template handling; different APIs." }, { "left": "issue:45229", "right": "issue:45823", "accept": false, - "reason": "CUDA OOM during Gemma4 inference is not the same as a device mismatch in Gemma4 PLE forward." - }, - { - "left": "issue:43844", - "right": "issue:44964", - "accept": false, - "reason": "ZeRO-3 gradient growth is unrelated to loading microsoft/Phi-4-multimodal-instruct." + "reason": "Both are Gemma4 inference issues, but OOM and device-map mismatch are different failure modes." }, { "left": "issue:43873", "right": "issue:46050", "accept": false, - "reason": "Quantized offloading behavior and sinq save/load breakage are both quantization-related but clearly different bugs." + "reason": "Quantization/offloading theme overlaps, but offload behavior and sinq save/load persistence are distinct bugs." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Flash Attention 2 CI failures and MLFlow reporting on training failure are unrelated." + "reason": "FlashAttention CI failures vs MLflow reporting on failed training; unrelated." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "Siglip2 bad outputs with AutoModel/pipeline and unexpected position-id keys are different failure modes." + "reason": "Siglip2 nonsensical outputs vs unexpected position-id key regression; unrelated." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "flash_attention.py crashing on s_aux=None is unrelated to quantized model persistence after save_pretrained/from_pretrained." + "reason": "Flash-attention sink-less-model crash vs sinq quantized model save/load breakage; unrelated." }, { "left": "issue:29159", "right": "issue:32090", "accept": false, - "reason": "Tokenizer slow/fast inconsistency and Trainer broadcast NoneType error do not share the same root cause." + "reason": "Tokenizer fast/slow inconsistency vs GPU broadcast TypeError; unrelated." }, { "left": "issue:42490", "right": "issue:43696", "accept": false, - "reason": "Trailing-slash import warning is unrelated to CUDA OOM." + "reason": "Trailing-slash import warning vs GPT-oss OOM; unrelated." }, { "left": "issue:44977", "right": "issue:46701", "accept": false, - "reason": "Qwen3.5 flash-attention generation issues and DiffusionGemma pipeline failure are separate model-specific bugs." - }, - { - "left": "issue:45479", - "right": "issue:46829", - "accept": false, - "reason": "Degenerate zero-loss classification behavior is unrelated to VideoPrism hidden_states being populated incorrectly." + "reason": "Qwen3.5 flash-attention generation issue vs DiffusionGemma image-text pipeline bug; unrelated." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Template-node compilation failure and Python 3.13 flash_attn import KeyError are different subsystems and symptoms." + "reason": "Template compiler error vs flash_attn import KeyError on Python 3.13; unrelated." }, { "left": "issue:28218", "right": "issue:34567", "accept": false, - "reason": "Tokenizer added-token spacing and TrainerState num_input_tokens_seen not updating are unrelated." - }, - { - "left": "issue:43065", - "right": "issue:46369", - "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and sam3-litetext memory overflow are both Sam3-related but not the same bug." + "reason": "Tokenizer adds extra space vs TrainerState token counter not updating; unrelated." }, { - "left": "issue:44554", - "right": "issue:45446", + "left": "issue:45479", + "right": "issue:46829", "accept": false, - "reason": "MPS attention correctness with mismatched head dims and a PyTorch version check bug in flex_attention are unrelated." + "reason": "Degenerate zero-loss classification vs VideoPrism output field mix-up; unrelated." } ] }, @@ -8330,7 +8352,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:15354", "issue:29870", @@ -8340,6 +8362,7 @@ "issue:39692", "issue:41720", "issue:42490", + "issue:43065", "issue:43232", "issue:43295", "issue:43450", @@ -8357,7 +8380,6 @@ "issue:44112", "issue:44186", "issue:44423", - "issue:44466", "issue:44554", "issue:44610", "issue:44734", @@ -8365,11 +8387,11 @@ "issue:44977", "issue:44987", "issue:45072", - "issue:45127", "issue:45198", "issue:45200", "issue:45356", "issue:45406", + "issue:45446", "issue:45520", "issue:45588", "issue:45759", @@ -8377,11 +8399,14 @@ "issue:46121", "issue:46178", "issue:46192", + "issue:46369", "issue:46710", "issue:46821" ], "soft_pairs": [ + "issue:43065|issue:46369", "issue:44423|issue:44734", + "issue:44554|issue:45446", "issue:15354|issue:38175", "issue:43493|issue:43701", "issue:41720|issue:46710", @@ -8393,233 +8418,238 @@ "issue:45588|issue:45759", "issue:43540|issue:45200", "issue:44186|issue:46192", + "issue:43295|issue:43873", "issue:43572|issue:46821", "issue:45200|issue:45823", - "issue:43295|issue:43873", "issue:45823|issue:46192", "issue:43295|issue:44610", "issue:29870|issue:29942", "issue:45520|issue:46178", "issue:43572|issue:44987", "issue:43232|issue:44849", - "issue:43450|issue:43575", "issue:43550|issue:43844", + "issue:43450|issue:43575", "issue:44554|issue:45356", "issue:37428|issue:43824", "issue:42490|issue:43493", "issue:39692|issue:46192", "issue:38175|issue:43493", - "issue:43653|issue:44112", - "issue:44466|issue:45127" + "issue:43653|issue:44112" ], - "prepared_review_unit_hash": "37193cc1a7a65bf5b1ba1b44bb95e682eaf6ebc3f3883521756b35b29d89c4ae", + "prepared_review_unit_hash": "4237fb223061a2fdc125d614f7d83e8b266b5890eaab1fc5f5c3a97e05348241", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 30, - "serialized_chars": 36531, - "estimated_input_tokens": 9133, - "estimated_eval_tokens": 18522 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37226, + "estimated_input_tokens": 9307, + "estimated_eval_tokens": 18870 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous. Most soft pairs share only a broad theme (serve, tokenizer, multimodal, flash-attn, or model-loading regressions) but not the same concrete bug or code path, so I would reject all pairwise merges. No single issue cleanly subsumes the rest; issue 43295 is the closest representative only because it is a concrete, well-scoped regression report.", - "confidence": 0.95, - "canonical_issue_reason": "No true canonical duplicate stands out; 43295 is the closest representative because it is a concrete regression with a clear repro, but the rest of the set spans unrelated failures.", + "summary": "The candidate pairs are mostly broad similarity matches by subsystem or model family, but they describe different bugs, regressions, or code paths. I don\u2019t see any true duplicate pairs here, so no soft edge should be accepted.", + "confidence": 0.92, + "canonical_issue_reason": "No clear canonical issue: the soft pairs are all false positives with different triggers, symptoms, or code paths.", "canonical_pr_reason": null, - "best_issue_reason": "43295 is the best single issue to keep as an exemplar of the cluster because it is specific and actionable, but it does not actually cover the other bugs.", + "best_issue_reason": "If one issue must be singled out, issue:45200 is the most concrete and actionable standalone report, but it does not act as a true duplicate hub for the set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3-related, but one is about a dummy Conv2d in the decoder and the other is memory overflow in a litetext model." + }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serve crashes, but one is a continuous-batching .to() AttributeError and the other is KV-cache continuation tensor indexing; different code paths and failures." + "reason": "Both hit transformers serve, but one crashes on multimodal continuous batching and the other on KV-cache continuation indexing." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "Different problems: MPS attention correctness vs a PyTorch version check for AuxRequest import in flex_attention." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "Different models and symptoms: JIT/script GeneratorExp export failure vs. SigLIP2 zero-probability outputs." + "reason": "Unrelated failures: TorchScript export of a ViT model vs zero probabilities in SigLIP2 inference." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 implementation discrepancy is unrelated to a resume_from_checkpoint key mismatch." + "reason": "SigLIP2 model discrepancy and a checkpoint resume key mismatch are unrelated bugs." }, { "left": "issue:41720", "right": "issue:46710", "accept": false, - "reason": "Different model families and failure modes: auto device mapping cuda assert vs. accuracy regression after upgrade." + "reason": "Different failure modes: CUDA assert with auto device mapping vs accuracy regression after a version upgrade." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Gemma4 serve processor _tokenizer crash is unrelated to rope-validation JSON list TypeError." + "reason": "Gemma4 processor/tokenizer crash and rope-parameter JSON list handling are separate code paths." }, { "left": "issue:44977", "right": "issue:46710", "accept": false, - "reason": "Flash-attention generation bug in Qwen3.5 is not the same as a DeepSeek accuracy regression." + "reason": "Flash-attention generation bug and a broad accuracy regression are not the same defect." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel is unrelated to Wav2Vec2 save/tokenization failure." + "reason": "One is a dtype mismatch in bfloat16 inference; the other is Wav2Vec2 save/tokenization failure." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "ImportError for a flash-attention helper symbol is a different issue from meta-device model loading breakage." + "reason": "Import error for a flash-attention helper and meta-device loading breakage are distinct issues." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor/tokenizer regression in multimodal text+image handling is distinct from batched video shape handling." + "reason": "Processor.tokenizer access in custom code and batched video processor shape mismatch are different regressions." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Flash-attention sink-less model crash and composite Qwen config unwrapping are different bugs." + "reason": "Flash-attention sink handling and Qwen multimodal config unwrapping do not describe the same bug." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Video input ValueError in Qwen3OmniMoe is unrelated to Gemma4 mm_token_type_ids defaults." + "reason": "Video-input processing in Qwen3OmniMoe is unrelated to Gemma4 mm_token_type_ids defaults." }, { "left": "issue:44186", "right": "issue:46192", "accept": false, - "reason": "LayoutLMv2 tokenizer NER/padding crash is not the same as return_tensors being ignored when text_kwargs is passed." + "reason": "Tokenizer crashes on NER/padding vs return_tensors being ignored are separate tokenizer behaviors." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Custom processor-tokenizer regression and quantization offloading failure are not the same underlying problem." }, { "left": "issue:43572", "right": "issue:46821", "accept": false, - "reason": "StableLmConfig pad_token_idx regression and TimesFM window_size AttributeError are different config issues." + "reason": "Both mention config arguments, but one is StableLm pad_token_idx and the other is TimesFM window_size." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Gemma4 mm_token_type_ids defaulting and PLE device mismatch under device_map='auto' are separate problems." - }, - { - "left": "issue:43295", - "right": "issue:43873", - "accept": false, - "reason": "Processor/tokenizer regression is not the same as offloading failing with quantization." + "reason": "Both are Gemma4-related, but one is token-type defaults for text-only fine-tuning and the other is a device_map forward mismatch." }, { "left": "issue:45823", "right": "issue:46192", "accept": false, - "reason": "Gemma4 device mismatch and ignored return_tensors are unrelated code paths." + "reason": "Gemma4 device placement and a generic return_tensors handling bug are unrelated." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Multimodal processor/tokenizer regression does not match OmDet-Turbo size-mismatch preprocessing." + "reason": "Processor/tokenizer access regression and OmDet-Turbo input-size mismatch are different code paths." }, { "left": "issue:29870", "right": "issue:29942", "accept": false, - "reason": "Bart generate UnboundLocalError and Flash Attention 2 test failures are different bugs." + "reason": "BART generation UnboundLocalError and Flash Attention 2 test failures are not the same bug." }, { "left": "issue:45520", "right": "issue:46178", "accept": false, - "reason": "flash_attn import KeyError on Python 3.13 is unrelated to BACKENDS_MAPPING rejecting tensorflow_text." + "reason": "Different import/backend failures: flash_attn on Python 3.13 vs tensorflow_text backend mapping." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "StableLmConfig missing pad_token_idx is not the same as a Qwen3.5 flash-attention generation issue." + "reason": "StableLm config field loss and Qwen3.5 flash-attention generation issues are unrelated." }, { "left": "issue:43232", "right": "issue:44849", "accept": false, - "reason": "Generation kwargs update after sync_gpus is a different path from Qwen3.5 output_hidden_states behavior." + "reason": "Generation kwargs sync handling and output_hidden_states behavior in Qwen3.5 are separate bugs." }, { - "left": "issue:43450", - "right": "issue:43575", + "left": "issue:43550", + "right": "issue:43844", "accept": false, - "reason": "Video processor batch-shape bug and Qwen2 TP OOM are unrelated." + "reason": "torch.compile/SDPA failure and ZeRO-3 gradient growth are different training issues." }, { - "left": "issue:43550", - "right": "issue:43844", + "left": "issue:43450", + "right": "issue:43575", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to DeepSpeed ZeRO-3 gradient inflation." + "reason": "Video processor batch-shape bug and Qwen2-57B-A14B-Instruct OOM are not duplicates." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "MPS attention correctness issue is not the same as Kimi-K2.5 tokenizer codec handling regression." + "reason": "MPS attention correctness and Kimi-K2.5 tokenizer regression are unrelated." }, { "left": "issue:37428", "right": "issue:43824", "accept": false, - "reason": "Flash-attention helper import error and Qwen2.5-VL import failure are different missing-symbol issues." + "reason": "Different import errors: flash-attention utility import vs Qwen2.5-VL class import." }, { "left": "issue:42490", "right": "issue:43493", "accept": false, - "reason": "Trailing-slash import warning is unrelated to SigLIP2 implementation discrepancy." + "reason": "Trailing-slash model-path warning and SigLIP2 HF/JAX discrepancy are unrelated." }, { "left": "issue:39692", "right": "issue:46192", "accept": false, - "reason": "SigLIP2 documentation example errors are not the same as return_tensors being ignored with text_kwargs." + "reason": "SigLIP2 documentation example errors and return_tensors handling are different problems." }, { "left": "issue:38175", "right": "issue:43493", "accept": false, - "reason": "Both involve SigLIP2, but one is zero probabilities and the other is a broader HF-vs-JAX discrepancy; not clearly the same concrete bug." + "reason": "Both are SigLIP2-related, but one reports zero probabilities while the other reports implementation divergence vs JAX." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "BigBirdTokenizer special-token bug and GraniteSpeech stale device override CI failure are unrelated." - }, - { - "left": "issue:44466", - "right": "issue:45127", - "accept": false, - "reason": "lm_head.weight serialization inconsistency and LoRA merge collapse are different tied-weights issues with different failure modes." + "reason": "Tokenizer special-token registration and a stale CI device override test are unrelated." } ] }, @@ -8641,7 +8671,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28218", "issue:35141", @@ -8667,6 +8697,7 @@ "issue:44112", "issue:44206", "issue:44442", + "issue:44466", "issue:44479", "issue:44521", "issue:44561", @@ -8675,6 +8706,7 @@ "issue:44857", "issue:44898", "issue:45083", + "issue:45127", "issue:45198", "issue:45200", "issue:45290", @@ -8684,24 +8716,23 @@ "issue:45563", "issue:45584", "issue:45593", - "issue:45759", "issue:46192", "issue:46209", "issue:46234", "issue:46285", - "issue:46512", - "issue:46821" + "issue:46512" ], "soft_pairs": [ + "issue:44466|issue:45127", "issue:44521|issue:45290", - "issue:45381|issue:46209", "issue:43278|issue:43856", + "issue:45381|issue:46209", "issue:28218|issue:35141", "issue:42915|issue:43824", "issue:43950|issue:46285", "issue:43824|issue:43975", - "issue:43334|issue:43525", "issue:43452|issue:46512", + "issue:43334|issue:43525", "issue:43262|issue:43450", "issue:44442|issue:45198", "issue:44792|issue:45563", @@ -8710,207 +8741,199 @@ "issue:43819|issue:44112", "issue:45200|issue:45325", "issue:43827|issue:43901", - "issue:44610|issue:44857", "issue:43388|issue:43526", - "issue:43335|issue:43550", "issue:44561|issue:46234", + "issue:43335|issue:43550", "issue:41720|issue:46209", "issue:43819|issue:45083", "issue:44857|issue:45593", - "issue:44206|issue:45584", - "issue:46512|issue:46821", - "issue:45759|issue:46285" + "issue:44610|issue:44857", + "issue:44206|issue:45584" ], - "prepared_review_unit_hash": "b385f635a6e543762021fcdc007584dfbbe8ba9492b14a3abe7566c860f71c93", + "prepared_review_unit_hash": "0eb28aace78bfeb96f05ac53c118aadcaf458b5783d1a5436d96e7c87e8daefe", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 36097, - "estimated_input_tokens": 9025, - "estimated_eval_tokens": 18306 + "soft_pair_count": 26, + "serialized_chars": 35889, + "estimated_input_tokens": 8973, + "estimated_eval_tokens": 18202 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "All proposed pairs look like false positives: they overlap only at a broad subsystem level (tokenization, multimodal processors, configs, docs, or GPU/runtime issues) but not the same concrete bug. No single artifact stands out as a valid duplicate canonical across this set.", - "confidence": 0.99, - "canonical_issue_reason": "No canonical issue: the items are heterogeneous standalone bugs across unrelated code paths, so picking one would be arbitrary.", - "canonical_pr_reason": "No PRs are present in the provided items.", - "best_issue_reason": "No best issue: there is no coherent duplicate cluster here, only loosely similar reports.", - "best_pr_reason": "No PRs are present in the provided items.", + "summary": "This cluster does not look like a duplicate set. It is a grab bag of unrelated transformer bugs, regressions, and docs issues that only share broad API or subsystem keywords.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44521", - "right": "issue:45290", + "left": "issue:44466", + "right": "issue:45127", "accept": false, - "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal inputs while the other is a crash on assistant tool-call messages with no content; different failures and fixes." + "reason": "Both concern tied weights, but one is device-dependent lm_head serialization and the other is LoRA merge collapse with extended vocabulary; different concrete bugs." }, { - "left": "issue:45381", - "right": "issue:46209", + "left": "issue:44521", + "right": "issue:45290", "accept": false, - "reason": "One is a Qwen2.5-VL video position_ids regression; the other is an FP8 inference failure on B200. They are unrelated bugs." + "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal input and the other is a crash on tool-call messages; not the same failure." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Embedding dtype drift in evaluate is unrelated to Qwen3 MoE training memory usage." + "reason": "Embedding dtype drift in evaluate and Qwen3 MoE memory usage are unrelated issues." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Qwen2.5-VL vision position ids bug and FP8 inference failure on B200 are different model/runtime problems." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer spacing after added tokens is a different problem from output embeddings being reinitialized during post_init." + "reason": "Tokenizer spacing after added tokens is unrelated to embedding reinitialization after resizing." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "FineGrainedFP8Config failure in Qwen3Moe and a Qwen2.5-VL import error are different code paths." + "reason": "Qwen3Moe FP8 failure and Qwen2.5-VL import error are different code paths and symptoms." }, { "left": "issue:43950", "right": "issue:46285", "accept": false, - "reason": "Non-persistent buffer corruption on from_pretrained is unrelated to CLIP text-model structural inconsistency." + "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to CLIP text model structural inconsistency." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "Importing Qwen2_5_VLForConditionalGeneration and incorrect detokenization in DeepSeek coder are unrelated." + "reason": "Missing import for a VL class and incorrect detokenization are separate bugs." }, { - "left": "issue:43334", - "right": "issue:43525", + "left": "issue:43452", + "right": "issue:46512", "accept": false, - "reason": "Both mention missing pad_token_id, but they affect different model configs (Qwen3-VL vs Llama4) and likely need separate fixes." + "reason": "gguf_file loading failures and stale config dtype after from_config are unrelated." }, { - "left": "issue:43452", - "right": "issue:46512", + "left": "issue:43334", + "right": "issue:43525", "accept": false, - "reason": "gguf_file loading failures are unrelated to from_config dtype metadata being stale." + "reason": "Both mention missing pad_token_id, but they affect different model configs and loading paths; not the same regression." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Last-batch label dropping in gather_for_metrics is unrelated to BeitImageProcessorFast reducing labels to one element." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and batched video shape handling are different processor bugs." + "reason": "Audio chat-template sampling-rate default and batched video shape handling are different processor bugs." }, { "left": "issue:44442", "right": "issue:45198", "accept": false, - "reason": "FastSpeech2 tokenizer loading and Wav2Vec2 save/tokenization failures are separate speech-model issues." + "reason": "Tokenizer loading failure and save_pretrained/tokenization failure in Wav2Vec2 are not the same issue." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "A janus image-generation test failure is unrelated to a stale paged-generate warning." + "reason": "Janus image generation test failure and stale paged-generate warning are unrelated." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Perceiver non-default resolution failure and RT-DETR memory not being released are unrelated." + "reason": "Perceiver interpolation failure and RT-DETR memory release bug are different models and failure modes." }, { "left": "issue:44479", "right": "issue:46192", "accept": false, - "reason": "Qwen video-input regression and return_tensors being ignored when text_kwargs is passed are different API regressions." + "reason": "Qwen video input regression and return_tensors being ignored are unrelated API bugs." }, { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "DAC forward/from_latents mismatch is unrelated to a GraniteSpeech CI test issue." + "reason": "DAC forward/from_latents mismatch and a stale GraniteSpeech CI test do not describe the same bug." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Gemma 4 missing mm_token_type_ids defaults and Qwen2.5-VL still-image temporal scaling are different multimodal bugs." + "reason": "Gemma 4 mm_token_type_ids defaulting and Qwen2.5-VL rope indexing are different multimodal bugs." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs problems, but one is about pipeline() removal examples and the other about return_all_scores wording; not the same change." - }, - { - "left": "issue:44610", - "right": "issue:44857", - "accept": false, - "reason": "OmDet-Turbo input-size mismatch and LwDetrImageLoss AMP crash are distinct detection-model bugs." - }, - { - "left": "issue:43388", - "right": "issue:43526", - "accept": false, - "reason": "Metric gathering dropping tuple labels and BeitImageProcessorFast reduce_labels truncation are different label-handling paths." + "reason": "Both are docs issues, but they cover different pipeline docs and different stale references." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and Bamba torch.compile SDPA failure are unrelated." - }, - { - "left": "issue:44561", - "right": "issue:46234", - "accept": false, - "reason": "Missing is_torch_fx_available and missing is_falcon_mamba_ssm_available are different utility/API issues." + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile SDPA failure are unrelated model/runtime issues." }, { "left": "issue:41720", "right": "issue:46209", "accept": false, - "reason": "A800 cudaErrorAssert under auto device mapping is unrelated to FP8 inference failure on B200." + "reason": "Auto device mapping cudaErrorAssert and FP8 inference on B200 are different hardware/runtime failures." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "DAC latent mismatch and qwen3_omni_moe feature-length helper behavior are different code paths." + "reason": "DAC latent mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "AMP/CUDA crash in LwDetrImageLoss and missing auxiliary losses in D-FINE are separate training bugs." + "reason": "AMP CUDA crash in LwDetrImageLoss is unrelated to D-FINE not using auxiliary losses." }, { - "left": "issue:44206", - "right": "issue:45584", + "left": "issue:44610", + "right": "issue:44857", "accept": false, - "reason": "Unsupported center argument in LasrFeatureExtractor and Whisper empty-transcription generation failure are unrelated." + "reason": "OmDet-Turbo output size mismatch and LwDetrImageLoss AMP crash are different bugs." }, { - "left": "issue:46512", - "right": "issue:46821", + "left": "issue:44206", + "right": "issue:45584", "accept": false, - "reason": "Stale config dtype after from_config and TimesFM window_size AttributeError are unrelated." + "reason": "Unsupported center arg in LasrFeatureExtractor and Whisper empty-transcription failure are unrelated." }, { - "left": "issue:45759", - "right": "issue:46285", + "left": "issue:44561", + "right": "issue:46234", "accept": false, - "reason": "Unwrapping text_config for composite Qwen multimodal configs and CLIP text-model structural inconsistency are different model-API issues." + "reason": "Both are missing import_utils helpers, but they involve different removed/absent functions and different breakages; not one underlying bug." } ] }, @@ -8932,7 +8955,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28282", "issue:29127", @@ -8953,37 +8976,38 @@ "issue:43611", "issue:43653", "issue:43720", - "issue:43819", "issue:43825", "issue:43854", "issue:43931", "issue:44112", "issue:44351", "issue:44448", - "issue:44493", "issue:44514", "issue:44521", "issue:44743", "issue:44855", - "issue:44857", "issue:45083", "issue:45127", "issue:45200", "issue:45245", "issue:45290", - "issue:45310", "issue:45412", - "issue:45478", "issue:45520", "issue:45588", "issue:45593", + "issue:45759", "issue:45859", "issue:46170", + "issue:46285", "issue:46326", "issue:46431", - "issue:46566" + "issue:46512", + "issue:46566", + "issue:46821" ], "soft_pairs": [ + "issue:46512|issue:46821", + "issue:45759|issue:46285", "issue:43606|issue:45083", "issue:45520|issue:46170", "issue:38175|issue:45412", @@ -9000,239 +9024,221 @@ "issue:43720|issue:44112", "issue:29127|issue:32090", "issue:43335|issue:45083", + "issue:44514|issue:44521", + "issue:43335|issue:45593", "issue:37428|issue:44351", "issue:43335|issue:44112", - "issue:43335|issue:45593", - "issue:44514|issue:44521", - "issue:45588|issue:45859", "issue:45290|issue:46326", + "issue:45588|issue:45859", "issue:46431|issue:46566", "issue:34689|issue:35141", "issue:45127|issue:45245", "issue:42490|issue:43854", - "issue:45200|issue:45588", - "issue:45310|issue:45478", - "issue:43819|issue:44493", - "issue:43606|issue:44857" + "issue:45200|issue:45588" ], - "prepared_review_unit_hash": "93eba44a1a4b4e253a411fef52a577c7a6b675d7831c234bd63db62ac7c2a8a6", + "prepared_review_unit_hash": "7c7dec748ebb22b47ca2980dae8bdec94424a46f6a536d209ec7e8af1c1c4d7b", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36887, - "estimated_input_tokens": 9222, - "estimated_eval_tokens": 18700 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36179, + "estimated_input_tokens": 9045, + "estimated_eval_tokens": 18346 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is largely heterogeneous; most soft matches are false positives across different bugs. The only clear duplicate pair is the Qwen3.5 MoE from_pretrained error (45310/45478).", - "confidence": 0.92, - "canonical_issue_reason": "issue:45310 is the earliest and more detailed report of the exact same Qwen3.5 MoE from_pretrained failure as issue:45478.", + "summary": "This is a loose similarity bucket of unrelated Transformers issues. The soft edges mostly connect superficially similar bug reports (imports, loading, chat templates, offload, config, CI) rather than the same underlying defect, so none should be merged.", + "confidence": 0.79, + "canonical_issue_reason": "No single canonical issue: the tickets span many distinct subsystems and failure modes, and the closest pairs are only title-level similar.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45310 is the best representative issue in the cluster because it is the clearest, earliest, and most complete duplicate target for the only true duplicate pair.", + "best_issue_reason": "No strong global representative stands out; the cluster is heterogeneous and not centered on one bug.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:46512", + "right": "issue:46821", + "accept": false, + "reason": "Different bugs in different models: stale config dtype vs TimesFM window_size AttributeError." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Both concern multimodal model/config structure, but they hit different code paths and failure modes." + }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "Different bugs: CPU offload device mismatch in bark-small vs qwen3_omni_moe feature-length helper behavior." + "reason": "Unrelated regressions: CPU offload device mismatch vs a helper length-calculation bug." }, { "left": "issue:45520", "right": "issue:46170", "accept": false, - "reason": "Both are import-related, but one is a Python 3.13 flash_attn KeyError and the other is a Windows ROCm torch.distributed dependency error." + "reason": "Both are import/dependency issues, but the platform, module, and root cause differ." }, { "left": "issue:38175", "right": "issue:45412", "accept": false, - "reason": "Unrelated symptoms and components: SigLIP2 zero probabilities vs RT-DETR memory not being released." + "reason": "Different subsystems entirely: zero probabilities in Siglip2 vs RT-DETR memory cleanup." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "Both mention loading/version issues, but one is a missing internal flash attention symbol and the other is Torch 2.10 compatibility." + "reason": "Version-compatibility problems, but one is a missing internal symbol import and the other is Torch 2.10 incompatibility." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Different models and failures: SwitchTransformers sparse-layer config bug vs BigBirdTokenizer special-token registration issue." + "reason": "Different model/tokenizer bugs: sparse layer construction vs special-token decode behavior." }, { "left": "issue:43550", "right": "issue:44112", "accept": false, - "reason": "Different areas: torch.compile/SDPA model failure vs a stale CI test for GraniteSpeech device override." + "reason": "Both are CI/runtime issues, but they affect different models and code paths." }, { "left": "issue:29942", "right": "issue:32090", "accept": false, - "reason": "Both are test/runtime failures, but Flash Attention 2 test failures are unrelated to Trainer NoneType broadcast errors." + "reason": "Flash Attention test failure vs Trainer GPU broadcast NoneType error are unrelated bugs." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "Different environments and root causes: missing PyTorch import vs MacOS bus error with a community CLIP model." + "reason": "Missing PyTorch import and MacOS bus error on CLIP are different failures." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Both are model-loading problems, but one is a base_model_prefix regression and the other is a Qwen3-VL weight-shape mismatch." + "reason": "Different model-loading breakages: base_model_prefix handling vs weight-shape mismatch." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Different model behavior regressions: Pegasus v4/v5 output drift vs recurrent state reset in modular_qwen3_5." + "reason": "Different regression types: output drift across versions vs recurrent-state reset in modular Qwen code." }, { "left": "issue:43335", "right": "issue:43504", "accept": false, - "reason": "Different bugs in different model families: SwitchTransformers config creation vs BEiT legacy-field loading failure." + "reason": "Separate bugs in different models; no shared underlying defect." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 indentation/script parsing issue in DebertaV2Model is unrelated to qwen3_omni_moe feature-length computation." + "reason": "Python 3.13 JIT parsing issue vs qwen3_omni_moe helper behavior are unrelated." }, { "left": "issue:43278", "right": "issue:43825", "accept": false, - "reason": "Different issues: embedding dtype drift during eval vs pipeline translation-task error messaging." + "reason": "Embedding dtype persistence and pipeline task messaging are different issues." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "BitNet packed-weight loading is unrelated to GraniteSpeech CI device-override flakiness." + "reason": "Accelerate loading of packed weights vs a stale CI device-override test are unrelated." }, { "left": "issue:29127", "right": "issue:32090", "accept": false, - "reason": "LayoutLMv3 error clarity and Trainer broadcast NoneType are separate bugs." + "reason": "Different errors in different components: LayoutLMv3 validation messaging vs Trainer broadcast handling." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug is unrelated to qwen3_omni_moe output-length helper behavior." + "reason": "No shared root cause; these are distinct model-specific bugs." }, { - "left": "issue:37428", - "right": "issue:44351", + "left": "issue:44514", + "right": "issue:44521", "accept": false, - "reason": "Different import failures: missing flash support symbol vs cannot import HybridCache." + "reason": "Both are chat-template related, but one is batched padding behavior and the other is assistant-mask generation for multimodal inputs." }, { "left": "issue:43335", - "right": "issue:44112", + "right": "issue:45593", "accept": false, - "reason": "Different models and failure modes; no shared underlying code path." + "reason": "Different models and code paths; no same underlying bug." }, { - "left": "issue:43335", - "right": "issue:45593", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "SwitchTransformers config bug is unrelated to D-FINE auxiliary-loss behavior." + "reason": "Both are import errors, but they involve different symbols and regressions." }, { - "left": "issue:44514", - "right": "issue:44521", + "left": "issue:45290", + "right": "issue:46326", "accept": false, - "reason": "Both involve chat templates, but one crashes on batched padding=False and the other returns all-zero assistant masks for multimodal inputs." + "reason": "Both touch chat templating, but they fail in different branches and with different inputs." }, { "left": "issue:45588", "right": "issue:45859", "accept": false, - "reason": "Different flash-attention-related failures: sink-less s_aux=None crash vs CPU offload incompatibility in Qwen3_5MoeTextRotaryEmbedding." - }, - { - "left": "issue:45290", - "right": "issue:46326", - "accept": false, - "reason": "Both touch apply_chat_template, but tool-call/no-content crashes and continue_final_message tokenizer behavior are different bugs." + "reason": "Distinct failures: flash_attention integration on s_aux=None vs CPU-offload incompatibility in Qwen3_5Moe." }, { "left": "issue:46431", "right": "issue:46566", "accept": false, - "reason": "No common underlying bug: missing chat template vs DiffusionGemma weights staying on meta under device_map=auto." + "reason": "Missing chat template support vs meta-device loading are unrelated." }, { "left": "issue:34689", "right": "issue:35141", "accept": false, - "reason": "Different loading/training issues: Llama 3.2 Vision load regression vs embedding reinitialization after resize_token_embeddings." + "reason": "Different model-loading problems: one breaks Llama 3.2 Vision loading, the other reinitializes embeddings after resize." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "Unrelated failures: LoRA merge collapse with extended vocab vs category-count upper-bound runtime error." + "reason": "LoRA merge collapse and category-count overflow are unrelated bugs." }, { "left": "issue:42490", "right": "issue:43854", "accept": false, - "reason": "A trailing-slash import warning is not the same as a GLM-4.7-Flash model-loading failure in unit tests." + "reason": "Local-path slash warning and GLM-4.7-Flash unit-test loading failure are different issues." }, { "left": "issue:45200", "right": "issue:45588", "accept": false, - "reason": "Gemma 4 mm_token_type_ids defaulting and flash_attention sink-less crash are different code paths." - }, - { - "left": "issue:45310", - "right": "issue:45478", - "accept": true, - "reason": "Same concrete bug: both report the Qwen3.5 MoE from_pretrained error on transformers>=5.4.0; 45478 is a duplicate of 45310." - }, - { - "left": "issue:43819", - "right": "issue:44493", - "accept": false, - "reason": "DAC forward/from_latents mismatch is unrelated to position-id unexpected-key behavior." - }, - { - "left": "issue:43606", - "right": "issue:44857", - "accept": false, - "reason": "Both mention device/mixed-precision issues, but they affect different models and failure mechanisms." + "reason": "Gemma token-type defaults and flash_attention s_aux handling are unrelated code paths." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "Grounded in the packet. The only clearly duplicated pair is issue:45310 / issue:45478, and the other soft-edge rejections stay appropriately conservative." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -9247,7 +9253,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:10000", "issue:15354", @@ -9258,7 +9264,6 @@ "issue:29942", "issue:31501", "issue:32090", - "issue:33290", "issue:33357", "issue:34689", "issue:35141", @@ -9272,7 +9277,9 @@ "issue:43526", "issue:43550", "issue:43575", + "issue:43606", "issue:43701", + "issue:43819", "issue:44206", "issue:44360", "issue:44448", @@ -9280,30 +9287,31 @@ "issue:44561", "issue:44617", "issue:44779", - "issue:44855", "issue:44857", "issue:44977", "issue:45072", "issue:45083", "issue:45229", "issue:45290", + "issue:45310", "issue:45381", "issue:45459", - "issue:45588", + "issue:45478", "issue:45823", "issue:45920", "issue:46192", "issue:46234", - "issue:46399", - "issue:46655", "issue:46697" ], "soft_pairs": [ + "issue:45310|issue:45478", + "issue:43819|issue:44493", "issue:44448|issue:44779", + "issue:43606|issue:44857", "issue:39692|issue:43526", "issue:26413|issue:31501", - "issue:44617|issue:46697", "issue:28218|issue:32090", + "issue:44617|issue:46697", "issue:45459|issue:46234", "issue:39692|issue:44493", "issue:44206|issue:45823", @@ -9314,8 +9322,8 @@ "issue:43550|issue:45083", "issue:44360|issue:44561", "issue:38175|issue:44493", - "issue:29942|issue:35141", "issue:27957|issue:34689", + "issue:29942|issue:35141", "issue:29870|issue:32090", "issue:44448|issue:44977", "issue:43278|issue:43329", @@ -9324,184 +9332,198 @@ "issue:45290|issue:45381", "issue:10000|issue:15354", "issue:44977|issue:45229", - "issue:43278|issue:43381", - "issue:44855|issue:46399", - "issue:44857|issue:45588", - "issue:29870|issue:33290", - "issue:45290|issue:46655" + "issue:43278|issue:43381" ], - "prepared_review_unit_hash": "7c7be759616b429e6a97d1092d6de99e4f011339c47ccdb71ba1d4ff01df893c", + "prepared_review_unit_hash": "77ae8d35f31b02d8c467ed3ba2f76c39322aef0ff7a9f43ad992e58f4536deb8", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36757, - "estimated_input_tokens": 9190, - "estimated_eval_tokens": 18636 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35775, + "estimated_input_tokens": 8944, + "estimated_eval_tokens": 18144 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the soft pairs span unrelated model, tokenizer, training, and inference bugs. I reject all soft edges. If one issue must represent the cluster, 45920 is the broadest regression-style tokenizer issue affecting multiple models.", - "confidence": 0.79, - "canonical_issue_reason": "issue:45920 is the broadest and most central bug report here: it describes a v5 tokenizer regression impacting multiple model families, so it is the best single representative issue.", + "summary": "This cluster is mostly heterogeneous: the only clear duplicate pair is the two identical Qwen3.5 MoE from_pretrained reports (45310/45478). The rest are broad text-similarity matches across unrelated bugs. No PRs are present.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45310 is the earliest of the only exact duplicate pair, and it cleanly represents the Qwen3.5 MoE from_pretrained regression also reported in 45478.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the strongest global issue candidate because it is broad, regression-oriented, and already framed as affecting multiple models, unlike the narrower single-model bugs.", + "best_issue_reason": "Issue 45920 is the strongest standalone issue in the set to use as a representative: it describes a broad v5 tokenizer regression affecting multiple named models and has the most cross-references.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same title and same Qwen3.5 MoE from_pretrained failure; these look like the same underlying bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different code paths: DAC.from_latents mismatch vs position-id key warnings across many models." + }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer/output regressions, but they affect different models and different failure modes (Pegasus v4/v5 output drift vs DeepSeek tokenizer IDs)." + "reason": "Both are v5 regressions, but one is Pegasus generation output drift and the other is DeepSeek tokenizer IDs; not the same bug." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "Both involve runtime/device issues, but Bark CPU offload mismatch and LwDetr AMP crash are unrelated." }, { "left": "issue:39692", "right": "issue:43526", "accept": false, - "reason": "SigLIP2 documentation example errors and BEiTImageProcessorFast label reduction are unrelated bugs in different code paths." + "reason": "SigLIP2 documentation example errors and BeitImageProcessorFast reduce_labels behavior are unrelated." }, { "left": "issue:26413", "right": "issue:31501", "accept": false, - "reason": "Both mention training startup/resume behavior, but one is an empty epoch iterator on resume and the other is FineWeb SLM training not starting; not the same bug." + "reason": "Checkpoint resume epoch_iterator failure and FineWeb training not starting are different training issues." }, { - "left": "issue:44617", - "right": "issue:46697", + "left": "issue:28218", + "right": "issue:32090", "accept": false, - "reason": "CUDA OOM in Sam3Video is unrelated to Swin2SR over-padding in an image processor." + "reason": "Tokenizer added-space bug and Trainer GPU broadcast NoneType error are unrelated." }, { - "left": "issue:28218", - "right": "issue:32090", + "left": "issue:44617", + "right": "issue:46697", "accept": false, - "reason": "Tokenizer spacing after added tokens and Trainer GPU broadcast NoneType errors are different subsystems and failure modes." + "reason": "CUDA OOM in Sam3Video and over-padding in Swin2SR image processing are different bugs." }, { "left": "issue:45459", "right": "issue:46234", "accept": false, - "reason": "One is tokenizer error masking when protobuf is missing; the other is a missing import utility symbol. Different bugs." + "reason": "Both touch import/utils, but hidden tokenizer decode errors vs a missing availability helper are different problems." }, { "left": "issue:39692", "right": "issue:44493", "accept": false, - "reason": "A broken SigLIP2 docs example is unrelated to unexpected position-id key warnings across models." + "reason": "Documentation/quantization example issue vs position-id key warnings; no shared bug." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "These are distinct: feature-extractor center-arg crash vs Gemma4 device-map forward mismatch." + "reason": "LasrFeatureExtractor center-arg crash and Gemma4 device-map mismatch are unrelated." }, { "left": "issue:43701", "right": "issue:45083", "accept": false, - "reason": "Checkpoint key mismatch and qwen3_omni_moe feature-length helper behavior are not the same underlying issue." + "reason": "Resume-from-checkpoint key mismatch and qwen3_omni_moe feature-length behavior are different code paths." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Both involve precision/dtype problems, but they hit different models and code paths; not mergeable as one fix." + "reason": "Both are dtype/runtime mismatches, but they affect different models and mechanisms; not the same fix." }, { "left": "issue:32090", "right": "issue:33357", "accept": false, - "reason": "NoneType GPU broadcast failure and MacOS bus error in CLIP loading are unrelated." + "reason": "Trainer NoneType broadcast error and MacOS CLIP bus error are unrelated failures." }, { "left": "issue:42490", "right": "issue:43575", "accept": false, - "reason": "Local-path import warning and Qwen2-57B tensor-parallel OOM are unrelated." + "reason": "Trailing-slash local model import warning and tensor-parallel OOM are unrelated." }, { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba and qwen3_omni_moe feature-length logic are unrelated." + "reason": "Bamba torch.compile/SDPA failure and qwen3_omni_moe feature-length issue are different bugs." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "A missing ReLU in DSA indexer and removal of is_torch_fx_available breaking trust_remote_code are different issues." + "reason": "DSA ReLU discussion and removal of is_torch_fx_available breaking trust_remote_code are unrelated." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "Zero probabilities in SigLIP2 and unexpected position-id keys are different bugs." + "reason": "SigLIP2 zero-probability issue and position-id key warnings are different regressions." }, { - "left": "issue:29942", - "right": "issue:35141", + "left": "issue:27957", + "right": "issue:34689", "accept": false, - "reason": "Failing Flash Attention 2 tests and embedding reinitialization after resize are unrelated." + "reason": "Both mention model loading/attention, but XLMRoberta Flash Attention 2 and Llama 3.2 Vision loading are distinct issues." }, { - "left": "issue:27957", - "right": "issue:34689", + "left": "issue:29942", + "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta and Llama 3.2 Vision loading breakage are unrelated." + "reason": "Flash Attention tests failing and embedding reinitialization on resize are unrelated." }, { "left": "issue:29870", "right": "issue:32090", "accept": false, - "reason": "Bart generate decoder_attention_mask assignment bug is unrelated to Trainer GPU broadcast NoneType handling." + "reason": "BART generate decoder_attention_mask bug and GPU broadcast NoneType error are different failures." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Pegasus v4/v5 output differences and Qwen3.5 flash-attention generation failures are different code paths." + "reason": "Both are generation regressions, but Pegasus v4/v5 output drift and Qwen3.5 flash-attention generation failure are not the same bug." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "Dtype drift between train/eval and undefined video-branch helper symbols are unrelated." + "reason": "Embedding dtype drift in evaluate and undefined video-token helper variables are unrelated." }, { "left": "issue:45920", "right": "issue:46192", "accept": false, - "reason": "Both concern tokenizer behavior, but one is wrong token IDs across multiple models while the other is return_tensors being ignored with explicit text_kwargs; not the same bug." + "reason": "Both are tokenizer/API regressions in v5, but wrong token IDs across models vs return_tensors being ignored are different underlying bugs." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "Qwen2.5-VL iterable multi-image finetuning failure and BEiT label reduction bug are unrelated." + "reason": "Qwen2.5-VL iterable dataset finetuning failure and BeitImageProcessorFast reduce_labels bug are unrelated." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Chat template crashes on tool-call assistant messages and Qwen2.5-VL video position IDs are different failures." + "reason": "Chat-template tool-call crash and qwen2.5-vl video vision_position_ids bug are different multimodal issues." }, { "left": "issue:10000", "right": "issue:15354", "accept": false, - "reason": "German DistilBertModel issue and torch.jit GeneratorExp export limitation are unrelated." + "reason": "German DistilBertModel issue and GeneratorExp/torch.jit.script export failure are unrelated." }, { "left": "issue:44977", @@ -9513,38 +9535,17 @@ "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Embedding dtype drift in eval and gradient checkpointing being disallowed in eval mode are distinct problems." - }, - { - "left": "issue:44855", - "right": "issue:46399", - "accept": false, - "reason": "Python 3.13 IndentationError from a comment between decorator and def and a missing decorator registration are not the same bug." - }, - { - "left": "issue:44857", - "right": "issue:45588", - "accept": false, - "reason": "AMP/CUDA crash in LwDetrImageLoss is unrelated to flash_attention AttributeError on s_aux=None." - }, - { - "left": "issue:29870", - "right": "issue:33290", - "accept": false, - "reason": "Bart generation mask bug and Adafactor/deepspeed OOM are unrelated." - }, - { - "left": "issue:45290", - "right": "issue:46655", - "accept": false, - "reason": "Chat template/tool-call crash and SmolVLM video padding calculation are different issues." + "reason": "BF16-to-FP32 dtype change in evaluation and gradient checkpointing in eval mode are different behaviors." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only clearly supported duplicate pair is 45310/45478 with identical titles, and the remaining soft-edge rejections do not overstate similarity. No PRs are present, so the PR-related fields are appropriately null." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -9559,7 +9560,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:25251", "issue:29127", @@ -9574,11 +9575,8 @@ "issue:41720", "issue:42175", "issue:42490", - "issue:42915", - "issue:42994", "issue:43278", "issue:43381", - "issue:43412", "issue:43575", "issue:43611", "issue:43653", @@ -9587,19 +9585,20 @@ "issue:43927", "issue:43976", "issue:44186", - "issue:44206", "issue:44361", "issue:44448", "issue:44466", - "issue:44479", "issue:44561", "issue:44568", "issue:44743", "issue:44779", + "issue:44855", + "issue:44857", "issue:44871", "issue:44977", "issue:45198", "issue:45229", + "issue:45290", "issue:45372", "issue:45440", "issue:45491", @@ -9608,26 +9607,32 @@ "issue:45698", "issue:45803", "issue:45823", - "issue:46121" + "issue:46121", + "issue:46399", + "issue:46655" ], "soft_pairs": [ + "issue:44855|issue:46399", + "issue:29870|issue:33290", + "issue:45290|issue:46655", + "issue:44857|issue:45588", "issue:43927|issue:45698", "issue:37428|issue:42175", "issue:32090|issue:33290", "issue:44561|issue:44779", "issue:43873|issue:45229", - "issue:44448|issue:44466", "issue:25251|issue:32090", + "issue:44448|issue:44466", "issue:43976|issue:45440", "issue:42490|issue:43873", - "issue:45588|issue:45823", "issue:44568|issue:45491", + "issue:45588|issue:45823", "issue:43278|issue:43575", "issue:41720|issue:45229", - "issue:44977|issue:45563", "issue:43381|issue:43701", - "issue:33357|issue:39692", + "issue:44977|issue:45563", "issue:44871|issue:44977", + "issue:33357|issue:39692", "issue:29127|issue:34567", "issue:45198|issue:45803", "issue:43653|issue:44743", @@ -9635,207 +9640,210 @@ "issue:44186|issue:45198", "issue:45372|issue:45588", "issue:43611|issue:46121", - "issue:29870|issue:34689", - "issue:42915|issue:42994", - "issue:43412|issue:45588", - "issue:44206|issue:44479" + "issue:29870|issue:34689" ], - "prepared_review_unit_hash": "70109cfefee97b942c779214500d59d2771e5bcc19a741c7cadc43908da28757", + "prepared_review_unit_hash": "f3c29caf80f3b2d23c98b385813afa525fbef9511ca04dbdd8b27818e10e635f", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 36299, - "estimated_input_tokens": 9075, - "estimated_eval_tokens": 18406 + "soft_pair_count": 29, + "serialized_chars": 36740, + "estimated_input_tokens": 9185, + "estimated_eval_tokens": 18626 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is a grab-bag of mostly unrelated issue reports; none of the soft pairs look like true duplicates or the same underlying bug.", - "confidence": 0.74, - "canonical_issue_reason": "issue:45698 is the best anchor: it is an open, concrete save/load custom-module regression with clear user impact and some inbound references, while the rest of the set is much more heterogeneous.", + "summary": "This cluster is mostly a heterogeneous mix of unrelated Transformers bugs. The soft-similarity pairs overlap lexically on broad areas like generation, serialization, tokenizers, or memory, but they do not appear to describe the same concrete defect, so none should be merged as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:45698 is the strongest single issue to represent the cluster because it has a specific reproducible failure and the broadest save/load relevance among these otherwise unrelated reports.", + "best_issue_reason": "issue:43873 is the strongest issue candidate by discussion/inbound activity and broad infrastructure relevance, but it is still only one bug report in a largely unrelated set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Both are import/registration failures, but they affect different model code paths (DebertaV2 vs Qwen3.5) and different root causes." + }, + { + "left": "issue:29870", + "right": "issue:33290", + "accept": false, + "reason": "Bart generate() UnboundLocalError and DeepSpeed/Adafactor OOM are unrelated training/runtime problems." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Chat template tool-call crash and SmolVLM video padding bug are different APIs and failure modes." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "One is an AMP/CUDA loss crash, the other is a flash-attention integration AttributeError on sink-less models." + }, { "left": "issue:43927", "right": "issue:45698", "accept": false, - "reason": "Both involve save/load behavior, but 43927 is a config/token-ID loss bug and 45698 is custom-module resolution after save_pretrained." + "reason": "Both involve save/load behavior, but one loses config token IDs and the other loads the wrong custom module; different underlying bugs." }, { "left": "issue:37428", "right": "issue:42175", "accept": false, - "reason": "One is a flash-attention import error; the other is a package/backend install issue." + "reason": "Flash-attention import error and missing TensorFlow backend in a torch install are unrelated." }, { "left": "issue:32090", "right": "issue:33290", "accept": false, - "reason": "Trainer broadcast NoneType vs Adafactor OOM in DeepSpeed are different failure modes." + "reason": "Trainer broadcast NoneType and Adafactor OOM are different failure classes with no shared code-path." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "One is a v5 API removal breaking trust_remote_code; the other is a tokenizer regression in DeepSeek." + "reason": "A removed utility breaking trust_remote_code models is unrelated to DeepSeek tokenizer output regressions." }, { "left": "issue:43873", "right": "issue:45229", "accept": false, - "reason": "Quantization offloading vs multi-GPU OOM are distinct memory/inference issues." + "reason": "Quantization/offloading behavior and Gemma4 multi-GPU OOM are both memory-adjacent, but not the same bug or fix." }, { - "left": "issue:44448", - "right": "issue:44466", + "left": "issue:25251", + "right": "issue:32090", "accept": false, - "reason": "Pegasus output regression vs lm_head serialization inconsistency are different code paths." + "reason": "Pipeline top_k nesting bug and Trainer GPU broadcast error are unrelated." }, { - "left": "issue:25251", - "right": "issue:32090", + "left": "issue:44448", + "right": "issue:44466", "accept": false, - "reason": "Pipeline top_k nesting bug vs Trainer GPU broadcast error are unrelated." + "reason": "Both are v5 regressions, but one is generation output drift and the other is serialization of lm_head.weight; different concrete issues." }, { "left": "issue:43976", "right": "issue:45440", "accept": false, - "reason": "Python version compatibility vs DeepSeekV3MoE implementation divergence are unrelated." + "reason": "Python version compatibility breakage and DeepSeekV3MoE divergence are not the same defect." }, { "left": "issue:42490", "right": "issue:43873", "accept": false, - "reason": "Local-path trailing-slash warning vs quantization offloading are unrelated." + "reason": "Trailing-slash import warning and quantization/offloading failure are unrelated." }, { - "left": "issue:45588", - "right": "issue:45823", + "left": "issue:44568", + "right": "issue:45491", "accept": false, - "reason": "Flash-attention s_aux=None crash vs Gemma4 PLE device mismatch are different forward-path bugs." + "reason": "Tokenizer special-token behavior and Gemma3 NaN embeddings from sliding-window attention are different problems." }, { - "left": "issue:44568", - "right": "issue:45491", + "left": "issue:45588", + "right": "issue:45823", "accept": false, - "reason": "Tokenizer special-token handling vs Gemma3 NaN embeddings are unrelated." + "reason": "Flash-attention sinkless-model crash and Gemma4 device_map mismatch are distinct forward-path bugs." }, { "left": "issue:43278", "right": "issue:43575", "accept": false, - "reason": "Train/eval dtype drift vs TP OOM are different issues." + "reason": "Training/eval dtype drift and Qwen2 tensor-parallel OOM are unrelated." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Qwen3 auto device-map assert vs Gemma4 multi-GPU OOM are unrelated model-specific failures." - }, - { - "left": "issue:44977", - "right": "issue:45563", - "accept": false, - "reason": "Qwen3.5 flash-attention generation bug vs paged-generate warning are unrelated." + "reason": "Both involve large-model inference failures, but one is cudaErrorAssert on auto device mapping and the other is plain OOM; not the same root cause." }, { "left": "issue:43381", "right": "issue:43701", "accept": false, - "reason": "Eval/gradient-checkpointing restriction vs resume_from_checkpoint key mismatch are unrelated." + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch do not share a code-path." }, { - "left": "issue:33357", - "right": "issue:39692", + "left": "issue:44977", + "right": "issue:45563", "accept": false, - "reason": "MacOS CLIP bus error vs SigLIP2 docs/example errors are different failures." + "reason": "Flash-attention generation failure and a stale num_return_sequences warning are not the same bug." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Gemma eos_token_id mismatch vs Qwen3.5 flash-attention regression are unrelated." + "reason": "Gemma eos_token_id config mismatch and Qwen3.5 flash-attention generation failure are unrelated model-specific issues." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS bus error on a CLIP model and SigLIP2 doc example errors are different scenarios and failure types." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 error-handling clarity vs TrainerState token counting are unrelated." + "reason": "LayoutLMv3 error-message clarity and TrainerState token counter not updating are unrelated." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Wav2Vec2 save/tokenization failure vs Fuyu tensor-conversion exception handling are different preprocessing bugs." + "reason": "Wav2Vec2 save/tokenization failure and a bare-except in Fuyu tensor conversion are different bugs." }, { "left": "issue:43653", "right": "issue:44743", "accept": false, - "reason": "BigBird special-token registration bug vs Qwen3.5 recurrent-state reset are unrelated." + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated tokenizer vs model-state issues." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "MLuke tokenizer AttributeError vs Wav2Vec2 save/tokenization failure are unrelated." + "reason": "Both touch tokenization, but MLuke AttributeError and Wav2Vec2 save_pretrained/tokenization failure are not the same defect." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "LayoutLMv2 NER/padding crash vs Wav2Vec2 save/tokenization failure are unrelated." + "reason": "LayoutLMv2 NER/padding crash and Wav2Vec2 save/tokenization failure are different tokenization pathways." }, { "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Gemma4 processor import failure vs flash-attention s_aux=None crash are unrelated." + "reason": "Gemma 4 processor import breakage and flash-attention sinkless crash are unrelated." }, { "left": "issue:43611", "right": "issue:46121", "accept": false, - "reason": "base_model_prefix loading breakage vs rope-validation list TypeError are unrelated." + "reason": "base_model_prefix loading regression and rope-validation TypeError involve different loading/validation code." }, { "left": "issue:29870", "right": "issue:34689", "accept": false, - "reason": "BART generation attention-mask bug vs Llama 3.2 model-loading regression are different code paths." - }, - { - "left": "issue:42915", - "right": "issue:42994", - "accept": false, - "reason": "FineGrainedFP8Config load failure vs quantized model saving failure are distinct quantization bugs." - }, - { - "left": "issue:43412", - "right": "issue:45588", - "accept": false, - "reason": "Executorch export guard/erfinv issue vs flash-attention s_aux=None crash are unrelated." - }, - { - "left": "issue:44206", - "right": "issue:44479", - "accept": false, - "reason": "Feature-extractor center-arg regression vs Qwen video-input regression are different preprocessing paths." + "reason": "Bart generate() UnboundLocalError and Llama 3.2 Vision model-loading breakage are different model-specific regressions." } ] }, @@ -9857,7 +9865,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:27957", "issue:29127", @@ -9865,17 +9873,16 @@ "issue:30064", "issue:33290", "issue:33357", - "issue:34689", "issue:35141", "issue:42490", "issue:42915", + "issue:42994", "issue:43065", "issue:43066", - "issue:43334", "issue:43381", + "issue:43412", "issue:43421", "issue:43526", - "issue:43531", "issue:43575", "issue:43653", "issue:43701", @@ -9884,11 +9891,13 @@ "issue:43931", "issue:43994", "issue:44186", + "issue:44206", "issue:44265", "issue:44279", "issue:44387", "issue:44442", "issue:44462", + "issue:44479", "issue:45072", "issue:45198", "issue:45245", @@ -9899,25 +9908,27 @@ "issue:45459", "issue:45491", "issue:45563", + "issue:45588", "issue:45593", "issue:45712", "issue:45834", - "issue:46121", "issue:46369", - "issue:46399", "issue:46693" ], "soft_pairs": [ + "issue:42915|issue:42994", + "issue:43412|issue:45588", + "issue:44206|issue:44479", "issue:45372|issue:45406", - "issue:29942|issue:30064", "issue:43756|issue:45440", + "issue:29942|issue:30064", "issue:43653|issue:44442", "issue:45491|issue:46693", "issue:43065|issue:45412", "issue:29127|issue:33357", - "issue:45459|issue:45712", "issue:45245|issue:45563", "issue:43065|issue:43994", + "issue:45459|issue:45712", "issue:44265|issue:45198", "issue:43575|issue:43701", "issue:43066|issue:44462", @@ -9925,249 +9936,227 @@ "issue:45440|issue:45593", "issue:29127|issue:35141", "issue:42490|issue:43381", - "issue:45412|issue:46369", "issue:43994|issue:45412", + "issue:45412|issue:46369", "issue:43421|issue:45491", "issue:43575|issue:44387", "issue:44442|issue:45072", - "issue:43526|issue:45834", "issue:42915|issue:43931", + "issue:43526|issue:45834", "issue:43526|issue:46693", "issue:27957|issue:29127", "issue:43526|issue:44186", "issue:43824|issue:44279", - "issue:46121|issue:46399", - "issue:27957|issue:33290", - "issue:43994|issue:46369", - "issue:43334|issue:43531", - "issue:33290|issue:34689" + "issue:27957|issue:33290" ], - "prepared_review_unit_hash": "fa02133ad5ded0e89f7bada7825209ff17816c3a1acf5d0e06fa79d94c1dd80e", + "prepared_review_unit_hash": "bd6c6f1a3424914df0f3692c45f33c29ccd44910e2a5197951ed8e0f6c347e4d", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 33, - "serialized_chars": 36933, - "estimated_input_tokens": 9234, - "estimated_eval_tokens": 18724 + "soft_pair_count": 32, + "serialized_chars": 36569, + "estimated_input_tokens": 9143, + "estimated_eval_tokens": 18542 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab bag of unrelated Transformers issues. The shared words are mostly superficial (e.g. Flash Attention, Qwen, tokenizer, memory), but the underlying bugs and code paths differ, so none of the soft pairs should be merged as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": "No canonical issue: the items span many unrelated subsystems and failure modes, so there is no single report that safely represents the cluster.", + "summary": "The soft pairs look like broad-topic false positives rather than true duplicates: they share surface themes such as tokenizers, quantization, Flash Attention, or memory, but each points to a different model, code path, or failure mode. No soft edge is strong enough to merge into one duplicate cluster.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45491 is the clearest standalone, user-facing regression with a tightly scoped reproducer; the rest of the items are mostly unrelated false-positive matches.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a strong cluster representative because the set is heterogeneous rather than one underlying bug with variants.", + "best_issue_reason": "45491 is the strongest issue to anchor on: it has a concrete failure mode, specific triggering conditions, and clear impact.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45372", - "right": "issue:45406", + "left": "issue:42915", + "right": "issue:42994", "accept": false, - "reason": "Different root causes: a dependency/import failure in Gemma processor loading vs a missing _tokenizer attribute in transformers serve." + "reason": "Different phases and bugs: FP8 config failure versus quantized model saving failure." }, { - "left": "issue:29942", - "right": "issue:30064", + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export crash and flash-attention integration crash are separate code paths." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Feature extractor arg regression is unrelated to the Qwen VL video-input regression." + }, + { + "left": "issue:45372", + "right": "issue:45406", "accept": false, - "reason": "Flash Attention 2 test failures and void segmentation map processing are unrelated bugs." + "reason": "One is a dependency import error; the other is a missing Gemma4Processor attribute in serve." }, { "left": "issue:43756", "right": "issue:45440", "accept": false, - "reason": "Both concern model implementation mismatches, but they affect different models and code paths; not the same bug." + "reason": "Different model-specific implementation bugs: RoPE layer dropping versus DeepSeekV3MoE divergence." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "FA2 test failures and void segmentation-map processing are unrelated." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "Tokenizer special-token registration bug vs AutoTokenizer failing to load a specific tokenizer; different failure modes." + "reason": "Tokenizer special-token registration bug does not match a FastSpeech2Conformer tokenizer load failure." }, { "left": "issue:45491", "right": "issue:46693", "accept": false, - "reason": "One is a NaN correctness bug with sliding-window attention, the other is a performance regression with FA2 long-context generation." + "reason": "Both mention FA2/long context, but one is NaN embeddings and the other is a performance regression." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "SAM3 pixel decoder dummy layer and RT-DETR memory leak are unrelated issues." + "reason": "Dummy Conv2d placeholder code is not the same as an RT-DETR memory leak." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 error messaging and a CLIP bus error on macOS are not the same problem." - }, - { - "left": "issue:45459", - "right": "issue:45712", - "accept": false, - "reason": "Tokenizer protobuf exception handling and leftover dummy class cleanup are unrelated." + "reason": "Error-message clarity for LayoutLMv3 is unrelated to a CLIP bus error on macOS." }, { "left": "issue:45245", "right": "issue:45563", "accept": false, - "reason": "Category-limit RuntimeError vs stale generate warning; no shared code-path bug." + "reason": "Category-count runtime error and stale generate warning are different problems." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "Different models and failures: dummy Conv2d in SAM3PixelDecoder vs SIGLIP2 loading/inference nonsense." + "reason": "A dummy class cleanup issue does not match SigLIP2 nonsensical outputs." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Protobuf error masking and leftover dummy classes are separate maintenance issues." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export failure with torch_compilable_check and Wav2Vec2 save/tokenization failure are distinct issues." + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "TP OOM when loading a large Qwen2 model vs resume_from_checkpoint key mismatch are unrelated." + "reason": "Qwen TP OOM and resume_from_checkpoint key mismatch do not share the same bug." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Both involve tokenizer infrastructure, but one is the wrong decoder type and the other ignores tokenizer.json; different bugs." + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are distinct tokenizer-loading issues." }, { "left": "issue:45372", "right": "issue:45459", "accept": false, - "reason": "Dependency import failure in Gemma processor loading vs protobuf-missing tokenizer error masking are different paths." + "reason": "Gemma 4 dependency import failure is not the same as protobuf exception handling." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "DeepSeekV3 implementation divergence and D-FINE auxiliary-loss behavior are unrelated model bugs." + "reason": "DeepSeekV3MoE divergence and D-FINE auxiliary-loss behavior are unrelated model bugs." }, { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "LayoutLMv3 box-clarity issue and embedding reinit after resize are unrelated." + "reason": "Message clarity for one error is not the same as embedding reinitialization logic." }, { "left": "issue:42490", "right": "issue:43381", "accept": false, - "reason": "Trailing-slash import warning and gradient-checkpointing in eval mode are unrelated." + "reason": "Path-trailing-slash import warning and gradient-checkpointing eval-mode restriction are unrelated." }, { - "left": "issue:45412", - "right": "issue:46369", + "left": "issue:43994", + "right": "issue:45412", "accept": false, - "reason": "Both mention memory, but one is a model deletion leak and the other is inference memory overflow; different causes." + "reason": "SigLIP2 bad outputs and RT-DETR memory leak do not share a concrete code path." }, { - "left": "issue:43994", - "right": "issue:45412", + "left": "issue:45412", + "right": "issue:46369", "accept": false, - "reason": "SIGLIP2 inference failure and RT-DETR memory leak are unrelated." + "reason": "Memory leak on deletion and memory overflow during sam3-litetext usage are different failures." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "Tokenizer post-processor update bug and Gemma3 NaN embeddings are unrelated." + "reason": "Tokenizer post-processor sync bug is unrelated to Gemma3 NaN embeddings." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "OOM from tensor parallel loading and int4 reserved-memory regression are both memory issues but not the same bug." + "reason": "Model-specific TP OOM and int4 reserved-memory regression are not the same bug." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "Tokenizer loading failure and dtype mismatch inference failures are unrelated." - }, - { - "left": "issue:43526", - "right": "issue:45834", - "accept": false, - "reason": "BeitImageProcessorFast label reduction bug and Kosmos2.5 OCR index error are unrelated." - }, - { - "left": "issue:42915", - "right": "issue:43931", - "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen3-VL weight-shape mismatch are different model/config issues." - }, - { - "left": "issue:43526", - "right": "issue:46693", - "accept": false, - "reason": "Label reduction bug in BeitImageProcessorFast vs FA2 long-context performance regression; unrelated." + "reason": "Tokenizer load failure and bfloat16 dtype mismatch are different subsystems." }, { "left": "issue:27957", "right": "issue:29127", "accept": false, - "reason": "XLMRoberta Flash Attention 2 support and LayoutLMv3 error-handling clarity are unrelated." - }, - { - "left": "issue:43526", - "right": "issue:44186", - "accept": false, - "reason": "BeitImageProcessorFast label reduction and LayoutLMv2Tokenizer NER/padding crashes are unrelated." + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to LayoutLMv3 error handling." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "Qwen2.5-VL import error and a generic dependency issue are not the same bug." - }, - { - "left": "issue:46121", - "right": "issue:46399", - "accept": false, - "reason": "Rope-parameter TypeError and Qwen3.5 kernel-registration ValueError are different validation paths." - }, - { - "left": "issue:27957", - "right": "issue:33290", - "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta and Adafactor/deepspeed OOM are unrelated." + "reason": "Missing Qwen2.5-VL import symbol and a generic dependency issue are not the same concrete problem." }, { - "left": "issue:43994", - "right": "issue:46369", + "left": "issue:43526", + "right": "issue:45834", "accept": false, - "reason": "SIGLIP2 inference problems and SAM3-litetext memory overflow are unrelated." + "reason": "BeitImageProcessorFast label reduction bug and Kosmos2.5 OCR index error are unrelated." }, { - "left": "issue:43334", - "right": "issue:43531", + "left": "issue:43526", + "right": "issue:46693", "accept": false, - "reason": "Both are Qwen3-family issues, but one is a missing pad_token_id and the other is a sliding-window bug; not the same concrete defect." + "reason": "Label-reduction bug and FA2 long-context performance regression are unrelated." }, { - "left": "issue:33290", - "right": "issue:34689", + "left": "issue:27957", + "right": "issue:33290", "accept": false, - "reason": "Adafactor/deepspeed OOM and Llama 3.2 Vision model-loading regression are unrelated." + "reason": "XLMRoberta FA2 support and deepspeed/adafactor OOM are different failures." } ] }, @@ -10189,40 +10178,38 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:27957", "issue:29870", "issue:33290", + "issue:34689", "issue:35141", "issue:42175", "issue:42490", "issue:43262", "issue:43316", + "issue:43334", "issue:43381", "issue:43493", "issue:43525", - "issue:43643", + "issue:43531", "issue:43646", "issue:43720", "issue:43723", - "issue:43749", "issue:43761", "issue:43827", "issue:43873", - "issue:43931", + "issue:43994", "issue:44062", "issue:44077", - "issue:44079", "issue:44206", "issue:44442", "issue:44479", "issue:44514", "issue:44743", "issue:44811", - "issue:44871", "issue:44898", - "issue:44913", "issue:44977", "issue:45003", "issue:45245", @@ -10233,23 +10220,29 @@ "issue:45479", "issue:45588", "issue:45600", - "issue:45759", "issue:45859", "issue:45907", + "issue:46121", "issue:46234", + "issue:46369", + "issue:46399", "issue:46640", "issue:46697" ], "soft_pairs": [ + "issue:46121|issue:46399", + "issue:43994|issue:46369", + "issue:43334|issue:43531", + "issue:33290|issue:34689", "issue:45250|issue:46640", "issue:33290|issue:35141", "issue:45588|issue:45600", - "issue:43761|issue:45859", "issue:43381|issue:43873", + "issue:43761|issue:45859", "issue:44514|issue:45381", "issue:43646|issue:44077", - "issue:43262|issue:44206", "issue:44977|issue:45245", + "issue:43262|issue:44206", "issue:44743|issue:44811", "issue:44442|issue:44898", "issue:42490|issue:44743", @@ -10262,203 +10255,192 @@ "issue:43493|issue:43827", "issue:44077|issue:44479", "issue:45325|issue:46697", - "issue:44062|issue:45907", - "issue:43873|issue:44871", - "issue:43761|issue:44079", - "issue:43643|issue:44913", - "issue:43316|issue:45759", - "issue:43749|issue:43931" + "issue:44062|issue:45907" ], - "prepared_review_unit_hash": "4dc90299678feb9b5c53d2cf7affcf9d791e127ddd6c35ddc585f63c38146499", + "prepared_review_unit_hash": "d508085053ccfefd6b03f4fbfabe7d7ebe449f2da56e71140feb1bbed09f0ec7", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 27, - "serialized_chars": 35858, - "estimated_input_tokens": 8965, - "estimated_eval_tokens": 18186 + "soft_pair_count": 26, + "serialized_chars": 35565, + "estimated_input_tokens": 8892, + "estimated_eval_tokens": 18040 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are unrelated regressions or feature issues, and none of the soft pairs look like the same underlying bug/change. The only vaguely overlapping themes (Flash Attention, Qwen/VL, config/load-path regressions) are still concrete enough to stay separate.", - "confidence": 0.91, - "canonical_issue_reason": "No single canonical issue cleanly fits the whole cluster because the items span many unrelated bugs; the closest broad anchor is the generic Flash Attention report.", - "canonical_pr_reason": "No pull requests are present in this cluster.", - "best_issue_reason": "issue:46640 is the broadest and most generic issue in the only loosely related subtheme (Flash Attention), so it is the best representative anchor if one must be chosen.", - "best_pr_reason": "No pull requests are present in this cluster.", + "summary": "The cluster is noisy and mostly heterogeneous; nearly all soft pairs are clearly unrelated. The only plausible duplicate-looking pair is the two very generic Flash Attention 2 reports, but the titles are too vague to confidently merge.", + "confidence": 0.58, + "canonical_issue_reason": "issue:46640 is the broadest and most current flash-attention-related report in the set, so it is the best umbrella candidate if any Flash Attention 2 duplicates are consolidated.", + "canonical_pr_reason": null, + "best_issue_reason": "Same as canonical: it is the most generic issue in the only apparent shared theme, and there is no better representative with clearer scope.", + "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "Different bugs: rope-validation TypeError vs a Qwen3.5 kernel registration crash." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "Unrelated models and symptoms: SigLIP2 output quality vs Sam3 LiteText memory overflow." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Both are Qwen3-family, but one is a missing pad_token_id attribute and the other is a sliding_window behavior bug; not the same issue." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Different subsystems entirely: DeepSpeed/Adafactor OOM vs model loading for Llama 3.2 Vision." + }, { "left": "issue:45250", "right": "issue:46640", "accept": false, - "reason": "Both mention Flash Attention 2, but one is a generic title and the other is an issue-specific report; they do not describe the same concrete failure path." + "reason": "Both are very generic Flash Attention 2 reports, but the available evidence is too thin to confirm the same underlying bug." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "Adafactor/DeepSpeed OOM and embedding resize/reinit are unrelated bugs in different code paths." + "reason": "OOM in optimizer training is unrelated to token embedding resize/post_init reinitialization." }, { "left": "issue:45588", "right": "issue:45600", "accept": false, - "reason": "One is a flash_attention runtime AttributeError; the other is a broken auto_mappings reference to removed configs. Different subsystems and failures." + "reason": "One is a flash_attention integration crash on s_aux=None; the other is an auto-mappings CI break from removed configs." }, { - "left": "issue:43761", - "right": "issue:45859", + "left": "issue:43381", + "right": "issue:43873", "accept": false, - "reason": "CLIPVision hidden_states regression and Qwen3_5Moe CPU-offload incompatibility are distinct model-specific issues." + "reason": "Gradient-checkpointing eval-mode behavior is unrelated to quantization/offloading." }, { - "left": "issue:43381", - "right": "issue:43873", + "left": "issue:43761", + "right": "issue:45859", "accept": false, - "reason": "Gradient checkpointing in eval mode and quantization offloading are separate behaviors with no shared code-path bug." + "reason": "Different model paths and failures: CLIPVision hidden_states regression vs Qwen3_5Moe CPU offload incompatibility." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen2.5-VL-related, but one is a chat-template batching crash and the other is a video rope/position-id regression." + "reason": "Both concern Qwen2.5-VL, but one is batched chat-template padding behavior and the other is video position-id scaling; different bugs." }, { "left": "issue:43646", "right": "issue:44077", "accept": false, - "reason": "AutoConfig missing fields under trust_remote_code and patchtsmixer post_init policy are unrelated." + "reason": "A general custom model init regression is not the same as patchtsmixer's post_init allowance issue." }, { - "left": "issue:43262", - "right": "issue:44206", + "left": "issue:44977", + "right": "issue:45245", "accept": false, - "reason": "Both touch audio preprocessing, but one is a sampling-rate default issue and the other is an unsupported center argument crash." + "reason": "Qwen3.5 flash-attention generation failure is unrelated to the categories > 2^24 runtime error." }, { - "left": "issue:44977", - "right": "issue:45245", + "left": "issue:43262", + "right": "issue:44206", "accept": false, - "reason": "Qwen3.5 flash-attention generation and the 2^24 categories runtime error are completely different failures." + "reason": "Audio chat-template sampling-rate defaulting is unrelated to LastFeatureExtractor's unsupported center argument." }, { "left": "issue:44743", "right": "issue:44811", "accept": false, - "reason": "A recurrent-state cache bug and Whisper batch_decode skipping special tokens are unrelated." + "reason": "Different bugs in different areas: recurrent-state reset vs Whisper batch_decode skipping special tokens." }, { "left": "issue:44442", "right": "issue:44898", "accept": false, - "reason": "FastSpeech2Conformer tokenizer loading and Perceiver interpolation failure are different model/input issues." + "reason": "Tokenizer loading failure and Perceiver image-resolution failure are unrelated." }, { "left": "issue:42490", "right": "issue:44743", "accept": false, - "reason": "A trailing-slash import warning and Qwen recurrent-state reset are not the same bug." + "reason": "Local-path import warnings do not match the Qwen recurrent-state cache bug." }, { "left": "issue:45003", "right": "issue:45440", "accept": false, - "reason": "Unsafe sys.modules access in modeling_utils and DeepSeekV3MoE divergence are separate code-path problems." + "reason": "Unsafe sys.modules access is unrelated to DeepSeekV3MoE divergence from remote implementation." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "BitNet packed-weight loading and tokenizer loading failures have no common underlying defect." + "reason": "BitNet accelerate-loading/unpacked-weights failure is unrelated to tokenizer loading." }, { "left": "issue:42175", "right": "issue:46234", "accept": false, - "reason": "Package extra/backend selection and a missing import_utils helper are different packaging/API issues." + "reason": "Packaging backend inclusion and a missing import_utils helper are different issues." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Gemma3TextConfig API discrepancy and missing pad_token_id on Llama4Config are both config-related, but they are not the same defect." + "reason": "Both are config attribute discrepancies, but they concern different model families and different missing fields." }, { "left": "issue:27957", "right": "issue:29870", "accept": false, - "reason": "XLM-RoBERTa Flash Attention 2 support and Bart decoder_attention_mask assignment error are unrelated." + "reason": "XLM-Roberta Flash Attention 2 support and BART decoder_attention_mask assignment are unrelated." }, { "left": "issue:43723", "right": "issue:45479", "accept": false, - "reason": "Tokenizer loading in v5 and degenerate zero-loss sequence classification are unrelated issues." + "reason": "Tokenizer loading in v5 is unrelated to the degenerate zero-loss classification bug." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation mismatch and outdated docs referencing pipeline() are unrelated." + "reason": "SigLIP2 implementation mismatch and docs still referencing pipeline() are unrelated." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "patchtsmixer post_init policy and Qwen video-input regression are separate model-level regressions." + "reason": "PatchTSMixer post_init policy and Qwen VL video-input regression are unrelated." }, { "left": "issue:45325", "right": "issue:46697", "accept": false, - "reason": "Qwen2.5-VL rope scaling and Swin2SR over-padding are different preprocessing/math bugs." + "reason": "Qwen2.5-VL temporal position-id scaling and Swin2SR over-padding are unrelated." }, { "left": "issue:44062", "right": "issue:45907", "accept": false, - "reason": "AddedToken special-arg duplication and list_repo_files revision forwarding are unrelated API bugs." - }, - { - "left": "issue:43873", - "right": "issue:44871", - "accept": false, - "reason": "Quantization/offloading behavior and Gemma eos_token_id inconsistency do not share the same underlying defect." - }, - { - "left": "issue:43761", - "right": "issue:44079", - "accept": false, - "reason": "CLIPVision hidden_states missing and ModelOutput key assignment semantics are not the same concrete bug." - }, - { - "left": "issue:43643", - "right": "issue:44913", - "accept": false, - "reason": "trust_remote_code config field loss and GPTNeoX rotary_pct reload reset are distinct config-serialization issues." - }, - { - "left": "issue:43316", - "right": "issue:45759", - "accept": false, - "reason": "Gemma3TextConfig API mismatch and unwrapping text_config for composite multimodal configs are different model-init behaviors." - }, - { - "left": "issue:43749", - "right": "issue:43931", - "accept": false, - "reason": "FSDP CPU-efficient loading breakage and Qwen3-VL weight-shape mismatch are separate loading failures." + "reason": "AddedToken special-argument handling is unrelated to list_repo_files revision forwarding." } ] }, @@ -10480,7 +10462,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:16998", "issue:22355", @@ -10497,22 +10479,23 @@ "issue:43550", "issue:43575", "issue:43606", - "issue:43646", + "issue:43643", "issue:43701", "issue:43716", "issue:43742", + "issue:43749", "issue:43761", "issue:43825", + "issue:43873", "issue:43931", "issue:44077", - "issue:44162", + "issue:44079", "issue:44188", "issue:44190", "issue:44442", "issue:44464", "issue:44496", "issue:44534", - "issue:44792", "issue:44855", "issue:44871", "issue:44913", @@ -10521,20 +10504,24 @@ "issue:45216", "issue:45584", "issue:45698", + "issue:45759", "issue:45803", - "issue:46050", "issue:46178", "issue:46234", "issue:46291", - "issue:46365", "issue:46421", "issue:46506", "issue:46821" ], "soft_pairs": [ + "issue:43873|issue:44871", + "issue:43761|issue:44079", + "issue:43643|issue:44913", "issue:44855|issue:45137", - "issue:44987|issue:46178", + "issue:43749|issue:43931", + "issue:43316|issue:45759", "issue:43761|issue:44188", + "issue:44987|issue:46178", "issue:43232|issue:43388", "issue:44190|issue:44496", "issue:43931|issue:45216", @@ -10555,202 +10542,210 @@ "issue:44871|issue:45584", "issue:43316|issue:43716", "issue:43316|issue:46821", - "issue:44077|issue:44534", - "issue:44871|issue:46050", - "issue:43646|issue:44913", - "issue:45803|issue:46365", - "issue:44162|issue:44792" + "issue:44077|issue:44534" ], - "prepared_review_unit_hash": "1d6fc33a95335da789cf3e601623c1970dccf7ae76471954392af720011856e8", + "prepared_review_unit_hash": "fc69c4ba0d71bceda346f6009aa817f5ff56613f1678fc22bc9d4fdb95a9d571", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 28, - "serialized_chars": 35957, - "estimated_input_tokens": 8990, - "estimated_eval_tokens": 18236 + "soft_pair_count": 29, + "serialized_chars": 36313, + "estimated_input_tokens": 9079, + "estimated_eval_tokens": 18414 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is very noisy: the items span unrelated bugs, regressions, environment errors, and model-specific issues. None of the soft pairs look like true duplicates; at best a few share a broad theme, but not the same concrete code path.", - "confidence": 0.29, - "canonical_issue_reason": "No true canonical duplicate emerges; the cluster is too heterogeneous to treat any one issue as the same bug as the rest.", + "summary": "This cluster is highly heterogeneous: the soft edges are mostly superficial topical overlaps, not the same underlying bug or change. I would not merge any of the listed pairs as duplicates. #43873 is the most active issue, but it is still too broad to serve as a real canonical duplicate hub.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue: the items span unrelated bugs; issue:43873 is only the most active thread, not a true duplicate hub.", "canonical_pr_reason": null, - "best_issue_reason": "If an anchor is needed, issue:45698 is the most representative because it is open, active, and the broadest save/load regression in the set.", + "best_issue_reason": "issue:43873 has the strongest engagement (discussion/inbound refs) and is the closest thing to a hub, though it is still too broad to represent the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Different problems: offloading/quantization handling vs Gemma-3 eos_token_id config inconsistency." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVision hidden_states regression and ModelOutput key assignment are unrelated code paths." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "AutoConfig trust_remote_code missing fields is unrelated to GPTNeoXConfig rotary_pct reload behavior." + }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 import/parsing failure vs DeepSpeed ZeRO3 deque underflow; different code paths and symptoms." + "reason": "DebertaV2 import/script parsing on Python 3.13 is unrelated to DeepSpeed ZeRO3 deque errors." }, { - "left": "issue:44987", - "right": "issue:46178", + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Qwen3-VL weight-shape mismatch are different loading failures." + }, + { + "left": "issue:43316", + "right": "issue:45759", "accept": false, - "reason": "Model loading regression for a specific repo vs missing BACKENDS_MAPPING entry for tensorflow_text; unrelated causes." + "reason": "Both involve multimodal/composite configs, but the concrete bugs and models differ." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "CLIPVision hidden_states regression vs attention-kernel divergence under torch.compile; not the same bug." + "reason": "Hidden_states regression in CLIPVision is unrelated to attention-kernel divergence under torch.compile." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "Physical-intelligence/fast loading failure is unrelated to the tensorflow_text backend-mapping error." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs sync issue vs metric gathering dropping tuple labels; different subsystems." + "reason": "Generation sync_gpus kwargs bug is unrelated to gather_for_metrics truncating tuple labels." }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Local dataset loading script error vs unrecognized model/config loading failure." + "reason": "Local dataset loading in an example script is unrelated to an unrecognized model/config loading error." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Model weight-shape mismatch on load vs save_pretrained checkpoint regression; related area but not the same concrete failure." + "reason": "Both touch Qwen3-VL/Qwen3.5, but one is a load-time shape mismatch and the other is a save_pretrained regression." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Dequantized model save failure vs custom-module resolution after save_pretrained/from_pretrained; different serialization bugs." + "reason": "Dequantized model saving failure and wrong custom-module loading are separate save/load bugs." }, { "left": "issue:44442", "right": "issue:45803", "accept": false, - "reason": "AutoTokenizer load failure vs Fuyu tensor conversion exception handling; unrelated." + "reason": "AutoTokenizer loading failure and a bare except in FuyuBatchFeature are unrelated." }, { "left": "issue:43701", "right": "issue:46506", "accept": false, - "reason": "Checkpoint key mismatch on resume vs CPU/CUDA output divergence; different problems." + "reason": "resume_from_checkpoint key mismatch is unrelated to CPU/CUDA divergence in PPDocLayoutV3." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "3d_parallel.py execution failure vs tensor-parallel OOM when loading a large model; not the same issue." + "reason": "3d_parallel.py runtime failure and tp-load OOM on Qwen2 are different issues." }, { "left": "issue:27957", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 support issue vs NoneType broadcast error in Trainer; unrelated." + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Trainer _gpu_broadcast_one NoneType errors." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "OOM during tensor-parallel load vs key error loading a different model; both loading-related but not the same bug." + "reason": "Qwen2 tensor-parallel OOM and MobileLLM config key error are different loading bugs." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "torch.compile/SDPA failure vs resume_from_checkpoint key mismatch; different code paths." + "reason": "torch.compile SDPA failure in Bamba is unrelated to resume_from_checkpoint key mismatches." }, { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Pipeline top_k output shape change vs missing PyTorch import error; unrelated." + "reason": "pipeline top_k nesting behavior is unrelated to missing PyTorch import errors." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch vs checkpoint key mismatch; not the same underlying bug." + "reason": "CPU offload device mismatch and checkpoint key mismatch are different failure modes." }, { "left": "issue:44464", "right": "issue:46421", "accept": false, - "reason": "Chunked generation output inconsistency vs spurious recompilation from cache lazy init; similar topic, different failure." + "reason": "Chunked-generation output inconsistency and lazy cache recompilation are related broadly to generation, but not the same concrete bug." }, { "left": "issue:46234", "right": "issue:46291", "accept": false, - "reason": "Missing import_utils helper vs kernels dependency version/revision import failure; unrelated." + "reason": "Missing import_utils symbol and kernels backend version/revision failure are unrelated import-time issues." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "model_max_length question vs missing transformers.onnx module; clearly unrelated." + "reason": "DeBERTa model_max_length question and transformers.onnx missing-module error are unrelated." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Custom model/processor.tokenizer regression vs pipeline translation error-message regression; different features." + "reason": "Processor/tokenizer regression and pipeline translation-task messaging are separate v5 issues." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Custom processor.tokenizer regression vs Gemma-3 eos_token_id config mismatch; unrelated." + "reason": "Processor/tokenizer regression is unrelated to Gemma-3 eos_token_id inconsistency." }, { "left": "issue:44871", "right": "issue:45584", "accept": false, - "reason": "Gemma-3 eos_token_id inconsistency vs Whisper empty transcription after align_special_tokens; different models and bugs." + "reason": "Gemma eos_token config mismatch and Whisper empty-transcription failure are different tokenization bugs." }, { "left": "issue:43316", "right": "issue:43716", "accept": false, - "reason": "Gemma3TextConfig API discrepancy vs Mistral-3 image preprocessor dtype mismatch; unrelated." + "reason": "Gemma3TextConfig API discrepancy is unrelated to Mistral-3 image-preprocessor dtype mismatch." }, { "left": "issue:43316", "right": "issue:46821", "accept": false, - "reason": "Gemma3TextConfig API discrepancy vs TimesFM window_size AttributeError; different components." + "reason": "Gemma3TextConfig API discrepancy is unrelated to TimesFM window_size AttributeError." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "patchtsmixer post_init validation vs non-persistent buffers junk-fill regression; not the same code path." - }, - { - "left": "issue:44871", - "right": "issue:46050", - "accept": false, - "reason": "Gemma-3 token-id config mismatch vs quantized save_pretrained/from_pretrained corruption; unrelated." - }, - { - "left": "issue:43646", - "right": "issue:44913", - "accept": false, - "reason": "Custom model init breakage in v5 vs GPTNeoX rotary_pct reload regression; different underlying defects." - }, - { - "left": "issue:45803", - "right": "issue:46365", - "accept": false, - "reason": "Bare except swallowing interrupts in Fuyu batching vs Trainer import blocked by datasets; unrelated." + "reason": "patchtsmixer post_init policy and non-persistent buffer junk initialization are different model/config issues." } ] }, @@ -10772,11 +10767,9 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:35141", "issue:41950", - "issue:42490", "issue:43065", "issue:43352", "issue:43388", @@ -10789,47 +10782,54 @@ "issue:43531", "issue:43540", "issue:43572", + "issue:43646", "issue:43650", "issue:43673", "issue:43688", "issue:43761", + "issue:44162", "issue:44190", - "issue:44479", "issue:44496", "issue:44514", "issue:44534", "issue:44610", "issue:44617", - "issue:44734", "issue:44743", "issue:44792", + "issue:44871", "issue:44913", "issue:44964", "issue:44977", "issue:45127", "issue:45200", - "issue:45406", "issue:45542", "issue:45600", "issue:45632", "issue:45698", "issue:45803", + "issue:46050", "issue:46170", "issue:46178", "issue:46202", "issue:46234", + "issue:46365", "issue:46369", "issue:46424", - "issue:46640" + "issue:46640", + "issue:46832" ], "soft_pairs": [ + "issue:44871|issue:46050", + "issue:43646|issue:44913", + "issue:44162|issue:44792", + "issue:45803|issue:46365", "issue:43452|issue:44496", - "issue:45803|issue:46170", "issue:43404|issue:43441", + "issue:45803|issue:46170", "issue:45632|issue:46424", "issue:44610|issue:46369", + "issue:43761|issue:46832", "issue:43650|issue:46202", - "issue:43761|issue:44479", "issue:43572|issue:46178", "issue:43540|issue:44514", "issue:45542|issue:46234", @@ -10838,188 +10838,198 @@ "issue:43388|issue:43688", "issue:44792|issue:44977", "issue:44792|issue:45600", - "issue:45127|issue:45200", "issue:43673|issue:44743", "issue:43352|issue:46640", + "issue:45127|issue:45200", "issue:41950|issue:43475", "issue:43525|issue:43531", "issue:44190|issue:44964", - "issue:44534|issue:44913", - "issue:35141|issue:42490", - "issue:44734|issue:45406" + "issue:44534|issue:44913" ], - "prepared_review_unit_hash": "b146725045dff085677e8fc31563e7eb3bd85d8f4d50a0272fb11b0208254399", + "prepared_review_unit_hash": "c768a69374aa9753a34cbfeefbdf8fab8ea34c052fdeb437afaa859fb8b76b65", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 24, - "serialized_chars": 34330, - "estimated_input_tokens": 8583, - "estimated_eval_tokens": 17422 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35389, + "estimated_input_tokens": 8848, + "estimated_eval_tokens": 17952 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a single duplicate set; it mixes many unrelated regressions across models, processors, cache behavior, and config/loading paths. All soft pairs look like false positives and should be kept separate.", - "confidence": 0.95, - "canonical_issue_reason": "No issue is a clear canonical target because the items are semantically unrelated rather than variations of one underlying bug.", + "summary": "This set is mostly a collection of distinct Transformers bug reports with only superficial thematic overlap (loading/saving, config regressions, flash-attn, multimodal/video, cache behavior). I would not merge any of the soft pairs as duplicates. No PRs are present.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45698 is the broadest and most active report here: it describes a general from_pretrained/save_pretrained custom-module loading regression, has strong discussion/inbound references, and is the best representative if one issue must anchor the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "There is no strong global best issue for deduping this cluster; the nearest matches are still different code paths and different product areas.", + "best_issue_reason": "issue:45698 is the strongest standalone issue to keep: it is open, well-discussed, and framed at the right abstraction level for a reusable loading/serialization bug rather than a model-specific symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43452", - "right": "issue:44496", + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "Both concern generation/model config after v5 changes, but one is an eos_token_id mismatch and the other is quantized-model roundtrip corruption; different underlying bugs." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Both are config/regression reports, but one is a broad v5 custom-model initialization breakage while the other is a specific GPTNeoX rotary_pct reload bug." + }, + { + "left": "issue:44162", + "right": "issue:44792", "accept": false, - "reason": "Both are load-time errors, but one is about `gguf_file` handling in tokenizer/model loading while the other is an unrecognized model/config.json problem; different failure modes and code paths." + "reason": "ESM2 being broken and janus image-generation test failure are unrelated model-specific failures, not the same bug." }, { "left": "issue:45803", - "right": "issue:46170", + "right": "issue:46365", + "accept": false, + "reason": "One is a bare-except bug in Fuyu tensor conversion; the other is an import failure when datasets is installed. Different code paths and failure modes." + }, + { + "left": "issue:43452", + "right": "issue:44496", "accept": false, - "reason": "Both mention error handling/import trouble, but one is a bare `except` swallowing interrupts in tensor conversion and the other is a Windows/ROCm `torch.distributed` dependency issue." + "reason": "Both are loading-related, but one is about gguf_file handling in from_pretrained and the other is an unrecognized model/config.json problem for a specific model repository." }, { "left": "issue:43404", "right": "issue:43441", "accept": false, - "reason": "One is a missing/tied `lm_head` weight bug, the other is a FlashAttention support/runtime failure for a different model family." + "reason": "Both mention model support constraints, but one is a missing weight tie in Mistral3 and the other is a FlashAttention compatibility issue for Ministral-3." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "KeyboardInterrupt-swallowing in Fuyu tensor conversion and a Windows/ROCm import dependency issue are unrelated." }, { "left": "issue:45632", "right": "issue:46424", "accept": false, - "reason": "Different cache bugs: one is a local `trust_remote_code` path collision, the other is `max_cache_length` only being honored for quantized cache." + "reason": "Cache-path collision for local models and max_cache_length only applying to Quantized Cache involve different cache subsystems and different bugs." }, { "left": "issue:44610", "right": "issue:46369", "accept": false, - "reason": "One is a processor/model input size mismatch, the other is a memory overflow in sam3-litetext; not the same underlying issue." + "reason": "Processor output size mismatch and Sam3-lite-text memory overflow are separate SAM/vision-model issues with different root causes." }, { - "left": "issue:43650", - "right": "issue:46202", + "left": "issue:43761", + "right": "issue:46832", "accept": false, - "reason": "The titles are unrelated; one is a malformed/meme issue, the other is a Gemma tokenizer roundtrip bug." + "reason": "CLIPVision hidden_states regression and RT-DETR feature-level mismatch are unrelated architecture-specific failures." }, { - "left": "issue:43761", - "right": "issue:44479", + "left": "issue:43650", + "right": "issue:46202", "accept": false, - "reason": "Both involve vision/video regressions, but one is CLIP hidden states not returned and the other is a video-input regression for several Qwen variants." + "reason": "A generic 'ADD THE DATA' issue is not the same as Gemma pipeline roundtrip failure on the underscore token; no shared bug semantics." }, { "left": "issue:43572", "right": "issue:46178", "accept": false, - "reason": "Both mention missing backend/config behavior, but one is `pad_token_idx` in StableLmConfig and the other is an undefined `tensorflow_text` backend mapping." + "reason": "StableLm pad_token_idx regression and a tensorflow_text backend mapping error are different config/backend issues." }, { "left": "issue:43540", "right": "issue:44514", "accept": false, - "reason": "Both are Qwen video-related, but one is a `ValueError` in Qwen3OmniMoe video processing and the other is a batched chat-template padding crash in Qwen2.5 VL." + "reason": "Qwen3OmniMoe video-processing ValueError and Qwen2_5_VLProcessor batched chat-template crash are both multimodal, but they fail in different code paths." }, { "left": "issue:45542", "right": "issue:46234", "accept": false, - "reason": "Different dependency/import problems: tensorboard without TensorFlow vs a missing `is_falcon_mamba_ssm_available` utility symbol." + "reason": "TensorBoard-without-TensorFlow backend error and missing is_falcon_mamba_ssm_available helper are unrelated dependency/API issues." }, { "left": "issue:43065", "right": "issue:44617", "accept": false, - "reason": "`Sam3PixelDecoder` dummy conv2d and Sam3Video CUDA OOM are not the same bug; one is architectural placeholder code, the other is runtime memory pressure." + "reason": "A dummy Conv2d in Sam3PixelDecoder and Sam3Video CUDA OOM are not the same bug; one is architectural placeholder code, the other is resource exhaustion." }, { "left": "issue:43504", "right": "issue:45698", "accept": false, - "reason": "One is a legacy field preventing Beit pretrained load; the other is a wrong custom module being loaded after save_pretrained, which is a different serialization/path issue." + "reason": "Beit legacy-field preset loading failure and wrong custom-module loading after save_pretrained are both loading problems, but they affect different mechanisms and code paths." }, { "left": "issue:43388", "right": "issue:43688", "accept": false, - "reason": "Both involve metric/loss behavior, but one is tuple label truncation in `gather_for_metrics` and the other is auxiliary loss normalization in MoE models." + "reason": "Batch-label dropping in gather_for_metrics and auxiliary-loss normalization are both training-related, but they are separate bugs in different subsystems." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Different Qwen-related failures: a janus image-generation test case vs flash-attention generation behavior in Qwen3.5." + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are distinct model/runtime failures." }, { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "A janus generation test failure and stale `auto_mappings.py` references are unrelated issues in different subsystems." - }, - { - "left": "issue:45127", - "right": "issue:45200", - "accept": false, - "reason": "Both touch embedding/token handling, but one is LoRA merge collapse with tied embeddings and the other is missing `mm_token_type_ids` defaults in Gemma 4." + "reason": "Janus test failure and auto_mappings.py referencing removed Sam3LiteText configs are unrelated; one is a runtime test bug, the other is CI/import mapping breakage." }, { "left": "issue:43673", "right": "issue:44743", "accept": false, - "reason": "Both mention cache/state handling, but one is GenerationMixin cache missing during chunked prefill and the other is recurrent states resetting in modular_qwen3_5 with seq_len>1." + "reason": "Generation cache missing during chunked_prefill and recurrent states resetting in modular_qwen3_5.py both involve state/caching, but they are different mechanisms and code paths." }, { "left": "issue:43352", "right": "issue:46640", "accept": false, - "reason": "Both are FlashAttention-related, but one is a concrete unsupported-model error for NemotronH and the other is a generic open flash-attention issue with no shared code path shown." + "reason": "NemotronH FlashAttention 2 support gap and a generic Flash Attention 2 issue are too vague and model-specific to treat as the same bug." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA/extended-vocab collapse with tied embeddings and Gemma-4 mm_token_type_ids defaulting are different model-training problems." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Video-classification processor lookup and a missing `fpn_position_embeddings` attribute in Sam3Video are different bugs in different components." + "reason": "Video-classification pipeline looking for image processors and Sam3VisionEncoderOutput missing fpn_position_embeddings are separate video/model API bugs." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "One is a missing `pad_token_id` on `Llama4Config`; the other is a `sliding_window` issue for Qwen3-MoE. Similar general area, but not the same defect." + "reason": "Missing pad_token_id in Llama4Config and sliding_window issues in Qwen3-MoE are both config-related, but they are not the same defect." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Both are loading-related, but one is local dataset loading in an example script and the other is model loading for Phi-4 multimodal; unrelated code paths." + "reason": "Local dataset loading in an example script and loading microsoft/Phi-4-multimodal-instruct with latest transformers are unrelated failures." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer junk in v5 and GPTNeoX `rotary_pct` reverting on reload are different configuration/serialization bugs." - }, - { - "left": "issue:35141", - "right": "issue:42490", - "accept": false, - "reason": "One is token/output embedding reinit after resizing; the other is import warnings for local model paths ending with a slash. Not the same issue." - }, - { - "left": "issue:44734", - "right": "issue:45406", - "accept": false, - "reason": "KV-cache continuation indexing in serve and Gemma4Processor missing `_tokenizer` are distinct runtime failures." + "reason": "Non-persistent buffers filled with junk in v5 and GPTNeoX rotary_pct silently reverting are both regressions, but they affect different internals." } ] }, @@ -11041,7 +11051,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:25251", "issue:27957", @@ -11049,10 +11059,9 @@ "issue:35141", "issue:39692", "issue:41720", + "issue:42490", "issue:43065", "issue:43335", - "issue:43526", - "issue:43540", "issue:43550", "issue:43644", "issue:43673", @@ -11068,8 +11077,8 @@ "issue:44610", "issue:44617", "issue:44734", - "issue:44779", "issue:44855", + "issue:44869", "issue:44877", "issue:44913", "issue:44933", @@ -11078,6 +11087,7 @@ "issue:45357", "issue:45362", "issue:45381", + "issue:45406", "issue:45563", "issue:45584", "issue:45588", @@ -11086,13 +11096,16 @@ "issue:46082", "issue:46369", "issue:46421", - "issue:46489", "issue:46617", "issue:46655", - "issue:46697" + "issue:46697", + "issue:46832" ], "soft_pairs": [ + "issue:44734|issue:45406", + "issue:35141|issue:42490", "issue:46082|issue:46697", + "issue:44869|issue:46832", "issue:45632|issue:46617", "issue:44855|issue:45588", "issue:43673|issue:46421", @@ -11107,8 +11120,8 @@ "issue:43335|issue:44855", "issue:44479|issue:44734", "issue:45381|issue:46369", - "issue:44077|issue:44877", "issue:43550|issue:44190", + "issue:44077|issue:44877", "issue:41720|issue:45245", "issue:43065|issue:44610", "issue:43335|issue:44190", @@ -11116,224 +11129,211 @@ "issue:45127|issue:45357", "issue:39692|issue:44190", "issue:45381|issue:46655", - "issue:44112|issue:45600", - "issue:43526|issue:46697", - "issue:44779|issue:46489", - "issue:44617|issue:46369", - "issue:43540|issue:44734" + "issue:44112|issue:45600" ], - "prepared_review_unit_hash": "8fac2d14d9f8425855007836b732b53637a025cefef84e80a737d5677867043d", + "prepared_review_unit_hash": "7639a545e4c9da51cf13be95e93ab7c8ed6c8370c0a5e5f3318a94cb198ca7d5", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35869, - "estimated_input_tokens": 8968, - "estimated_eval_tokens": 18192 + "soft_pair_count": 28, + "serialized_chars": 35624, + "estimated_input_tokens": 8906, + "estimated_eval_tokens": 18068 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is almost entirely unrelated issue-level bugs. The only likely duplicate pair is the DeepSeek tokenizer regression reports (#44779 and #46489), where #46489 looks like a narrower follow-up to the broader v5 tokenizer breakage in #44779.", - "confidence": 0.79, - "canonical_issue_reason": "#44779 is the broadest and most representative report: it describes the v5 DeepSeek tokenizer regression in general, and it also has inbound references. It works well as the umbrella issue for the more specific DeepSeek-Coder follow-up.", + "summary": "No soft-edge pair looks like the same underlying bug/change; these issues span unrelated subsystems (generation, model configs, docs, vision/video, caching, and CI).", + "confidence": 0.94, + "canonical_issue_reason": "issue:43644 is the broadest, most infrastructure-level report here (a v5.0.0 regression affecting non-persistent buffers across models), so it is the best single issue to anchor the cluster even though the cluster is not a true duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "#44779 is the best issue to anchor the duplicate set because it captures the general regression rather than a model-specific subset, making it the most useful canonical issue for triage.", + "best_issue_reason": "issue:43644 is the strongest representative issue because it is cross-cutting, concrete, and has downstream impact beyond one model family; the rest are narrower, model-specific, or documentation/CI-only reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Different failure modes and code paths: KV-cache continuation tensor indexing in serve vs. Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "One is output embedding reinit after resize; the other is a local-path trailing-slash import warning. No shared bug." + }, { "left": "issue:46082", "right": "issue:46697", "accept": false, - "reason": "Unrelated model-config validation bug vs. image processor over-padding bug; different code paths and symptoms." + "reason": "LlamaConfig head_dim validation and Swin2SR image over-padding are unrelated." + }, + { + "left": "issue:44869", + "right": "issue:46832", + "accept": false, + "reason": "Whisper timestamp decode crash and RT-DETR backbone feature-level TypeError are distinct model bugs." }, { "left": "issue:45632", "right": "issue:46617", "accept": false, - "reason": "Both involve trust_remote_code/cache, but one is a leaf-directory collision and the other is a FileNotFoundError while loading from cache; not clearly the same underlying bug." + "reason": "Both involve trust_remote_code/cache, but one is leaf-directory cache collision and the other is FileNotFoundError from cache loading; not the same root cause." }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Python 3.13 TorchScript parsing error vs. flash_attention crash on s_aux=None; different subsystems and failure modes." + "reason": "Python 3.13 JIT parsing/IndentationError is unrelated to flash_attention.py crashing on s_aux=None." }, { "left": "issue:43673", "right": "issue:46421", "accept": false, - "reason": "Both mention chunked prefill/cache, but one is missing GenerationMixin cache in v5.0.0 and the other is lazy cache init causing recompilation; related area, not the same bug." + "reason": "Both mention chunked prefill/cache, but one is missing cache in v5.0.0 while the other is lazy cache init causing recompilation." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Weight-shape mismatch on model load is unrelated to the tied-weights warning in Qwen3-Next." + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-Next tied-weights warning are different issues in different model families." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs referencing removed pipeline() APIs is not the same bug as SigLIP2 producing bad outputs with AutoModel/pipeline." + "reason": "Documentation references removed pipeline APIs; the other is a SigLIP2 inference correctness problem." }, { "left": "issue:29870", "right": "issue:35141", "accept": false, - "reason": "Bart generate() decoder_attention_mask error and output embedding reinitialization after resize are different generation/model-init bugs." + "reason": "Bart generate() decoder_attention_mask bug is unrelated to token-embedding reinitialization after resize." }, { "left": "issue:44617", "right": "issue:45584", "accept": false, - "reason": "CUDA OOM in Sam3Video is unrelated to Whisper failing on empty transcription after align_special_tokens." + "reason": "Sam3Video OOM is not the same as Whisper empty-transcription/alignment crash." }, { "left": "issue:45563", "right": "issue:45584", "accept": false, - "reason": "A stale warning about num_return_sequences is not the same as an empty-transcription generation failure." + "reason": "A stale warning about num_return_sequences is a different bug from Whisper failing on empty transcription after align_special_tokens." }, { "left": "issue:43644", "right": "issue:44913", "accept": false, - "reason": "Non-persistent buffer initialization junk and GPTNeoXConfig rotary_pct reset are different configuration/serialization issues." + "reason": "Generic v5 buffer corruption and GPTNeoX rotary_pct reload regression are unrelated config/runtime bugs." }, { "left": "issue:44933", "right": "issue:45362", "accept": false, - "reason": "Missing import from image_utils is unrelated to Qwen3.5 chat crashes." + "reason": "Nonexistent import from image_utils and Qwen3.5 chat crash are different problems." }, { "left": "issue:25251", "right": "issue:27957", "accept": false, - "reason": "Pipeline top_k nesting behavior and XLMRoberta Flash Attention 2 are unrelated." + "reason": "Pipeline top_k nesting bug and XLMRoberta Flash Attention 2 support are unrelated." }, { "left": "issue:43335", "right": "issue:44855", "accept": false, - "reason": "Sparse-layer config bug and a Python 3.13 TorchScript import failure are unrelated." + "reason": "SwitchTransformers sparse-layer creation on zero layers has nothing to do with DebertaV2Model import parsing on Python 3.13." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "Qwen VL video-input regression and serve KV-cache tensor-indexing crash are different code paths." + "reason": "Qwen VL video input regression and serve KV-cache continuation crash affect different paths and symptoms." }, { "left": "issue:45381", "right": "issue:46369", "accept": false, - "reason": "Qwen2.5-VL video-position-id issue and Sam3-litetext memory overflow are not the same bug." + "reason": "Wrong vision_position_ids in Qwen2.5-VL is unrelated to Sam3-LiteText memory overflow." }, { - "left": "issue:44077", - "right": "issue:44877", + "left": "issue:43550", + "right": "issue:44190", "accept": false, - "reason": "Optional post_init handling and strict config loading for granite_speech are different problems." + "reason": "torch.compile + SDPA failure in Bamba-9B-v2 is unrelated to loading a local dataset in the image-classification script." }, { - "left": "issue:43550", - "right": "issue:44190", + "left": "issue:44077", + "right": "issue:44877", "accept": false, - "reason": "torch.compile/SDPA failure in Bamba is unrelated to dataset loading in the example script." + "reason": "PatchTSMixer post_init policy and granite_speech config loading are different config-validation issues." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "Auto device mapping cudaErrorAssert and category-count overflow are unrelated." + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to the category-count limit RuntimeError." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo input-size mismatch are different model/processor issues." + "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo processor size mismatch are distinct model/processor bugs." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "Sparse-layer config creation and local dataset loading failure are unrelated." + "reason": "SwitchTransformers sparse-layer bug is unrelated to dataset loading in the no_trainer example." }, { "left": "issue:44464", "right": "issue:45563", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is not the same as a stale warning emission." + "reason": "Chunked generation inconsistency with compiled forward and a stale warning about num_return_sequences are not the same bug." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA merging with extended vocab collapse and incorrect visual encoder keys on save_pretrained are different serialization/model-weight bugs." + "reason": "LoRA merge collapse with extended vocabulary and incorrect visual-encoder keys on save_pretrained are different regressions." }, { "left": "issue:39692", "right": "issue:44190", "accept": false, - "reason": "SigLIP2 docs example errors are unrelated to local dataset loading in image classification." + "reason": "SigLIP2 docs/model mismatch and local dataset loading failure are unrelated." }, { "left": "issue:45381", "right": "issue:46655", "accept": false, - "reason": "Both are video-related, but Qwen2.5-VL vision_position_ids and SmolVLM padding calculation are different bugs in different processors." + "reason": "Qwen2.5-VL video position IDs bug and SmolVLM padding calculation bug affect different processors." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "GraniteSpeech CI device-override test flakiness and removed Sam3LiteText auto-mapping references are separate CI issues." - }, - { - "left": "issue:43526", - "right": "issue:46697", - "accept": false, - "reason": "BeitImageProcessorFast reduce_labels returning one label and Swin2SR over-padding are different image-processor bugs." - }, - { - "left": "issue:44779", - "right": "issue:46489", - "accept": true, - "reason": "Both report DeepSeek tokenizer regressions on transformers v5+, and #46489 reads like a narrower follow-up/gap in the fix for the broader issue described in #44779." - }, - { - "left": "issue:44617", - "right": "issue:46369", - "accept": false, - "reason": "CUDA OOM in Sam3Video is unrelated to memory overflow when calling sam3-litetext." - }, - { - "left": "issue:43540", - "right": "issue:44734", - "accept": false, - "reason": "Qwen3OmniMoe video-input ValueError and serve KV-cache continuation tensor indexing are different failures." + "reason": "GraniteSpeech stale CI test and auto_mappings referencing removed Sam3LiteText configs are separate CI breakages." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The assessment is well grounded in the packet and stays conservative on the soft pairs. The only accepted duplicate, #44779/#46489, is supported by the shared DeepSeek tokenizer regression on transformers v5+ and the fact that #46489 reads as a narrower follow-up to the broader #44779 report. The remaining pairs are appropriately rejected as different bugs or only loosely related." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -11348,12 +11348,11 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:25251", "issue:27957", "issue:29870", - "issue:32090", "issue:35141", "issue:36246", "issue:38617", @@ -11361,8 +11360,8 @@ "issue:42175", "issue:42796", "issue:43295", - "issue:43404", - "issue:43479", + "issue:43526", + "issue:43540", "issue:43643", "issue:43650", "issue:43854", @@ -11371,20 +11370,19 @@ "issue:44220", "issue:44246", "issue:44279", - "issue:44295", "issue:44462", "issue:44496", "issue:44521", + "issue:44617", "issue:44734", "issue:44737", + "issue:44779", "issue:44938", "issue:44964", "issue:45003", - "issue:45216", "issue:45245", "issue:45276", "issue:45290", - "issue:45375", "issue:45381", "issue:45542", "issue:45584", @@ -11395,16 +11393,22 @@ "issue:46082", "issue:46121", "issue:46234", - "issue:46506", + "issue:46369", + "issue:46489", + "issue:46697", "issue:46752" ], "soft_pairs": [ + "issue:44779|issue:46489", + "issue:43526|issue:46697", + "issue:44617|issue:46369", + "issue:43540|issue:44734", "issue:43295|issue:45276", "issue:45290|issue:46752", "issue:44737|issue:45803", "issue:44462|issue:45907", - "issue:45276|issue:45584", "issue:44521|issue:45381", + "issue:45276|issue:45584", "issue:44938|issue:45542", "issue:25251|issue:29870", "issue:45632|issue:46234", @@ -11420,203 +11424,199 @@ "issue:42796|issue:44521", "issue:44186|issue:44734", "issue:36246|issue:41628", - "issue:36246|issue:38617", "issue:43650|issue:44220", - "issue:32090|issue:35141", - "issue:43404|issue:45216", - "issue:43479|issue:45375", - "issue:44295|issue:46506" + "issue:36246|issue:38617" ], - "prepared_review_unit_hash": "230e7a041992b996af438838fd072c851f7bdba130c4eb09ec60324963781ece", + "prepared_review_unit_hash": "d4f020aa79548737dccd1bdba78efa9230ab0a1bdda7023a411e13b92d449c27", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35472, - "estimated_input_tokens": 8868, - "estimated_eval_tokens": 17992 + "serialized_chars": 35350, + "estimated_input_tokens": 8838, + "estimated_eval_tokens": 17932 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly a set of near-miss similarity matches, not true duplicates. The only thematically close groups are around chat templating and model loading, but the paired items still describe different failure modes or code paths, so I would not merge any of them.", - "confidence": 0.84, - "canonical_issue_reason": "issue:44521 is the best anchor for the cluster because it is an active, user-facing multimodal `apply_chat_template` bug with inbound references and the broadest surrounding discussion; it is the clearest representative of the small chat-template subtheme here.", + "summary": "The cluster is highly heterogeneous: the soft-similarity pairs mostly share broad subsystem themes or surface wording, but they do not look like the same underlying bug/change. No PRs are present.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest representative issue here: it is open, has the most inbound references, and describes a concrete, well-scoped multimodal apply_chat_template bug.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44521 is the most suitable global issue because it is open, has multiple references, and sits at the center of the most coherent subset of related items; the rest of the cluster is too heterogeneous to justify a tighter canonical choice.", + "best_issue_reason": "issue:44521 is the most central and actionable issue in this set, with clear impact and active references; it is the best single issue to anchor the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44779", + "right": "issue:46489", + "accept": false, + "reason": "Same tokenizer family and both mention v5 regressions, but one is DeepSeek generally and the other is DeepSeek-Coder v1 with a separate PR-gap claim; not clearly the same concrete bug." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "Both are image-processor edge cases, but they affect different models and different behaviors (label reduction vs over-padding)." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3 memory-related, but they concern different model variants and likely different causes; not one duplicate bug." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Different code paths: Qwen3OmniMoe video input validation versus serve KV-cache continuation indexing." + }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Both mention embedding-related regressions, but one is about `processor.tokenizer`/image passing behavior and the other is about `resize_token_embeddings` reinitializing outputs; different bugs and code paths." + "reason": "Both involve regressions around model/processor behavior, but the concrete failures are unrelated (processor.tokenizer access vs resize_token_embeddings semantics)." }, { "left": "issue:45290", "right": "issue:46752", "accept": false, - "reason": "Both hit `apply_chat_template`, but one is about assistant messages with tool calls and no content, while the other is an empty-list crash; close in surface area, not the same underlying bug." + "reason": "Both are apply_chat_template crashes, but the triggering inputs differ materially: assistant tool-call messages with no content versus an empty messages list." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "XLNet device placement in positional encoding and Fuyu's bare `except:` swallowing errors are unrelated failures despite both touching runtime behavior." + "reason": "Different subsystems and failure modes: CPU tensor allocation in XLNet versus a broad exception handler in FuyuBatchFeature." }, { "left": "issue:44462", "right": "issue:45907", "accept": false, - "reason": "One is `AutoTokenizer` ignoring `tokenizer.json`; the other is `list_repo_files` not forwarding `revision`. Same repo area, different issues." + "reason": "One is AutoTokenizer ignoring tokenizer.json, the other is hub file listing not forwarding revision; related repo-loading area but not the same bug." }, { - "left": "issue:45276", - "right": "issue:45584", + "left": "issue:44521", + "right": "issue:45381", "accept": false, - "reason": "Embedding resize/tie-weight behavior and Whisper empty-transcription generation are distinct model bugs with no shared code-path evidence." + "reason": "Both are multimodal-related, but one is assistant mask generation and the other is vision_position_ids for Qwen2.5-VL video inputs." }, { - "left": "issue:44521", - "right": "issue:45381", + "left": "issue:45276", + "right": "issue:45584", "accept": false, - "reason": "Both are multimodal/chat-template adjacent, but one is all-zero assistant masks and the other is wrong `vision_position_ids` for Qwen2.5-VL video input; not the same defect." + "reason": "Unrelated bugs: Gemma4 embedding resizing versus Whisper empty-transcription generation failure." }, { "left": "issue:44938", "right": "issue:45542", "accept": false, - "reason": "Python 3.14 import failure and missing TensorFlow backend from a Torch install are separate dependency/runtime problems." + "reason": "Python 3.14 import failure is unrelated to missing TensorFlow backend wiring for the torch extra." }, { "left": "issue:25251", "right": "issue:29870", "accept": false, - "reason": "Pipeline `top_k` nesting behavior and Bart `decoder_attention_mask` initialization are unrelated bugs." + "reason": "Different generation/pipeline bugs with different symptoms and code paths (nested-list output vs decoder_attention_mask assignment)." }, { "left": "issue:45632", "right": "issue:46234", "accept": false, - "reason": "`trust_remote_code` cache-path collisions and a missing `is_falcon_mamba_ssm_available` export are different import/runtime issues." + "reason": "One is a trust_remote_code cache-path collision, the other is a missing import_utils helper symbol." }, { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Slow import timing and unsafe `sys.modules[]` access are not the same bug, even though both involve import-time behavior." + "reason": "Import slowness and unsafe sys.modules access are not the same concrete defect." }, { "left": "issue:43854", "right": "issue:44964", "accept": false, - "reason": "Loading `zai-org/GLM-4.7-Flash` in tests and loading `microsoft/Phi-4-multimodal-instruct` in latest transformers are separate model-specific failures." + "reason": "Both are model-loading issues, but they involve different models and likely different root causes." }, { "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "A model-loading regression and stale `auto_mappings.py` references are different failure modes." + "reason": "Different failures: model loading regression versus stale auto_mappings references to removed configs." }, { "left": "issue:43643", "right": "issue:44496", "accept": false, - "reason": "`trust_remote_code=True` dropping fields from `AutoConfig` and an unrecognized model/config error are distinct configuration problems." + "reason": "Both are config-loading problems, but one is trust_remote_code returning incomplete configs while the other is an unrecognized model missing model_type." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "A stale CI device override test and a Phi-4 multimodal loading failure are unrelated." + "reason": "CI failure in GraniteSpeech device override is unrelated to a Phi-4-multimodal load regression." }, { "left": "issue:27957", "right": "issue:35141", "accept": false, - "reason": "Flash Attention 2 support for XLM-Roberta and `post_init` reinitializing embeddings when `tie_word_embedding=False` are different model behaviors." + "reason": "Flash Attention support for XLMRoberta and embedding reinitialization after resize_token_embeddings are unrelated issues." }, { "left": "issue:46082", "right": "issue:46121", "accept": false, - "reason": "Llama `head_dim` validation and Rope-validation list handling are separate config-validation bugs." + "reason": "Different config-validation failures: Llama head_dim rejection versus JSON-loaded rope-ignore key handling." }, { "left": "issue:45245", "right": "issue:45584", "accept": false, - "reason": "`number of categories cannot exceed 2^24` and Whisper empty-transcription generation are unrelated." + "reason": "Category-cardinality overflow is unrelated to Whisper empty-transcription generation failure." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "A packaging/backend-selection issue and a generic dependency issue are too broad and not the same underlying defect." + "reason": "One is a specific backend-install packaging complaint; the other is a vague dependency issue without a matching concrete defect." }, { "left": "issue:42796", "right": "issue:44521", "accept": false, - "reason": "Tokenizer BPE marker handling and multimodal assistant-mask generation are different tokenizer/template bugs." + "reason": "Tokenizer BPE-marker handling and multimodal assistant-mask generation are different bugs in different components." }, { "left": "issue:44186", "right": "issue:44734", "accept": false, - "reason": "LayoutLMv2 NER padding/truncation crashes and `transformers serve` KV-cache continuation indexing are unrelated code paths." + "reason": "LayoutLMv2Tokenizer padding/NER crash and serve KV-cache continuation indexing are unrelated." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but for different symbols (`Qwen2_5_VLImageProcessor` vs `AutoImageProcessor`) and likely different export gaps." - }, - { - "left": "issue:36246", - "right": "issue:38617", - "accept": false, - "reason": "A missing vision processor export and a missing `layer_type_validation` import are unrelated import failures." + "reason": "Both are ImportErrors, but they concern different missing exports in different modules." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "The placeholder/data issue and `_torch_extract_fbank_features()` bug do not describe the same problem." + "reason": "The titles are too generic to infer a shared underlying bug; the described areas are unrelated." }, { - "left": "issue:32090", - "right": "issue:35141", - "accept": false, - "reason": "`_gpu_broadcast_one` rejecting `None` in Trainer and embedding reinitialization after resizing are different failures." - }, - { - "left": "issue:43404", - "right": "issue:45216", - "accept": false, - "reason": "Mistral3 lm_head tying and Qwen3.5 save_pretrained regression affect different model internals." - }, - { - "left": "issue:43479", - "right": "issue:45375", - "accept": false, - "reason": "Phi4 multimodal config defaults being initialized incorrectly and Qwen3_5MoeVisionConfig dropping a field under `@strict` are different config-serialization bugs." - }, - { - "left": "issue:44295", - "right": "issue:46506", + "left": "issue:36246", + "right": "issue:38617", "accept": false, - "reason": "Reading `position_ids` after buffer registration and divergent CPU/CUDA outputs for PPDocLayoutV3 are not the same bug." + "reason": "Different missing symbols in different modules; no evidence they stem from the same change or bug." } ] }, @@ -11638,15 +11638,18 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:39692", + "issue:32090", + "issue:35141", "issue:41950", "issue:42907", "issue:43295", "issue:43299", "issue:43381", + "issue:43404", "issue:43441", + "issue:43479", "issue:43526", "issue:43531", "issue:43673", @@ -11655,16 +11658,14 @@ "issue:43976", "issue:44016", "issue:44062", - "issue:44075", "issue:44220", "issue:44279", + "issue:44295", "issue:44327", "issue:44336", - "issue:44368", "issue:44496", "issue:44521", "issue:44683", - "issue:44779", "issue:44792", "issue:44933", "issue:44987", @@ -11672,17 +11673,15 @@ "issue:45083", "issue:45216", "issue:45335", - "issue:45405", + "issue:45375", "issue:45446", "issue:45584", "issue:45588", "issue:45593", "issue:45600", "issue:45759", - "issue:46082", "issue:46123", "issue:46369", - "issue:46399", "issue:46424", "issue:46506", "issue:46515", @@ -11690,6 +11689,10 @@ "issue:46752" ], "soft_pairs": [ + "issue:43404|issue:45216", + "issue:32090|issue:35141", + "issue:43479|issue:45375", + "issue:44295|issue:46506", "issue:43673|issue:46424", "issue:41950|issue:44016", "issue:43531|issue:45600", @@ -11697,252 +11700,234 @@ "issue:43976|issue:44987", "issue:43295|issue:45335", "issue:45083|issue:46506", - "issue:44683|issue:45446", "issue:43381|issue:45593", "issue:42907|issue:43441", + "issue:44683|issue:45446", + "issue:43673|issue:44336", "issue:43526|issue:43901", "issue:44496|issue:45042", - "issue:43673|issue:44336", "issue:43901|issue:44496", "issue:43976|issue:44279", + "issue:44792|issue:46424", "issue:43441|issue:45588", "issue:44496|issue:45759", - "issue:44792|issue:46424", "issue:44016|issue:44062", "issue:43824|issue:44933", "issue:44327|issue:46752", "issue:46123|issue:46369", "issue:44521|issue:46693", - "issue:44279|issue:46515", "issue:43299|issue:45216", - "issue:45335|issue:45584", - "issue:46082|issue:46399", - "issue:44279|issue:45405", - "issue:43295|issue:44779", - "issue:44075|issue:44368", - "issue:39692|issue:43901", - "issue:43976|issue:45405" + "issue:44279|issue:46515", + "issue:45335|issue:45584" ], - "prepared_review_unit_hash": "b4b9ea9f33887aac05b134899ca835f2e902813cc98270497c352c7dcc7d8421", + "prepared_review_unit_hash": "d1f8dba1ba9608292d5783de308416d83d07b9edd833e5cae15403c23fff0191", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 37199, - "estimated_input_tokens": 9300, - "estimated_eval_tokens": 18856 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36076, + "estimated_input_tokens": 9019, + "estimated_eval_tokens": 18294 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is a grab bag of unrelated Transformers issues across models, pipelines, docs, cache/generation, and backend-specific regressions. None of the soft pairs look like the same underlying bug/change, so all are rejected.", - "confidence": 0.93, - "canonical_issue_reason": "No single canonical issue stands out: the items span many distinct subsystems and model-specific failures, so there is no clear umbrella duplicate to anchor the cluster.", + "summary": "These items are a grab bag of distinct Transformers bugs and regressions rather than a duplicate set. The soft-similarity pairs mostly share broad themes like configs, generation, or model loading, but they target different code paths and should stay separate.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44521 is the strongest representative overall: it is open, has the most inbound references and discussion, and describes a concrete multimodal API regression with a clear repro.", "canonical_pr_reason": null, - "best_issue_reason": "No global best issue candidate: the soft pairs are heterogeneous false positives, and no issue cleanly represents the whole cluster.", + "best_issue_reason": "issue:44521 is the best global issue candidate because it is active, well-scoped, and appears most likely to benefit from continued tracking or follow-up.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both concern model weight tying/save behavior, but they hit different models and different failure modes; not the same underlying bug." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "One is a Trainer GPU broadcast TypeError; the other is post-init reinitialization after resizing embeddings. Related area, different bug." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Both are config-handling issues, but one is default init for Phi4Multimodal and the other is a missing field silently dropped for Qwen3_5MoeVisionConfig." + }, + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "Unrelated failures: buffer access for position_ids versus CPU/CUDA result divergence in PPDocLayoutV3." + }, { "left": "issue:43673", "right": "issue:46424", "accept": false, - "reason": "Both mention cache behavior, but one is a missing GenerationMixin cache in chunked prefill while the other is cache length handling for quantized cache; different code paths." + "reason": "Both involve cache settings, but one is a missing GenerationMixin cache in chunked prefill while the other is max_cache_length being honored only for quantized cache." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Video-classification processor selection and a notebook syntax error are unrelated bugs in different surfaces." + "reason": "A video-classification processor lookup bug and a notebook syntax error are completely different artifacts." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and stale auto_mappings references are distinct issues with no shared fix." + "reason": "Qwen3-MoE sliding_window behavior and stale auto_mappings references are unrelated." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error and _torch_extract_fbank_features() failure are different failures in different components." + "reason": "Notebook syntax error versus an audio feature extraction issue; no shared underlying code path." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "A Python version compatibility report and a model-loading failure are not the same underlying defect." + "reason": "One is a Python-version compatibility problem; the other is a specific model load failure for physical-intelligence/fast." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "Both involve processors/config, but one is processor.tokenizer regression and the other is resize_token_embeddings not updating decoder embeddings." + "reason": "Both mention token/embedding handling, but one is a multimodal processor regression and the other is a t5gemma decoder embedding update bug." }, { "left": "issue:45083", "right": "issue:46506", "accept": false, - "reason": "Helper length computation in qwen3_omni_moe and CPU/CUDA result mismatch in PPDocLayoutV3 are unrelated." - }, - { - "left": "issue:44683", - "right": "issue:45446", - "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 and a PyTorch version check bug for AuxRequest import are separate compatibility problems." + "reason": "Different helper behavior in qwen3_omni_moe versus device-dependent output drift in PPDocLayoutV3." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing in eval mode and D-FINE auxiliary losses when denoising is off affect different model/training paths." + "reason": "Gradient checkpointing eval-mode restriction is unrelated to D-FINE auxiliary-loss behavior." }, { "left": "issue:42907", "right": "issue:43441", "accept": false, - "reason": "Saving dequantized Ministral/Devstral models and Ministral-3 FlashAttention crash are different failures." + "reason": "Both touch Ministral/Devstral areas, but one is save/export of dequantized models and the other is FlashAttention loading failure." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both mention flex_attention, but one is a torch>=2.9 compiled failure and the other is a version-check/import bug for AuxRequest." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Cache/memory-related naming aside, chunked-prefill cache handling and ANSI output formatting are unrelated." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "BeitImageProcessorFast label reduction and text classification docs around return_all_scores are unrelated." + "reason": "ImageProcessor label reduction and text-classification docs mismatch are different subsystems and failures." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Unrecognized model loading and PIL image processors requiring torchvision are different class of issues." - }, - { - "left": "issue:43673", - "right": "issue:44336", - "accept": false, - "reason": "Generation cache missing and ANSI code emission in loading_report have no shared bug." + "reason": "An unrecognized-model config issue is not the same as PIL image processors incorrectly requiring torchvision." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Docs mismatch for return_all_scores and model config recognition failure are unrelated." + "reason": "Documentation mismatch for pipelines and model config import failure do not share an underlying bug." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "Python support mismatch and a generic dependency issue do not describe the same bug." + "reason": "The former is a Python compatibility regression; the latter is a vague dependency problem." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "A failed janus image-generation test and a cache-length configuration issue are different failures." }, { "left": "issue:43441", "right": "issue:45588", "accept": false, - "reason": "FlashAttention crash for Ministral-3 and sink-less s_aux=None AttributeError are separate code-path failures." + "reason": "Ministral FlashAttention loading and sink-less flash_attention aux handling are separate code paths." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "Generic unrecognized model error and composite Qwen multimodal config unwrapping are different model-loading problems." - }, - { - "left": "issue:44792", - "right": "issue:46424", - "accept": false, - "reason": "A janus image-generation test failure is unrelated to cache max length handling." + "reason": "Unrecognized model config versus from_config not unwrapping text_config for composite multimodal configs are related only at a very high level." }, { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error and AddedToken keyword conflict are not the same defect." + "reason": "Notebook syntax error is unrelated to the AddedToken keyword-argument TypeError." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Qwen2.5-VL import error and nonexistent image_utils import are distinct import/export issues." + "reason": "Both are import errors, but they involve different missing symbols from different modules." }, { "left": "issue:44327", "right": "issue:46752", "accept": false, - "reason": "QA decode_spans top-k bounds error and empty apply_chat_template list crash are different pipeline edge cases." + "reason": "QA decode_spans top-k bounds error and empty-list apply_chat_template crash are unrelated." }, { "left": "issue:46123", "right": "issue:46369", "accept": false, - "reason": "MaskGenerationPipeline batch-finalization bug and sam3-litetext memory overflow are unrelated." + "reason": "Partial-batch mask generation dropping results is not the same as memory overflow in sam3-litetext." }, { "left": "issue:44521", "right": "issue:46693", "accept": false, - "reason": "All-zero assistant masks for multimodal chat templates and FA2 long-context performance regression are different generation issues." - }, - { - "left": "issue:44279", - "right": "issue:46515", - "accept": false, - "reason": "A dependency issue with transformers and a torch 2.4.1 incompatibility are not the same bug." + "reason": "Assistant-mask generation for multimodal chat templates is unrelated to FA2 long-context performance regression." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Qwen3VL MoE loading regression and Qwen3.5 save_pretrained checkpoint regression are different model lifecycle problems." - }, - { - "left": "issue:45335", - "right": "issue:45584", - "accept": false, - "reason": "Embedding resize not propagating to decoder.embed_tokens and Whisper empty-transcription failure are unrelated model-specific bugs." - }, - { - "left": "issue:46082", - "right": "issue:46399", - "accept": false, - "reason": "LlamaConfig head_dim validation and a missing decorator for apply_rotary_pos_emb registration are separate configuration/codegen issues." + "reason": "Qwen3VL MoE loading breakage and Qwen3.5 checkpoint save corruption are different model/version regressions." }, { "left": "issue:44279", - "right": "issue:45405", - "accept": false, - "reason": "A vague dependency issue and a MIN_PEFT_VERSION bump to an unreleased version are not the same underlying problem." - }, - { - "left": "issue:43295", - "right": "issue:44779", - "accept": false, - "reason": "processor.tokenizer regression and Deepseek tokenizer incorrect results are different tokenizer-related failures." - }, - { - "left": "issue:44075", - "right": "issue:44368", - "accept": false, - "reason": "SGD args being ignored and a tie_word_embeddings warning are unrelated; one is optimizer behavior, the other is config warning noise." - }, - { - "left": "issue:39692", - "right": "issue:43901", + "right": "issue:46515", "accept": false, - "reason": "SigLIP2 docs example mismatch/quantization failure and text classification docs mentioning return_all_scores are different documentation issues." + "reason": "A generic dependency issue is not the same as a specific torch 2.4.1 incompatibility." }, { - "left": "issue:43976", - "right": "issue:45405", + "left": "issue:45335", + "right": "issue:45584", "accept": false, - "reason": "Python version support and an unreleased PEFT minimum version bump are distinct packaging/dependency concerns." + "reason": "t5gemma embedding resize and Whisper empty-transcription generation failures are unrelated." } ] }, @@ -11964,14 +11949,14 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:33290", + "issue:39692", "issue:41950", "issue:42796", - "issue:42994", + "issue:43295", "issue:43299", - "issue:43317", "issue:43441", "issue:43479", "issue:43525", @@ -11979,18 +11964,19 @@ "issue:43650", "issue:43653", "issue:43854", + "issue:43901", + "issue:43976", "issue:44016", + "issue:44075", "issue:44242", - "issue:44261", - "issue:44263", + "issue:44279", "issue:44291", "issue:44351", "issue:44361", + "issue:44368", "issue:44462", "issue:44466", - "issue:44623", "issue:44683", - "issue:44749", "issue:44779", "issue:45020", "issue:45081", @@ -12001,20 +11987,25 @@ "issue:45278", "issue:45357", "issue:45372", + "issue:45405", "issue:45542", "issue:45593", - "issue:45600", "issue:45656", - "issue:45712", "issue:45812", "issue:45823", "issue:45834", - "issue:45907", "issue:46082", "issue:46209", + "issue:46399", "issue:46512" ], "soft_pairs": [ + "issue:46082|issue:46399", + "issue:44279|issue:45405", + "issue:43295|issue:44779", + "issue:39692|issue:43901", + "issue:44075|issue:44368", + "issue:43976|issue:45405", "issue:46082|issue:46512", "issue:43577|issue:45216", "issue:44351|issue:45278", @@ -12035,197 +12026,198 @@ "issue:44361|issue:45593", "issue:43441|issue:45834", "issue:45103|issue:45372", - "issue:42994|issue:43317", - "issue:33290|issue:45656", - "issue:45081|issue:45907", - "issue:45600|issue:45712", - "issue:44623|issue:44749", - "issue:44261|issue:44263" + "issue:33290|issue:45656" ], - "prepared_review_unit_hash": "688d05e3bac42eea6617d15d9eb5d39fcf7f41080188d2bfff3e14c6251727f4", + "prepared_review_unit_hash": "fd88ccb349a0862e77543d4e69f05b72fbcf524b6bb16087b091b32e554cee3f", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 26, - "serialized_chars": 35044, - "estimated_input_tokens": 8761, - "estimated_eval_tokens": 17778 + "soft_pair_count": 27, + "serialized_chars": 35388, + "estimated_input_tokens": 8847, + "estimated_eval_tokens": 17950 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous transformer bug issues with no true duplicate group. The only near-match is the save_pretrained/serialization regression pair, but they still look like different bugs, so I rejected all soft edges.", - "confidence": 0.33, - "canonical_issue_reason": "No single issue cleanly represents the whole cluster; if one anchor is required, issue 45216 is the closest fit because it is a concrete v5 save_pretrained regression and is the nearest soft match to another serialization-related issue.", + "summary": "This set is mostly a grab-bag of unrelated model/tokenizer/config regressions and docs/infra issues, not a single duplicate cluster. No soft edge looks like the same underlying bug/change.", + "confidence": 0.88, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 45216 is the strongest representative of the only loosely related subgroup (save_pretrained/serialization regressions), but the rest of the cluster is not about the same bug.", + "best_issue_reason": "issue:45020 is the broadest umbrella report here because it covers recent-version breakage in model loading, but the overall set is too heterogeneous to treat as one duplicate family.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "Different bugs: LlamaConfig head_dim validation vs Qwen3.5 rotary-kernel registration failure." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Unrelated: generic transformers dependency issue vs PEFT minimum-version bump." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Both tokenizer regressions, but different models and failure modes (processor/tokenizer images vs Deepseek token IDs)." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Different docs issues in different pipelines; no shared code-path bug." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "SGD optimizer args vs Qwen3.5 tie_word_embeddings warning are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python-version compatibility break is not the same as a PEFT version requirement change." + }, { "left": "issue:46082", "right": "issue:46512", "accept": false, - "reason": "Different bugs: LlamaConfig head_dim validation vs stale config.dtype after from_config. Same config area, but not the same underlying issue." + "reason": "Config validation bug vs stale dtype metadata after from_config; different problem." }, { "left": "issue:43577", "right": "issue:45216", "accept": false, - "reason": "Both involve save_pretrained/dtype regressions, but one is BLIP2 submodules staying float32 while the other is Qwen3.5 saving wrong checkpoint keys. Different failure modes and code paths." + "reason": "BLIP2 dtype-loading issue vs Qwen3.5 save_pretrained key serialization bug." }, { "left": "issue:44351", "right": "issue:45278", "accept": false, - "reason": "A missing HybridCache import is a specific symbol break, while the other is broad import failures after upgrade. Not the same bug." + "reason": "HybridCache import failure is only one possible import error; not the same as the broad import-breakage report." }, { "left": "issue:42796", "right": "issue:43650", "accept": false, - "reason": "Unrelated: tokenizer BPE marker handling vs a vague 'ADD THE DATA' issue." + "reason": "Tokenizer bug vs non-bug placeholder issue; no relation." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "Video-classification pipeline processor lookup and remote_code breakage are different problems." + "reason": "Video-classification processor lookup bug is separate from remote_code model-loading regressions." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but for different model families and likely different code paths; not one concrete bug." + "reason": "Both tokenizer regressions, but different model families and likely different root causes." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Model loading failure for Qwen3VL MoE vs save_pretrained serialization of Qwen3.5 visual encoder keys. Different surfaces." + "reason": "Qwen3VL load failure vs Qwen3.5 save_pretrained serialization issue; different code paths." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Different config issues: missing default vision/audio subconfigs vs missing pad_token_id on Llama4Config." + "reason": "Different config bugs: multimodal default initialization vs missing pad_token_id attribute." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "Different logic: router load-balancing loss gating vs Gemma 4 mm_token_type_ids defaulting." + "reason": "MoE load-balancing loss bug is unrelated to Gemma 4 mm_token_type_ids defaults." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Different models and load failures; one is Qwen3VL MoE, the other GLM-4.7-Flash unit-test loading." + "reason": "Different model load failures with different stacks and symptoms." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Both touch weights/embeddings, but one is lm_head.weight serialization inconsistency and the other is resize_token_embeddings not updating Gemma4 embedding/output layers." + "reason": "Both touch weights/embeddings, but one is serialization on device and the other is resize_token_embeddings propagation." }, { "left": "issue:41950", "right": "issue:45823", "accept": false, - "reason": "Completely different bugs: processor lookup in a video pipeline vs Gemma4 device mismatch during forward." + "reason": "Pipeline processor-selection bug vs Gemma4 device-mapping mismatch; unrelated." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "Tokenizer special-token decode bug vs D-FINE auxiliary loss behavior. No overlap." + "reason": "Tokenizer special-token registration vs D-FINE auxiliary-loss omission are unrelated." }, { "left": "issue:44683", "right": "issue:45542", "accept": false, - "reason": "Compiled flex_attention on torch>=2.9 is unrelated to tensorboard without TensorFlow backend errors." + "reason": "Torch 2.9 flex_attention failure is unrelated to the TensorFlow/tensorboard backend error." }, { "left": "issue:43650", "right": "issue:46209", "accept": false, - "reason": "Unrelated issue text vs FP8 inference failure on B200." + "reason": "Placeholder issue vs FP8/B200 inference failure; no duplicate relationship." }, { "left": "issue:44462", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-related, but AutoTokenizer ignoring tokenizer.json and _patch_mistral_regex backend_tokenizer crashes are different bugs." + "reason": "Both tokenizer-loading issues, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error is unrelated to the init_empty_weights unexpected _is_hf_initialized argument regression." + "reason": "Notebook syntax error vs init_empty_weights TypeError; different layers entirely." }, { "left": "issue:44361", "right": "issue:45593", "accept": false, - "reason": "MLukeTokenizer AttributeError and D-FINE auxiliary-loss bug are unrelated." + "reason": "MLukeTokenizer task error is unrelated to D-FINE loss handling." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "Different model-specific failures: Ministral-3 FlashAttention vs Kosmos2.5 OCR index error." + "reason": "Different model-specific runtime bugs: FlashAttention on Ministral vs OCR index error on Kosmos2.5." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "Auto-docstring future-annotations crash is unrelated to ReasoningEffort import failure in processor loading." - }, - { - "left": "issue:42994", - "right": "issue:43317", - "accept": false, - "reason": "Both involve quantization/offload, but one is saving quantized models and the other is loading dequantized models with device_map=auto. Not the same underlying bug." + "reason": "auto_docstring annotations crash is unrelated to Gemma 4 processor-loading import failure." }, { "left": "issue:33290", "right": "issue:45656", "accept": false, - "reason": "Deepspeed Adafactor OOM and optimizer.step being called twice are different optimizer integration bugs." - }, - { - "left": "issue:45081", - "right": "issue:45907", - "accept": false, - "reason": "Tokenizer regex patch crash vs list_repo_files not forwarding revision. Different components." - }, - { - "left": "issue:45600", - "right": "issue:45712", - "accept": false, - "reason": "Both are repo-maintenance/CI-related, but one is auto_mappings referencing removed configs and the other is leftover dummy classes." - }, - { - "left": "issue:44623", - "right": "issue:44749", - "accept": false, - "reason": "processor.save_pretrained missing files is unrelated to the reported tokenizer slowdown after upgrading to 5.3.0." - }, - { - "left": "issue:44261", - "right": "issue:44263", - "accept": false, - "reason": "Different issues: MLA rms_norm_eps precision handling vs torch.split return values in GlmMoeDsaIndexer." + "reason": "Different DeepSpeed problems: Adafactor OOM vs optimizer.step being called twice." } ] }, @@ -12247,15 +12239,14 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:41628", "issue:42175", - "issue:42491", - "issue:43299", + "issue:42994", + "issue:43317", "issue:43441", "issue:43525", - "issue:43526", "issue:43643", "issue:43673", "issue:43723", @@ -12268,8 +12259,9 @@ "issue:44016", "issue:44162", "issue:44188", - "issue:44242", "issue:44246", + "issue:44261", + "issue:44263", "issue:44297", "issue:44336", "issue:44373", @@ -12278,34 +12270,39 @@ "issue:44623", "issue:44671", "issue:44683", + "issue:44749", "issue:44779", - "issue:45003", "issue:45072", - "issue:45103", + "issue:45081", "issue:45362", "issue:45397", "issue:45542", "issue:45561", "issue:45563", - "issue:45593", + "issue:45600", + "issue:45712", "issue:45800", "issue:45823", + "issue:45907", "issue:46291", "issue:46399", - "issue:46614", "issue:46655", - "issue:46665", "issue:46726", "issue:46829" ], "soft_pairs": [ + "issue:42994|issue:43317", + "issue:45081|issue:45907", + "issue:45600|issue:45712", + "issue:44623|issue:44749", + "issue:44261|issue:44263", "issue:42175|issue:43976", "issue:44623|issue:45397", "issue:44671|issue:44779", "issue:43901|issue:44246", "issue:43723|issue:44162", - "issue:43441|issue:44683", "issue:43673|issue:43761", + "issue:43441|issue:44683", "issue:45542|issue:45800", "issue:43824|issue:43976", "issue:43525|issue:43881", @@ -12314,168 +12311,193 @@ "issue:44188|issue:44336", "issue:44589|issue:45362", "issue:43525|issue:43643", - "issue:46655|issue:46726", "issue:44336|issue:44373", + "issue:46655|issue:46726", "issue:41628|issue:43976", "issue:43784|issue:45542", "issue:44246|issue:44496", "issue:44297|issue:45397", - "issue:45072|issue:46829", "issue:45561|issue:45563", - "issue:45103|issue:46399", - "issue:46614|issue:46665", - "issue:44242|issue:45593", - "issue:43526|issue:45003", - "issue:42491|issue:43299" + "issue:45072|issue:46829" ], - "prepared_review_unit_hash": "f8090855bbab8e0c0f1d3566d5aed04c3892e46af6b5d326eb918036eedbc6fe", + "prepared_review_unit_hash": "57057f61a2e5a4f54aca6f87bca76ee33dd40661b6f24b9abfdeeff209343871", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36031, - "estimated_input_tokens": 9008, - "estimated_eval_tokens": 18272 + "serialized_chars": 36097, + "estimated_input_tokens": 9025, + "estimated_eval_tokens": 18306 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a mix of unrelated Transformers issues rather than a true duplicate set. None of the soft pairs appear to describe the same underlying bug or change, so all are rejected.", - "confidence": 0.96, - "canonical_issue_reason": "issue:43824 is the strongest standalone report in the set: a concrete, user-facing import failure with detailed reproduction and inbound references.", + "summary": "This cluster is mostly a loose collection of unrelated Transformers v5 regressions, packaging/import issues, and model-specific bugs. I do not see any pair that is clearly the same underlying defect, so all soft edges should be rejected.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43723 is the broadest user-facing regression in the set (tokenizer loading in v5) and is the closest thing to a representative umbrella issue, but it is not a true duplicate hub for the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43824 is the most actionable and highest-signal issue here; it has clear impact and more evidence than the other unrelated reports.", + "best_issue_reason": "43723 has the clearest general bug report and the best chance of representing the cluster\u2019s overall v5 loading-regression theme, even though the rest of the items are not duplicates of it.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both involve quantization/offload, but one is saving a quantized model and the other is loading a dequantized model with device_map=auto; different code paths and failure modes." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regex patching crashes versus list_repo_files revision forwarding; these are unrelated utilities and not the same bug." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both touch generated/mapping files, but one is stale auto_mappings config references and the other is leftover dummy classes leaking into exports; distinct cleanup bugs." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "processor.save_pretrained missing files is unrelated to the reported slowdown in data filtering after upgrade." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "These affect different model internals: MLA q_a_layernorm config precision versus GlmMoeDsaIndexer torch.split return handling." + }, { "left": "issue:42175", "right": "issue:43976", "accept": false, - "reason": "Packaging/backend selection vs Python-version compatibility; different failure modes and fixes." + "reason": "One is backend extras installation for TensorFlow, the other is Python version compatibility for Transformers 5.1.0; not the same defect." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Missing files on processor save is unrelated to a zero3/from_pretrained loading crash." + "reason": "Saving processor artifacts missing files and Gemma4 ZeRO3 from_pretrained are separate load/save paths with no concrete shared bug." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "Different components: masked-LM prediction bug vs tokenizer regression." + "reason": "CamemBERT MLM prediction regression and Deepseek tokenizer regression are different model-specific failures." }, { "left": "issue:43901", "right": "issue:44246", "accept": false, - "reason": "Docs mismatch for a pipeline argument vs slow import performance; not the same bug." + "reason": "Docs mismatch for return_all_scores versus slow import time are unrelated issues." }, { "left": "issue:43723", "right": "issue:44162", "accept": false, - "reason": "AutoTokenizer loading issue and ESM2 model breakage are separate model-specific regressions." + "reason": "Tokenizer loading regression in v5 and ESM2 being broken are different model/loading problems." }, { - "left": "issue:43441", - "right": "issue:44683", + "left": "issue:43673", + "right": "issue:43761", "accept": false, - "reason": "Both mention attention kernels, but one is FlashAttention on Ministral and the other is compiled flex_attention on torch>=2.9; different code paths." + "reason": "Generation cache missing during chunked prefill and CLIPVisionModel hidden_states behavior are different code paths and outputs." }, { - "left": "issue:43673", - "right": "issue:43761", + "left": "issue:43441", + "right": "issue:44683", "accept": false, - "reason": "Generation cache issue during chunked prefill is unrelated to CLIPVisionModel hidden_states output." + "reason": "Both are attention-kernel related, but one is Ministral flash-attention failure and the other is compiled flex_attention failure on newer torch; not the same bug." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "TensorBoard/TF backend installation bug is unrelated to a torch/transformers version incompatibility." + "reason": "TensorFlow backend installation gap and torch/Transformers version incompatibility are different packaging/version problems." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "Missing Qwen2.5-VL class import is not the same as Python 3.9/3.10 support regression." + "reason": "Missing Qwen2_5_VLForConditionalGeneration export is unrelated to Python 3.9+/3.10+ compatibility." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Different model loading failures with different root causes; no shared concrete code path." + "reason": "Llama4Config missing pad_token_id and glm-4v-9b loading failure are model-specific and do not share a concrete code path." }, { "left": "issue:44016", "right": "issue:45823", "accept": false, - "reason": "Notebook syntax error and Gemma4 device-mapping mismatch are unrelated." + "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated issues." }, { "left": "issue:46291", "right": "issue:46399", "accept": false, - "reason": "Kernel package metadata/version requirement issue vs a missing-decorator registration crash; different fixes." + "reason": "Kernels v0.15 revision/version requirement when importing pipelines is different from a Qwen3.5 kernel registration crash." }, { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "Attention kernel divergence on torch.compile is unrelated to ANSI codes in loading_report." + "reason": "torch.compile attention-kernel divergence and ANSI codes in loading_report are entirely different subsystems." }, { "left": "issue:44589", "right": "issue:45362", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to Qwen3.5 chat crashing." + "reason": "Float8 storage lookup failure and Qwen3.5 chat crash are different failures with no shared root cause shown." }, { "left": "issue:43525", "right": "issue:43643", "accept": false, - "reason": "A missing model config field and a trust_remote_code config hydration bug are related only broadly, not the same concrete defect." + "reason": "A missing pad_token_id field on one config is not the same as trust_remote_code returning an object with missing fields." }, { - "left": "issue:46655", - "right": "issue:46726", + "left": "issue:44336", + "right": "issue:44373", "accept": false, - "reason": "Video padding math bug vs missing vision-backend error; different underlying problems." + "reason": "Terminal ANSI output bug and wrong docstring for position_ids are unrelated." }, { - "left": "issue:44336", - "right": "issue:44373", + "left": "issue:46655", + "right": "issue:46726", "accept": false, - "reason": "Terminal ANSI output bug and a wrong docstring are not duplicates." + "reason": "Incorrect video padding calculation and missing-vision-backend image processor error are different processor bugs." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "AutoImageProcessor import failure and Python version support issue are unrelated." + "reason": "Cannot import AutoImageProcessor and Transformers Python-version support are unrelated compatibility issues." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "Import-time NameError in sentence-transformers is not the same as missing TensorFlow backend installation." + "reason": "Import-time NameError from sentence-transformers integration and missing TensorFlow backend packaging are different dependency problems." }, { "left": "issue:44246", "right": "issue:44496", "accept": false, - "reason": "Import slowness and unrecognized model config errors have different causes and remedies." + "reason": "Slow import timing and unrecognized model/config loading are not the same underlying defect." }, { "left": "issue:44297", @@ -12483,47 +12505,17 @@ "accept": false, "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma4 zero3 loading." }, - { - "left": "issue:45072", - "right": "issue:46829", - "accept": false, - "reason": "Dtype mismatch in inference is unrelated to hidden_states being returned in the wrong output field." - }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "xdist race in testing utils is unrelated to a stale generation warning." - }, - { - "left": "issue:45103", - "right": "issue:46399", - "accept": false, - "reason": "Auto-docstring annotation crash and kernel registration error are unrelated." - }, - { - "left": "issue:46614", - "right": "issue:46665", - "accept": false, - "reason": "Stale phonemizer language state is a tokenizer/backend caching bug, not a generic from_pretrained inconsistency." - }, - { - "left": "issue:44242", - "right": "issue:45593", - "accept": false, - "reason": "Both concern missing auxiliary losses, but they occur in different models and different conditional branches; not the same fix." - }, - { - "left": "issue:43526", - "right": "issue:45003", - "accept": false, - "reason": "BeitImageProcessorFast label reduction bug is unrelated to unsafe sys.modules access." + "reason": "pytest-xdist capture-file race and stale generation warning are separate issues in different parts of the test/runtime stack." }, { - "left": "issue:42491", - "right": "issue:43299", + "left": "issue:45072", + "right": "issue:46829", "accept": false, - "reason": "Both involve Qwen3 MoE loading, but one is LoRA compatibility across branches and the other is a Qwen3VL MoE load regression; too different to merge." + "reason": "dtype mismatches in bfloat16 inference and incorrect hidden_states assignment are different model output bugs." } ] }, @@ -12545,19 +12537,19 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:36246", "issue:41950", + "issue:42491", "issue:42757", "issue:42886", "issue:43257", - "issue:43262", - "issue:43475", + "issue:43299", + "issue:43526", "issue:43531", "issue:43540", "issue:43550", - "issue:43596", "issue:43673", "issue:43901", "issue:44162", @@ -12573,11 +12565,10 @@ "issue:44496", "issue:44561", "issue:44734", - "issue:44749", "issue:44779", - "issue:44857", "issue:44871", "issue:45003", + "issue:45103", "issue:45137", "issue:45200", "issue:45292", @@ -12585,257 +12576,257 @@ "issue:45375", "issue:45397", "issue:45561", + "issue:45593", "issue:45759", - "issue:45859", "issue:46132", - "issue:46369", "issue:46399", "issue:46424", "issue:46506", + "issue:46614", "issue:46655", + "issue:46665", "issue:46752", "issue:46821" ], "soft_pairs": [ + "issue:45103|issue:46399", + "issue:46614|issue:46665", + "issue:44242|issue:45593", + "issue:43526|issue:45003", + "issue:42491|issue:43299", "issue:43540|issue:46752", "issue:43531|issue:45362", - "issue:44246|issue:44871", "issue:44295|issue:45397", + "issue:44246|issue:44871", "issue:44336|issue:44485", "issue:45003|issue:46821", "issue:44451|issue:46752", "issue:44336|issue:46424", "issue:42886|issue:46132", "issue:43550|issue:46506", - "issue:44779|issue:45292", "issue:41950|issue:44220", + "issue:44779|issue:45292", "issue:36246|issue:42757", "issue:36246|issue:44351", "issue:43901|issue:45003", - "issue:43257|issue:45137", "issue:44734|issue:46424", + "issue:43257|issue:45137", "issue:45003|issue:45561", - "issue:44496|issue:45003", "issue:43540|issue:46655", + "issue:44496|issue:45003", "issue:43673|issue:44561", "issue:44162|issue:45200", "issue:44479|issue:46655", - "issue:45859|issue:46399", - "issue:44242|issue:44857", - "issue:45375|issue:45759", - "issue:43262|issue:44246", - "issue:44295|issue:44749", - "issue:43475|issue:46369", - "issue:36246|issue:43596" + "issue:45375|issue:45759" ], - "prepared_review_unit_hash": "86748a0b9f41f92d66e70f292d455a5c722edb17c99bd7ab60830191e7168785", + "prepared_review_unit_hash": "7d1cab9cc16afe5004f536cb904784ac16c76b85e94cdefe7b6ea95249aeb787", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36588, - "estimated_input_tokens": 9147, - "estimated_eval_tokens": 18550 + "soft_pair_count": 29, + "serialized_chars": 36318, + "estimated_input_tokens": 9080, + "estimated_eval_tokens": 18416 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of unrelated Transformers bug reports spanning different models, cache/offline behavior, pipelines, and runtime environments. None of the soft-linked pairs look like the same underlying defect, so I would not merge any of them as duplicates. The strongest representative issue is the broad Qwen video regression report.", - "confidence": 0.91, - "canonical_issue_reason": "issue:44479 is the broadest, most central report here: it describes a concrete regression in v5.3.0 affecting multiple Qwen video/multimodal variants, with a shared code-path problem rather than a narrow model-specific edge case.", + "summary": "This cluster is mostly heterogeneous issues with only loose topical similarity; I would not merge any soft edge pairs as duplicates. The closest theme is Qwen/multimodal/video-regression reports, but the concrete failures differ.", + "confidence": 0.74, + "canonical_issue_reason": "Issue 44479 is the broadest concrete regression report in the set, covering multiple Qwen multimodal/video-input models under a single v5.3.0 regression. It is the best anchor only in the sense of breadth, not because the whole cluster is one duplicate thread.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the best standalone issue because it is specific, high-signal, and covers a wider affected surface than the other candidates.", + "best_issue_reason": "Issue 44479 is the most representative issue to anchor the cluster because it spans several related models and a clear version regression, making it the strongest single candidate among otherwise unrelated reports.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "Different failure modes: auto-docstring parsing with future annotations vs a Qwen3.5 kernel-registration crash." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Unrelated subsystems: stale phonemizer language state vs from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Both concern missing losses, but they affect different models and code paths (router loss vs D-FINE auxiliary losses)." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "Different areas entirely: Beit image processor label reduction vs unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen MoE loading, but the concrete breakage and model families differ." + }, { "left": "issue:43540", "right": "issue:46752", "accept": false, - "reason": "Different failures: Qwen3OmniMoe video input processing vs empty-list handling in apply_chat_template." + "reason": "Video-input ValueError in Qwen3OmniMoe is unrelated to empty-list chat template handling." }, { "left": "issue:43531", "right": "issue:45362", "accept": false, - "reason": "Different subsystems and symptoms: sliding_window behavior in Qwen3-MoE vs transformers chat crash for Qwen3.5-35B." + "reason": "Sliding-window behavior in Qwen3-MoE is not the same bug as a chat crash in Qwen3.5-35B." }, { - "left": "issue:44246", - "right": "issue:44871", + "left": "issue:44295", + "right": "issue:45397", "accept": false, - "reason": "MoE router loss computation is unrelated to Gemma eos_token_id config mismatch." + "reason": "Different bugs: position_ids buffer access vs gemma-4 ZeRO3 from_pretrained." }, { - "left": "issue:44295", - "right": "issue:45397", + "left": "issue:44246", + "right": "issue:44871", "accept": false, - "reason": "Buffer/position_ids access bug is not the same as Gemma-4 ZeRO3 from_pretrained failure." + "reason": "Import slowness is unrelated to inconsistent eos_token_id configuration." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI color output in loading_report is unrelated to GLM-5 RoPE implementation discussion." + "reason": "ANSI codes in loading_report is unrelated to GLM-5 RoPE implementation." }, { "left": "issue:45003", "right": "issue:46821", "accept": false, - "reason": "sys.modules access safety bug is unrelated to TimesFM window_size AttributeError." + "reason": "Unsafe sys.modules access is unrelated to a TimesFM window_size AttributeError." }, { "left": "issue:44451", "right": "issue:46752", "accept": false, - "reason": "Model loading regression for vesteinn/ScandiBERT is unrelated to empty chat-template inputs." + "reason": "Model loading failure for ScandiBERT is unrelated to apply_chat_template on an empty list." }, { "left": "issue:44336", "right": "issue:46424", "accept": false, - "reason": "Terminal coloring in loading_report and cache-length handling are different concerns." + "reason": "Terminal color output bug is unrelated to cache max length behavior." }, { "left": "issue:42886", "right": "issue:46132", "accept": false, - "reason": "Offline cache/tokenizer loading bug is not the same as AttentionInterface.register behavior changes." + "reason": "Offline cache loading is unrelated to AttentionInterface.register behavior changes." }, { "left": "issue:43550", "right": "issue:46506", "accept": false, - "reason": "torch.compile/SDPA crash in Bamba is unrelated to PPDocLayoutV3 CPU/CUDA result differences." + "reason": "Torch.compile/SDPA failure in Bamba is unrelated to CPU vs CUDA divergence in PPDocLayoutV3." }, { - "left": "issue:44779", - "right": "issue:45292", + "left": "issue:41950", + "right": "issue:44220", "accept": false, - "reason": "DeepSeek tokenizer regression is unrelated to resize_token_embeddings not updating output embeddings." + "reason": "Video-classification image-processor lookup and fbank feature extraction are different pipelines." }, { - "left": "issue:41950", - "right": "issue:44220", + "left": "issue:44779", + "right": "issue:45292", "accept": false, - "reason": "Video-classification pipeline choosing image processors is unrelated to _torch_extract_fbank_features()." + "reason": "Tokenizer correctness regression is unrelated to resize_token_embeddings not updating output embeddings." }, { "left": "issue:36246", "right": "issue:42757", "accept": false, - "reason": "Missing Qwen2_5_VLImageProcessor import and missing is_offline_mode are different import errors." + "reason": "Both are import errors, but for different symbols and different libraries/modules." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Different missing symbols: Qwen2_5_VLImageProcessor vs HybridCache." + "reason": "Missing Qwen2_5_VLImageProcessor export is not the same as missing HybridCache export." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "TextClassificationPipeline docs mismatch is unrelated to sys.modules access in modeling_utils." + "reason": "Docs mentioning return_all_scores is unrelated to sys.modules access." }, { - "left": "issue:43257", - "right": "issue:45137", + "left": "issue:44734", + "right": "issue:46424", "accept": false, - "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed is not the same as ZeRO3 deque IndexError." + "reason": "KV-cache continuation crash and cache max_length semantics are distinct cache bugs." }, { - "left": "issue:44734", - "right": "issue:46424", + "left": "issue:43257", + "right": "issue:45137", "accept": false, - "reason": "KV cache continuation crash in transformers serve is unrelated to cache max length configuration semantics." + "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is unrelated to ZeRO3 deque underflow." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "sys.modules access safety issue is unrelated to pytest-xdist captured_info.txt race conditions." + "reason": "Modeling_utils import safety and pytest-xdist capture-file races are unrelated." }, { - "left": "issue:44496", - "right": "issue:45003", + "left": "issue:43540", + "right": "issue:46655", "accept": false, - "reason": "Unrecognized OLMo-Hybrid model config is unrelated to unsafe sys.modules lookup." + "reason": "Both are video-related, but Qwen3OmniMoe input validation and SmolVLM padding calculation are different code paths." }, { - "left": "issue:43540", - "right": "issue:46655", + "left": "issue:44496", + "right": "issue:45003", "accept": false, - "reason": "Qwen3OmniMoe video processing ValueError and SmolVLM padding calculation bug are separate video bugs." + "reason": "Unrecognized model/config loading is unrelated to sys.modules access." }, { "left": "issue:43673", "right": "issue:44561", "accept": false, - "reason": "GenerationMixin cache missing in chunked_prefill is unrelated to trust_remote_code breakage from removed is_torch_fx_available." + "reason": "Chunked-prefill cache handling and removal of is_torch_fx_available are separate regressions." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "ESM2 model breakage is unrelated to Gemma 4 mm_token_type_ids defaulting." + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are unrelated." }, { "left": "issue:44479", "right": "issue:46655", "accept": false, - "reason": "Both are video-related, but they affect different models and different code paths: Qwen video regression vs SmolVLM padding." - }, - { - "left": "issue:45859", - "right": "issue:46399", - "accept": false, - "reason": "CPU offload incompatibility in Qwen3_5MoeTextRotaryEmbedding is not the same as missing decorator for apply_rotary_pos_emb registration." + "reason": "Both involve video models, but the reported regressions are model-specific and not the same underlying bug." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:44485", + "right": "issue:46655", "accept": false, - "reason": "Router logits/load balancing loss is unrelated to LwDetrImageLoss AMP/CUDA crashes." + "reason": "GLM-5 RoPE implementation is unrelated to SmolVLM video padding." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Missing deepstack_visual_indexes in config is unrelated to unwrapping text_config for composite multimodal configs." - }, - { - "left": "issue:43262", - "right": "issue:44246", - "accept": false, - "reason": "Audio chat-template sampling-rate default bug is unrelated to load balancing loss/router logits behavior." - }, - { - "left": "issue:44295", - "right": "issue:44749", - "accept": false, - "reason": "Position_ids buffer error and post-upgrade filtering slowdown are different bugs with different symptoms." - }, - { - "left": "issue:43475", - "right": "issue:46369", - "accept": false, - "reason": "SAM3 video encoder missing attribute and memory overflow in sam3-litetext are separate issues." - }, - { - "left": "issue:36246", - "right": "issue:43596", - "accept": false, - "reason": "Missing Qwen2_5_VLImageProcessor import is unrelated to ZeRO3/BertModel index errors." + "reason": "Both are Qwen multimodal config issues, but one drops a config field while the other mishandles text_config unwrapping." } ] }, @@ -12857,31 +12848,34 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:29870", "issue:30333", + "issue:36246", "issue:42886", - "issue:43065", + "issue:43262", "issue:43295", "issue:43352", - "issue:43408", "issue:43475", "issue:43540", + "issue:43596", "issue:43653", - "issue:43723", - "issue:43931", "issue:44038", "issue:44060", "issue:44077", "issue:44162", + "issue:44242", + "issue:44246", + "issue:44295", "issue:44368", "issue:44466", "issue:44479", "issue:44561", "issue:44623", + "issue:44749", + "issue:44857", "issue:44877", - "issue:44898", "issue:44977", "issue:44987", "issue:44995", @@ -12892,23 +12886,26 @@ "issue:45292", "issue:45335", "issue:45375", - "issue:45397", "issue:45412", "issue:45459", - "issue:45563", "issue:45584", - "issue:45588", "issue:45600", - "issue:45656", + "issue:45859", "issue:46123", "issue:46170", + "issue:46369", "issue:46399", "issue:46506", - "issue:46710", "issue:46726", - "issue:46829" + "issue:46832" ], "soft_pairs": [ + "issue:45859|issue:46399", + "issue:44242|issue:44857", + "issue:44295|issue:44749", + "issue:43262|issue:44246", + "issue:43475|issue:46369", + "issue:36246|issue:43596", "issue:45137|issue:45600", "issue:44877|issue:45375", "issue:45292|issue:45584", @@ -12919,214 +12916,208 @@ "issue:44038|issue:46506", "issue:29870|issue:30333", "issue:44987|issue:45459", - "issue:44479|issue:46170", "issue:43352|issue:45250", + "issue:44479|issue:46170", "issue:45412|issue:46123", "issue:43653|issue:44162", "issue:44060|issue:44077", "issue:44368|issue:44977", - "issue:44877|issue:46170", + "issue:46170|issue:46832", "issue:44995|issue:45200", "issue:42886|issue:45229", - "issue:43475|issue:43540", - "issue:45397|issue:45656", - "issue:43065|issue:43408", - "issue:44368|issue:45563", - "issue:43931|issue:44368", - "issue:44995|issue:45588", - "issue:43723|issue:46710", - "issue:44898|issue:46829" + "issue:44877|issue:46170", + "issue:43475|issue:43540" ], - "prepared_review_unit_hash": "2bd34c9e0d127f5f084d94fe72bd898fc6dd5abeae1083e31223ff9ffa13f434", + "prepared_review_unit_hash": "bd61730bb4129f94c104e8e891a67bededb78bc99ba51a47d64f575ad566e882", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35843, - "estimated_input_tokens": 8961, - "estimated_eval_tokens": 18178 + "serialized_chars": 36001, + "estimated_input_tokens": 9001, + "estimated_eval_tokens": 18258 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is loosely related by recurring Transformers v5/v6 regression reports, but the paired items are mostly distinct bugs across different models and subsystems. None of the soft edges look like true duplicates.", - "confidence": 0.34, - "canonical_issue_reason": "issue:44479 is the broadest concrete regression report in the set, spanning multiple Qwen VL variants and a shared v5.3.0 video-input failure mode.", + "summary": "This cluster is a grab-bag of unrelated issues across tokenizers, multimodal processors, model configs, generation, memory, and infrastructure. The soft pair similarities are topical rather than duplicate-level, so none should be merged.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe here: the items are heterogeneous and the candidate pairs do not share the same concrete bug path.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the strongest representative of the cluster\u2019s main theme: a real, user-facing Transformers regression with clear scope and affected models.", + "best_issue_reason": "No issue is a strong global representative because none of the candidates are true duplicates; they span different models and failure modes.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:45137", - "right": "issue:45600", + "left": "issue:45859", + "right": "issue:46399", "accept": false, - "reason": "DeepSpeed ZeRO3 deque underflow is unrelated to a removed Sam3LiteText config reference." + "reason": "Both involve Qwen3.5/MoE rotary-attention code, but one is CPU-offload incompatibility and the other is a kernel registration crash; different failure paths." }, { - "left": "issue:44877", - "right": "issue:45375", + "left": "issue:44242", + "right": "issue:44857", "accept": false, - "reason": "Both involve loading/config validation, but they target different models and different missing-config problems." + "reason": "Mixture-of-experts load-balancing loss and RT-DETR AMP CUDA crashes are unrelated bugs." }, { - "left": "issue:45292", - "right": "issue:45584", + "left": "issue:44295", + "right": "issue:44749", "accept": false, - "reason": "resize_token_embeddings behavior is unrelated to Whisper empty-transcription generation after align_special_tokens." + "reason": "One is a position_ids buffer access error; the other is a slowdown regression after upgrade." }, { - "left": "issue:43295", - "right": "issue:45292", + "left": "issue:43262", + "right": "issue:44246", "accept": false, - "reason": "Processor/tokenizer regression is a different code path from output-embedding resize propagation." + "reason": "Audio chat-template sampling-rate defaults and slow imports are unrelated issues." }, { - "left": "issue:44561", - "right": "issue:46399", + "left": "issue:43475", + "right": "issue:46369", "accept": false, - "reason": "trust_remote_code breakage from removing is_torch_fx_available is unrelated to Qwen3.5 kernel registration." + "reason": "Both are SAM3-related, but one is a missing attribute in the vision encoder output and the other is a memory overflow in sam3-litetext." }, { - "left": "issue:44623", - "right": "issue:46726", + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "Qwen2.5-VL image processor import failure and DeepSpeed ZeRO3/BertModel index error do not share a code path." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "Empty-deque ZeRO3 failure and stale auto_mappings references are different kinds of breakage." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both are config-related, but they concern different models and different bugs: strict loading vs a missing field being dropped." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Resize-token-embeddings behavior and Whisper empty-transcription generation are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "A processor.tokenizer regression with image inputs is not the same bug as resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:44561", + "right": "issue:46399", + "accept": false, + "reason": "Removal of is_torch_fx_available affecting trust_remote_code is unrelated to Qwen3.5 rotary kernel registration." + }, + { + "left": "issue:44623", + "right": "issue:46726", "accept": false, - "reason": "Missing processor.save_pretrained files is not the same as AutoProcessor failing when a vision backend is absent." + "reason": "Saving processor files and missing vision backends causing AutoProcessor errors are different loader/saver problems." }, { "left": "issue:44466", "right": "issue:45335", "accept": false, - "reason": "Both mention tied/resized embeddings, but one is device-dependent serialization and the other is a model-specific resize propagation bug." + "reason": "Both mention tied/resize embeddings, but one is serialization inconsistency and the other is a model-specific resize propagation bug." }, { "left": "issue:44038", "right": "issue:46506", "accept": false, - "reason": "Different models, different symptoms, and no shared concrete bug path." + "reason": "Qwen3-VL-Moe and PPDocLayoutV3 issues are entirely separate model families and symptoms." }, { "left": "issue:29870", "right": "issue:30333", "accept": false, - "reason": "BART generate() decoder_attention_mask crash is unrelated to MLFlow reporting after failed training jobs." + "reason": "Bart generate UnboundLocalError and MLFlow job-status reporting are unrelated." }, { "left": "issue:44987", "right": "issue:45459", "accept": false, - "reason": "Both concern tokenizer loading, but one is a version-specific model load failure and the other is exception masking when protobuf is missing." + "reason": "Tokenizer load failure on transformers>=5.1.0 and protobuf error masking are both loading-related but not the same defect." }, { - "left": "issue:44479", - "right": "issue:46170", + "left": "issue:43352", + "right": "issue:45250", "accept": false, - "reason": "Video-input regression in Qwen models is unrelated to a Windows/ROCm torch.distributed import problem." + "reason": "A model not supporting Flash Attention 2.0 and a generic Flash Attention 2.0 issue are not the same concrete bug." }, { - "left": "issue:43352", - "right": "issue:45250", + "left": "issue:44479", + "right": "issue:46170", "accept": false, - "reason": "A model not supporting Flash Attention 2.0 is not the same as a generic flash-attention topic issue." + "reason": "A video regression in Qwen models and a Windows ROCm import/dependency issue are unrelated." }, { "left": "issue:45412", "right": "issue:46123", "accept": false, - "reason": "RT-DETR memory release and MaskGenerationPipeline partial-batch handling are distinct bugs." + "reason": "Memory not released on RT-DETR deletion and MaskGenerationPipeline partial-batch handling are different subsystems." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token registration is unrelated to the broad ESM2 breakage report." + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same underlying problem." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "Tied-weights warning logic and patchtsmixer post_init allowance are different validation/config issues." + "reason": "A tied-weights warning bug and patchtsmixer post_init policy are unrelated." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "Same family name, but one is a tied-embeddings warning while the other is a flash-attention generation failure." + "reason": "Both mention Qwen3.5 and flash attention/tied weights, but one is a warning about config tying and the other is a generation failure." }, { - "left": "issue:44877", - "right": "issue:46170", + "left": "issue:46170", + "right": "issue:46832", "accept": false, - "reason": "Strict config loading and Windows ROCm import/dependency failure are unrelated." + "reason": "Windows ROCm torch.distributed dependency issues and RTDetrModel feature-level shape errors do not overlap." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "GlmMoeDsa stale cache crash is unrelated to Gemma 4 mm_token_type_ids defaults." + "reason": "A stale cache crash in GlmMoeDsa and Gemma4 mm_token_type_ids defaults are unrelated." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Offline tokenizer cache loading and Gemma 4 multi-GPU OOM are clearly different problems." - }, - { - "left": "issue:43475", - "right": "issue:43540", - "accept": false, - "reason": "Both are multimodal/video issues, but they affect different models and different internal objects." - }, - { - "left": "issue:45397", - "right": "issue:45656", - "accept": false, - "reason": "Gemma-4 Zero3 loading and double optimizer stepping under deepspeed are not the same bug." - }, - { - "left": "issue:43065", - "right": "issue:43408", - "accept": false, - "reason": "SAM3 dummy Conv2d implementation and sam3_video vs sam3_tracker mismatch are related only by model family, not by defect." - }, - { - "left": "issue:44368", - "right": "issue:45563", - "accept": false, - "reason": "These are different warnings with different triggers and code paths." - }, - { - "left": "issue:43931", - "right": "issue:44368", - "accept": false, - "reason": "Qwen3-VL weight-shape mismatch is unrelated to tie_word_embeddings warnings." - }, - { - "left": "issue:44995", - "right": "issue:45588", - "accept": false, - "reason": "Second-forward stale cache crash is unrelated to a flash_attention AttributeError on s_aux=None." + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are different problems." }, { - "left": "issue:43723", - "right": "issue:46710", + "left": "issue:44877", + "right": "issue:46170", "accept": false, - "reason": "Tokenizer loading regression and accuracy regression on DeepSeek are not the same concrete issue." + "reason": "Strict config loading for granite_speech and Windows ROCm dependency/import errors are unrelated." }, { - "left": "issue:44898", - "right": "issue:46829", + "left": "issue:43475", + "right": "issue:43540", "accept": false, - "reason": "Perceiver interpolation failure and VideoPrism output field mix-up are different vision-classification bugs." + "reason": "Both are multimodal/video issues, but they affect different models and fail for different reasons." } ] }, @@ -13148,13 +13139,12 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:28282", - "issue:30990", "issue:42994", "issue:43065", "issue:43232", + "issue:43408", "issue:43441", "issue:43475", "issue:43482", @@ -13162,278 +13152,289 @@ "issue:43646", "issue:43673", "issue:43701", + "issue:43723", "issue:43761", "issue:43906", - "issue:43986", + "issue:43931", "issue:43994", "issue:44162", "issue:44188", "issue:44242", "issue:44297", "issue:44361", + "issue:44368", "issue:44483", - "issue:44610", "issue:44623", "issue:44749", - "issue:44869", "issue:44871", - "issue:44964", - "issue:45137", - "issue:45356", + "issue:44898", + "issue:44995", "issue:45375", + "issue:45397", "issue:45478", - "issue:45584", + "issue:45563", + "issue:45588", "issue:45656", "issue:45685", "issue:45987", "issue:46082", "issue:46123", "issue:46132", - "issue:46170", "issue:46202", "issue:46421", "issue:46431", + "issue:46710", "issue:46726", "issue:46752", "issue:46821", "issue:46829" ], "soft_pairs": [ + "issue:45397|issue:45656", + "issue:43065|issue:43408", + "issue:44368|issue:45563", + "issue:43931|issue:44368", + "issue:43723|issue:46710", + "issue:44995|issue:45588", "issue:44188|issue:46202", "issue:45987|issue:46082", - "issue:43906|issue:46202", + "issue:44898|issue:46829", "issue:44483|issue:46202", - "issue:44610|issue:46829", + "issue:43906|issue:46202", "issue:44188|issue:46421", "issue:45685|issue:46431", - "issue:45584|issue:46170", "issue:43065|issue:46123", "issue:44297|issue:44623", "issue:44361|issue:44749", - "issue:43761|issue:44242", "issue:43994|issue:46726", + "issue:43761|issue:44242", "issue:43646|issue:44162", "issue:44749|issue:45656", "issue:43701|issue:44749", "issue:44871|issue:46132", "issue:44623|issue:45656", - "issue:42994|issue:43482", "issue:43232|issue:43673", + "issue:42994|issue:43482", "issue:43441|issue:46752", "issue:45375|issue:45478", "issue:43475|issue:43643", - "issue:43475|issue:46821", - "issue:28282|issue:30990", - "issue:43986|issue:46726", - "issue:44869|issue:45356", - "issue:44964|issue:45137" + "issue:43475|issue:46821" ], - "prepared_review_unit_hash": "fdb447c38eb281fe7707ddc2d7b7867a9961731690db7483ac0193b380801976", + "prepared_review_unit_hash": "b2bb76cb17a4e238c139115fdff818abc37a8fa8c0a35055952c13711e46b5ad", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 28, - "serialized_chars": 35546, - "estimated_input_tokens": 8887, - "estimated_eval_tokens": 18030 + "soft_pair_count": 29, + "serialized_chars": 35860, + "estimated_input_tokens": 8965, + "estimated_eval_tokens": 18186 }, "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 across different models and subsystems. The only clear duplicate-like pair is the Qwen3.5 MoE config regression: one report identifies the missing `deepstack_visual_indexes` field and the other reports the resulting `from_pretrained` failure.", - "confidence": 0.78, - "canonical_issue_reason": "Issue 45478 is the most representative user-facing report of the Qwen3.5 MoE regression, while 45375 looks like the root-cause diagnosis of the same missing-config-field bug.", + "summary": "The cluster is very heterogeneous: mostly unrelated Transformers v5 regressions, model-specific loading bugs, and a few generation/cache issues. There is no strong duplicate hub, so all soft links should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "No single issue cleanly covers this cluster; if forced to pick one representative, issue 43646 is the broadest Transformers v5 breakage, but it still only matches a subset of the items.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45478 is the broadest and clearest issue for the underlying Qwen3.5 MoE `from_pretrained` regression and is the best single issue to keep as the canonical entry.", + "best_issue_reason": "Issue 43646 is the best umbrella candidate because it is broad and version-wide ('Transformers 5.0.0 breaks custom model initialization'), making it the closest representative of the cluster's general theme.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Different bugs: one is a Gemma-4 ZeRO/from_pretrained loading issue, the other is a deepspeed optimizer-step duplication problem." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both mention SAM3, but one is a dummy Conv2d implementation detail and the other is a model-class mismatch warning." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Both are warnings, but about unrelated behaviors: tie_word_embeddings vs a stale num_return_sequences warning." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Different subsystems and failures: Qwen3-VL weight-shape mismatch vs a Qwen3.5 LoRA warning." + }, + { + "left": "issue:43723", + "right": "issue:46710", + "accept": false, + "reason": "One is tokenizer loading in v5; the other is an accuracy regression in a specific model after upgrade." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Different runtime paths: stale indexer cache on second forward pass vs flash-attention crashing on s_aux=None." + }, { "left": "issue:44188", "right": "issue:46202", "accept": false, - "reason": "Different bugs: attention-kernel divergence under `torch.compile` vs Gemma roundtrip/tokenization behavior." + "reason": "Torch.compile attention-kernel branching is unrelated to Gemma token roundtrip failure on \u2581." }, { "left": "issue:45987", "right": "issue:46082", "accept": false, - "reason": "Unrelated failures: cache sequence length return type vs Llama head-dimension validation." + "reason": "Cache return-type bug vs LlamaConfig head_dim validation; same broad area but not the same defect." }, { - "left": "issue:43906", - "right": "issue:46202", + "left": "issue:44898", + "right": "issue:46829", "accept": false, - "reason": "One is an isolated reproduction of another issue; the other is a Gemma pipeline roundtrip bug." + "reason": "Different vision-model bugs: Perceiver resolution handling vs VideoPrism hidden_states/output mapping." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "Chat-completions request rejection is unrelated to Gemma pipeline roundtrip behavior." + "reason": "API request rejection in chat/completions is unrelated to tokenizer/pipeline roundtrip behavior." }, { - "left": "issue:44610", - "right": "issue:46829", + "left": "issue:43906", + "right": "issue:46202", "accept": false, - "reason": "Different model/processor bugs: input-size mismatch vs incorrect hidden_states output mapping." + "reason": "An isolated reproduction of another issue is not the same as Gemma pipeline token roundtrip failure." }, { "left": "issue:44188", "right": "issue:46421", "accept": false, - "reason": "Both mention cache/prefill, but one is a torch.compile kernel divergence and the other is recompilation from lazy cache init." + "reason": "Both touch generation/chunked prefill, but one is kernel branching divergence and the other is cache lazy-init recompilation." }, { "left": "issue:45685", "right": "issue:46431", "accept": false, - "reason": "Unrelated: MPS histogram support error vs missing chat template in a processor." - }, - { - "left": "issue:45584", - "right": "issue:46170", - "accept": false, - "reason": "Whisper empty-transcription crash is unrelated to Windows ROCm distributed dependency import/type errors." + "reason": "MPS histogram/int kernel error is unrelated to missing chat-template support in a processor." }, { "left": "issue:43065", "right": "issue:46123", "accept": false, - "reason": "Different features entirely: Sam3PixelDecoder dummy conv vs MaskGenerationPipeline final-batch handling." + "reason": "SAM3 pixel decoder implementation detail is unrelated to MaskGenerationPipeline's final-partial-batch logic." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Both involve save_pretrained, but one is tokenizer metadata and the other is missing processor files; not the same bug." + "reason": "Both are save_pretrained bugs, but tokenizer metadata mismatch and processor file omission are different code paths." }, { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Tokenizer AttributeError is unrelated to the training slowdown regression after upgrade." + "reason": "Tokenizer AttributeError and a multilingual performance regression are not the same underlying bug." }, { - "left": "issue:43761", - "right": "issue:44242", + "left": "issue:43994", + "right": "issue:46726", "accept": false, - "reason": "CLIP hidden_states regression and MoE load-balancing loss omission are different code paths and symptoms." + "reason": "One is wrong outputs from AutoModel/pipeline, the other is a missing-vision-backend processor error." }, { - "left": "issue:43994", - "right": "issue:46726", + "left": "issue:43761", + "right": "issue:44242", "accept": false, - "reason": "Both touch AutoProcessor/vision, but one is a bad SigLIP2 inference result and the other is missing-backend error handling." + "reason": "CLIPVision hidden_states regression is unrelated to MoE load-balancing loss gating." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Custom model initialization breakage is not the same as ESM2 being broken." + "reason": "Both are broad regressions, but one is custom model initialization under v5 and the other is an ESM2-specific breakage." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "Performance regression during filtering is unrelated to deepspeed optimizer stepping twice." + "reason": "Performance slowdown after upgrade is not the same as deepspeed stepping the optimizer twice." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "Checkpoint key mismatch is a distinct bug from the upgrade slowdown report." + "reason": "Checkpoint-key mismatch and a runtime throughput regression are different failure modes." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma EOS config mismatch is unrelated to AttentionInterface registration behavior changes." + "reason": "EOS token config inconsistency is unrelated to AttentionInterface.register changing function behavior." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "Missing processor save files and double optimizer stepping are unrelated." + "reason": "Processor save-pretrained file omission is unrelated to duplicate optimizer stepping." }, { - "left": "issue:42994", - "right": "issue:43482", + "left": "issue:43232", + "right": "issue:43673", "accept": false, - "reason": "Quantized save failure and GGUF load failure are opposite directions and different failure modes." + "reason": "Both involve generation/caching, but one is sync_gpus/update_model_kwargs ordering and the other is missing cache in chunked_prefill." }, { - "left": "issue:43232", - "right": "issue:43673", + "left": "issue:42994", + "right": "issue:43482", "accept": false, - "reason": "Both mention generation/cache, but the concrete bugs differ: sync_gpus update handling vs missing cache during chunked prefill." + "reason": "Quantized model saving and Qwen2.5-GGUF loading are opposite operations with different failure modes." }, { "left": "issue:43441", "right": "issue:46752", "accept": false, - "reason": "FlashAttention failure for Ministral-3 is unrelated to empty-list chat-template crashes." + "reason": "FlashAttention crash in Ministral-3 is unrelated to apply_chat_template failing on an empty list." }, { "left": "issue:45375", "right": "issue:45478", - "accept": true, - "reason": "Same underlying Qwen3.5 MoE bug: 45375 identifies the missing config field and 45478 reports the resulting `from_pretrained` regression." + "accept": false, + "reason": "Both involve Qwen3.5 variants, but one is a strict-field drop in config and the other is a from_pretrained loading error." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM3 output attribute missing and AutoConfig missing fields are different issues in different components." + "reason": "Missing fpn_position_embeddings in SAM3 video output is unrelated to trust_remote_code AutoConfig field loss." }, { "left": "issue:43475", "right": "issue:46821", "accept": false, - "reason": "SAM3 vision output bug is unrelated to TimesFM `window_size` AttributeError." - }, - { - "left": "issue:28282", - "right": "issue:30990", - "accept": false, - "reason": "PyTorch-not-found import error is not the same as Sentence Transformers getting stuck loading." - }, - { - "left": "issue:43986", - "right": "issue:46726", - "accept": false, - "reason": "Both concern AutoProcessor and missing vision support, but one is a video-model crash without torchvision and the other is a generic unrecognized-image-processor error; not enough evidence they are the same concrete bug." + "reason": "SAM3 video encoder attribute loss and TimesFM window_size AttributeError are different model-specific bugs." }, { - "left": "issue:44869", - "right": "issue:45356", - "accept": false, - "reason": "Whisper timestamp decode crash and Kimi tokenizer regression are distinct failures." - }, - { - "left": "issue:44964", - "right": "issue:45137", + "left": "issue:43475", + "right": "issue:46821", "accept": false, - "reason": "Phi-4 multimodal loading failure is unrelated to DeepSpeed ZeRO3 deque underflow." + "reason": "Same as above: no shared underlying code path or fix." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The review is grounded in the packet and stays mostly conservative. The only accepted soft edge, issue:45375 \u2194 issue:45478, is plausibly the same Qwen3.5 MoE config regression based on the titles and references. The rest of the pairwise rejections are appropriately cautious given the available evidence." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -13448,16 +13449,13 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:30333", + "issue:28282", "issue:30990", - "issue:33290", - "issue:35141", "issue:36010", "issue:39692", "issue:41628", - "issue:42175", "issue:42371", "issue:42673", "issue:42831", @@ -13471,51 +13469,59 @@ "issue:43784", "issue:43824", "issue:43873", - "issue:43976", + "issue:43986", "issue:44038", "issue:44188", "issue:44222", + "issue:44242", "issue:44351", - "issue:44496", "issue:44623", "issue:44792", + "issue:44869", "issue:44933", "issue:44945", + "issue:44964", "issue:45081", "issue:45092", "issue:45103", "issue:45137", "issue:45230", "issue:45310", + "issue:45356", "issue:45375", "issue:45397", "issue:45406", "issue:45561", - "issue:45600", "issue:45753", "issue:46202", "issue:46326", "issue:46399", "issue:46701", - "issue:46726" + "issue:46726", + "issue:46832" ], "soft_pairs": [ - "issue:41628|issue:44933", + "issue:28282|issue:30990", + "issue:43986|issue:46726", "issue:45310|issue:45375", + "issue:41628|issue:44933", + "issue:44964|issue:45137", + "issue:44869|issue:45356", "issue:45230|issue:45397", + "issue:44242|issue:46832", "issue:39692|issue:46726", "issue:44038|issue:44623", "issue:42994|issue:43525", "issue:43723|issue:46326", "issue:42371|issue:45092", - "issue:43784|issue:46399", "issue:43232|issue:44188", + "issue:43784|issue:46399", "issue:45406|issue:46202", "issue:43873|issue:44945", "issue:44945|issue:45753", "issue:41628|issue:44351", - "issue:42994|issue:43824", "issue:42673|issue:43704", + "issue:42994|issue:43824", "issue:45081|issue:45103", "issue:44038|issue:45310", "issue:46202|issue:46701", @@ -13524,84 +13530,108 @@ "issue:36010|issue:41628", "issue:42831|issue:44222", "issue:43232|issue:44792", - "issue:43504|issue:45561", - "issue:45561|issue:45600", - "issue:30333|issue:33290", - "issue:42175|issue:44933", - "issue:44496|issue:45561", - "issue:43976|issue:44933", - "issue:30990|issue:35141" + "issue:43504|issue:45561" ], - "prepared_review_unit_hash": "0fbacba030712d7370352e75b0f8758f46f9cd78a38992b34bc92d4c905c8acf", + "prepared_review_unit_hash": "9081bb5291f795e841863d37cb265ade26c25a1c2b6e5483c60b427394e7c994", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 31, - "serialized_chars": 36715, - "estimated_input_tokens": 9179, - "estimated_eval_tokens": 18614 + "soft_pair_count": 30, + "serialized_chars": 36475, + "estimated_input_tokens": 9119, + "estimated_eval_tokens": 18494 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of Transformers v5 / model-specific regression reports rather than one duplicate bug. The only clear duplicate soft edge is the repeated Qwen3ForCausalLM multi-thread VRAM leak report; the rest are related at a broad theme level but not the same concrete code-path problem.", - "confidence": 0.68, - "canonical_issue_reason": "Issue 41628 is the best representative of the cluster\u2019s broadest pattern: top-level import/API breakage in Transformers. It is more generic and central than the many model-specific or unrelated regressions in the cluster.", + "summary": "This set is mostly heterogeneous issue reports around unrelated transformers/model bugs, so nearly all soft matches should be rejected. The only clear duplicate-looking pair is the repeated Qwen3ForCausalLM VRAM leak report (42673/43704), which appears to be the same underlying bug.", + "confidence": 0.92, + "canonical_issue_reason": "issue:42673 is the stronger canonical issue because it is the earlier, more discussed report of the exact same Qwen3ForCausalLM VRAM leak later repeated as issue:43704.", "canonical_pr_reason": null, - "best_issue_reason": "41628 is the strongest single issue to anchor the cluster because it captures the common import-surface regression theme without overfitting to one model family.", + "best_issue_reason": "issue:42673 is the best representative issue in this set because it has the clearest duplicate twin, the most discussion, and a concrete reproducible bug description.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:41628", - "right": "issue:44933", + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "Missing PyTorch dependency error is unrelated to a sentence-transformers loading hang." + }, + { + "left": "issue:43986", + "right": "issue:46726", "accept": false, - "reason": "Both are import failures, but they involve different missing symbols/modules (AutoImageProcessor vs image_utils) and different fixes." + "reason": "Both involve processor/backend availability, but one is a video-model torchvision crash and the other is a generic unrecognized image processor error." }, { "left": "issue:45310", "right": "issue:45375", "accept": false, - "reason": "Different bugs: a Qwen3.5 MoE from_pretrained error versus a missing deepstack_visual_indexes config field being dropped." + "reason": "Qwen3.5 MoE from_pretrained failure is not the same bug as a missing deepstack_visual_indexes config field." }, { - "left": "issue:45230", - "right": "issue:45397", + "left": "issue:41628", + "right": "issue:44933", "accept": false, - "reason": "45230 is too generic to establish identity, and 45397 is a specific gemma-4 zero3 loading failure." + "reason": "Different missing-import failures from different modules; not the same underlying symbol/export bug." }, { - "left": "issue:39692", + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Model loading regression and DeepSpeed ZeRO3 deque underflow are unrelated code paths." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding crash is unrelated to the Kimi tokenizer codec regression." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A placeholder bug report does not match a specific gemma-4 zero3 from_pretrained failure." + }, + { + "left": "issue:44242", + "right": "issue:46832", + "accept": false, + "reason": "Load-balancing loss logic and RT-DETR backbone feature-level mismatch are different bugs." + }, + { + "left": "issue:39692", "right": "issue:46726", "accept": false, - "reason": "Both touch vision/documentation flows, but one is a SigLIP2 example/quantization mismatch and the other is an AutoProcessor backend-missing error." + "reason": "Documentation example errors are not the same as the AutoProcessor backend-missing crash." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Different code paths: Qwen3-VL-Moe model bug versus processor.save_pretrained missing files." + "reason": "Qwen3-VL-Moe import/runtime issue is unrelated to processor.save_pretrained missing files." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized model saving failure is unrelated to the Llama4Config pad_token_id attribute error." + "reason": "Quantized model saving failure and missing pad_token_id on Llama4Config are different issues." }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Tokenizer loading in v5 and continue_final_message on mistral_common are separate tokenizer issues with different symptoms." + "reason": "Tokenizer loading in v5 and continue_final_message behavior for mistral_common are different failure modes." }, { "left": "issue:42371", @@ -13610,142 +13640,112 @@ "reason": "TF32 API guidance is unrelated to the InternVL2 meta-initialization incompatibility." }, { - "left": "issue:43784", - "right": "issue:46399", + "left": "issue:43232", + "right": "issue:44188", "accept": false, - "reason": "Importing sentence-transformers with latest transformers and a Qwen3.5 kernel-registration ValueError are different failures." + "reason": "Generation kwargs update after sync_gpus is not the same as attention-kernel divergence under torch.compile." }, { - "left": "issue:43232", - "right": "issue:44188", + "left": "issue:43784", + "right": "issue:46399", "accept": false, - "reason": "One is a generation-kwargs update issue after sync_gpus; the other is a torch.compile attention-kernel branching problem." + "reason": "sentence-transformers import NameError is unrelated to a missing decorator on apply_rotary_pos_emb registration." }, { "left": "issue:45406", "right": "issue:46202", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve is not the same as the Gemma pipeline roundtrip failure on \u2581." + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on \u2581 are different bugs." }, { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Quantization/offloading behavior and pipeline-parallelism output correctness are different bugs." + "reason": "Quantization offloading behavior is unrelated to incorrect outputs under pipeline parallelism." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Both involve generation on Qwen models, but one is pipeline-parallel incorrect output and the other is an infinite loop on a specific image." + "reason": "Pipeline-parallel output corruption is not the same as an infinite loop on a specific image." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "Both are top-level import errors, but they are different missing APIs (AutoImageProcessor vs HybridCache) and likely distinct fixes." - }, - { - "left": "issue:42994", - "right": "issue:43824", - "accept": false, - "reason": "Quantized saving failure and Qwen2_5_VLForConditionalGeneration import failure are unrelated." + "reason": "Both are import errors, but they concern different missing transformers symbols and different fixes." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Exact same reported bug: Qwen3ForCausalLM leaks VRAM when used in multiple dataloader threads." + "reason": "These have the same title and describe the same Qwen3ForCausalLM VRAM leak in multiple dataloader threads; 43704 is a duplicate report." + }, + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Quantized saving failure and missing Qwen2_5_VLForConditionalGeneration import are unrelated." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Mistral tokenizer backend_tokenizer crash and auto_docstring future-annotations AttributeError are separate failures." + "reason": "Mistral regex patch crash and auto-docstring future-annotations crash are different AttributeError sites." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen MoE, but the concrete failures and model versions differ." + "reason": "Both mention Qwen MoE, but one is a Qwen3-VL-Moe crash and the other is a Qwen3.5-Moe from_pretrained error." }, { "left": "issue:46202", "right": "issue:46701", "accept": false, - "reason": "Both are Gemma-related pipeline issues, but they are different outputs and different failure modes." + "reason": "Gemma tokenizer roundtrip failure and DiffusionGemma image-text pipeline failure are different concrete problems." }, { "left": "issue:44222", "right": "issue:45310", "accept": false, - "reason": "FP8 save_pretrained moe is not the same as the Qwen3.5 MoE from_pretrained error." + "reason": "FP8 save_pretrained on moe models is not the same as the Qwen3.5-Moe from_pretrained error." }, { "left": "issue:43278", "right": "issue:45137", "accept": false, - "reason": "Embedding dtype drift in evaluate and DeepSpeed ZeRO3 deque underflow are unrelated." + "reason": "Embedding dtype drift between train/eval is unrelated to the DeepSpeed ZeRO3 deque error." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Both are import regressions, but they concern different symbols and modules, so they should not be merged." + "reason": "GenerationMixin import failure and AutoImageProcessor import failure are separate missing-symbol regressions." }, { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "FP8 accuracy degradation and FP8 save_pretrained failure are related only at a broad quantization level, not the same bug." + "reason": "FineGrainedFP8 accuracy regression is not the same as FP8 save_pretrained behavior." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Generation kwargs handling after sync_gpus is unrelated to the janus image generation test failure." + "reason": "Generation state update bug and janus image generation test failure are unrelated." }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "A legacy-field loading bug for BeitForSemanticSegmentation is unrelated to the pytest-xdist captured_info.txt race." - }, - { - "left": "issue:45561", - "right": "issue:45600", - "accept": false, - "reason": "Test-worker file race and removed Sam3LiteText config references are separate CI issues." - }, - { - "left": "issue:30333", - "right": "issue:33290", - "accept": false, - "reason": "MLFlow reporting on failed training jobs and Adafactor OOM in deepspeed are different training issues." - }, - { - "left": "issue:42175", - "right": "issue:44933", - "accept": false, - "reason": "Package backend installation coverage and a missing image_utils import are unrelated." - }, - { - "left": "issue:44496", - "right": "issue:45561", - "accept": false, - "reason": "Unrecognized model/config loading and xdist file-race CI failures are not the same bug." - }, - { - "left": "issue:43976", - "right": "issue:44933", - "accept": false, - "reason": "Python version compatibility and a missing image_utils import are unrelated." + "reason": "BEiT legacy-field loading failure is unrelated to the pytest-xdist captured_info.txt race." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The only accepted pair is the exact duplicate Qwen3ForCausalLM VRAM-leak report, and the rejected pairs are appropriately treated as similar-but-distinct bugs rather than merged issues. The choice of 41628 as a central import/API regression anchor is reasonable given the listed issues." + "feedback": "The summary is well grounded in the packet and stays conservative on the soft matches. The only accepted duplicate pair, issue:42673 and issue:43704, is supported by identical titles, earlier date, and higher discussion on 42673. No PR-mergeability concerns apply here since these are issues, not PRs." }, "evaluator_used": true, "retried": false, @@ -13764,12 +13764,12 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:22355", "issue:27957", - "issue:28282", "issue:30333", + "issue:30990", + "issue:33290", "issue:34689", "issue:35141", "issue:42175", @@ -13778,21 +13778,18 @@ "issue:42898", "issue:43116", "issue:43295", - "issue:43377", "issue:43479", "issue:43723", "issue:43784", "issue:43824", "issue:43874", + "issue:43976", "issue:44016", "issue:44060", - "issue:44075", "issue:44079", "issue:44190", - "issue:44222", - "issue:44242", "issue:44295", - "issue:44297", + "issue:44496", "issue:44568", "issue:44623", "issue:44734", @@ -13806,6 +13803,8 @@ "issue:45310", "issue:45356", "issue:45405", + "issue:45561", + "issue:45600", "issue:45656", "issue:45712", "issue:45812", @@ -13816,24 +13815,29 @@ "issue:46829" ], "soft_pairs": [ + "issue:45561|issue:45600", + "issue:30333|issue:33290", + "issue:42175|issue:44933", + "issue:44496|issue:45561", + "issue:43976|issue:44933", + "issue:30990|issue:35141", "issue:42175|issue:45712", "issue:42898|issue:44568", "issue:43479|issue:46829", "issue:46077|issue:46123", - "issue:43377|issue:44242", "issue:44623|issue:45310", "issue:43479|issue:44869", "issue:42673|issue:46693", - "issue:44016|issue:46231", "issue:42898|issue:45812", "issue:43723|issue:44568", + "issue:44016|issue:46231", "issue:42898|issue:44779", - "issue:42222|issue:45200", "issue:43479|issue:44079", - "issue:43784|issue:43824", + "issue:42222|issue:45200", "issue:44734|issue:45103", - "issue:27957|issue:30333", + "issue:43784|issue:43824", "issue:43874|issue:45092", + "issue:27957|issue:30333", "issue:30333|issue:34689", "issue:42898|issue:45356", "issue:42898|issue:43295", @@ -13841,220 +13845,222 @@ "issue:43116|issue:44190", "issue:44060|issue:45125", "issue:45405|issue:45712", - "issue:30333|issue:35141", - "issue:22355|issue:28282", - "issue:44222|issue:44297", - "issue:44933|issue:45712", - "issue:44075|issue:46231" + "issue:30333|issue:35141" ], - "prepared_review_unit_hash": "98fb26a3692e521e3838bc50a9492a59fd419aa3437cf33328bade29494b3546", + "prepared_review_unit_hash": "d4f39c3f2eafdfadb4689e209bdb7706044e6817d5849cf5b8bf9e9f00bda5d5", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36971, - "estimated_input_tokens": 9243, - "estimated_eval_tokens": 18742 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36727, + "estimated_input_tokens": 9182, + "estimated_eval_tokens": 18620 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous and does not look like a true duplicate set. It mixes unrelated packaging/import issues, tokenizer-v5 regressions, multimodal model bugs, examples/docs problems, and training/runtime bugs. If one issue has to represent the cluster, issue 42898 is the broadest tokenizer-regression report, but it is still not a duplicate of most others.", - "confidence": 0.24, - "canonical_issue_reason": "Issue 42898 is the most central and broadest report in the weak tokenizer-regression subset, with the most soft links to other v5 tokenizer complaints.", + "summary": "The set is highly heterogeneous: tokenizer regressions, multimodal model bugs, training/runtime issues, CI/docs problems, and example-script failures. No single canonical duplicate stands out, and every soft-edge pair is a distinct issue rather than the same underlying bug/change.", + "confidence": 0.97, + "canonical_issue_reason": "No coherent shared bug/change spans the cluster; the issues cover many unrelated subsystems and failure modes.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 42898 is the best representative of the cluster's small tokenizer-regression theme, but the surrounding issues are too diverse for a single definitive duplicate.", + "best_issue_reason": "No issue is a good global canonical for this set; at best, some tokenizer-related reports are broadly similar, but they do not represent the same defect.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:45561", + "right": "issue:45600", + "accept": false, + "reason": "Different areas: pytest-xdist temp-file race vs a broken auto_mappings config reference." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLflow reporting after failed training is unrelated to deepspeed Adafactor OOM." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging/backend dependency issue vs a bad import in image_utils." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading failure is unrelated to a testing-utils race condition." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility bug vs nonexistent image_utils import." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-transformers loading hang is unrelated to embedding resize/post_init reinit." + }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Both relate to torch/dummy backend packaging, but one is a missing TensorFlow backend and the other is leftover dummy classes; not the same bug." + "reason": "Missing tensorflow backend in extras is unrelated to leftover dummy PT classes." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both concern tokenizer behavior in v5, but one is global cleanup-space behavior and the other is BOS/EOS insertion for a specific tokenizer." + "reason": "Both are tokenizer-related regressions, but they describe different concrete behaviors and code paths." }, { "left": "issue:43479", "right": "issue:46829", "accept": false, - "reason": "Different multimodal bugs: one is default config initialization, the other is incorrect output field assignment." + "reason": "Different multimodal output/config initialization bug vs classifier output field mix-up." }, { "left": "issue:46077", "right": "issue:46123", "accept": false, - "reason": "Unrelated model attention-mask bug vs pipeline batching bug." - }, - { - "left": "issue:43377", - "right": "issue:44242", - "accept": false, - "reason": "Different code paths: padding-mask sensitivity in MIMI encoder vs MoE load-balancing loss gating." + "reason": "Incorrect attention mask construction is unrelated to final-batch pipeline bookkeeping." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "One is processor save_pretrained file emission, the other is Qwen3.5 MoE loading regression." + "reason": "Processor save_pretrained file omission is unrelated to Qwen3.5 MoE from_pretrained failure." }, { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Different multimodal components and symptoms; config defaults vs Whisper timestamp decode crash." + "reason": "Multimodal config defaults vs Whisper timestamp decode crash are unrelated." }, { "left": "issue:42673", "right": "issue:46693", "accept": false, - "reason": "Both mention generation/performance, but one is a VRAM leak in dataloader threads and the other is a long-context FA2 regression." - }, - { - "left": "issue:44016", - "right": "issue:46231", - "accept": false, - "reason": "Notebook syntax error and documentation parameter mismatch are unrelated." + "reason": "VRAM leak in multi-thread dataloading is not the same as FA2 long-context performance regression." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Both are v5 tokenizer regressions, but one is cleanup-space behavior and the other is wrong Granite token IDs." + "reason": "Both involve tokenizers in v5, but the reported defects and affected models differ." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "Different tokenizer problems: loading failure in v5 vs add_special_tokens behavior for one tokenizer." + "reason": "AutoTokenizer loading issue is distinct from missing BOS/EOS insertion for a specific tokenizer." }, { - "left": "issue:42898", - "right": "issue:44779", + "left": "issue:44016", + "right": "issue:46231", "accept": false, - "reason": "Both are tokenizer regressions, but the reported failure modes and affected code paths differ." + "reason": "Notebook syntax error is unrelated to a documentation parameter-name mismatch." }, { - "left": "issue:42222", - "right": "issue:45200", + "left": "issue:42898", + "right": "issue:44779", "accept": false, - "reason": "Different model families and failures: vitpose breakage vs Gemma4 mm_token_type_ids defaulting." + "reason": "Generic clean_up_tokenization_spaces behavior change is not the same defect as DeepSeek tokenizer regression." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Config initialization bug and ModelOutput key assignment bug are unrelated." + "reason": "Phi4MultimodalConfig initialization bug is unrelated to ModelOutput key assignment logic." }, { - "left": "issue:43784", - "right": "issue:43824", + "left": "issue:42222", + "right": "issue:45200", "accept": false, - "reason": "Both are import errors, but they involve different symbols and different failure causes." + "reason": "Broken vitpose models and Gemma 4 mm_token_type_ids defaulting are different model-specific issues." }, { "left": "issue:44734", "right": "issue:45103", "accept": false, - "reason": "Serving KV-cache indexing crash is unrelated to auto-docstring annotation handling." + "reason": "KV-cache tensor indexing crash is unrelated to auto-docstring annotation handling." }, { - "left": "issue:27957", - "right": "issue:30333", + "left": "issue:43784", + "right": "issue:43824", "accept": false, - "reason": "XLMRoberta Flash Attention issue and MLflow reporting issue are unrelated." + "reason": "Sentence-transformers import NameError and Qwen2.5-VL import error are distinct failures." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Different multimodal model bugs: missing processor method vs remote-code meta-init incompatibility." + "reason": "Missing image-patch helper in Glm46VImageProcessorFast is not the same as remote-code meta-init incompatibility." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 support for XLM-R is unrelated to MLflow reporting on failed jobs." }, { "left": "issue:30333", "right": "issue:34689", "accept": false, - "reason": "Training-job reporting failure is unrelated to Llama vision model loading regression." + "reason": "Training-job MLflow reporting and Llama 3.2 Vision model loading are unrelated." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but they describe different mechanisms and model-specific symptoms." + "reason": "Both are tokenizer regressions, but they concern different models and different underlying codec/regex behavior." }, { "left": "issue:42898", "right": "issue:43295", "accept": false, - "reason": "Processor/tokenizer API regression is different from cleanup-space behavior changes." + "reason": "Processor.tokenizer access breakage is a different compatibility issue than clean_up_tokenization_spaces behavior." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Position_ids buffer read error and double optimizer-step bug are unrelated." + "reason": "Position_ids buffer access error is unrelated to deepspeed optimizer double-step behavior." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Different example scripts and different failures; one returns empty classification results, the other cannot load a local dataset." + "reason": "Example script returns empty multi-label results vs local dataset loading failure are different bugs." }, { "left": "issue:44060", "right": "issue:45125", "accept": false, - "reason": "Same model family name, but different bugs: tied-weights warning vs missing tensor-parallel plan." + "reason": "Incorrect tied-weights warning in Qwen3-Next is unrelated to missing _tp_plan in Qwen3.5 MoE." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "PEFT version pinning issue and dummy class cleanup are unrelated." + "reason": "Dependency version bump policy issue is unrelated to dummy class cleanup." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "MLflow reporting and output-embedding reinitialization are unrelated bugs." - }, - { - "left": "issue:22355", - "right": "issue:28282", - "accept": false, - "reason": "Different import failures with different missing modules/libraries." - }, - { - "left": "issue:44222", - "right": "issue:44297", - "accept": false, - "reason": "Both involve save_pretrained, but one is FP8 MoE model saving and the other is tokenizer class metadata." - }, - { - "left": "issue:44933", - "right": "issue:45712", - "accept": false, - "reason": "Nonexistent image_utils import and leftover dummy classes are separate import/cleanup issues." - }, - { - "left": "issue:44075", - "right": "issue:46231", - "accept": false, - "reason": "Optimizer argument handling and docs parameter naming are unrelated." + "reason": "MLflow failure reporting and embedding reinitialization after resize are unrelated." } ] }, @@ -14076,26 +14082,25 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:30990", - "issue:38175", + "issue:22355", + "issue:28282", "issue:41628", "issue:42757", "issue:42796", "issue:42831", - "issue:42907", "issue:43122", "issue:43232", "issue:43475", "issue:43479", "issue:43644", "issue:43723", + "issue:44075", "issue:44190", "issue:44206", "issue:44222", "issue:44263", - "issue:44279", "issue:44295", "issue:44297", "issue:44360", @@ -14105,21 +14110,21 @@ "issue:44485", "issue:44623", "issue:44734", + "issue:44933", "issue:44977", "issue:45071", "issue:45092", "issue:45103", "issue:45245", "issue:45310", - "issue:45381", "issue:45656", + "issue:45712", "issue:45834", - "issue:46050", "issue:46082", "issue:46132", "issue:46209", + "issue:46231", "issue:46305", - "issue:46328", "issue:46421", "issue:46424", "issue:46431", @@ -14128,6 +14133,10 @@ "issue:46762" ], "soft_pairs": [ + "issue:44222|issue:44297", + "issue:22355|issue:28282", + "issue:44075|issue:46231", + "issue:44933|issue:45712", "issue:44360|issue:46762", "issue:42831|issue:44623", "issue:44448|issue:46132", @@ -14137,8 +14146,8 @@ "issue:43479|issue:43644", "issue:45245|issue:46132", "issue:41628|issue:42757", - "issue:45071|issue:46082", "issue:44373|issue:44485", + "issue:45071|issue:46082", "issue:44297|issue:45310", "issue:46132|issue:46710", "issue:44222|issue:44484", @@ -14150,251 +14159,233 @@ "issue:43232|issue:45834", "issue:44263|issue:45092", "issue:44734|issue:46421", - "issue:43475|issue:46305", "issue:46209|issue:46431", + "issue:43475|issue:46305", "issue:46305|issue:46655", "issue:44295|issue:44623", "issue:44263|issue:44360", - "issue:30990|issue:38175", - "issue:44263|issue:44373", - "issue:42831|issue:45310", - "issue:45381|issue:46305", - "issue:42907|issue:46050", - "issue:44279|issue:46328", - "issue:43644|issue:46305" + "issue:44263|issue:44373" ], - "prepared_review_unit_hash": "48565d7cb14c411dcaf75e1c05f55d1a4d4fd31ae95c230f7fbb83c7469807d8", + "prepared_review_unit_hash": "b48cfbdd4f307026e5571f7b753f901364b8fbd60e09b7cd85da812645552ef3", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37486, - "estimated_input_tokens": 9372, - "estimated_eval_tokens": 19000 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36373, + "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 not a true duplicate set; it mixes many unrelated Transformer issues (tokenization, generation, config validation, save/load, FP8, multimodal/video bugs). The soft-similarity pairs are mostly broad subsystem overlap, not the same underlying defect.", - "confidence": 0.91, - "canonical_issue_reason": "No single canonical issue stands out because the cluster is heterogeneous and the apparent matches are mostly superficial similarity rather than the same bug.", + "summary": "Mostly a heterogeneous set of unrelated Transformers bug reports. The only mild local cohesion is around the DSA indexer and a few save_pretrained/tokenizer issues, but none of the soft pairs look like true duplicates. No PRs are present.", + "confidence": 0.34, + "canonical_issue_reason": "issue:44263 is the strongest concrete bug report in the set and sits near the small DSA-indexer mini-cluster, so it is the best standalone representative among otherwise mixed topics.", "canonical_pr_reason": null, - "best_issue_reason": "If one representative issue is needed, issue:44297 is the clearest, most self-contained report, but it does not subsume the others.", + "best_issue_reason": "issue:44263 is the most substantive, code-path-specific issue here; the neighboring DSA-indexer reports are related enough to make it the best issue candidate, even though they are not duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both involve save_pretrained, but one is an FP8 MoE save issue and the other is tokenizer_class metadata mismatch; different bugs and failure modes." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Both are import errors, but for unrelated symbols and causes: transformers.onnx vs missing PyTorch in the environment." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "Optimizer-related only at a broad level; one is ignored SGD args, the other is a docs/parameter-name mismatch in set_training." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both touch imports/dummy objects, but one is a missing image_utils import and the other is leftover dummy_pt_objects cleanup." + }, { "left": "issue:44360", "right": "issue:46762", "accept": false, - "reason": "Different bugs in different attention/indexing implementations; one is about a DSA indexer, the other about Minimax MSA numerical differences." + "reason": "Both are numerical/implementation complaints, but they target different attention/indexing code paths and are not the same bug." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FP8 accuracy regression vs missing files in processor.save_pretrained; unrelated failure modes." + "reason": "FP8 accuracy regression and processor.save_pretrained missing files are unrelated defects." }, { "left": "issue:44448", "right": "issue:46132", "accept": false, - "reason": "Different code paths: Pegasus output regression versus AttentionInterface.register changing callable behavior." + "reason": "One reports version-to-version output differences; the other is a behavior change in AttentionInterface.register." }, { "left": "issue:44297", "right": "issue:45656", "accept": false, - "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to deepspeed causing optimizer.step to run twice." + "reason": "Tokenizer metadata save bug vs optimizer step being called twice under deepspeed; no shared code-path bug." }, { "left": "issue:42796", "right": "issue:43723", "accept": false, - "reason": "Both involve tokenizers, but one is BPE marker handling and the other is an AutoTokenizer.from_pretrained import/load problem." + "reason": "Both are tokenizer/import-adjacent, but one is BPE marker handling and the other is AutoTokenizer loading failure in v5." }, { "left": "issue:44222", "right": "issue:44623", "accept": false, - "reason": "Both are save_pretrained-related, but one is FP8 MoE serialization and the other is processor file emission; different bugs." + "reason": "Both mention save_pretrained, but one is FP8 MoE model saving and the other is processor file emission; different artifacts." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Multimodal config default initialization is not the same as non-persistent buffers being filled with junk." + "reason": "Config default initialization with None and non-persistent buffer junk are different model-initialization problems." }, { "left": "issue:45245", "right": "issue:46132", "accept": false, - "reason": "Category-count runtime limit and AttentionInterface.register behavior are unrelated." + "reason": "Category-count overflow and AttentionInterface.register behavior are unrelated." }, { "left": "issue:41628", "right": "issue:42757", "accept": false, - "reason": "Two distinct missing-symbol import errors affecting different packages and APIs." + "reason": "Both are import errors, but they concern different packages/symbols and different root causes." }, { - "left": "issue:45071", - "right": "issue:46082", + "left": "issue:44373", + "right": "issue:44485", "accept": false, - "reason": "PretrainedConfig type-checking regression is different from LlamaConfig head_dim validation logic." + "reason": "Docstring mismatch for position_ids vs GLM-5 RoPE implementation; one is documentation, the other is model logic." }, { - "left": "issue:44373", - "right": "issue:44485", + "left": "issue:45071", + "right": "issue:46082", "accept": false, - "reason": "Docstring mismatch for position_ids is not the same as the GLM-5 RoPE implementation discussion." + "reason": "Type-checking regression in v5.4.0 vs LlamaConfig head_dim validation; not the same underlying bug." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer save_pretrained mismatch and Qwen3.5 MoE from_pretrained error are separate issues." + "reason": "Tokenizer save metadata bug and Qwen3.5 MoE from_pretrained error are separate issues." }, { "left": "issue:46132", "right": "issue:46710", "accept": false, - "reason": "AttentionInterface.register behavior change is unrelated to a model accuracy regression." + "reason": "Attention registration semantics and DeepSeek accuracy regression do not share the same concrete bug." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "FP8 MoE save_pretrained bug is unrelated to a question about max_shard_size defaults." + "reason": "FP8 model save bug vs a question about max_shard_size defaults; no duplicate relationship." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Tokenizer loading/import issue and flash-attention generation failure are different problems." + "reason": "Tokenizer loading failure and flash-attention generation issue are unrelated." }, { "left": "issue:44190", "right": "issue:46082", "accept": false, - "reason": "Example script failing to load a local dataset is unrelated to LlamaConfig head_dim validation." + "reason": "Dataset loading script issue vs config validation on head_dim; different subsystem and bug." }, { "left": "issue:43232", "right": "issue:46424", "accept": false, - "reason": "Generation kwargs update after sync_gpus and cache max length behavior are different cache/generation bugs." + "reason": "Generation kwargs after sync_gpus vs cache max_length semantics are separate cache/generation behaviors." }, { "left": "issue:43122", "right": "issue:46209", "accept": false, - "reason": "Tokenization changes across versions and FP8 inference on B200 are unrelated." + "reason": "Tokenization regression across versions and FP8 inference on B200 are unrelated." }, { "left": "issue:44206", "right": "issue:45103", "accept": false, - "reason": "Feature extractor center-arg crash and auto_docstring annotation crash are unrelated modules." + "reason": "Feature-extractor center-arg crash vs auto_docstring annotations crash; different code paths." }, { "left": "issue:43232", "right": "issue:45834", "accept": false, - "reason": "Generation kwarg handling is not the same as the Kosmos2.5 long-OCR index error." + "reason": "Generation kwarg handling vs Kosmos2.5 OCR index error are unrelated." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "DSA indexer split behavior and remote-code checkpoint meta-init incompatibility are different bugs." + "reason": "DSA indexer torch.split return values and remote-code checkpoint meta-init incompatibility are different bugs." }, { "left": "issue:44734", "right": "issue:46421", "accept": false, - "reason": "Both mention cache, but one is a serve KV-cache indexing crash and the other is spurious recompilation from lazy cache init." + "reason": "KV-cache continuation indexing crash vs lazy cache initialization causing recompilation are not the same defect." }, { - "left": "issue:43475", - "right": "issue:46305", + "left": "issue:46209", + "right": "issue:46431", "accept": false, - "reason": "SAM 3 Video missing attribute and MoonshineStreaming mask off-by-one are unrelated video bugs." + "reason": "FP8 inference on B200 and missing chat template on a processor are unrelated." }, { - "left": "issue:46209", - "right": "issue:46431", + "left": "issue:43475", + "right": "issue:46305", "accept": false, - "reason": "FP8 inference on B200 and missing chat template in processor.save_pretrained are unrelated." + "reason": "SAM3 vision encoder attribute error vs MoonshineStreaming sliding-window off-by-one are different model bugs." }, { "left": "issue:46305", "right": "issue:46655", "accept": false, - "reason": "Both are video-related, but one is a sliding-window mask off-by-one and the other is incorrect padding in a video processor." + "reason": "Both are sequence/padding style bugs, but in different processors with different off-by-one mechanics." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Position_ids buffer access error is unrelated to processor.save_pretrained missing files." + "reason": "Buffer access error for position_ids vs missing processor save_pretrained files are unrelated." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Same broad DSA area, but the concrete defects differ: torch.split return handling versus missing ReLU." - }, - { - "left": "issue:30990", - "right": "issue:38175", - "accept": false, - "reason": "Sentence-transformers loading hang and zero probabilities in SigLIP2 are unrelated." + "reason": "Both mention the DSA indexer, but one is a torch.split return-value issue and the other is a missing ReLU; related area, not the same bug." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "DSA indexer split behavior and a docstring error for position_ids are not the same bug." - }, - { - "left": "issue:42831", - "right": "issue:45310", - "accept": false, - "reason": "Both are model/precision regressions, but one is FineGrainedFP8 accuracy and the other is a Qwen3.5 MoE loading error." - }, - { - "left": "issue:45381", - "right": "issue:46305", - "accept": false, - "reason": "Qwen2.5-VL video vision_position_ids and MoonshineStreaming mask logic are different models and different defects." - }, - { - "left": "issue:42907", - "right": "issue:46050", - "accept": false, - "reason": "Dequantized Ministrals save failure and sinq quantized save/load breakage are related only at a very broad save/load level." - }, - { - "left": "issue:44279", - "right": "issue:46328", - "accept": false, - "reason": "A dependency issue with transformers is unrelated to a SequenceFeatureExtractor.pad performance problem." - }, - { - "left": "issue:43644", - "right": "issue:46305", - "accept": false, - "reason": "Non-persistent buffer junk in v5 and a sliding-window mask off-by-one are unrelated bugs." + "reason": "DSA indexer bug vs a docstring error for position_ids; clearly not duplicates." } ] }, @@ -14416,13 +14407,14 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:22355", "issue:27957", "issue:28218", "issue:30990", "issue:33357", + "issue:38175", "issue:39401", "issue:41628", "issue:42371", @@ -14437,38 +14429,43 @@ "issue:43232", "issue:43408", "issue:43576", + "issue:43644", "issue:43646", - "issue:43756", "issue:43784", "issue:43824", - "issue:43976", "issue:44038", - "issue:44060", "issue:44188", "issue:44261", "issue:44263", - "issue:44295", - "issue:44297", + "issue:44279", "issue:44373", "issue:44483", - "issue:44488", "issue:44568", "issue:44829", "issue:44945", "issue:45092", "issue:45278", + "issue:45310", "issue:45357", + "issue:45381", "issue:45406", "issue:45593", "issue:45753", "issue:46002", - "issue:46172", + "issue:46050", + "issue:46305", + "issue:46328", "issue:46369", - "issue:46402", - "issue:46762" + "issue:46402" ], "soft_pairs": [ + "issue:30990|issue:38175", + "issue:42831|issue:45310", + "issue:44279|issue:46328", + "issue:45381|issue:46305", + "issue:42907|issue:46050", "issue:45092|issue:45357", + "issue:43644|issue:46305", "issue:44829|issue:46002", "issue:22355|issue:27957", "issue:44261|issue:45092", @@ -14482,141 +14479,166 @@ "issue:27957|issue:30990", "issue:42886|issue:43066", "issue:42757|issue:45278", - "issue:42886|issue:42994", "issue:44188|issue:44373", + "issue:42886|issue:42994", "issue:43408|issue:46369", "issue:43232|issue:45593", "issue:44261|issue:44373", - "issue:43646|issue:46002", "issue:42796|issue:43122", + "issue:43646|issue:46002", "issue:42907|issue:44945", - "issue:43576|issue:43784", "issue:30990|issue:33357", - "issue:43756|issue:44060", - "issue:42371|issue:44263", - "issue:44295|issue:44297", - "issue:39401|issue:44488", - "issue:43976|issue:44945", - "issue:46172|issue:46762" + "issue:43576|issue:43784", + "issue:42371|issue:44263" ], - "prepared_review_unit_hash": "49940d5749f091da59f222427eeedbaa21065e2e600812f59bb6351e49030e44", + "prepared_review_unit_hash": "ddd044e59eba6fae79851d2b0b2fa7e4d3639d460107d0b435629d2194c2d757", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 30, - "serialized_chars": 36224, - "estimated_input_tokens": 9056, - "estimated_eval_tokens": 18368 + "soft_pair_count": 31, + "serialized_chars": 36488, + "estimated_input_tokens": 9122, + "estimated_eval_tokens": 18500 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: most items are unrelated tokenizer/import/regression/model-specific issues, so there is no strong duplicate core. I rejected all soft edges as not the same underlying bug or change.", - "confidence": 0.21, - "canonical_issue_reason": "Issue 45278 is the broadest umbrella candidate in this set, covering a version-upgrade import-regression theme that is closer to several other import-related reports than the rest of the cluster.", + "summary": "This cluster is very heterogeneous: most pairs are different bug reports with only superficial topical overlap. The clearest umbrella issue is the broad import-regression report, but there is no PR in the set.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45278 is the broadest umbrella here: it captures multiple post-upgrade import failures and can represent the import-error subgroup better than the narrower, model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best global issue candidate because it is the most general and could reasonably absorb adjacent import-error reports if any consolidation were needed.", + "best_issue_reason": "If forced to pick one representative issue, issue:45278 is the best single anchor for the cluster because it is broad and cross-cutting, though the overall set is still not a true duplicate cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Different symptoms and different code paths: loading hangs in Sentence-Transformers vs zero probabilities in a SigLIP2 model." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "Both mention model/precision problems, but one is a FineGrainedFP8 accuracy regression and the other is a Qwen3.5-MoE from_pretrained error." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "A vague dependency issue is not the same bug as a SequenceFeatureExtractor.pad performance regression." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Both are offset/masking style bugs, but they affect different models and different code paths." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both involve quantized model persistence, but one is a dequantized Ministral/Devstral save failure and the other is a sinq save/load regression." + }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Different models and different failure modes: meta-initialization incompatibility vs incorrect visual encoder keys on save." + "reason": "Different bugs: remote-code/meta-init incompatibility for old InternVL2 vs incorrect visual encoder keys on Qwen3.5 save_pretrained." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Unrelated regressions: non-persistent buffer initialization junk vs a MoonshineStreaming sliding-window off-by-one." }, { "left": "issue:44829", "right": "issue:46002", "accept": false, - "reason": "Both are training/initialization problems, but the concrete bugs differ: flash-attention training degradation vs NaNs from checkout+init." + "reason": "Different failure modes: degenerate training with flash_attention_3 vs NaNs at checkout/init in classification heads." }, { "left": "issue:22355", "right": "issue:27957", "accept": false, - "reason": "Unrelated: an import-path failure for transformers.onnx versus an XLMRoberta flash-attention issue." + "reason": "Import failure for transformers.onnx is unrelated to an XLMRoberta Flash Attention 2 issue." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "Different code paths and symptoms: MLA RMS norm epsilon precision issue vs remote-code/meta-init incompatibility." + "reason": "Separate problems: missing rms_norm_eps in MLA layernorm vs remote-code meta-initialization incompatibility." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Serve/API request parsing failure and Gemma4Processor attribute crash are different concrete bugs." + "reason": "One is an API request rejection in chat/completions; the other is a processor attribute crash in transformers serve." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "Different models and likely different causes; one is a FineGrainedFP8 accuracy regression, the other is a Qwen3-VL-Moe bug report." + "reason": "Both mention Qwen/MoE, but the reported failures are different and do not point to the same code path." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and different modules; not the same underlying bug." + "reason": "Both are import errors, but for different symbols and likely different missing-export regressions." }, { "left": "issue:43408", "right": "issue:46402", "accept": false, - "reason": "A model-type warning and missing text-encoder weights are distinct SAM3 issues, not one duplicate defect." + "reason": "Both involve sam3, but a model-type warning is not the same as missing text-encoder weights loading." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "FineGrainedFP8 accuracy loss and a Qwen3_5 image infinite-loop issue are unrelated." + "reason": "Different issues: FP8 accuracy regression vs infinite loop on a specific image in Qwen3_5." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "ONNX import failure versus tokenizer spacing after added tokens; no shared underlying bug." + "reason": "An ONNX import error is unrelated to tokenizer adding an extra space after an added token." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Different tokenizer regressions on different models: wrong offset mappings vs missing BOS/EOS addition." + "reason": "Both are tokenizer bugs, but one is offset_mapping on Qwen3 and the other is BOS/EOS handling on mdeberta-v3-base." }, { "left": "issue:27957", "right": "issue:30990", "accept": false, - "reason": "Flash Attention 2 support for XLMRoberta is not the same bug as Sentence Transformers getting stuck loading." + "reason": "Different subsystems and symptoms: Flash Attention 2 on XLMRoberta vs Sentence-Transformers loading hang." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline-cache tokenizer loading failure and wrong decoder type in v5 are different tokenizer issues." + "reason": "Offline cache loading and wrong decoder type are different tokenizer/version regressions." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "45278 is a broad import-regression report, but 42757 is a specific missing import from huggingface_hub; not the same concrete bug." - }, - { - "left": "issue:42886", - "right": "issue:42994", - "accept": false, - "reason": "Tokenizer offline-cache loading and quantized model saving are unrelated." + "reason": "A specific huggingface_hub import error is not the same underlying bug as a broad set of transformers import failures." }, { "left": "issue:44188", @@ -14624,89 +14646,65 @@ "accept": false, "reason": "Attention-kernel divergence under torch.compile is unrelated to a wrong docstring for position_ids." }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Tokenizer offline cache loading and quantized model saving are different problems." + }, { "left": "issue:43408", "right": "issue:46369", "accept": false, - "reason": "Both mention SAM3, but a memory overflow in sam3-litetext is not the same as a model-type warning for sam3_tracker." + "reason": "Same model family, but a model-type warning is not the same as a memory overflow in sam3-litetext." }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Generation kwargs handling and D-FINE auxiliary-loss behavior are unrelated code paths." + "reason": "Generation sync_gpus state handling is unrelated to D-FINE auxiliary-loss behavior." }, { "left": "issue:44261", "right": "issue:44373", "accept": false, - "reason": "A numerical precision/config bug is not a docstring issue." - }, - { - "left": "issue:43646", - "right": "issue:46002", - "accept": false, - "reason": "Custom model initialization breakage and NaNs in classification heads are different failures, even if both involve initialization." + "reason": "A precision/config bug in MLA q_a_layernorm is not the same as a documentation mismatch." }, { "left": "issue:42796", "right": "issue:43122", "accept": false, - "reason": "Tokenizer BPE-marker handling for Ministral and version-to-version tokenization differences are not clearly the same bug." + "reason": "Tokenizer BPE-marker handling for a specific model is not the same as a version-wide tokenization change report." }, { - "left": "issue:42907", - "right": "issue:44945", + "left": "issue:43646", + "right": "issue:46002", "accept": false, - "reason": "Quantized model saving failure and incorrect LLM output under pipeline parallelism are unrelated." + "reason": "Custom model initialization breakage and NaNs in classification heads are different failure modes." }, { - "left": "issue:43576", - "right": "issue:43784", + "left": "issue:42907", + "right": "issue:44945", "accept": false, - "reason": "Broken env command and a NameError importing sentence-transformers are different regression surfaces." + "reason": "Quantized save failures are unrelated to incorrect outputs under pipeline parallelism." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "Both are loading/runtime failures, but one is Sentence Transformers hanging and the other is a MacOS bus error on a CLIP model." + "reason": "Loading stuck in one model stack is not the same as a MacOS bus error in another." }, { - "left": "issue:43756", - "right": "issue:44060", + "left": "issue:43576", + "right": "issue:43784", "accept": false, - "reason": "Smollm3 RoPE-layer dropping and Qwen3-Next tied-weights warning are different architecture-specific issues." + "reason": "A broken CLI env command is not the same bug as a NameError when importing sentence-transformers." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 configuration guidance and torch.split return values in GlmMoeDsaIndexer are unrelated." - }, - { - "left": "issue:44295", - "right": "issue:44297", - "accept": false, - "reason": "Different tokenizer serialization problems: reading registered position_ids buffer vs mismatched tokenizer_class in config." - }, - { - "left": "issue:39401", - "right": "issue:44488", - "accept": false, - "reason": "Wrong offset_mapping for Qwen3 tokenizer and failure to load cjvt/sleng-bert are separate tokenizer/model issues." - }, - { - "left": "issue:43976", - "right": "issue:44945", - "accept": false, - "reason": "Python-version compatibility in 5.1.0 is unrelated to pipeline-parallelism output corruption." - }, - { - "left": "issue:46172", - "right": "issue:46762", - "accept": false, - "reason": "Both concern numerical behavior, but they are different models and different attention implementations." + "reason": "TF32 configuration guidance and a torch.split return-value issue are unrelated." } ] }, @@ -14728,18 +14726,16 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:22355", "issue:29127", "issue:29159", "issue:36296", - "issue:38617", "issue:39401", "issue:41628", "issue:42175", "issue:42757", - "issue:42898", "issue:42907", "issue:43116", "issue:43122", @@ -14747,19 +14743,21 @@ "issue:43408", "issue:43452", "issue:43475", - "issue:43493", + "issue:43756", "issue:43784", "issue:43824", "issue:43937", + "issue:43976", "issue:44038", + "issue:44060", "issue:44117", - "issue:44188", "issue:44222", "issue:44295", "issue:44297", "issue:44351", "issue:44360", "issue:44485", + "issue:44488", "issue:44521", "issue:44625", "issue:44704", @@ -14774,30 +14772,35 @@ "issue:45290", "issue:45478", "issue:45600", - "issue:45753", + "issue:46172", "issue:46326", - "issue:46328", - "issue:46614" + "issue:46614", + "issue:46762" ], "soft_pairs": [ + "issue:44295|issue:44297", + "issue:43756|issue:44060", + "issue:39401|issue:44488", + "issue:43976|issue:44945", + "issue:46172|issue:46762", "issue:42907|issue:44964", "issue:42907|issue:43937", - "issue:43784|issue:45600", "issue:44117|issue:44704", + "issue:43784|issue:45600", "issue:41628|issue:45278", - "issue:44625|issue:44829", "issue:44704|issue:45290", + "issue:44625|issue:44829", "issue:39401|issue:44521", "issue:44360|issue:44485", "issue:43408|issue:43475", "issue:43122|issue:44521", - "issue:43122|issue:46326", "issue:44038|issue:44297", + "issue:43122|issue:46326", "issue:42907|issue:43452", "issue:39401|issue:46614", "issue:22355|issue:29159", - "issue:45216|issue:45478", "issue:44038|issue:44295", + "issue:45216|issue:45478", "issue:43116|issue:45200", "issue:44038|issue:44222", "issue:43824|issue:44351", @@ -14806,233 +14809,228 @@ "issue:36296|issue:43116", "issue:43278|issue:44945", "issue:22355|issue:29127", - "issue:42175|issue:44938", - "issue:43493|issue:45753", - "issue:44188|issue:46328", - "issue:42898|issue:46614", - "issue:43937|issue:44485", - "issue:38617|issue:45278" + "issue:42175|issue:44938" ], - "prepared_review_unit_hash": "e3b631de1b14fc2eef82b1807f8905d89c6ff5134743096a9b935cb4e1295658", + "prepared_review_unit_hash": "e3a1cd42e9a7200d5fc949132cac12d3fb49c5cff2c23d700438fdde92b97732", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37023, - "estimated_input_tokens": 9256, - "estimated_eval_tokens": 18768 + "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 mostly a grab-bag of unrelated Transformers issues. The only mild thematic overlap is around import/version regressions, but the candidate pairs do not look like the same concrete bug or change, so I would not merge any of them.", - "confidence": 0.47, - "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report in the set (\u201cmany import errors after update\u201d), making it the most reasonable anchor among otherwise unrelated issues.", + "summary": "This cluster is very noisy and mostly consists of unrelated model-, tokenizer-, import-, and training-regression reports. I do not see any soft edge that clearly represents the same underlying bug/change.", + "confidence": 0.24, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella in the set: it captures a general import-regression report after a version upgrade, which is the closest central theme among the otherwise heterogeneous issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best global representative because it is the most general and could plausibly subsume several of the import-regression reports, even though it is not a duplicate of the others.", + "best_issue_reason": "Issue 45278 is the best available anchor for this cluster because it is the most general and could subsume some of the narrower import-error reports, though the overall cluster is not a single coherent duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Different bugs: buffer registration/position_ids access versus tokenizer_class metadata saved in tokenizer_config.json." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Both are model-specific, but one is RoPE-layer dropping and the other is a tied-weights warning in Qwen3-Next; no shared code-path bug." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Both involve tokenizers, but one is wrong offset_mapping for Qwen3 and the other is a model loading failure for cjvt/sleng-bert." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Unrelated failures: Python-version compatibility versus incorrect output under pipeline parallelism." + }, + { + "left": "issue:46172", + "right": "issue:46762", + "accept": false, + "reason": "Different model implementations and symptoms; no evidence they are the same numerical bug." + }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Both are model-related errors, but one is about saving dequantized Ministrals/Devstrals and the other about loading Phi-4 multimodal; different code paths and likely different bugs." + "reason": "Different model families and different failure modes: save-dequantized Minstrals/Devstrals versus loading Phi-4 multimodal." }, { "left": "issue:42907", "right": "issue:43937", "accept": false, - "reason": "Different models and different failure modes: dequantized save_pretrained vs invalid GenerationConfig." + "reason": "Different regressions: dequantized save failure versus GLM-5 GenerationConfig validation." }, { - "left": "issue:43784", - "right": "issue:45600", + "left": "issue:44117", + "right": "issue:44704", "accept": false, - "reason": "Both are import-time breakages, but the missing `nn` in sentence-transformers and broken `auto_mappings.py` references are distinct causes." + "reason": "Tokenizer mapping fallback bug versus AutoProcessor not forwarding kwargs to cached_file; separate code paths." }, { - "left": "issue:44117", - "right": "issue:44704", + "left": "issue:43784", + "right": "issue:45600", "accept": false, - "reason": "These touch tokenizer/processor plumbing, but one is a tokenizer mapping None-handling bug and the other is kwargs not forwarded to `cached_file`." + "reason": "Both are import/config issues, but one is a NameError in sentence-transformers integration and the other is stale auto_mappings references for Sam3LiteText." }, { "left": "issue:41628", "right": "issue:45278", "accept": false, - "reason": "`AutoImageProcessor` missing and \u201cmany import errors\u201d are related at a very high level, but the evidence is too broad to treat them as the same concrete bug." + "reason": "Both are import-related, but 45278 is a broad report of many import errors and 41628 is one specific missing export; not enough to call them the same bug." }, { - "left": "issue:44625", - "right": "issue:44829", + "left": "issue:44704", + "right": "issue:45290", "accept": false, - "reason": "Unrelated topics: config propagation for Qwen3.5 labels versus degenerate training with flash_attention_3." + "reason": "AutoProcessor kwargs forwarding is unrelated to apply_chat_template crashing on assistant tool-call messages." }, { - "left": "issue:44704", - "right": "issue:45290", + "left": "issue:44625", + "right": "issue:44829", "accept": false, - "reason": "Different API paths: processor kwargs forwarding versus chat template crashes on tool-call assistant messages." + "reason": "Different issues: num_labels propagation in Qwen3.5 versus degenerate training with flash_attention_3." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Both mention tokenization-related outputs, but one is wrong offset mappings and the other is multimodal assistant masks; not the same bug." + "reason": "Tokenizer offset_mapping bug versus multimodal assistant_mask generation; different behavior and code paths." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "Completely different components and problems: DSA indexer ReLU discussion vs GLM-5 RoPE implementation." + "reason": "Completely different topics: DSA indexer ReLU discussion versus GLM-5 RoPE implementation." }, { "left": "issue:43408", "right": "issue:43475", "accept": false, - "reason": "Both are SAM 3 Video-related, but a model-type warning and a missing `fpn_position_embeddings` attribute are different defects." + "reason": "Both are SAM 3 Video-related, but one is a model-type warning and the other is a missing attribute in vision encoder output." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "Both are tokenizer/template regressions, but one is tokenization differences across versions and the other is zero assistant masks for multimodal inputs." + "reason": "Tokenization regression versus multimodal chat-template assistant masks; not the same underlying defect." }, { - "left": "issue:43122", - "right": "issue:46326", + "left": "issue:44038", + "right": "issue:44297", "accept": false, - "reason": "Both involve tokenization behavior, but they concern different tokenizer families and distinct regressions." + "reason": "Qwen3-VL-Moe loading/training issue versus tokenizer save_pretrained metadata mismatch." }, { - "left": "issue:44038", - "right": "issue:44297", + "left": "issue:43122", + "right": "issue:46326", "accept": false, - "reason": "Different issues: Qwen3-VL-Moe bug versus tokenizer_class metadata mismatch on save_pretrained." + "reason": "Different tokenizer bugs: version-to-version tokenization differences versus continue_final_message on mistral_common tokenizers." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Both involve model I/O, but gguf_file loading failures and dequantized-model save failures are not the same underlying bug." + "reason": "Saving dequantized Minstrals/Devstrals is unrelated to gguf_file breaking AutoTokenizer/AutoModel loading." }, { "left": "issue:39401", "right": "issue:46614", "accept": false, - "reason": "Both are tokenizer-related, but offset_mapping behavior and stale phonemizer backend state are separate code paths." + "reason": "Both mention tokenizers, but one is Qwen3 offset_mapping and the other is stale phonemizer language state in Wav2Vec2PhonemeCTCTokenizer." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "An import failure for `transformers.onnx` is unrelated to slow-vs-fast tokenizer inconsistency." + "reason": "No module named transformers.onnx is unrelated to slow-vs-fast tokenizer inconsistencies." }, { - "left": "issue:45216", - "right": "issue:45478", + "left": "issue:44038", + "right": "issue:44295", "accept": false, - "reason": "Both mention Qwen3.5 and version regressions, but one is save_pretrained checkpoint correctness and the other is from_pretrained loading failure." + "reason": "Different problems: Qwen3-VL-Moe bug versus position_ids buffer access error." }, { - "left": "issue:44038", - "right": "issue:44295", + "left": "issue:45216", + "right": "issue:45478", "accept": false, - "reason": "Different symptoms and likely different subsystems: Qwen3-VL-Moe bug versus position_ids buffer access error." + "reason": "Both involve Qwen3.5 and newer versions, but one is save_pretrained checkpoint corruption and the other is from_pretrained loading failure; not clearly the same bug." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Classification pipeline output emptiness and Gemma 4 mm_token_type_ids defaults are unrelated." + "reason": "Example-script multi-label output bug versus Gemma 4 mm_token_type_ids defaulting; unrelated." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Both mention MoE/model saving, but they are different bugs with different triggers and symptoms." + "reason": "Qwen3-VL-Moe loading issue versus FP8 save_pretrained moe issue; related only by broad model family." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but they are for different exported symbols and there is no sign of the same code-path breakage." + "reason": "Different missing symbols from different APIs; both are import errors but not the same concrete bug." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions across versions, but they affect different models and different tokenization behavior." + "reason": "Both are tokenizer regressions, but one is generic tokenization changes in v5 and the other is Deepseek-specific incorrect results." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Import errors in different packages (`huggingface_hub` vs `transformers`) are not the same bug." + "reason": "Different packages and failure modes: huggingface_hub missing is_offline_mode versus transformers missing HybridCache." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor parallel training and multi-label classification pipeline output are unrelated." + "reason": "Tensor-parallel training bug is unrelated to the example script returning empty multi-label results." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Embedding dtype drift in evaluate and incorrect output under pipeline parallelism are different problems." + "reason": "Dtype mismatch during evaluate is unrelated to incorrect output under pipeline parallelism." }, { "left": "issue:22355", "right": "issue:29127", "accept": false, - "reason": "Different areas and symptoms: missing `transformers.onnx` module vs LayoutLMv3 error-message clarity." + "reason": "No module import failure versus a LayoutLMv3 error-message clarity issue; not duplicates." }, { "left": "issue:42175", "right": "issue:44938", "accept": false, - "reason": "TensorFlow backend packaging and Python 3.14 loading failures are not the same underlying bug." - }, - { - "left": "issue:43493", - "right": "issue:45753", - "accept": false, - "reason": "Both are multimodal model issues, but different models and different failure modes; not a duplicate pair." - }, - { - "left": "issue:44188", - "right": "issue:46328", - "accept": false, - "reason": "Attention-kernel divergence under torch.compile and SequenceFeatureExtractor.pad inefficiency are unrelated." - }, - { - "left": "issue:42898", - "right": "issue:46614", - "accept": false, - "reason": "Tokenization cleanup behavior and phonemizer backend state are separate issues." - }, - { - "left": "issue:43937", - "right": "issue:44485", - "accept": false, - "reason": "Same model family (GLM-5) but different concrete bugs: generation config validation vs RoPE implementation." - }, - { - "left": "issue:38617", - "right": "issue:45278", - "accept": false, - "reason": "Both are import regressions, but the specific missing symbol and the broader \u201cmany import errors\u201d report are not the same concrete defect." + "reason": "Backend packaging/install issue versus Python 3.14 load failure; different root causes." } ] }, @@ -15054,23 +15052,20 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:28348", - "issue:33453", - "issue:39401", + "issue:38617", "issue:42222", "issue:42371", "issue:42673", - "issue:42796", - "issue:43064", + "issue:42898", "issue:43066", "issue:43116", "issue:43441", "issue:43493", "issue:43611", "issue:43618", - "issue:43673", "issue:43825", "issue:43827", "issue:43874", @@ -15079,9 +15074,10 @@ "issue:43992", "issue:44038", "issue:44075", + "issue:44188", "issue:44263", "issue:44355", - "issue:44380", + "issue:44485", "issue:44625", "issue:44704", "issue:44779", @@ -15090,13 +15086,14 @@ "issue:45072", "issue:45083", "issue:45092", + "issue:45278", "issue:45357", "issue:45446", "issue:45478", - "issue:45563", + "issue:45753", "issue:46002", - "issue:46077", "issue:46231", + "issue:46328", "issue:46614", "issue:46640", "issue:46697", @@ -15106,231 +15103,231 @@ "issue:46829" ], "soft_pairs": [ + "issue:43493|issue:45753", + "issue:44188|issue:46328", + "issue:42898|issue:46614", + "issue:43937|issue:44485", + "issue:38617|issue:45278", "issue:43066|issue:44779", "issue:43611|issue:46002", "issue:44355|issue:46231", "issue:43116|issue:43493", - "issue:43976|issue:45446", "issue:44038|issue:45478", + "issue:43976|issue:45446", "issue:42673|issue:45072", - "issue:44625|issue:46002", "issue:45357|issue:45478", + "issue:44625|issue:46002", "issue:45083|issue:45446", "issue:42371|issue:43874", "issue:43618|issue:46829", - "issue:44075|issue:44936", "issue:43116|issue:43825", - "issue:42222|issue:43441", + "issue:44075|issue:44936", "issue:43992|issue:44704", + "issue:42222|issue:43441", "issue:44995|issue:46697", "issue:43066|issue:46710", "issue:28348|issue:46640", - "issue:46614|issue:46752", "issue:44263|issue:46762", + "issue:46614|issue:46752", "issue:45092|issue:46762", - "issue:43827|issue:45446", - "issue:43064|issue:43673", - "issue:42673|issue:42796", - "issue:43937|issue:45563", - "issue:33453|issue:39401", - "issue:44380|issue:46077" + "issue:43827|issue:45446" ], - "prepared_review_unit_hash": "770f370684b8b80705daf9b7896bc13609d40cb4ff81207fd9816e7034e6439f", + "prepared_review_unit_hash": "8d73576e22545e149b0fedf0670b1e5be32b4f8cf2405e0dac39de7044780ab7", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, + "node_count": 47, + "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35786, - "estimated_input_tokens": 8947, - "estimated_eval_tokens": 18150 + "serialized_chars": 35186, + "estimated_input_tokens": 8797, + "estimated_eval_tokens": 17850 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are unrelated v5-era regressions across tokenization, model loading, training, generation, attention, and docs. None of the soft-edge pairs looks like the same underlying bug or change, so I would not merge or dedupe them.", - "confidence": 0.91, - "canonical_issue_reason": "No single canonical issue fits the cluster well; the items span many distinct subsystems and failure modes rather than one shared bug.", + "summary": "This cluster is mostly a loose theme match around Transformers v5 regressions, but the individual issues point to different concrete bugs across models, tokenizers, processors, import paths, and docs. I don\u2019t see any soft-edge pair that is clearly the same underlying defect.", + "confidence": 0.88, + "canonical_issue_reason": "No single issue cleanly represents the whole cluster; if one must be chosen as the broadest umbrella, issue 45278 is the closest because it covers general import regressions after upgrading Transformers, but it is still not a true duplicate of the rest.", "canonical_pr_reason": null, - "best_issue_reason": "No issue is a safe global duplicate target here. The titles point to different concrete problems, so picking one as the cluster representative would be misleading.", + "best_issue_reason": "Issue 45278 is the best global representative only in a very loose sense because it is the broadest regression report in the set; however, the cluster is heterogeneous and it should not be treated as a duplicate anchor for the others.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Different models and symptoms: SigLIP2 implementation discrepancy vs a Qwen3_5 infinite-loop bug on one image." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "Unrelated code paths: attention-kernel branching under torch.compile vs a feature-extractor padding performance issue." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "Different subsystems: tokenizer-space cleanup behavior in v5 vs stale espeak backend state in phonemization." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Both are GLM-5-related, but one is GenerationConfig validation and the other is RoPE implementation; not the same bug." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "45278 is a broad import-error report, but 38617 is one specific missing symbol import; not enough evidence they are the same underlying defect." + }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer-related, but one is a decoder-type change in v5 and the other is a DeepSeek tokenization regression; different underlying bugs." + "reason": "Both involve tokenizers, but one is a decoder-type regression and the other is incorrect DeepSeek tokenization output." }, { "left": "issue:43611", "right": "issue:46002", "accept": false, - "reason": "Model loading with base_model_prefix is unrelated to NaNs in classification heads during init/checkout." + "reason": "Model loading via base_model_prefix is unrelated to NaNs in classification heads during checkout/init." }, { "left": "issue:44355", "right": "issue:46231", "accept": false, - "reason": "Compiled Python file errors and a set_training() doc parameter typo are not the same issue." + "reason": "Compiled-Python runtime errors are unrelated to a documentation parameter-name mismatch." }, { "left": "issue:43116", "right": "issue:43493", "accept": false, - "reason": "Example script multi-label output failure is unrelated to a SigLIP2 HF-vs-JAX implementation discrepancy." + "reason": "Example-script multi-label output bug is unrelated to a SigLIP2 implementation discrepancy." }, { - "left": "issue:43976", - "right": "issue:45446", + "left": "issue:44038", + "right": "issue:45478", "accept": false, - "reason": "Python version support bug and a PyTorch version check bug are different compatibility problems." + "reason": "Related Qwen/MoE area, but different models and different failure modes: one bug in Qwen3-VL-Moe, the other a from_pretrained error in Qwen3.5 Moe." }, { - "left": "issue:44038", - "right": "issue:45478", + "left": "issue:43976", + "right": "issue:45446", "accept": false, - "reason": "Both involve Qwen3-family v5 issues, but one is Qwen3-VL-Moe and the other is Qwen3.5-Moe from_pretrained loading; not the same concrete bug." + "reason": "Python-version incompatibility is a different problem from a flex_attention import/version check bug." }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "VRAM leakage in multithreaded loading and bfloat16 dtype mismatches are different failure modes." + "reason": "VRAM leak in Qwen3ForCausalLM is unrelated to bfloat16 dtype mismatches in different models." }, { - "left": "issue:44625", - "right": "issue:46002", + "left": "issue:45357", + "right": "issue:45478", "accept": false, - "reason": "num_labels propagation in config is unrelated to NaNs appearing in classification heads." + "reason": "Both touch Qwen3.5 model handling, but save_pretrained key mapping and from_pretrained loading failure are distinct bugs." }, { - "left": "issue:45357", - "right": "issue:45478", + "left": "issue:44625", + "right": "issue:46002", "accept": false, - "reason": "Both are Qwen3.5 save/load regressions, but save_pretrained key mapping and from_pretrained loading errors are distinct bugs." + "reason": "num_labels propagation is a config-wiring bug, not the same as NaNs in classification-head initialization." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "A helper length-computation issue in qwen3_omni_moe is unrelated to an import/version-check problem in flex_attention." + "reason": "A helper length-calculation bug in qwen3_omni_moe is unrelated to a PyTorch version-check import issue." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 configuration settings and a missing image-patch helper in GLM46V are unrelated." + "reason": "TF32 API guidance and missing image-patch-count method are unrelated." }, { "left": "issue:43618", "right": "issue:46829", "accept": false, - "reason": "Both concern output dataclasses, but they affect different model outputs and fields; not the same bug." - }, - { - "left": "issue:44075", - "right": "issue:44936", - "accept": false, - "reason": "Unused SGD args in the optimizer and trainer.evaluate() failing after train() are different training bugs." + "reason": "Both are output-field regressions, but they affect different model output classes and are not clearly the same code-path bug." }, { "left": "issue:43116", "right": "issue:43825", "accept": false, - "reason": "Multi-label example-script failure and a pipeline() error-message regression are unrelated." + "reason": "A bad example-script result is not the same as an incorrect pipeline error message about translation support." }, { - "left": "issue:42222", - "right": "issue:43441", + "left": "issue:44075", + "right": "issue:44936", "accept": false, - "reason": "Vitpose breakage and Ministral FlashAttention failures are completely different model-specific issues." + "reason": "Optimizer argument handling and trainer evaluate-after-train failure are different issues." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "Missing embed_tokens.weight in UMT5 loading is unrelated to AutoProcessor not forwarding kwargs to cached_file." + "reason": "A missing embed_tokens weight in UMT5Encoder loading is unrelated to AutoProcessor kwargs forwarding." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "VitPose breakage and Ministral FlashAttention failure are different model-specific regressions." }, { "left": "issue:44995", "right": "issue:46697", "accept": false, - "reason": "A stale indexer cache crash in GlmMoeDsa is unrelated to Swin2SR image over-padding." + "reason": "A stale GlmMoeDsa indexer cache bug is unrelated to Swin2SR image over-padding." }, { "left": "issue:43066", "right": "issue:46710", "accept": false, - "reason": "Tokenizer decoder-type regression and a DeepSeek accuracy regression are different problems despite both being v5-related." + "reason": "Tokenizer decoder-type regression is not the same as a downstream accuracy regression in DeepSeek-R1-Distill-Llama-8B." }, { "left": "issue:28348", "right": "issue:46640", "accept": false, - "reason": "A feature request for GPT2 FlashAttention support is not the same as a generic Flash Attention 2 issue." + "reason": "One is a feature request for GPT2 flash-attention support; the other is an unspecified Flash Attention 2 issue." }, { - "left": "issue:46614", - "right": "issue:46752", + "left": "issue:44263", + "right": "issue:46762", "accept": false, - "reason": "Stale espeak backend state in phonemize is unrelated to apply_chat_template crashing on an empty list." + "reason": "torch.split return handling in GlmMoeDsaIndexer is unrelated to Minimax MSA numerical-difference regressions." }, { - "left": "issue:44263", - "right": "issue:46762", + "left": "issue:46614", + "right": "issue:46752", "accept": false, - "reason": "GlmMoeDsa torch.split() behavior and Minimax attention numerical differences are different model/code-path issues." + "reason": "Stale phonemizer language state and empty-list chat-template IndexError are unrelated bugs." }, { "left": "issue:45092", "right": "issue:46762", "accept": false, - "reason": "InternVL2 meta-init incompatibility is unrelated to Minimax attention numerical discrepancies." + "reason": "Remote-code checkpoint/meta-init incompatibility is unrelated to attention numerical differences." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Docs still mentioning pipeline() and a flex_attention import version check are not the same bug." - }, - { - "left": "issue:43064", - "right": "issue:43673", - "accept": false, - "reason": "Distributed optimizer state corruption in training and missing GenerationMixin cache in chunked prefill are unrelated." - }, - { - "left": "issue:42673", - "right": "issue:42796", - "accept": false, - "reason": "A VRAM leak in Qwen3ForCausalLM is unrelated to a tokenizer BPE-marker handling bug in Ministral." - }, - { - "left": "issue:43937", - "right": "issue:45563", - "accept": false, - "reason": "Invalid GenerationConfig and a stale num_return_sequences warning are different generation-path issues." - }, - { - "left": "issue:33453", - "right": "issue:39401", - "accept": false, - "reason": "Generic tokenizer loading regression and Qwen3 offset_mapping errors are related only broadly; not the same concrete bug." - }, - { - "left": "issue:44380", - "right": "issue:46077", - "accept": false, - "reason": "GPT2 attention scaling being ignored and Gemma4 bidirectional masks still being causal are different attention bugs in different models." + "reason": "Docs still referencing pipeline() after removals is unrelated to the flex_attention import/version-check bug." } ] }, @@ -15352,15 +15349,17 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ + "issue:33453", "issue:38617", "issue:39401", "issue:41628", "issue:42222", - "issue:42371", "issue:42617", "issue:42673", + "issue:42796", + "issue:43064", "issue:43122", "issue:43408", "issue:43450", @@ -15369,12 +15368,11 @@ "issue:43937", "issue:43986", "issue:44117", - "issue:44261", "issue:44291", + "issue:44380", "issue:44410", "issue:44466", "issue:44484", - "issue:44496", "issue:44779", "issue:44829", "issue:44877", @@ -15388,67 +15386,66 @@ "issue:45362", "issue:45405", "issue:45446", - "issue:45478", "issue:45561", + "issue:45563", "issue:45588", "issue:46002", + "issue:46077", "issue:46178", - "issue:46209", "issue:46231", "issue:46298", "issue:46328", "issue:46512", "issue:46614", - "issue:46617", "issue:46655", "issue:46752" ], "soft_pairs": [ + "issue:43064|issue:43673", + "issue:42673|issue:42796", + "issue:43937|issue:45563", + "issue:33453|issue:39401", "issue:45356|issue:46614", "issue:43673|issue:46328", + "issue:44380|issue:46077", "issue:42673|issue:46655", - "issue:44291|issue:45071", "issue:42222|issue:46752", + "issue:44291|issue:45071", "issue:45405|issue:45446", - "issue:44117|issue:45588", "issue:44466|issue:46002", - "issue:44410|issue:45310", + "issue:44117|issue:45588", "issue:39401|issue:44779", + "issue:44410|issue:45310", "issue:43937|issue:44977", "issue:45071|issue:46512", + "issue:45341|issue:46328", "issue:44829|issue:45216", "issue:45341|issue:45561", - "issue:45341|issue:46328", "issue:43450|issue:43986", "issue:44117|issue:45081", "issue:42617|issue:45341", "issue:38617|issue:41628", "issue:43122|issue:46298", - "issue:45362|issue:46328", "issue:43408|issue:43742", + "issue:45362|issue:46328", "issue:44877|issue:46231", "issue:42222|issue:46178", - "issue:44484|issue:45310", - "issue:42371|issue:44261", - "issue:39401|issue:46209", - "issue:43937|issue:44496", - "issue:44410|issue:45478", - "issue:42371|issue:46617" + "issue:44484|issue:45310" ], - "prepared_review_unit_hash": "934391b43d37b8a9823ddc2ba968c4506460367e6e948180fc44ae5701ba2d0c", + "prepared_review_unit_hash": "99973ae48a4e236842c23c33e178d5946e5cf8a981849a4894d17f6686e7ad24", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36689, + "serialized_chars": 36692, "estimated_input_tokens": 9173, "estimated_eval_tokens": 18602 }, @@ -15457,192 +15454,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The proposed soft links are mostly loose lexical matches across different subsystems, models, or failure modes. I would reject all of them; this cluster does not look like one duplicate family. If forced to name a representative issue, #45310 is the strongest issue-style report in the set.", - "confidence": 0.62, - "canonical_issue_reason": "#45310 is the clearest, most specific user-facing regression report in the set: it names a concrete model family, version gate, and loading failure. It also sits in the most obviously duplicate-looking subthread among the Qwen3.5 MoE reports.", + "summary": "This cluster is mostly a loose set of distinct Transformers issues around tokenizers, model loading, generation, and testing utilities. I did not find any soft pair that looks like the same underlying bug/change.", + "confidence": 0.36, + "canonical_issue_reason": "issue:43122 is the most representative broad regression here: a generic tokenizer behavior change across versions, with some nearby tokenizer issues but no exact duplicate in the provided set.", "canonical_pr_reason": null, - "best_issue_reason": "#45310 is the best representative issue because it is narrowly scoped, versioned, and actionable. The rest of the items are too heterogeneous to serve as a better cluster anchor.", + "best_issue_reason": "issue:43122 is the best global issue anchor because it is broad, version-related, and close to the main tokenizer-regression theme that many other items in the cluster orbit around.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Different areas and failure modes: FSDP/PEFT optimizer state handling vs generation cache missing during chunked prefill." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "Unrelated bugs: VRAM leak in Qwen3ForCausalLM multithreaded dataloaders vs tokenizer BPE-marker handling." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Both mention generation, but one is invalid GenerationConfig for GLM-5 and the other is a stale warning about num_return_sequences." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer issues, but one is a generic loading regression and the other is Qwen3 offset_mapping correctness; different bugs." + }, { "left": "issue:45356", "right": "issue:46614", "accept": false, - "reason": "Different tokenizer stacks and different bug classes: codec/warning regression vs stale phonemizer backend state." + "reason": "Different tokenizers and code paths: Kimi-K2.5 codec/fix_mistral_regex regression vs Wav2Vec2 phonemizer backend state bug." }, { "left": "issue:43673", "right": "issue:46328", "accept": false, - "reason": "Generation cache missing during chunked prefill is unrelated to a padding performance issue in SequenceFeatureExtractor." + "reason": "Generation cache bug vs SequenceFeatureExtractor.pad performance issue; not the same underlying defect." }, { - "left": "issue:42673", - "right": "issue:46655", + "left": "issue:44380", + "right": "issue:46077", "accept": false, - "reason": "VRAM leak in Qwen3ForCausalLM is a different problem from incorrect video padding calculation in SmolVLM." + "reason": "Both are attention-related configs, but one ignores GPT2 attention scaling under SDPA/FlashAttention while the other builds the wrong mask for Gemma4 bidirectional attention." }, { - "left": "issue:44291", - "right": "issue:45071", + "left": "issue:42673", + "right": "issue:46655", "accept": false, - "reason": "One is an unexpected init_empty_weights argument error; the other is a PretrainedConfig type-checking regression." + "reason": "Different subsystems: VRAM leak in model threading vs video processor padding calculation." }, { "left": "issue:42222", "right": "issue:46752", "accept": false, - "reason": "VitPose model breakage and apply_chat_template empty-list crash are unrelated code paths." + "reason": "Vitpose model breakage is unrelated to apply_chat_template crashing on an empty list." }, { - "left": "issue:45405", - "right": "issue:45446", + "left": "issue:44291", + "right": "issue:45071", "accept": false, - "reason": "A dependency version bump issue is not the same underlying bug as a PyTorch version guard in flex_attention." + "reason": "Both are v5 compatibility regressions, but one is unexpected _is_hf_initialized in init_empty_weights and the other is PretrainedConfig type-checking behavior." }, { - "left": "issue:44117", - "right": "issue:45588", + "left": "issue:45405", + "right": "issue:45446", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES returning None is unrelated to an AttributeError on s_aux in flash_attention integration." + "reason": "Version-check issues in different dependencies/code paths: PEFT minimum version bump vs PyTorch version check for flex_attention AuxRequest." }, { "left": "issue:44466", "right": "issue:46002", "accept": false, - "reason": "Serialization of tied lm_head weights is a different issue from NaNs appearing in classification heads." + "reason": "Serialization of tied weights depending on device vs NaNs in classification heads after checkout/init are different bugs." }, { - "left": "issue:44410", - "right": "issue:45310", + "left": "issue:44117", + "right": "issue:45588", "accept": false, - "reason": "These are different Qwen model families with different failures: missing projections vs from_pretrained load error." + "reason": "TOKENIZER_MAPPING_NAMES None handling is unrelated to a flash_attention.py crash on s_aux=None." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer issues, but they affect different models and different tokenization defects." + "reason": "Both are tokenizer regressions, but they involve different model families and distinct symptoms (offset mapping vs incorrect tokenization results)." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing qwen3next projections and Qwen3.5 MoE from_pretrained error are separate loading/model-definition issues." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "GenerationConfig validation failure is not the same as a flash-attention generation correctness bug." + "reason": "Different generation bugs: invalid GenerationConfig for GLM-5 vs Qwen3.5 flash-attention generation failure." }, { "left": "issue:45071", "right": "issue:46512", "accept": false, - "reason": "Type checking in PretrainedConfig and stale config.dtype after from_config are distinct configuration regressions." + "reason": "PretrainedConfig type-checking breakage is not the same as stale dtype propagation from from_config(dtype=...)." }, { - "left": "issue:44829", - "right": "issue:45216", + "left": "issue:45341", + "right": "issue:46328", "accept": false, - "reason": "Degenerate training with flash_attention_3 is not the same underlying bug as incorrect save_pretrained checkpoint serialization." + "reason": "One is a vague testing_utils bug report; the other is a specific SequenceFeatureExtractor.pad inefficiency in array conversion." }, { - "left": "issue:45341", - "right": "issue:45561", + "left": "issue:44829", + "right": "issue:45216", "accept": false, - "reason": "A generic testing_utils bug and an xdist race on captured_info.txt are not the same concrete defect." + "reason": "Different failure modes: flash_attention_3 causes degenerate training vs save_pretrained saving an incorrect Qwen3.5 checkpoint." }, { "left": "issue:45341", - "right": "issue:46328", + "right": "issue:45561", "accept": false, - "reason": "Testing utility behavior and SequenceFeatureExtractor padding performance are unrelated." + "reason": "The second is a specific xdist race on captured_info.txt; the first is too generic and does not clearly describe the same bug." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Incorrect batched video processor shapes and missing torchvision crash are separate video-processing failures." + "reason": "Video processor batch shape bug vs AutoProcessor crash without torchvision are related broadly to video, but not the same underlying defect." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "A tokenizer mapping lookup bug is unrelated to the Mistral regex patch AttributeError." + "reason": "Different tokenizer-loading failures: missing mapping entry handling vs Mistral regex patch crashing on backend_tokenizer absence." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "A failure to run 3d_parallel.py does not match a testing_utils bug." + "reason": "3d_parallel.py runtime failure is unrelated to a testing_utils bug." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols and different API surfaces." + "reason": "Both are import errors, but they concern different symbols and code paths; no sign of one shared regression." }, { "left": "issue:43122", "right": "issue:46298", "accept": false, - "reason": "Tokenizer behavior regression and noisy logging message are not the same issue." + "reason": "Tokenizer behavior regression vs noisy logging message; clearly unrelated." }, { - "left": "issue:45362", - "right": "issue:46328", + "left": "issue:43408", + "right": "issue:43742", "accept": false, - "reason": "Qwen3.5 chat crashes and feature extractor padding inefficiency are unrelated." + "reason": "SAM3 tracker/video warning mismatch is unrelated to MobileLLM-125M key-loading failure." }, { - "left": "issue:43408", - "right": "issue:43742", + "left": "issue:45362", + "right": "issue:46328", "accept": false, - "reason": "A sam3 model-type warning is unrelated to a key error while loading MobileLLM-125M." + "reason": "Qwen3.5 chat crash and SequenceFeatureExtractor.pad inefficiency are unrelated." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Strict config loading for granite_speech and a docstring parameter name mismatch are unrelated." + "reason": "Strict config loading for granite_speech is unrelated to a documentation parameter-name typo." }, { "left": "issue:42222", "right": "issue:46178", "accept": false, - "reason": "VitPose breakage and BACKENDS_MAPPING rejecting tensorflow_text are different problem areas." + "reason": "Vitpose model breakage and BACKENDS_MAPPING tensorflow_text error are different issues." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "save_pretrained shard-size behavior is unrelated to the Qwen3.5 MoE from_pretrained error." - }, - { - "left": "issue:42371", - "right": "issue:44261", - "accept": false, - "reason": "TF32 API settings guidance and a q_a_layernorm precision/config bug are not the same defect." - }, - { - "left": "issue:39401", - "right": "issue:46209", - "accept": false, - "reason": "A tokenizer offset_mapping bug is unrelated to FP8 inference failing on B200." - }, - { - "left": "issue:43937", - "right": "issue:44496", - "accept": false, - "reason": "Invalid GenerationConfig and unrecognized model/config.json are different loading/validation problems." - }, - { - "left": "issue:44410", - "right": "issue:45478", - "accept": false, - "reason": "Different Qwen3-related model-loading failures; the titles overlap only at a very broad model-family level." - }, - { - "left": "issue:42371", - "right": "issue:46617", - "accept": false, - "reason": "TF32 behavior settings and trust_remote_code cache loading FileNotFoundError are unrelated." + "reason": "save_pretrained shard-size question is unrelated to Qwen3.5 MoE from_pretrained loading errors." } ] }, @@ -15664,12 +15661,14 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:27957", "issue:38617", + "issue:39401", "issue:41628", "issue:42222", + "issue:42371", "issue:43012", "issue:43066", "issue:43097", @@ -15677,15 +15676,14 @@ "issue:43352", "issue:43441", "issue:43452", - "issue:43531", "issue:43577", "issue:43673", "issue:43874", "issue:43931", + "issue:43937", "issue:43986", "issue:44038", "issue:44075", - "issue:44117", "issue:44220", "issue:44261", "issue:44263", @@ -15695,31 +15693,35 @@ "issue:44484", "issue:44485", "issue:44493", + "issue:44496", "issue:44829", "issue:44933", - "issue:44987", "issue:45003", "issue:45071", "issue:45278", "issue:45341", - "issue:45440", "issue:45464", "issue:45478", "issue:45542", - "issue:45588", "issue:45656", "issue:45753", - "issue:46123", "issue:46172", + "issue:46209", "issue:46231", "issue:46369", + "issue:46617", "issue:46762" ], "soft_pairs": [ + "issue:42371|issue:44261", + "issue:39401|issue:46209", + "issue:43937|issue:44496", + "issue:44410|issue:45478", + "issue:42371|issue:46617", "issue:44261|issue:46762", "issue:38617|issue:44220", - "issue:44933|issue:45278", "issue:44075|issue:45656", + "issue:44933|issue:45278", "issue:44263|issue:44485", "issue:43066|issue:43122", "issue:43097|issue:44493", @@ -15732,8 +15734,8 @@ "issue:44483|issue:45464", "issue:44263|issue:46172", "issue:44360|issue:46172", - "issue:43012|issue:43577", "issue:43874|issue:44261", + "issue:43012|issue:43577", "issue:27957|issue:43352", "issue:43931|issue:44038", "issue:43441|issue:44829", @@ -15742,240 +15744,228 @@ "issue:43986|issue:45341", "issue:44933|issue:46231", "issue:44261|issue:44485", - "issue:45753|issue:46369", - "issue:44117|issue:44987", - "issue:43531|issue:45341", - "issue:43874|issue:44263", - "issue:43673|issue:46123", - "issue:45440|issue:46172", - "issue:43097|issue:45588" + "issue:45753|issue:46369" ], - "prepared_review_unit_hash": "265e348a2fdfa15461acf19b9a9e14494cf219798d978e07c0a30db99114af07", + "prepared_review_unit_hash": "c54c6bf4aa66979e2ce519b3ed53d8dc848ca7e160505ee92a67d88d3808bfb5", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37108, - "estimated_input_tokens": 9277, - "estimated_eval_tokens": 18810 + "soft_pair_count": 32, + "serialized_chars": 36741, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is noisy and mostly \u0441\u043e\u0441\u0442\u043e\u0438\u0442 of unrelated Transformers regressions across models, tokenizers, imports, attention backends, and generation code. None of the soft pairs look like the same concrete bug/change, so no duplicate merges are warranted.", - "confidence": 0.84, - "canonical_issue_reason": "Issue 45278 is the broadest representative of the set: it describes the recurring import-regression pattern after upgrading to v5.5.0, which is the closest shared theme among the candidates.", + "summary": "This cluster is mostly a grab bag of Transformers regressions across v5: import/API breakages, tokenizer changes, model-specific loading bugs, and a few unrelated numerical/runtime issues. I don\u2019t see any soft pair that clearly describes the same underlying bug or a mergeable duplicate PR-sized change.", + "confidence": 0.93, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here: it describes multiple import failures after upgrading to v5.5.0, which is the closest fit for the cluster\u2019s import/API regression subset.", "canonical_pr_reason": null, - "best_issue_reason": "45278 is the most general and useful umbrella issue in the cluster; the other issues are narrower, model-specific, or orthogonal regressions.", + "best_issue_reason": "issue:45278 is the best single representative because it is the most general and central regression report among these issues; the rest are narrower symptom-specific bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API warning/deprecation vs MLA layernorm epsilon precision bug; different subsystem and failure mode." + }, + { + "left": "issue:39401", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer offset mapping bug vs FP8 inference failure on B200; unrelated behavior and code path." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "GLM-5 generation config validation error vs unrecognized model/config loading error; different root causes." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Missing qkv/gate projections in qwen3next layer 0 vs Qwen3.5 Moe from_pretrained error; distinct model families and symptoms." + }, + { + "left": "issue:42371", + "right": "issue:46617", + "accept": false, + "reason": "TF32 settings warning vs cache FileNotFoundError for trust_remote_code models; not the same bug." + }, { "left": "issue:44261", "right": "issue:46762", "accept": false, - "reason": "Both concern attention/numerical behavior, but they describe different models and different failure modes; not the same underlying bug." + "reason": "Layernorm epsilon precision issue vs numerical differences in a different attention implementation; same broad area, but not the same concrete bug." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "Different import/runtime failures in unrelated code paths (`layer_type_validation` vs audio fbank extraction)." + "reason": "Missing import from configuration_utils vs audio fbank feature extraction issue; unrelated code paths." }, { - "left": "issue:44933", - "right": "issue:45278", + "left": "issue:44075", + "right": "issue:45656", "accept": false, - "reason": "Both are import-related, but one is a specific missing symbol from `image_utils` and the other is a broad upgrade breakage; not a single concrete bug." + "reason": "Optimizer SGD args ignored vs optimizer step called twice with deepspeed; different optimizer behavior." }, { - "left": "issue:44075", - "right": "issue:45656", + "left": "issue:44933", + "right": "issue:45278", "accept": false, - "reason": "Both touch optimizers, but one is about SGD arguments being ignored and the other about Deepspeed causing double stepping." + "reason": "Nonexistent image_utils import and broad import-error report overlap thematically, but the specific missing symbol and upgrade scope are not the same issue." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "Different GLM-related bugs: DSA indexer split semantics vs RoPE implementation." + "reason": "torch.split return values bug in GlmMoeDsaIndexer vs GLM-5 RoPE implementation; both GLM-related but different concrete failures." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both are tokenizer regressions, but one is about decoder type and the other about changed tokenization output; too broad to merge." + "reason": "Wrong tokenizer decoder type vs different tokenization output across versions; related area, but one is type selection and the other is behavioral regression." }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Unrelated regressions: removed config API vs unexpected position-id keys." + "reason": "Removed tie_embeddings_and_encoder_decoder API vs unexpected position-id keys in model outputs; unrelated breakages." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "Generation cache regression vs a small testing utility bug; unrelated." + "reason": "Generation cache missing during chunked prefill vs a testing_utils bug; no shared underlying defect." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "Different subsystems: missing `AutoImageProcessor` import vs `PreTrainedConfig` type-checking breakage." + "reason": "AutoImageProcessor import failure vs PreTrainedConfig type checking breakage; both API regressions, but not the same concrete one." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Both are Qwen-family model issues, but one is a weight-shape mismatch and the other missing projections in layer 0; not the same bug." + "reason": "Qwen3-VL-30B-A3B-Instruct weight-shape mismatch vs qwen3next missing projections; different model load failures." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "VitPose model breakage and TensorFlow backend detection are unrelated." + "reason": "VitPose model breakage vs TensorBoard/TensorFlow backend detection bug; unrelated subsystems." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "Type-checking regression vs backend import error; no shared code path." + "reason": "Config type-checking regression vs undefined tf backend when only tensorboard is installed; different API paths." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "Completely different failures: sharding-size question vs Qwen3.5 MoE loading error." + "reason": "save_pretrained sharding-size question vs Qwen3.5 Moe from_pretrained error; not the same bug." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "Both involve chat/completions, but one is a streaming API failure and the other is a request-acceptance regression; not clearly the same bug." + "reason": "chat/completions endpoint rejection vs streaming inference failure on Qwen3.5-0.8B; both API-level, but distinct failures." }, { "left": "issue:44263", "right": "issue:46172", "accept": false, - "reason": "Different model internals and different numerical/shape issues." + "reason": "GlmMoeDsaIndexer split behavior vs deepseek-v4 dtype/transpose bug; different model internals." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "DSA indexer ReLU omission and DeepSeek V4 transpose bug are unrelated." + "reason": "DSA indexer missing ReLU vs deepseek-v4 comb transpose bug; unrelated implementation issues." }, { - "left": "issue:43012", - "right": "issue:43577", + "left": "issue:43874", + "right": "issue:44261", "accept": false, - "reason": "Different dtype/precision problems in different workflows." + "reason": "Missing get_number_of_image_patches in image processor vs MLA layernorm eps precision bug; different components." }, { - "left": "issue:43874", - "right": "issue:44261", + "left": "issue:43012", + "right": "issue:43577", "accept": false, - "reason": "Missing image-patch helper vs MLA layernorm epsilon mismatch; unrelated." + "reason": "PyTorch warning when compiling bf16 models vs BLIP2 dtype remaining float32; related only by dtype, not the same defect." }, { "left": "issue:27957", "right": "issue:43352", "accept": false, - "reason": "Both mention Flash Attention 2, but they are model-specific support issues for different architectures." + "reason": "XLMRoberta Flash Attention 2 support vs NemotronHForCausalLM Flash Attention 2.0 support; same feature area, but separate model-specific support requests." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Both involve Qwen3-VL, but one is a weight-shape mismatch and the other a broader model-loading failure; not the same concrete fix." + "reason": "Qwen3-VL weight mismatch vs Qwen3-VL-Moe bug; both VL models, but the reported load failures are different." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "Different models and different FlashAttention 3 problems; no clear shared bug." + "reason": "Ministral-3 FlashAttention failure vs flash_attention_3 training degeneration; both attention-related, but different code paths and symptoms." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "Config API removal and gguf loading breakage are unrelated." + "reason": "Removed config API vs gguf_file loading breakage; unrelated regression types." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "Unsafe `sys.modules` access and a testing utility bug are unrelated." + "reason": "Unsafe sys.modules access in modeling_utils vs testing_utils bug; different modules and bug classes." }, { "left": "issue:43986", "right": "issue:45341", "accept": false, - "reason": "Video model / torchvision dependency crash vs testing utility bug; unrelated." + "reason": "AutoProcessor crash without torchvision vs testing_utils bug; no shared root cause." }, { "left": "issue:44933", "right": "issue:46231", "accept": false, - "reason": "Missing import and documentation parameter mismatch are unrelated." + "reason": "Missing image_utils import vs docs mismatch for set_training(); one is a runtime import error, the other documentation only." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Both mention GLM/MLA-related math, but they are different bugs in different components." + "reason": "MLA layernorm epsilon precision issue vs GLM-5 RoPE implementation; both numeric, but not the same bug." }, { "left": "issue:45753", "right": "issue:46369", "accept": false, - "reason": "Infinite-loop image generation issue vs memory overflow in a different model; not the same bug." - }, - { - "left": "issue:44117", - "right": "issue:44987", - "accept": false, - "reason": "Both are from_pretrained failures, but the root causes are different and model-specific." - }, - { - "left": "issue:43531", - "right": "issue:45341", - "accept": false, - "reason": "Qwen3-MoE sliding-window issue and a testing utility bug are unrelated." - }, - { - "left": "issue:43874", - "right": "issue:44263", - "accept": false, - "reason": "Different GLM-related helper bugs in different code paths." - }, - { - "left": "issue:43673", - "right": "issue:46123", - "accept": false, - "reason": "Cache handling in generation and MaskGenerationPipeline batch-finalization are unrelated." - }, - { - "left": "issue:45440", - "right": "issue:46172", - "accept": false, - "reason": "Both are DeepSeek-family model issues, but they describe different implementations and different numerical mismatches." - }, - { - "left": "issue:43097", - "right": "issue:45588", - "accept": false, - "reason": "Removed tie-embeddings config API and a flash-attention crash on sink-less models are unrelated." + "reason": "Qwen3_5 infinite loop on a specific image vs sam3-litetext memory overflow; entirely different failure modes." } ] }, @@ -15997,7 +15987,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:38617", "issue:42757", @@ -16005,24 +15995,20 @@ "issue:42907", "issue:43012", "issue:43064", - "issue:43066", "issue:43097", - "issue:43208", - "issue:43352", "issue:43502", - "issue:43550", - "issue:43825", + "issue:43531", + "issue:43673", "issue:43873", - "issue:43906", - "issue:43976", + "issue:43874", "issue:43986", "issue:43992", "issue:44038", "issue:44062", "issue:44117", + "issue:44263", "issue:44315", "issue:44351", - "issue:44368", "issue:44380", "issue:44410", "issue:44484", @@ -16030,12 +16016,16 @@ "issue:44704", "issue:44936", "issue:44938", + "issue:44987", "issue:45103", + "issue:45341", "issue:45372", "issue:45375", - "issue:45712", + "issue:45440", + "issue:45588", "issue:45753", "issue:46050", + "issue:46123", "issue:46172", "issue:46178", "issue:46298", @@ -16049,11 +16039,17 @@ "issue:46821" ], "soft_pairs": [ + "issue:44117|issue:44987", + "issue:43531|issue:45341", + "issue:43874|issue:44263", + "issue:45440|issue:46172", + "issue:43673|issue:46123", + "issue:43097|issue:45588", "issue:44410|issue:45375", "issue:44038|issue:44410", - "issue:43986|issue:45372", "issue:38617|issue:42757", "issue:43064|issue:44315", + "issue:43986|issue:45372", "issue:43097|issue:46050", "issue:43986|issue:44484", "issue:46655|issue:46664", @@ -16065,257 +16061,223 @@ "issue:43012|issue:46821", "issue:44062|issue:44117", "issue:44117|issue:46701", - "issue:44936|issue:46365", "issue:43502|issue:43992", + "issue:44936|issue:46365", "issue:44704|issue:45103", "issue:44380|issue:46421", "issue:44938|issue:46178", "issue:43986|issue:44117", "issue:42907|issue:44655", - "issue:42907|issue:45753", - "issue:43825|issue:44655", - "issue:43208|issue:43550", - "issue:43012|issue:43550", - "issue:45712|issue:46298", - "issue:43012|issue:44368", - "issue:43992|issue:44484", - "issue:43066|issue:43906", - "issue:43352|issue:43976", - "issue:43352|issue:44380", - "issue:43097|issue:46298" + "issue:42907|issue:45753" ], - "prepared_review_unit_hash": "1491e9dd80b00b639df47e95e79d2c537ac088f1c77fee8bcd4318c5ef398b47", + "prepared_review_unit_hash": "60cbce8c38a3bce797a88feb3491014384f683070ab0e785364a4185176f477f", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 34, - "serialized_chars": 37771, - "estimated_input_tokens": 9443, - "estimated_eval_tokens": 19142 + "soft_pair_count": 30, + "serialized_chars": 36464, + "estimated_input_tokens": 9116, + "estimated_eval_tokens": 18488 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This looks like a noisy mix of unrelated Transformers issues rather than a true duplicate cluster. I rejected every soft edge; the only strong commonality is broad subsystem overlap, not the same underlying bug/change.", - "confidence": 0.89, - "canonical_issue_reason": "No single canonical issue stands out: the items span unrelated bugs across imports, model configs, training, serialization, processors, and backend behavior.", + "summary": "This cluster is highly heterogeneous: it mixes unrelated import errors, model-specific regressions, trainer/distributed bugs, processor issues, and deprecation complaints. None of the soft-edge pairs look like the same underlying defect, so no pair should be merged as duplicates. If one issue must anchor the cluster, the broadest and most user-visible is the v5 breaking-change report around `tie_embeddings_and_encoder_decoder` removal.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43097 is the strongest representative: it describes a concrete v5 breaking change with broader user impact and inbound references, unlike the other narrowly scoped or one-off model/runtime bugs.", "canonical_pr_reason": null, - "best_issue_reason": "#43502 is the clearest standalone report here: it describes a concrete, user-facing bug with a straightforward failure mode (API calls happen even with local_files_only=True).", + "best_issue_reason": "issue:43097 is the best global issue candidate because it is a clear, high-impact regression/deprecation complaint that is more representative of the cluster than the many unrelated niche failures.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Both involve tokenizer/model loading, but one is a mapping-lookup bug and the other is a model load failure; not the same concrete failure path." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Unrelated topics: Qwen3-MoE sliding-window behavior vs a testing_utils bug." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Different GLM-related code paths; image processor multimodal token counting vs a torch.split return-value issue." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Both mention DeepSeek, but they describe different model implementations and different tensor/code-path bugs." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Cache initialization during chunked prefill is not the same as a pipeline batch-finalization bug." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "A deprecation/removal complaint is unrelated to a flash-attention AttributeError on sink-less models." + }, { "left": "issue:44410", "right": "issue:45375", "accept": false, - "reason": "Both are Qwen-related, but one is missing attention projections and the other is a missing config field; different bugs and fixes." + "reason": "Different Qwen3 variants and different missing fields/projections; not one shared bug." }, { "left": "issue:44038", "right": "issue:44410", "accept": false, - "reason": "Same broad model family, but the failures differ: one is a Qwen3-VL-Moe issue, the other is qwen3next layer projection metadata." - }, - { - "left": "issue:43986", - "right": "issue:45372", - "accept": false, - "reason": "Both involve loading processors, but one is a torchvision dependency crash and the other is an import error from mistral_common." + "reason": "Both are Qwen3-family issues, but one is a transformers version bug and the other is a missing projection in layer 0." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are ImportErrors, but for different symbols in different modules; no shared underlying code-path bug is evident." + "reason": "Both are ImportErrors, but for different missing symbols from different packages and likely different fixes." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "Different training-path issues: optimizer state corruption under FSDP2/PEFT versus Liger Kernel not being applied with model_init." + "reason": "Distributed optimizer-state corruption and Liger-kernel application are separate training pipeline problems." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Both are loading crashes, but one is torchvision-related video processor loading and the other is mistral_common/Gemma processor loading." }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "Both touch model saving/loading, but one is about a removed config attribute and the other about broken quantized save/load round-tripping." + "reason": "Saving a dequantized model is a different defect from sinq quantization breakage across save/load." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "Processor loading crash versus save_pretrained shard-size behavior; unrelated bugs." + "reason": "Processor loading without torchvision is unrelated to the `max_shard_size` default question." }, { "left": "issue:46655", "right": "issue:46664", "accept": false, - "reason": "Same SmolVLM video processor area, but padding calculation and num_frames semantics are distinct defects." + "reason": "Same SmolVLM video processor area, but padding math and `num_frames` semantics are distinct bugs." }, { "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "A flaky test missing set_seed is unrelated to a noisy logging message." + "reason": "A flaky test setup issue and a noisy logging message are unrelated." }, { "left": "issue:46298", "right": "issue:46752", "accept": false, - "reason": "Logging noise and an apply_chat_template IndexError are unrelated failures." + "reason": "Noisy logging and an `apply_chat_template` empty-list crash have no shared failure path." }, { "left": "issue:46172", "right": "issue:46710", "accept": false, - "reason": "DeepSeek-V4 tensor operation bug versus an accuracy regression after upgrade; not the same underlying issue." + "reason": "DeepSeek tensor-shape code is unrelated to the reported accuracy regression in a different model family." }, { "left": "issue:38617", "right": "issue:44351", "accept": false, - "reason": "Both are import-related, but they concern different exported names and different components." + "reason": "Different missing imports (`layer_type_validation` vs `HybridCache`) and likely different version-skew fixes." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "One is a removed embedding-tie option in v5; the other is an offloading/quantization behavior bug." + "reason": "A v5 removal/deprecation issue is not the same as offloading failing under quantization." }, { "left": "issue:43012", "right": "issue:46821", "accept": false, - "reason": "A bfloat16 compile warning is unrelated to a TimesFM window_size AttributeError." + "reason": "PyTorch warning behavior for bfloat16 compilation is unrelated to a TimesFM `window_size` AttributeError." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "Both are tokenizer loading bugs, but one is an AddedToken constructor error and the other is a missing TOKENIZER_MAPPING_NAMES assumption." + "reason": "Tokenizer construction error vs tokenizer-mapping `None` handling; different code paths." }, { "left": "issue:44117", "right": "issue:46701", "accept": false, - "reason": "Tokenizer mapping None handling is unrelated to the DiffusionGemma pipeline failure." + "reason": "A tokenizer mapping assumption bug is not the same as a DiffusionGemma image-text pipeline failure." }, { - "left": "issue:44936", - "right": "issue:46365", + "left": "issue:43502", + "right": "issue:43992", "accept": false, - "reason": "Trainer evaluate/train lifecycle failure versus import-time conflict with datasets; different code paths." + "reason": "Both are from_pretrained-related, but one is offline/download policy leakage and the other is a missing weight in UMT5 loading." }, { - "left": "issue:43502", - "right": "issue:43992", + "left": "issue:44936", + "right": "issue:46365", "accept": false, - "reason": "Both happen during loading, but one is offline/file-only network leakage and the other is a missing embed_tokens weight." + "reason": "Trainer evaluate-after-train behavior is unrelated to an import failure caused by datasets." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "AutoProcessor kwargs forwarding and auto_docstring annotation parsing are unrelated bugs." + "reason": "Passing kwargs to cached_file and auto_docstring annotation handling are entirely separate bugs." }, { "left": "issue:44380", "right": "issue:46421", "accept": false, - "reason": "Both mention backend/attention behavior, but one ignores GPT2 attention scaling while the other triggers recompilation via lazy cache init." + "reason": "Attention-scaling config ignored in GPT2 backends is not the same as chunked-prefill cache recompilation." }, { "left": "issue:44938", "right": "issue:46178", "accept": false, - "reason": "Python 3.14 import/load failure and a tensorflow_text backend mapping error are unrelated." + "reason": "Python 3.14 import/load failure is unrelated to the BACKENDS_MAPPING tensorflow_text error." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Different processor-loading and tokenizer-mapping failures; no shared bug." + "reason": "Optional-dependency processor loading and tokenizer-mapping `None` handling are different failures." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Saving dequantized models and saving Pipeline objects are different save_pretrained problems with different affected objects." + "reason": "Saving a dequantized model and saving Pipeline objects involve different serialization paths and bugs." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Dequantized model save/load failure is unrelated to a Qwen3_5 infinite-loop inference bug." - }, - { - "left": "issue:43825", - "right": "issue:44655", - "accept": false, - "reason": "Pipeline task-support messaging and pipeline serialization are different issues." - }, - { - "left": "issue:43208", - "right": "issue:43550", - "accept": false, - "reason": "xLSTM training blockers and a Bamba torch.compile SDPA failure are separate model-specific bugs." - }, - { - "left": "issue:43012", - "right": "issue:43550", - "accept": false, - "reason": "A bfloat16 precision warning is unrelated to a compile-time SDPA failure." - }, - { - "left": "issue:45712", - "right": "issue:46298", - "accept": false, - "reason": "Leftover dummy classes and a noisy logging message are not the same bug." - }, - { - "left": "issue:43012", - "right": "issue:44368", - "accept": false, - "reason": "A compile warning is unrelated to a tie_word_embeddings warning during LoRA fine-tuning." - }, - { - "left": "issue:43992", - "right": "issue:44484", - "accept": false, - "reason": "Missing weight loading and max_shard_size default behavior are unrelated save/load concerns." - }, - { - "left": "issue:43066", - "right": "issue:43906", - "accept": false, - "reason": "Tokenizer decoder type mismatch and a separate reproduced issue are different failures." - }, - { - "left": "issue:43352", - "right": "issue:43976", - "accept": false, - "reason": "Flash Attention support for NemotronH versus Python version compatibility is unrelated." - }, - { - "left": "issue:43352", - "right": "issue:44380", - "accept": false, - "reason": "Both touch attention/flash-attn, but one is unsupported-model gating and the other is GPT2 attention scaling being ignored." - }, - { - "left": "issue:43097", - "right": "issue:46298", - "accept": false, - "reason": "A removed config option and a noisy logging message are unrelated." + "reason": "Model save/load breakage and a Qwen3_5 image infinite loop are unrelated defects." } ] }, @@ -16337,9 +16299,8 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:36010", "issue:36683", "issue:38617", "issue:42175", @@ -16348,51 +16309,61 @@ "issue:42994", "issue:43012", "issue:43064", + "issue:43066", + "issue:43097", "issue:43208", "issue:43299", "issue:43352", - "issue:43475", "issue:43502", "issue:43526", + "issue:43550", "issue:43582", - "issue:43611", "issue:43618", "issue:43756", "issue:43761", "issue:43824", + "issue:43825", "issue:43867", - "issue:43901", - "issue:43937", + "issue:43906", "issue:43976", + "issue:43992", "issue:44038", "issue:44077", "issue:44162", "issue:44222", "issue:44230", + "issue:44368", "issue:44380", "issue:44483", + "issue:44484", "issue:44485", "issue:44514", "issue:44625", "issue:44655", - "issue:44938", - "issue:45216", - "issue:45276", "issue:45478", "issue:45479", "issue:45584", - "issue:46077", - "issue:46082", + "issue:45712", "issue:46132", "issue:46172", - "issue:46365", - "issue:46369" + "issue:46298", + "issue:46365" ], "soft_pairs": [ + "issue:43825|issue:44655", + "issue:43208|issue:43550", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:43012|issue:44368", + "issue:43992|issue:44484", + "issue:43066|issue:43906", + "issue:43352|issue:43976", + "issue:43352|issue:44380", + "issue:43097|issue:46298", "issue:44162|issue:46172", - "issue:44077|issue:46365", "issue:44380|issue:46132", "issue:45584|issue:46365", + "issue:44077|issue:46365", "issue:44222|issue:45478", "issue:36683|issue:43582", "issue:44380|issue:45479", @@ -16404,261 +16375,234 @@ "issue:44483|issue:44514", "issue:43618|issue:44380", "issue:38617|issue:43502", - "issue:43618|issue:43761", "issue:44038|issue:44230", + "issue:43618|issue:43761", "issue:43756|issue:44485", "issue:42994|issue:44655", "issue:43299|issue:43867", - "issue:44625|issue:46365", - "issue:45276|issue:46077", - "issue:36683|issue:43475", - "issue:36010|issue:42757", - "issue:43352|issue:43937", - "issue:44483|issue:46369", - "issue:43611|issue:44655", - "issue:43901|issue:44655", - "issue:43867|issue:45216", - "issue:36683|issue:44077", - "issue:43976|issue:44938", - "issue:44625|issue:46077", - "issue:46077|issue:46082" + "issue:44625|issue:46365" ], - "prepared_review_unit_hash": "9e4f7ad0fac89e700a17f9a588a50253dfadaff5a68a787c76aee06cfe0deb47", + "prepared_review_unit_hash": "562963a2bd591cfbca562c05b6bbdd9c081a2bae40be221f79520531021adfd4", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 37308, - "estimated_input_tokens": 9327, - "estimated_eval_tokens": 18910 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36149, + "estimated_input_tokens": 9038, + "estimated_eval_tokens": 18332 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous issue cluster with similar wording but different bugs across unrelated subsystems. No single canonical issue spans the full set; the only plausible duplicate mini-cluster is the CLIP v5 regression pair.", - "confidence": 0.66, - "canonical_issue_reason": "No single canonical issue fits the full cluster because the items cover unrelated import errors, model regressions, training bugs, and API/docs issues. The only close duplicate signal is a narrow CLIPVisionModel v5 regression overlap.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues. The soft pairs are similar at a surface level (imports, attention, save/load, Qwen/CLIP, etc.) but do not look like the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "issue:46365 is the best representative issue here: it is currently open, user-facing, and describes a concrete import conflict affecting Trainer. The other issues are either narrower, model-specific, already closed, or about different code paths.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43761 is the strongest representative of the only plausible duplicate mini-cluster because it describes a concrete v5 regression in CLIPVisionModel.forward and is broader than the narrower attentions-missing symptom in 43618.", + "best_issue_reason": "issue:46365 is the strongest overall issue to anchor triage because it is broad, actionable, and high-impact. It is a clearer canonical candidate than the other reports in this set, which are mostly isolated regressions or feature-specific bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44162", - "right": "issue:46172", + "left": "issue:43825", + "right": "issue:44655", "accept": false, - "reason": "Different models and failures: ESM2 training breakage vs a DeepSeek-V4 tensor transpose issue." + "reason": "Different failures: one is a pipeline task-support error message, the other is save_pretrained failing for Pipeline objects. Same area, but not the same bug." }, { - "left": "issue:44077", - "right": "issue:46365", + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "Different models and code paths: xLSTM training blockers vs Bamba torch.compile/SDPA failure." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A PyTorch precision warning on bfloat16 compilation is not the same as a model-specific torch.compile SDPA crash." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "One is about leftover dummy classes leaking into the API surface; the other is just a noisy logging message. No shared bug." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Both mention warnings, but they come from different causes: compiler precision behavior vs tie_word_embeddings config warnings in Qwen3.5 LoRA." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "UMT5 missing embed_tokens.weight on load is unrelated to the max_shard_size default in save_pretrained." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type and an isolated reproduction of another issue are not the same underlying defect." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Nemotron Flash Attention gating and Python 3.9/3.10 compatibility are separate issues." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Both touch attention backends, but one is unsupported Flash Attention for a model and the other is GPT2 attention-scaling config being ignored. Different bugs." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "Removal of tie_embeddings_and_encoder_decoder is unrelated to a noisy logging message." + }, + { + "left": "issue:44162", + "right": "issue:46172", "accept": false, - "reason": "Unrelated problems: patchtsmixer post_init handling vs Trainer import failure linked to datasets." + "reason": "ESM2 breakage and a DeepSeek-V4 tensor transpose issue are different model-specific failures." }, { "left": "issue:44380", "right": "issue:46132", "accept": false, - "reason": "Different code paths: GPT2 attention scaling backend behavior vs AttentionInterface.register semantics." + "reason": "GPT2 backend config loss is not the same as AttentionInterface.register changing the registered function behavior." }, { "left": "issue:45584", "right": "issue:46365", "accept": false, - "reason": "Whisper empty-transcription generation bug is unrelated to the Trainer/datasets import problem." + "reason": "Whisper empty-transcription failure after token alignment is unrelated to the Trainer import conflict." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "patchtsmixer post_init validation and importing Trainer with datasets installed are unrelated." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "Same model family, but one is FP8 save_pretrained and the other is from_pretrained loading; not the same concrete bug." + "reason": "Both are MoE/save/load themed, but one is FP8 save_pretrained and the other is a Qwen3.5 Moe from_pretrained regression on newer transformers. Not the same bug." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3Config missing vocab_size and Apple Silicon caching_allocator_warmup TypeError are unrelated." + "reason": "Gemma3Config missing vocab_size and an Apple Silicon TypeError in caching_allocator_warmup are unrelated." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "GPT2 attention scaling backend issue is unrelated to the sequence-classification zero-loss bug." + "reason": "GPT2 attention scaling under SDPA/FlashAttention is unrelated to the degenerate zero-loss classification bug." }, { "left": "issue:42175", "right": "issue:46365", "accept": false, - "reason": "Package/backend dependency issue versus a Trainer import conflict; different failures." + "reason": "TensorFlow backend packaging and a Trainer import conflict involving datasets are different dependency issues." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "Distributed optimizer-state corruption and GPT2 attention scaling are different bugs." + "reason": "FSDP2/PEFT optimizer state corruption is a distributed training bug, not an attention-scaling regression." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but for different symbols in different modules with different fixes." + "reason": "Both are ImportErrors, but they involve different symbols, packages, and likely causes." }, { "left": "issue:42490", "right": "issue:43012", "accept": false, - "reason": "Local path trailing-slash warning and bfloat16 compile warning are unrelated." + "reason": "Local path trailing-slash warning is unrelated to a bfloat16 precision warning during compile." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training blockers and BeitImageProcessorFast label reduction are different subsystems and symptoms." + "reason": "xLSTM training blockers and BeitImageProcessorFast reduce_labels returning one label are unrelated." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "Chat-completions request validation and batched processor padding crash are unrelated API bugs." + "reason": "A chat-completions API rejection is not the same as Qwen2_5_VLProcessor crashing on batched apply_chat_template input." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIP output field assignment regression versus GPT2 attention scaling behavior are unrelated." + "reason": "CLIPOutput attentions assignment and GPT2 attention scaling under backends are different regressions." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Import error from configuration_utils and local_files_only network requests are different issues." - }, - { - "left": "issue:43618", - "right": "issue:43761", - "accept": true, - "reason": "Both describe the same CLIPVisionModel.forward v5 regression where expected outputs are not populated (attentions/hidden_states)." + "reason": "Missing layer_type_validation import and local_files_only still triggering network calls are unrelated import/network bugs." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Both mention Qwen3-VL/Qwen3.5 MoE FP8, but one is a bug report and the other is inference support; not the same bug." + "reason": "One is a general Qwen3-VL-Moe breakage on transformers 5.0, the other is a request for FP8 inference support. Not the same change." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both involve CLIP, but one loses attentions and the other loses hidden_states. Different outputs, different failure modes." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Smollm3 RoPE layer mismatch and GLM-5 RoPE implementation discussion are related only by theme, not the same change." + "reason": "Smollm3 RoPE layer dropping and GLM-5 RoPE implementation are model-specific and not the same bug." }, { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Quantized model saving failure and Pipeline.save_pretrained failure are different serialization bugs." + "reason": "Quantized model saving failure and saving Pipeline objects are both save-related, but they are different object types and failure paths." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Qwen3VL MoE loading regression and sorted state_dict load error do not point to the same code path." + "reason": "Qwen3VL-Moe loading regression and a state_dict sorting load error are not the same underlying issue." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "Qwen3.5 num_labels propagation and Trainer import conflict are unrelated." - }, - { - "left": "issue:45276", - "right": "issue:46077", - "accept": false, - "reason": "Gemma4 resize_token_embeddings propagation and Gemma4 bidirectional attention mask construction are different bugs." - }, - { - "left": "issue:36683", - "right": "issue:43475", - "accept": false, - "reason": "Gemma3Config vocab_size missing and SAM3 output missing fpn_position_embeddings are unrelated." - }, - { - "left": "issue:36010", - "right": "issue:42757", - "accept": false, - "reason": "Both are import errors, but for different symbols from different packages; no shared bug." - }, - { - "left": "issue:43352", - "right": "issue:43937", - "accept": false, - "reason": "Nemotron Flash Attention support and GLM-5 GenerationConfig validation are unrelated." - }, - { - "left": "issue:44483", - "right": "issue:46369", - "accept": false, - "reason": "Chat endpoint validation and a SAM3 litetext memory overflow are unrelated." - }, - { - "left": "issue:43611", - "right": "issue:44655", - "accept": false, - "reason": "base_model_prefix loading regression and Pipeline.save_pretrained failure are different code paths." - }, - { - "left": "issue:43901", - "right": "issue:44655", - "accept": false, - "reason": "Docs mentioning return_all_scores and save_pretrained pipeline persistence are unrelated." - }, - { - "left": "issue:43867", - "right": "issue:45216", - "accept": false, - "reason": "State_dict sorting load error and Qwen3.5 checkpoint regression are different loading/saving bugs." - }, - { - "left": "issue:36683", - "right": "issue:44077", - "accept": false, - "reason": "Gemma3Config attribute error and patchtsmixer post_init restriction are unrelated." - }, - { - "left": "issue:43976", - "right": "issue:44938", - "accept": false, - "reason": "Python version support failure and Python 3.14 load failure are not the same issue." - }, - { - "left": "issue:44625", - "right": "issue:46077", - "accept": false, - "reason": "num_labels propagation and Gemma4 attention-mask behavior are unrelated." - }, - { - "left": "issue:46077", - "right": "issue:46082", - "accept": false, - "reason": "Gemma4 attention-mask construction and LlamaConfig head_dim validation are different configuration bugs." + "reason": "num_labels propagation in Qwen3.5 is unrelated to the Trainer import conflict with datasets." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative on the mostly unrelated pairs. The CLIPVisionModel/CLIPOutput pair is the only credible mini-cluster signal, and the rest of the false verdicts are appropriately cautious." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -16673,58 +16617,70 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:28282", "issue:29870", "issue:33453", + "issue:36010", "issue:36296", "issue:36683", "issue:41669", + "issue:42757", "issue:42890", - "issue:42898", "issue:43010", "issue:43064", "issue:43208", - "issue:43502", + "issue:43352", + "issue:43475", "issue:43525", "issue:43575", + "issue:43611", "issue:43618", "issue:43867", + "issue:43901", + "issue:43937", "issue:43976", "issue:43992", - "issue:44062", "issue:44075", "issue:44077", - "issue:44112", "issue:44246", "issue:44355", - "issue:44360", + "issue:44483", "issue:44492", "issue:44625", - "issue:44734", + "issue:44655", "issue:44855", "issue:44938", - "issue:44964", "issue:45071", "issue:45216", + "issue:45276", "issue:45278", "issue:45310", "issue:45357", "issue:45375", - "issue:45440", "issue:45478", "issue:45561", - "issue:45600", "issue:45850", + "issue:46077", + "issue:46082", "issue:46202", - "issue:46285", "issue:46365", - "issue:46617", - "issue:46664", + "issue:46369", "issue:46829" ], "soft_pairs": [ + "issue:45276|issue:46077", + "issue:36683|issue:43475", + "issue:36010|issue:42757", + "issue:43352|issue:43937", + "issue:44483|issue:46369", + "issue:43611|issue:44655", + "issue:43901|issue:44655", + "issue:43867|issue:45216", + "issue:36683|issue:44077", + "issue:43976|issue:44938", + "issue:44625|issue:46077", + "issue:46077|issue:46082", "issue:43208|issue:46829", "issue:43618|issue:45216", "issue:36683|issue:43525", @@ -16747,237 +16703,246 @@ "issue:43010|issue:43976", "issue:43618|issue:44077", "issue:43992|issue:45478", - "issue:43867|issue:45357", - "issue:28282|issue:33453", - "issue:43618|issue:46285", - "issue:44360|issue:45440", - "issue:42890|issue:44964", - "issue:42890|issue:44112", - "issue:42890|issue:45600", - "issue:42898|issue:46664", - "issue:43502|issue:44062", - "issue:44734|issue:46617" + "issue:43867|issue:45357" ], - "prepared_review_unit_hash": "ca5c0ee1afa9a4e9beaab18592e5842b1e8192b9cf04f170d26e1ead1839e242", + "prepared_review_unit_hash": "3e159928e082047efcf8955938674f73f2b73d94936a85b82c69d9ba37a79027", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36981, - "estimated_input_tokens": 9246, - "estimated_eval_tokens": 18748 + "soft_pair_count": 35, + "serialized_chars": 37819, + "estimated_input_tokens": 9455, + "estimated_eval_tokens": 19166 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a loose mix of unrelated Transformers bug reports, mostly centered on regressions, model-specific config/loading failures, and test flakiness. The soft links are mostly superficial and should not be merged as duplicates. No PRs are present.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 45278 is the broadest and most representative post-upgrade regression report ('Many import errors after update from 4.57.0 to 5.5.0'), so it works best as the nominal issue anchor for this heterogeneous cluster.", + "summary": "This cluster is mostly a loose collection of unrelated Transformers regressions; the soft pairs share broad themes like loading, importing, or config handling, but not the same concrete bug. No soft edge looks safe to merge as a duplicate/combined PR candidate.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45278 is the broadest representative of the set: it captures the general post-upgrade import-regression theme that recurs across several items, though the cluster is still too heterogeneous for a tight duplicate grouping.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45278 is the best overall representative of the cluster because it captures the general version-regression theme more cleanly than the highly model- or subsystem-specific reports.", + "best_issue_reason": "issue:45278 is the best single issue to anchor the cluster because it is the most umbrella-like and closest to the repeated Transformers upgrade breakage reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43208", - "right": "issue:46829", + "left": "issue:45276", + "right": "issue:46077", "accept": false, - "reason": "xLSTM training blockers vs VideoPrism output-field mismatch; different models and failure modes." + "reason": "Different Gemma4 bugs: token embedding resizing vs causal attention mask construction." }, { - "left": "issue:43618", - "right": "issue:45216", + "left": "issue:36683", + "right": "issue:43475", "accept": false, - "reason": "CLIP attentions regression vs Qwen3.5 save_pretrained regression; unrelated code paths." + "reason": "Different models and code paths: Gemma3 config vocab_size vs SAM3 vision output missing field." }, { - "left": "issue:36683", - "right": "issue:43525", + "left": "issue:36010", + "right": "issue:42757", "accept": false, - "reason": "Different config attributes on different model families (Gemma3 vocab_size vs Llama4 pad_token_id)." + "reason": "Both are import errors, but from different packages and causes: GenerationMixin vs huggingface_hub is_offline_mode." }, { - "left": "issue:36683", - "right": "issue:44625", + "left": "issue:43352", + "right": "issue:43937", "accept": false, - "reason": "Gemma3 vocab_size bug and Qwen3.5 num_labels propagation bug are distinct config issues." + "reason": "Unrelated failures: Flash Attention support in Nemotron vs invalid GenerationConfig in GLM-5." }, { - "left": "issue:29870", - "right": "issue:36296", + "left": "issue:44483", + "right": "issue:46369", "accept": false, - "reason": "Bart generate UnboundLocalError vs tensor-parallel training bug; not the same bug." + "reason": "Different problems: chat/completions request handling vs Trainer import being blocked by datasets." }, { - "left": "issue:43064", - "right": "issue:44492", + "left": "issue:43611", + "right": "issue:44655", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption vs a cache-strategy typo; unrelated." + "reason": "Different code paths: model loading with base_model_prefix vs saving Pipeline objects." }, { - "left": "issue:43992", - "right": "issue:45310", + "left": "issue:43901", + "right": "issue:44655", "accept": false, - "reason": "UMT5 missing embed_tokens.weight vs Qwen3.5 Moe from_pretrained error; different model/load paths." + "reason": "Docs mismatch for return_all_scores vs pipeline serialization; not the same bug." }, { - "left": "issue:43010", - "right": "issue:45278", + "left": "issue:43867", + "right": "issue:45216", "accept": false, - "reason": "Cache update no_grad fix vs broad import-errors-after-upgrade report; different symptoms and fixes." + "reason": "Both involve model I/O, but one is a sorted-state_dict load error and the other is a Qwen3.5 save_pretrained regression." }, { - "left": "issue:44075", - "right": "issue:46365", + "left": "issue:36683", + "right": "issue:44077", "accept": false, - "reason": "Unused SGD args vs Trainer import failure with datasets installed; unrelated." + "reason": "Different model/config bugs: Gemma3 vocab_size vs patchtsmixer post_init policy." }, { - "left": "issue:29870", - "right": "issue:33453", + "left": "issue:43976", + "right": "issue:44938", "accept": false, - "reason": "Both are regressions, but Bart generate masking and tokenizer loading are different problems." + "reason": "Both are Python-version compatibility reports, but for different versions and failure modes." }, { - "left": "issue:44355", - "right": "issue:44855", + "left": "issue:44625", + "right": "issue:46077", "accept": false, - "reason": "Compiled-file import errors vs Python 3.13 TorchScript parsing/IndentationError; distinct root causes." + "reason": "Qwen3.5 num_labels propagation is unrelated to Gemma4 attention masking." }, { - "left": "issue:44938", - "right": "issue:45071", + "left": "issue:46077", + "right": "issue:46082", "accept": false, - "reason": "Python 3.14 load failure vs PretrainedConfig type-checking regression; not the same issue." + "reason": "Different config validations: Gemma4 bidirectional attention vs LlamaConfig head_dim constraints." }, { - "left": "issue:43867", - "right": "issue:46202", + "left": "issue:43208", + "right": "issue:46829", "accept": false, - "reason": "State_dict loading error vs Gemma pipeline roundtrip on \u2581; different model behavior." + "reason": "Unrelated model families and symptoms: xLSTM training blockers vs VideoPrism output field mismatch." }, { - "left": "issue:36683", - "right": "issue:43064", + "left": "issue:43618", + "right": "issue:45216", "accept": false, - "reason": "Gemma3 config field missing vs FSDP2/PEFT optimizer-state bug; unrelated." + "reason": "CLIPOutput attentions regression is unrelated to Qwen3.5 checkpoint saving." }, { - "left": "issue:43867", - "right": "issue:45375", + "left": "issue:36683", + "right": "issue:43525", "accept": false, - "reason": "Load-model error from sorted state_dict vs Qwen3_5MoeVisionConfig field omission; different bug classes." + "reason": "Different missing-config-attribute bugs in different models: vocab_size vs pad_token_id." }, { - "left": "issue:42890", - "right": "issue:45561", + "left": "issue:36683", + "right": "issue:44625", "accept": false, - "reason": "Seed-related integration-test flakiness vs xdist race in testing utils; both test issues but not the same bug." + "reason": "Both are config propagation issues, but for different fields and models." }, { - "left": "issue:41669", - "right": "issue:44246", + "left": "issue:29870", + "right": "issue:36296", "accept": false, - "reason": "Import-star performance cleanup vs intermittent slow import; related area, but not the same underlying issue." + "reason": "Bart generate UnboundLocalError is unrelated to the tensor-parallel training bug." }, { - "left": "issue:43010", - "right": "issue:45850", + "left": "issue:43064", + "right": "issue:44492", "accept": false, - "reason": "Cache update decoration bug vs broad 'too many models break' complaint; too generic to be a duplicate." + "reason": "Wrong optimizer states under FSDP2/PEFT is unrelated to a typo in cache strategies." }, { - "left": "issue:36296", - "right": "issue:43575", + "left": "issue:43992", + "right": "issue:45310", "accept": false, - "reason": "Tensor-parallel training correctness bug vs TP out-of-memory on Qwen2-57B; different failure types." + "reason": "Both are from_pretrained-related, but they affect different models and different missing-key behavior." }, { "left": "issue:43010", - "right": "issue:43976", + "right": "issue:45278", "accept": false, - "reason": "Cache no_grad bug vs Python-version incompatibility; unrelated." + "reason": "Cache update no_grad behavior is unrelated to general import errors after upgrade." }, { - "left": "issue:43618", - "right": "issue:44077", + "left": "issue:44075", + "right": "issue:46365", "accept": false, - "reason": "CLIP attentions missing vs patchtsmixer post_init policy change; different models and semantics." + "reason": "Optimizer SGD args not used is unrelated to Trainer import resolution with datasets installed." }, { - "left": "issue:43992", - "right": "issue:45478", + "left": "issue:29870", + "right": "issue:33453", "accept": false, - "reason": "UMT5 encoder weight-missing bug vs Qwen3.5 Moe from_pretrained regression; distinct model families." + "reason": "Different regressions: Bart generate mask error vs tokenizer loading regression." }, { - "left": "issue:43867", - "right": "issue:45357", + "left": "issue:44355", + "right": "issue:44855", "accept": false, - "reason": "Generic load error vs Qwen3.5 save_pretrained saving wrong visual encoder keys; load vs save path mismatch." + "reason": "Both are import-time failures, but one is compiled-file handling and the other is a Python 3.13 torch.jit parse issue." }, { - "left": "issue:28282", - "right": "issue:33453", + "left": "issue:44938", + "right": "issue:45071", "accept": false, - "reason": "Missing PyTorch ImportError vs tokenizer-loading regression; unrelated." + "reason": "Different compatibility breakages: Python 3.14 load failure vs PretrainedConfig type checking." }, { - "left": "issue:43618", - "right": "issue:46285", + "left": "issue:43867", + "right": "issue:46202", "accept": false, - "reason": "CLIP attentions regression vs CLIP text-model structure inconsistency; same family, but different concrete bug." + "reason": "Different model I/O bugs: sorted state_dict load error vs Gemma pipeline roundtrip failure." }, { - "left": "issue:44360", - "right": "issue:45440", + "left": "issue:36683", + "right": "issue:43064", "accept": false, - "reason": "DSA indexer ReLU discussion vs DeepseekV3MoE divergence; unrelated subsystems." + "reason": "Gemma3 config attribute bug is unrelated to FSDP2 optimizer-state corruption." }, { - "left": "issue:42890", - "right": "issue:44964", + "left": "issue:43867", + "right": "issue:45375", "accept": false, - "reason": "SAM HQ test seeding flake vs Phi-4 multimodal loading failure; different artifacts." + "reason": "Both touch config/model loading, but the actual missing-field and save/load issues are different." }, { "left": "issue:42890", - "right": "issue:44112", + "right": "issue:45561", "accept": false, - "reason": "Test nondeterminism due missing set_seed vs GraniteSpeech CI stale device override; different test failures." + "reason": "Both are test flakiness reports, but with different causes: missing set_seed vs xdist file race." }, { - "left": "issue:42890", - "right": "issue:45600", + "left": "issue:41669", + "right": "issue:44246", "accept": false, - "reason": "SAM HQ test flakiness vs auto_mappings referencing removed configs; unrelated." + "reason": "Similar broad import-performance theme, but not enough to show the same concrete underlying cause." }, { - "left": "issue:42898", - "right": "issue:46664", + "left": "issue:43010", + "right": "issue:45850", "accept": false, - "reason": "Tokenization-space behavior change vs SmolVLM num_frames semantics; different domains." + "reason": "A generic complaint that 5.0 breaks many models is too broad and not the same bug as cache update behavior." }, { - "left": "issue:43502", - "right": "issue:44062", + "left": "issue:36296", + "right": "issue:43575", "accept": false, - "reason": "local_files_only being ignored vs AddedToken duplicate 'special' kwarg; different API layers." + "reason": "Both mention tensor parallelism, but one is a training bug and the other is OOM during model loading." }, { - "left": "issue:44734", - "right": "issue:46617", + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache no_grad behavior is unrelated to Python-version compatibility." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different model issues: CLIP attentions regression vs patchtsmixer post_init validation." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Similar title shape, but these are different models and likely different loading failures." + }, + { + "left": "issue:43867", + "right": "issue:45357", "accept": false, - "reason": "KV-cache continuation indexing crash in serve vs trust_remote_code cache FileNotFoundError; unrelated paths." + "reason": "Different checkpoint/load issues: sorted state_dict load error vs wrong visual encoder keys on save_pretrained." } ] }, @@ -16999,8 +16964,9 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ + "issue:28282", "issue:29870", "issue:30333", "issue:30990", @@ -17008,25 +16974,24 @@ "issue:33453", "issue:34689", "issue:36296", - "issue:38617", - "issue:41628", "issue:42548", "issue:42757", + "issue:42890", + "issue:42898", "issue:43010", "issue:43065", "issue:43232", "issue:43299", - "issue:43475", "issue:43502", "issue:43576", + "issue:43618", "issue:43723", "issue:43824", "issue:43827", "issue:43874", - "issue:43901", + "issue:44062", + "issue:44112", "issue:44222", - "issue:44230", - "issue:44315", "issue:44351", "issue:44355", "issue:44360", @@ -17034,23 +16999,32 @@ "issue:44485", "issue:44556", "issue:44655", + "issue:44734", "issue:44749", - "issue:44908", - "issue:44938", "issue:44961", + "issue:44964", "issue:45071", "issue:45125", "issue:45375", - "issue:45468", + "issue:45440", "issue:45478", + "issue:45600", "issue:46132", - "issue:46143", - "issue:46144", + "issue:46285", "issue:46617", "issue:46664", "issue:46726" ], "soft_pairs": [ + "issue:28282|issue:33453", + "issue:43618|issue:46285", + "issue:44360|issue:45440", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:42898|issue:46664", + "issue:43502|issue:44062", + "issue:44734|issue:46617", "issue:43723|issue:44556", "issue:44387|issue:44556", "issue:44222|issue:45125", @@ -17077,277 +17051,252 @@ "issue:33453|issue:34689", "issue:44355|issue:44655", "issue:42757|issue:43576", - "issue:43576|issue:45071", - "issue:44315|issue:44360", - "issue:31795|issue:38617", - "issue:43576|issue:43824", - "issue:44961|issue:45468", - "issue:44351|issue:46617", - "issue:44485|issue:44961", - "issue:43576|issue:44938", - "issue:36296|issue:43901", - "issue:44908|issue:44961", - "issue:46143|issue:46144", - "issue:43475|issue:46617", - "issue:44222|issue:44230", - "issue:41628|issue:42548" + "issue:43576|issue:45071" ], - "prepared_review_unit_hash": "71cc4fd42b1b6fcdda288aefa71df405fb5b1304bef2c9c4fe9e02f1456fb909", + "prepared_review_unit_hash": "08049a09d4ef2fe197d3fc119c3886ee2e71defb1e9c20bab610a59cadb0f277", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 40, - "serialized_chars": 39162, - "estimated_input_tokens": 9791, - "estimated_eval_tokens": 19838 + "soft_pair_count": 36, + "serialized_chars": 38189, + "estimated_input_tokens": 9548, + "estimated_eval_tokens": 19352 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous: most items are unrelated loading, import, docs, or model-specific regressions. No soft pair looks like the same underlying bug/change, so all soft edges are rejected.", - "confidence": 0.82, - "canonical_issue_reason": "issue:43576 is the broadest, most reference-linked v5 regression in the set, so it is the closest anchor even though this is not a true duplicate cluster.", + "summary": "This cluster is noisy and mostly contains unrelated loading/import/regression issues; none of the soft pairs look like confident duplicates.", + "confidence": 0.31, + "canonical_issue_reason": "Issue 33453 is the broadest and most central loading-regression report in the set, and it best represents the tokenizer/model-loading subset of the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43576 is the best representative of the cluster's dominant theme of broad v5 breakage reports and has the strongest external signal among the issues.", + "best_issue_reason": "33453 is the most generally framed issue and the best candidate to anchor triage for the loading-regression subgroup, even though the cluster overall is heterogeneous.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "PyTorch-missing import error vs tokenizer loading regression; different failure modes and causes." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Both touch CLIP internals, but one is an attentions assignment bug and the other is a model-structure inconsistency." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "DSA indexer/ReLU discussion is unrelated to DeepSeekV3 implementation divergence." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "CI flake from missing set_seed is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "A flaky SAM HQ integration test is not the same bug as a GraniteSpeech stale device override failure." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Test seeding issue vs stale auto_mappings references; different code paths." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenization cleanup behavior change is unrelated to video processor num_frames semantics." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "local_files_only network requests and AddedToken keyword collision are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation crash and trust_remote_code cache FileNotFound are different loading bugs." + }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Tokenizer-loading regression vs generic checkpoint reload failure; different code paths." + "reason": "Both are loading-related, but one is tokenizer loading and the other is checkpoint reload after version upgrade." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Int4/OOM memory regression vs model reload failure; unrelated symptoms." + "reason": "Int4 quantization OOM is a different problem from reload failure after upgrade." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "FP8 save_pretrained bug vs missing tensor-parallel plan; different problems." + "reason": "FP8 save_pretrained moe and missing _tp_plan for tensor parallelism are distinct issues." }, { "left": "issue:43824", "right": "issue:46617", "accept": false, - "reason": "Missing import symbol vs cache FileNotFoundError; not the same bug." + "reason": "Cannot import a specific Qwen2.5 VL class is unrelated to cache file lookup failure." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "local_files_only network leak vs GLM RoPE implementation issue; unrelated." + "reason": "Network requests despite local_files_only and GLM-5 RoPE implementation are unrelated." }, { "left": "issue:43874", "right": "issue:46617", "accept": false, - "reason": "Missing processor method vs cache file lookup failure; different failures." + "reason": "Missing image-processor method and cache FileNotFound are different bugs." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "no_grad decorator issue vs dummy Conv2d discussion; unrelated." + "reason": "Cache update no_grad and dummy Conv2d in Sam3PixelDecoder do not share the same code-path problem." }, { "left": "issue:43576", "right": "issue:46132", "accept": false, - "reason": "CLI/env breakage vs AttentionInterface behavior change; different code paths." + "reason": "Broken env command and changed AttentionInterface.register behavior are separate API issues." }, { "left": "issue:45125", "right": "issue:45478", "accept": false, - "reason": "Tensor-parallel plan omission vs generic Qwen3.5 MoE loading error; too different." + "reason": "Same model family, but one is missing tensor-parallel metadata and the other is a generic from_pretrained error; not enough to merge." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLFlow reporting on failure vs tokenizer loading regression; unrelated." + "reason": "Failed MLFlow reporting and tokenizer loading regression are unrelated." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled-file errors vs post-upgrade performance slowdown; not the same bug." + "reason": "Compiled Python file errors and inference slowdown after upgrade are different problems." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Different missing exports from transformers; separate import issues." + "reason": "Different missing symbols with different root causes; both are import errors but not the same bug." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Missing tensor-parallel plan vs missing config field; different model bugs." + "reason": "Missing _tp_plan and missing deepstack_visual_indexes are separate Qwen3.5 MoE config issues." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Cache update no_grad issue vs generation sync_gpus bug; distinct paths." + "reason": "Cache update decoration and generation kwargs sync_gpus handling are distinct generation internals." }, { "left": "issue:46664", "right": "issue:46726", "accept": false, - "reason": "Video num_frames semantics vs AutoProcessor backend-missing error; unrelated." + "reason": "Video num_frames semantics and missing vision backend handling are unrelated processor issues." }, { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "Loading hang vs docs confusion; not the same underlying issue." + "reason": "Sentence Transformers loading stall and forward() documentation confusion are not the same bug." }, { "left": "issue:43502", "right": "issue:44222", "accept": false, - "reason": "local_files_only network leakage vs fp8 save_pretrained; unrelated." + "reason": "local_files_only networking bug and FP8 save_pretrained moe are unrelated." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "local_files_only leakage vs unrelated placeholder issue." + "reason": "No meaningful overlap beyond being issues; racoon is unrelated." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Pipeline docs after removals vs unrelated placeholder issue." + "reason": "Docs still referencing pipeline() and racoon are unrelated." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLFlow failure reporting vs tensor-parallel training bug; different areas." + "reason": "MLFlow reporting and tensor parallel training bug are different subsystems." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Specific Qwen3VL MoE load regression vs generic checkpoint reload issue; not one bug." + "reason": "Both mention version upgrades and loading, but they are different models and different failure modes." }, { "left": "issue:29870", "right": "issue:31795", "accept": false, - "reason": "Runtime UnboundLocalError in generate vs docs confusion; unrelated." + "reason": "Generate() UnboundLocalError in BART evaluation and documentation ambiguity are unrelated." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled-file errors vs PretrainedConfig type checking; different regressions." + "reason": "Compiled-file runtime errors and PretrainedConfig type checking are different issues." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Tokenizer-loading regression vs Llama 3.2 Vision model loading breakage; different failures." + "reason": "Both are loading-related, but one is a tokenizer regression and the other is a specific multimodal model-loading breakage." }, { "left": "issue:44355", "right": "issue:44655", "accept": false, - "reason": "Compiled-file errors vs Pipeline save_pretrained issue; unrelated." + "reason": "Compiled Python file errors and inability to save Pipeline objects are unrelated." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "huggingface_hub import error vs transformers env CLI breakage; different components." + "reason": "Hub import failure and broken env command are different v5 regressions." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "CLI/env breakage vs PretrainedConfig type checking; not the same bug." - }, - { - "left": "issue:44315", - "right": "issue:44360", - "accept": false, - "reason": "Liger Kernel application issue vs DSA indexer ReLU discussion; unrelated." - }, - { - "left": "issue:31795", - "right": "issue:38617", - "accept": false, - "reason": "Documentation question vs missing import symbol; unrelated." - }, - { - "left": "issue:43576", - "right": "issue:43824", - "accept": false, - "reason": "CLI/env command breakage vs missing model import; unrelated." - }, - { - "left": "issue:44961", - "right": "issue:45468", - "accept": false, - "reason": "Placeholder issue vs Gemma-4 audio positional encoding bug; unrelated." - }, - { - "left": "issue:44351", - "right": "issue:46617", - "accept": false, - "reason": "Missing HybridCache import vs cache FileNotFoundError; different failures." - }, - { - "left": "issue:44485", - "right": "issue:44961", - "accept": false, - "reason": "GLM-5 RoPE implementation issue vs placeholder issue; unrelated." - }, - { - "left": "issue:44908", - "right": "issue:44961", - "accept": false, - "reason": "Scheduler kwarg bug vs placeholder issue; unrelated." - }, - { - "left": "issue:46143", - "right": "issue:46144", - "accept": false, - "reason": "Both RoFormer-related, but one is kwargs propagation and the other attention-interface wiring; not the same bug." - }, - { - "left": "issue:43475", - "right": "issue:46617", - "accept": false, - "reason": "SAM3 video attribute error vs cache FileNotFoundError; unrelated." - }, - { - "left": "issue:44222", - "right": "issue:44230", - "accept": false, - "reason": "FP8 save_pretrained bug vs fp8 inference support request; same topic but different changes." - }, - { - "left": "issue:41628", - "right": "issue:42548", - "accept": false, - "reason": "Different missing transformer exports; separate import errors." + "reason": "Broken env command and type-checking regression are unrelated." } ] }, @@ -17369,28 +17318,27 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ "issue:31795", - "issue:33453", + "issue:36296", + "issue:38617", + "issue:41628", "issue:41669", "issue:42371", - "issue:42890", - "issue:43352", + "issue:42548", "issue:43408", + "issue:43475", "issue:43525", "issue:43572", "issue:43576", "issue:43673", - "issue:43827", - "issue:43976", + "issue:43824", + "issue:43901", "issue:44038", - "issue:44077", + "issue:44222", "issue:44230", - "issue:44246", "issue:44273", - "issue:44295", - "issue:44297", "issue:44315", "issue:44351", "issue:44360", @@ -17399,31 +17347,45 @@ "issue:44492", "issue:44556", "issue:44623", - "issue:44704", - "issue:44861", "issue:44908", + "issue:44938", + "issue:44961", "issue:45003", "issue:45071", "issue:45092", "issue:45125", "issue:45230", "issue:45278", - "issue:45341", "issue:45362", - "issue:45478", + "issue:45468", "issue:45542", "issue:45753", "issue:45850", "issue:46095", + "issue:46143", + "issue:46144", "issue:46209", "issue:46218", - "issue:46328", + "issue:46617", "issue:46664" ], "soft_pairs": [ - "issue:44315|issue:46209", - "issue:44485|issue:46664", - "issue:43673|issue:44492", + "issue:44315|issue:44360", + "issue:31795|issue:38617", + "issue:43576|issue:43824", + "issue:44961|issue:45468", + "issue:44351|issue:46617", + "issue:44485|issue:44961", + "issue:44908|issue:44961", + "issue:43576|issue:44938", + "issue:36296|issue:43901", + "issue:46143|issue:46144", + "issue:43475|issue:46617", + "issue:44222|issue:44230", + "issue:41628|issue:42548", + "issue:44315|issue:46209", + "issue:44485|issue:46664", + "issue:43673|issue:44492", "issue:45362|issue:45850", "issue:44273|issue:45542", "issue:44623|issue:45230", @@ -17435,308 +17397,253 @@ "issue:43576|issue:44556", "issue:41669|issue:43673", "issue:43525|issue:45850", - "issue:43572|issue:45850", "issue:44556|issue:45092", + "issue:43572|issue:45850", "issue:44492|issue:45850", "issue:45278|issue:45850", "issue:42371|issue:44492", "issue:44230|issue:46095", "issue:41669|issue:46218", - "issue:43408|issue:44315", - "issue:42890|issue:44908", - "issue:44246|issue:44273", - "issue:43576|issue:44908", - "issue:43352|issue:44393", - "issue:43976|issue:44393", - "issue:44704|issue:44908", - "issue:44908|issue:45071", - "issue:44297|issue:45230", - "issue:44295|issue:45230", - "issue:41669|issue:44351", - "issue:44492|issue:46328", - "issue:41669|issue:45341", - "issue:44393|issue:45478", - "issue:43976|issue:45850", - "issue:43976|issue:46218", - "issue:44077|issue:44861", - "issue:43827|issue:44908", - "issue:31795|issue:33453", - "issue:43576|issue:44273" + "issue:43408|issue:44315" ], - "prepared_review_unit_hash": "4860f3d35cb84559d4590fb43301ed31bac28b9fbdb88dbde1a5cbb3d153022e", + "prepared_review_unit_hash": "0df14c4cf991f3ace882a2ef49280f9568e0a681a1e8225357d676eed3ed5200", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 41, - "serialized_chars": 39288, - "estimated_input_tokens": 9822, - "estimated_eval_tokens": 19900 + "soft_pair_count": 35, + "serialized_chars": 37294, + "estimated_input_tokens": 9324, + "estimated_eval_tokens": 18904 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This set is mostly heterogeneous: docs, tokenizer/processor APIs, lazy-loading/import performance, cache semantics, model-specific config bugs, and several broad v5 regression reports. The soft pairs share at most broad subsystem similarity; none look like the same concrete bug or change.", - "confidence": 0.7, - "canonical_issue_reason": "Issue 45850 is the broadest umbrella-style report in the set and best matches the recurring 'v5 breaks many models' theme, but the overall cluster is too mixed to be a strong duplicate group.", + "summary": "This cluster is noisy and mostly spans unrelated bugs. A few issues share a generic v5 import/export-breakage theme, but none of the soft pairs look like the same concrete fix. Issue 45278 is the best umbrella representative.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45278 is the clearest broad-but-concrete umbrella: it describes many import errors after upgrading to 5.5.0, which matches the largest recurring theme in the set better than the narrower or more speculative items.", "canonical_pr_reason": null, - "best_issue_reason": "If one issue must represent the cluster, 45850 is the most general and central summary of the repeated version-regression complaints.", + "best_issue_reason": "issue:45278 is the most suitable global representative because it captures the shared post-v5 import/regression pattern without being as vague as the catch-all complaints.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44315", - "right": "issue:46209", - "accept": false, - "reason": "Different model/features and different failure modes; shared ML theme only." - }, - { - "left": "issue:44485", - "right": "issue:46664", - "accept": false, - "reason": "RoPE implementation discussion vs video frame semantics are unrelated." - }, - { - "left": "issue:43673", - "right": "issue:44492", - "accept": false, - "reason": "Cache regression during chunked prefill is not the same as a typo in cache-strategy naming." - }, - { - "left": "issue:45362", - "right": "issue:45850", - "accept": false, - "reason": "A specific chat crash is not the same concrete bug as a broad 'many models break in v5' complaint." - }, - { - "left": "issue:44273", - "right": "issue:45542", + "right": "issue:44360", "accept": false, - "reason": "Lazy loading/import behavior and tf-backend detection are unrelated." + "reason": "Different subsystems and failures: Liger Kernel/model_init vs a DSA indexer ReLU discussion." }, { - "left": "issue:44623", - "right": "issue:45230", + "left": "issue:31795", + "right": "issue:38617", "accept": false, - "reason": "Missing processor save files is too specific to equate with a generic bug report." + "reason": "Documentation confusion about forward args is unrelated to an import error for layer_type_validation." }, { - "left": "issue:44360", - "right": "issue:46664", + "left": "issue:43576", + "right": "issue:43824", "accept": false, - "reason": "DSA indexer behavior has no connection to SmolVLM video frame handling." + "reason": "Broken `transformers env` command and missing Qwen2_5_VLForConditionalGeneration are different problems." }, { - "left": "issue:44038", - "right": "issue:45125", + "left": "issue:44961", + "right": "issue:45468", "accept": false, - "reason": "Different Qwen model families and different problems; not the same code-path bug." + "reason": "No shared bug: placeholder issue `racoon` versus a Gemma-4 audio positional encoding bug." }, { - "left": "issue:44273", - "right": "issue:45003", + "left": "issue:44351", + "right": "issue:46617", "accept": false, - "reason": "Both touch import/lazy loading, but one is a generic lazy-loading failure and the other is sys.modules access; not enough evidence of one bug." + "reason": "Both are import/load failures, but one is a missing symbol export and the other is a cache FileNotFoundError." }, { - "left": "issue:44393", - "right": "issue:45753", + "left": "issue:44485", + "right": "issue:44961", "accept": false, - "reason": "Same model family, but bbox hallucination/error and infinite-loop behavior are different concrete failures." + "reason": "GLM-5 RoPE implementation discussion is unrelated to the `racoon` issue." }, { - "left": "issue:44393", - "right": "issue:45071", + "left": "issue:44908", + "right": "issue:44961", "accept": false, - "reason": "Model output bug vs PretrainedConfig typing regression are unrelated." + "reason": "Scheduler kwarg handling has nothing to do with the `racoon` report." }, { "left": "issue:43576", - "right": "issue:44556", - "accept": false, - "reason": "Both are v5 breakages, but the env command and checkpoint reload problems are different code paths." - }, - { - "left": "issue:41669", - "right": "issue:43673", + "right": "issue:44938", "accept": false, - "reason": "Import-star performance issue is unrelated to a cache-missing runtime regression." + "reason": "`transformers env` breakage and Python 3.14 load failure are distinct compatibility issues." }, { - "left": "issue:43525", - "right": "issue:45850", + "left": "issue:36296", + "right": "issue:43901", "accept": false, - "reason": "A missing pad_token_id attribute in one config is not the same as a broad v5 model-breakage report." + "reason": "Tensor parallel training bug is unrelated to a TextClassificationPipeline docs mismatch." }, { - "left": "issue:43572", - "right": "issue:45850", + "left": "issue:46143", + "right": "issue:46144", "accept": false, - "reason": "StableLmConfig pad token regression is model-specific and distinct from a generic v5 complaint." + "reason": "Both involve RoFormer, but one is kwargs passthrough and the other is attention-interface usage; not the same concrete bug." }, { - "left": "issue:44556", - "right": "issue:45092", + "left": "issue:43475", + "right": "issue:46617", "accept": false, - "reason": "Both involve v5 compatibility, but checkpoint reload failure and remote-code meta-init incompatibility are not the same bug." + "reason": "SAM 3 video attribute error is unrelated to a trust_remote_code cache FileNotFoundError." }, { - "left": "issue:44492", - "right": "issue:45850", + "left": "issue:44222", + "right": "issue:44230", "accept": false, - "reason": "A cache-strategy typo is not the same as a broad version-regression issue." + "reason": "Both mention FP8/moe, but one is save_pretrained behavior and the other is inference support." }, { - "left": "issue:45278", - "right": "issue:45850", + "left": "issue:41628", + "right": "issue:42548", "accept": false, - "reason": "Generic upgrade import errors overlap only at a high level with 'many models break'; no shared concrete failure path." + "reason": "Two different missing top-level exports; no evidence they stem from one identical regression." }, { - "left": "issue:42371", - "right": "issue:44492", + "left": "issue:44315", + "right": "issue:46209", "accept": false, - "reason": "TF32 API guidance and cache-strategy typo are unrelated." + "reason": "Liger Kernel/model_init and FP8 inference on B200 are unrelated code paths." }, { - "left": "issue:44230", - "right": "issue:46095", + "left": "issue:44485", + "right": "issue:46664", "accept": false, - "reason": "FP8 inference failure and a weight conversion-script question are different topics." + "reason": "GLM-5 RoPE implementation is unrelated to SmolVLM video-frame semantics." }, { - "left": "issue:41669", - "right": "issue:46218", + "left": "issue:43673", + "right": "issue:44492", "accept": false, - "reason": "Import performance and a specific module import failure are not the same bug." + "reason": "A GenerationMixin cache bug and a cache-strategy typo are not the same underlying issue." }, { - "left": "issue:43408", - "right": "issue:44315", + "left": "issue:45362", + "right": "issue:45850", "accept": false, - "reason": "Model-type mismatch warning and Liger Kernel application are unrelated." + "reason": "Specific Qwen3.5 chat crash versus a broad claim that v5 breaks many models; too different to merge." }, { - "left": "issue:42890", - "right": "issue:44908", + "left": "issue:44273", + "right": "issue:45542", "accept": false, - "reason": "Missing test seeds and scheduler keyword handling are unrelated." + "reason": "Lazy loading and TensorFlow backend detection are unrelated." }, { - "left": "issue:44246", - "right": "issue:44273", + "left": "issue:44623", + "right": "issue:45230", "accept": false, - "reason": "Both mention import/lazy-loading, but one is slow import and the other is correctness of lazy loading; not a clear duplicate." + "reason": "Missing processor.save_pretrained files is not the same as a generic bug report." }, { - "left": "issue:43576", - "right": "issue:44908", + "left": "issue:44360", + "right": "issue:46664", "accept": false, - "reason": "Env command breakage and inverse_sqrt scheduler kwargs are different features." + "reason": "DSA indexer ReLU discussion is unrelated to SmolVLM num_frames behavior." }, { - "left": "issue:43352", - "right": "issue:44393", + "left": "issue:44038", + "right": "issue:45125", "accept": false, - "reason": "Flash-Attention support for Nemotron and Qwen3-VL bounding-box errors are unrelated." + "reason": "Both mention Qwen/MoE, but one is a vague v5 bug report and the other is a missing tensor-parallel plan." }, { - "left": "issue:43976", - "right": "issue:44393", + "left": "issue:44273", + "right": "issue:45003", "accept": false, - "reason": "Python version compatibility and model hallucination/output errors are unrelated." + "reason": "Lazy loading complaints and unsafe sys.modules access may be adjacent, but they are not clearly the same bug." }, { - "left": "issue:44704", - "right": "issue:44908", + "left": "issue:44393", + "right": "issue:45753", "accept": false, - "reason": "AutoProcessor cached_file kwargs and scheduler kwargs are different APIs." + "reason": "Both are Qwen3-VL/Qwen3.5 vision issues, but one is 2D box hallucination and the other is an infinite loop on a specific image." }, { - "left": "issue:44908", + "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwarg propagation and config type checking are unrelated." + "reason": "A model-specific vision output bug is unrelated to PretrainedConfig type checking." }, { - "left": "issue:44297", - "right": "issue:45230", - "accept": false, - "reason": "Tokenizer save metadata bug and generic bug report are not the same issue." - }, - { - "left": "issue:44295", - "right": "issue:45230", + "left": "issue:43576", + "right": "issue:44556", "accept": false, - "reason": "Position_ids buffer handling is unrelated to a generic bug report." + "reason": "An env command regression and checkpoint reload incompatibility after upgrade are different symptoms." }, { "left": "issue:41669", - "right": "issue:44351", + "right": "issue:43673", "accept": false, - "reason": "Import-star performance is unrelated to a missing HybridCache symbol." + "reason": "Import-star removal concerns are unrelated to a missing GenerationMixin cache in chunked prefill." }, { - "left": "issue:44492", - "right": "issue:46328", + "left": "issue:43525", + "right": "issue:45850", "accept": false, - "reason": "Cache-strategy naming and SequenceFeatureExtractor padding performance are unrelated." + "reason": "Missing `pad_token_id` on Llama4Config is one model/config bug, not the broad v5 breakage complaint." }, { - "left": "issue:41669", - "right": "issue:45341", + "left": "issue:44556", + "right": "issue:45092", "accept": false, - "reason": "Import performance and a small testing_utils bug are not the same problem." + "reason": "Both are upgrade regressions, but one is checkpoint reload failure and the other is remote-code meta-init incompatibility." }, { - "left": "issue:44393", - "right": "issue:45478", + "left": "issue:43572", + "right": "issue:45850", "accept": false, - "reason": "Different Qwen3.5 issues with different failure modes and likely different code paths." + "reason": "StableLmConfig missing `pad_token_idx` is too specific to merge with a generic v5 breakage claim." }, { - "left": "issue:43976", + "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "Python support regression is not the same concrete issue as broad model breakage." + "reason": "A cache-strategy typo is not the same as a broad version-breakage report." }, { - "left": "issue:43976", - "right": "issue:46218", + "left": "issue:45278", + "right": "issue:45850", "accept": false, - "reason": "Python-version compatibility and module import failure are unrelated." + "reason": "Both mention v5 regressions, but 45278 is a concrete import-error umbrella while 45850 is too vague to be the same underlying bug." }, { - "left": "issue:44077", - "right": "issue:44861", + "left": "issue:42371", + "right": "issue:44492", "accept": false, - "reason": "post_init allowance in patchtsmixer and tied-weights-key handling are unrelated." + "reason": "TF32 API settings and cache-strategy typo are unrelated." }, { - "left": "issue:43827", - "right": "issue:44908", + "left": "issue:44230", + "right": "issue:46095", "accept": false, - "reason": "Docs still referencing pipeline() and scheduler kwarg propagation are unrelated." + "reason": "FP8 inference support for Qwen models is unrelated to a DeepSeek weight conversion question." }, { - "left": "issue:31795", - "right": "issue:33453", + "left": "issue:41669", + "right": "issue:46218", "accept": false, - "reason": "Documentation confusion and tokenizer-loading regression are different issues." + "reason": "Removing import-star usage is not the same as a missing Qwen3_5ForCausalLM import." }, { - "left": "issue:43576", - "right": "issue:44273", + "left": "issue:43408", + "right": "issue:44315", "accept": false, - "reason": "Broken env command and lazy-loading correctness are not the same bug." + "reason": "SAM3 video/tracker class mismatch is unrelated to Liger Kernel application in model_init." } ] }, @@ -17758,58 +17665,76 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", + "cluster_id": "cluster-10000-487", "nodes": [ - "issue:28348", - "issue:29870", - "issue:30333", - "issue:31515", "issue:31795", + "issue:33453", "issue:34689", - "issue:38617", "issue:41669", "issue:42371", - "issue:42913", - "issue:43066", + "issue:42890", "issue:43298", "issue:43352", "issue:43502", "issue:43519", - "issue:43525", - "issue:43572", "issue:43576", "issue:43606", - "issue:43618", "issue:43704", - "issue:44230", + "issue:43827", + "issue:43976", + "issue:44077", + "issue:44246", "issue:44261", "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:44938", - "issue:44998", + "issue:45071", "issue:45230", - "issue:45250", - "issue:45290", "issue:45310", + "issue:45341", "issue:45468", + "issue:45478", "issue:45850", "issue:46018", "issue:46095", "issue:46172", "issue:46218", + "issue:46328", "issue:46655" ], "soft_pairs": [ + "issue:42890|issue:44908", + "issue:44246|issue:44273", + "issue:43576|issue:44908", + "issue:43352|issue:44393", + "issue:43976|issue:44393", + "issue:44908|issue:45071", + "issue:44704|issue:44908", + "issue:44295|issue:45230", + "issue:44297|issue:45230", + "issue:41669|issue:44351", + "issue:44492|issue:46328", + "issue:41669|issue:45341", + "issue:44393|issue:45478", + "issue:43976|issue:45850", + "issue:43976|issue:46218", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "issue:31795|issue:33453", + "issue:43576|issue:44273", "issue:44315|issue:44829", "issue:44360|issue:45468", "issue:43519|issue:46655", @@ -17825,8 +17750,8 @@ "issue:43704|issue:44485", "issue:44485|issue:45468", "issue:43502|issue:43519", - "issue:45850|issue:46218", "issue:42371|issue:43704", + "issue:45850|issue:46218", "issue:43519|issue:44623", "issue:44263|issue:45468", "issue:43606|issue:45468", @@ -17834,472 +17759,605 @@ "issue:43298|issue:46095", "issue:43519|issue:43704", "issue:44351|issue:46218", - "issue:43352|issue:46218", - "issue:42913|issue:43066", - "issue:28348|issue:45250", - "issue:44908|issue:44998", - "issue:44492|issue:44998", - "issue:43618|issue:44998", - "issue:44998|issue:45850", - "issue:44938|issue:44998", - "issue:44861|issue:45290", - "issue:31515|issue:38617", - "issue:43572|issue:44861", - "issue:43525|issue:44861", - "issue:44230|issue:44623", - "issue:28348|issue:43352", - "issue:29870|issue:31515", - "issue:30333|issue:31515" + "issue:43352|issue:46218" ], - "prepared_review_unit_hash": "e3819e3c79cfa2d63a37af26f94362cbc4c35170e2372dd5db6a232474a4a2e5", + "prepared_review_unit_hash": "24b19f5a7499098d5aae8f1deec9c7b8d04ad1c5bbb18f749e06e21237d67f33", "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 40, - "serialized_chars": 38935, - "estimated_input_tokens": 9734, - "estimated_eval_tokens": 19724 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 44, + "serialized_chars": 39635, + "estimated_input_tokens": 9909, + "estimated_eval_tokens": 20074 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is very heterogeneous: most items are unrelated model-specific regressions, API breakages, or feature requests. I do not see any strong duplicate pairs; the best representative issue is the broad v5 regression umbrella.", - "confidence": 0.84, - "canonical_issue_reason": "issue:45850 is the broadest umbrella in the set and best captures the common theme of multiple Transformers v5 regressions, even though it is not a precise duplicate of the others.", + "summary": "This cluster is highly noisy: the soft edges are mostly title-level similarity across unrelated bugs (loading, processors, docs, model-specific regressions, and utilities). I did not find any convincing duplicate pairs, and there are no PRs in the cluster.", + "confidence": 0.27, + "canonical_issue_reason": "issue:44908 is the most connected node in the similarity graph and the closest thing to a hub, but the cluster is too heterogeneous for it to be a strong duplicate canonical.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45850 is the most suitable cluster representative because it is broad enough to summarize the recurring v5 breakage reports without overfitting to one model or subsystem.", + "best_issue_reason": "issue:44908 is the least-bad anchor only because it has the broadest linkage; it still does not cleanly represent a single underlying bug family.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test flakiness from missing seeds vs scheduler kwargs being ignored." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import-time slowness vs lazy-loading malfunction; related theme, different bug." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Broken env command vs inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash-Attention support gating vs Qwen3-VL bbox output error." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python-version compatibility regression vs vision-model output bug." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwarg propagation vs PretrainedConfig type-checking regression." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor cached_file kwargs bug vs scheduler kwargs bug." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "position_ids buffer access issue vs an unspecified generic bug report." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "tokenizer save metadata mismatch vs an unspecified generic bug report." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star cleanup/perf issue vs missing HybridCache import." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Cache-strategy typo vs SequenceFeatureExtractor.pad performance bug." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import-star cleanup/perf issue vs testing_utils bug." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5/Qwen3-VL model failures, not the same code-path." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "Python support regression vs vague complaint about many broken models." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python-version regression vs missing Qwen3_5 import." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "PatchTSMixer post_init policy vs tied-weight key AttributeError." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still mention pipeline() vs scheduler kwargs bug." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Forward-doc confusion vs tokenizer loading regression." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Env command breakage vs lazy-loading behavior." + }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Different bugs: Liger Kernel not applied with model_init vs flash_attention_3 causing degenerate training." + "reason": "Liger Kernel application with model_init vs flash_attention_3 training degeneration." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "Different code paths and symptoms: DSA indexer ReLU issue vs Gemma-4 audio positional encoding bug." + "reason": "DSA indexer activation issue vs Gemma-4 audio positional encoding bug." }, { "left": "issue:43519", "right": "issue:46655", "accept": false, - "reason": "Different processor bugs: Qwen3VL timestamp calculation vs SmolVLM video padding calculation." + "reason": "Qwen3VL timestamp math vs SmolVLM video padding calculation." }, { "left": "issue:44273", "right": "issue:44908", "accept": false, - "reason": "Lazy loading failure and inverse_sqrt scheduler kwargs handling are unrelated." + "reason": "Lazy-loading bug vs scheduler kwargs bug." }, { "left": "issue:44655", "right": "issue:44861", "accept": false, - "reason": "save_pretrained/pipeline file saving issue vs _get_tied_weight_keys AttributeError are separate failures." + "reason": "Pipeline save_pretrained issue vs tied-weight key crash." }, { "left": "issue:43576", "right": "issue:46218", "accept": false, - "reason": "Different issues: broken transformers env command vs missing Qwen3_5ForCausalLM import." + "reason": "v5 env command regression vs missing Qwen3_5 import." }, { "left": "issue:45230", "right": "issue:45310", "accept": false, - "reason": "Generic bug report is not the same concrete regression as the Qwen3.5 MoE from_pretrained error." + "reason": "Generic bug report vs specific Qwen3.5 MoE from_pretrained regression." }, { "left": "issue:43519", "right": "issue:44485", "accept": false, - "reason": "Different models and mechanics: timestamp calculation vs RoPE implementation." + "reason": "Timestamp calculation bug vs GLM-5 RoPE implementation." }, { "left": "issue:43519", "right": "issue:44908", "accept": false, - "reason": "Processor timestamp bug and scheduler kwargs bug do not match." + "reason": "Processor timestamp math vs scheduler kwargs bug." }, { "left": "issue:41669", "right": "issue:43704", "accept": false, - "reason": "Import * cleanup performance issue vs VRAM leak in multi-dataloader threads are unrelated." + "reason": "Import-star cleanup/perf issue vs VRAM leak in multi-thread use." }, { "left": "issue:31795", "right": "issue:34689", "accept": false, - "reason": "Documentation confusion about inputs/past_key_values is not the same as a model-loading breakage." + "reason": "Documentation confusion vs model-loading failure." }, { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "DeepSeek-V4 gating logic and comb.to(...).transpose(...) are different concrete bugs." + "reason": "Shared-expert gating issue vs tensor transpose/weight-conversion bug." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "VRAM leak and RoPE implementation bug are unrelated." + "reason": "VRAM leak vs RoPE implementation." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Different model-specific math issues in different components." + "reason": "GLM-5 RoPE implementation vs Gemma-4 audio positional encoding." }, { "left": "issue:43502", "right": "issue:43519", "accept": false, - "reason": "local_files_only API calls is unrelated to Qwen3VL timestamp computation." + "reason": "local_files_only network leak vs timestamp calculation." }, { - "left": "issue:45850", - "right": "issue:46218", + "left": "issue:42371", + "right": "issue:43704", "accept": false, - "reason": "Broad v5 breakage complaint and a specific import failure are not the same bug." + "reason": "TF32 settings API change vs VRAM leak." }, { - "left": "issue:42371", - "right": "issue:43704", + "left": "issue:45850", + "right": "issue:46218", "accept": false, - "reason": "TF32 settings change is unrelated to the VRAM leak report." + "reason": "Broad complaint about broken models vs missing Qwen3_5 import." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Processor timestamp bug and missing files on save_pretrained are different subsystems." + "reason": "Timestamp calculation bug vs processor save_pretrained missing files." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "torch.split return-value issue is unrelated to Gemma-4 audio positional encoding." + "reason": "torch.split return-value issue vs audio positional encoding bug." }, { "left": "issue:43606", "right": "issue:45468", "accept": false, - "reason": "CPU offload device mismatch and audio positional encoding are different bugs." + "reason": "CPU offload device mismatch vs audio positional encoding." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "MLA q_a_layernorm precision issue is unrelated to Gemma-4 audio encoding." + "reason": "RMS norm epsilon precision issue vs audio positional encoding." }, { "left": "issue:43298", "right": "issue:46095", "accept": false, - "reason": "Continuous batching audio-model support and deepseekv4 weight conversion are different requests." + "reason": "Continuous batching support gap vs DeepSeek-V4 weight-conversion question." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Timestamp calculation bug and VRAM leak are unrelated." + "reason": "Timestamp calculation bug vs VRAM leak." }, { "left": "issue:44351", "right": "issue:46218", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different missing-export bugs." + "reason": "Missing HybridCache import vs missing Qwen3_5 import." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Flash Attention support limitation and Qwen3_5ForCausalLM import failure are not the same issue." - }, + "reason": "Flash-Attention 2.0 support gating vs missing Qwen3_5 import." + } + ] + }, + "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": 24, + "review_unit_index": 58, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28348", + "issue:29870", + "issue:30333", + "issue:31515", + "issue:31795", + "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:44492", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44938", + "issue:44998", + "issue:45250", + "issue:45290", + "issue:45310", + "issue:45850", + "issue:46018", + "issue:46144" + ], + "soft_pairs": [ + "issue:42913|issue:43066", + "issue:28348|issue:45250", + "issue:44908|issue:44998", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44998|issue:45850", + "issue:44938|issue:44998", + "issue:44861|issue:45290", + "issue:31515|issue:38617", + "issue:43572|issue:44861", + "issue:43525|issue:44861", + "issue:44230|issue:44623", + "issue:28348|issue:43352", + "issue:29870|issue:31515", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "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": "6f7dd66e9e9849d541d061ecfe67e15f1d8f2cfa58a094a2b56837ce15260787", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 34, + "item_count": 34, + "soft_pair_count": 31, + "serialized_chars": 28240, + "estimated_input_tokens": 7060, + "estimated_eval_tokens": 14376 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No genuine duplicates detected; this cluster is a grab bag of unrelated tokenizer, model, training, and docs/config issues, with only loose topical overlap in a few version-regression titles. All soft edges should be rejected.", + "confidence": 0.77, + "canonical_issue_reason": "issue:42913 is the broadest version-regression/tokenizer anchor in the set, but the cluster is too heterogeneous for a true duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42913 is the most representative high-level regression report among the issues shown; it is the closest thing to a cluster anchor even though the set is not a real duplicate group.", + "best_pr_reason": null, + "soft_edge_verdicts": [ { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both mention tokenizer v5 behavior, but they describe different concrete problems." + "reason": "Related tokenizer/v5 symptoms, but one is generic behavior change and the other is a decoder-type bug; not the same fix." }, { "left": "issue:28348", "right": "issue:45250", "accept": false, - "reason": "GPT2 Flash Attention 2.0 support request is not the same as the generic Flash Attention 2.0 topic." + "reason": "Both mention Flash Attention 2.0, but they target different model families and likely different implementation gaps." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs bug and unrelated placeholder issue do not match." + "reason": "Inverse-sqrt scheduler kwargs bug is unrelated to the spurious 'Unemployment' issue." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "Typo in cache strategies is unrelated to the placeholder issue." + "reason": "Cache-strategy typo and the unrelated 'Unemployment' issue are not duplicates." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions regression is unrelated to the placeholder issue." + "reason": "CLIPOutput attentions regression is unrelated to the spurious 'Unemployment' issue." }, { "left": "issue:44998", "right": "issue:45850", "accept": false, - "reason": "The placeholder issue is not the same as the broad Transformers v5 regression report." + "reason": "'Unemployment' is unrelated to the broad '5.0 breaks too many models' complaint." }, { "left": "issue:44938", "right": "issue:44998", "accept": false, - "reason": "Python 3.14 load failure is unrelated to the placeholder issue." + "reason": "Python 3.14 import/load failure is unrelated to the spurious 'Unemployment' issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "Tied-weight-key crash and apply_chat_template crash are different code paths." + "reason": "Tied-weight key handling and chat-template tool-call crashing are different code paths and bugs." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Slow from_pretrained loading and missing layer_type_validation import are different problems." + "reason": "Slow from_pretrained loading is unrelated to the ImportError for layer_type_validation." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_idx in config is unrelated to tied-weight-key handling." + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to tied-weight key attribute handling." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_id in Llama4Config is unrelated to tied-weight-key handling." + "reason": "Llama4 pad_token_id missing is a config regression, not the tied-weight crash." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "FP8 Qwen3-VL support request is unrelated to processor save_pretrained file omission." + "reason": "FP8 inference support and missing processor.save_pretrained files are unrelated issues." }, { "left": "issue:28348", "right": "issue:43352", "accept": false, - "reason": "One requests Flash Attention 2.0 support, the other reports it is unsupported for a different model." + "reason": "Same Flash Attention 2.0 wording, but GPT2 support and Nemotron support are separate model-specific changes." }, { "left": "issue:29870", "right": "issue:31515", "accept": false, - "reason": "BART generate decoder_attention_mask crash and slow checkpoint loading are unrelated." + "reason": "BART generate() decoder_attention_mask bug is unrelated to from_pretrained slowness." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow job tracking failure and slow from_pretrained loading 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": 24, - "review_unit_index": 58, - "review_unit_count": 58, - "status": "reviewed", - "reason": null, - "source": "llm", - "cache_hit": false, - "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-10000-486", - "nodes": [ - "issue:31515", - "issue:31795", - "issue:36296", - "issue:37934", - "issue:42936", - "issue:43240", - "issue:43298", - "issue:43668", - "issue:43717", - "issue:43813", - "issue:44031", - "issue:44230", - "issue:45310", - "issue:46018", - "issue:46144" - ], - "soft_pairs": [ - "issue:31515|issue:36296", - "issue:31515|issue:31795", - "issue:44230|issue:45310", - "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": "cd732519a5bee28fa6c6e5d39e75e5ad477216115743e3bbaf75f0e704dc9128", - "component_budget": { - "node_count": 486, - "item_count": 486, - "soft_pair_count": 1791, - "serialized_chars": 928270, - "estimated_input_tokens": 232068, - "estimated_eval_tokens": 464392 - }, - "budget": { - "node_count": 15, - "item_count": 15, - "soft_pair_count": 16, - "serialized_chars": 13117, - "estimated_input_tokens": 3280, - "estimated_eval_tokens": 6816 - }, - "overflow_policy": "truncate_then_skip", - "trimmed": false, - "aggressively_trimmed": false, - "split": true, - "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes unrelated issues across loading performance, docs, tensor parallelism, model-specific normalization, loss plumbing, tokenizer warnings, fp8 support, and attention/interface bugs. None of the soft edges look like the same underlying bug or change.", - "confidence": 0.95, - "canonical_issue_reason": "No clear canonical issue: the items cover many different subsystems and failure modes, so any single issue would be arbitrary.", - "canonical_pr_reason": "No pull requests are present in this cluster.", - "best_issue_reason": "No single best issue stands out as a global representative because the cluster does not center on one concrete bug or fix.", - "best_pr_reason": "No pull requests are present in this cluster.", - "soft_edge_verdicts": [ + "reason": "MLflow run-finalization on failure is unrelated to from_pretrained slowness." + }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "Checkpoint loading slowdown in from_pretrained is unrelated to a tensor parallel training bug." + "reason": "Loading slowdown and tensor-parallel training bug are different problems." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "One is about load speed; the other is documentation confusion for forward arguments." + "reason": "Loading slowdown and forward() documentation confusion are unrelated." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "FP8 inference support request is different from a Qwen3.5 MoE from_pretrained error on newer transformers." + "reason": "FP8 support request and Qwen3.5 from_pretrained error are different model/version issues." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Continuous batching support for audio models is a different feature gap than Qwen fp8 inference support." + "reason": "Continuous batching for audio models and FP8 inference support are different feature gaps." }, { "left": "issue:43717", "right": "issue:46144", "accept": false, - "reason": "Mamba init_weights behavior is unrelated to RoFormer not using the attention interface." + "reason": "Mamba init_weights behavior and RoFormer attention interface usage are unrelated model implementation issues." }, { "left": "issue:42936", "right": "issue:46018", "accept": false, - "reason": "Mask2Former ignore_value handling is a different bug than DeepSeek-V4 shared expert gating." + "reason": "Mask2Former ignore_value handling and DeepSeek-V4 shared expert gating are unrelated." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "Incorrect type hint for a config field is unrelated to tokenizer regex warning behavior." + "reason": "A config type-hint bug is unrelated to the tokenizer regex warning regression." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Unused ignore_value in Mask2Former is not the same as kwargs not being forwarded to loss calculation." + "reason": "Mask2Former ignore_value usage and loss-kwargs propagation are different code paths." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Loss-function kwargs plumbing is unrelated to RoFormer attention interface usage." + "reason": "Loss-function kwargs propagation and RoFormer attention interface are unrelated bugs." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2Former ignore_value usage does not match a tokenizer regex warning regression." + "reason": "Mask2Former ignore_value handling and tokenizer regex warnings do not share a concrete bug." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "These are distinct model-specific issues affecting different code paths." + "reason": "Mask2Former ignore_value handling and RoFormer attention interface usage are unrelated." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Unused ignore_value and a typo in quantization_operation are clearly different defects." + "reason": "Mask2Former ignore_value handling and the quantization_operation typo are unrelated." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Passing kwargs into loss computation is unrelated to a config type-hint mistake." + "reason": "Loss kwargs propagation and ModernBERT type-hint correction are different bug classes." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "A type annotation bug is not the same as a documentation/code typo in quantization logic." + "reason": "ModernBERT norm_eps typing and a quantization-operation typo are unrelated." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Questioning whether a norm is RMS norm is a model-design issue, not a config type-hint error." + "reason": "Llama4 norm behavior question and ModernBERT type-hint issue are unrelated." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former ignore_value handling and ModernBERT norm_eps typing do not describe the same bug." + "reason": "Mask2Former ignore_value handling and ModernBERT type-hint correction are unrelated." } ] }, @@ -18332,12 +18390,12 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "2addad07a04e93090a872b62e655a0a8750aec8886365934036a1214c8a0be7e", + "prepared_review_unit_hash": "7723390dbce22cccff85211912e5a1819596d2dda385322f93151d2b0fab26b4", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2878, + "serialized_chars": 2877, "estimated_input_tokens": 720, "estimated_eval_tokens": 1696 }, @@ -18345,7 +18403,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2878, + "serialized_chars": 2877, "estimated_input_tokens": 720, "estimated_eval_tokens": 1696 }, @@ -18354,30 +18412,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster is a false positive: the three issues are about unrelated feature requests in different parts of the library, and the observed similarity is only superficial wording overlap.", - "confidence": 0.98, - "canonical_issue_reason": "No true duplicate stands out. If one issue must be used as the representative anchor, issue 10519 is the clearest and most concrete feature request, but it is not semantically related to the other two.", + "summary": "These three issues are unrelated and should not be clustered as duplicates. They cover different topics: tokenizer truncation behavior, learning-rate scheduler functionality, and Wav2Vec2 word boundary offsets.", + "confidence": 0.99, + "canonical_issue_reason": "Issue 10519 is the most concrete and self-contained request, but it is not actually equivalent to the others, so there is no true canonical duplicate in this set.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 10519 is the best standalone representative because it has the most specific scope and the strongest discussion around a concrete API change; the others concern entirely different functionality.", + "best_issue_reason": "10519 is the clearest issue to use as a representative only because it states a specific feature/behavior request; the other two are unrelated concerns.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Different problems: truncation behavior in tokenization versus word start/end offsets in Wav2Vec2. No shared code path or same underlying bug." + "reason": "Tokenizer truncation direction and Wav2Vec2 word-offset extraction are different features in different code paths." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "Unrelated feature requests: tokenizer truncation direction versus ReduceLROnPlateau-like scheduler behavior. Similarity is only textual, not functional." + "reason": "Truncation behavior for tokenizers is unrelated to ReduceLROnPlateau-like scheduler functionality." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "Different subsystems and goals: learning-rate scheduling versus audio word boundary offsets. They do not describe the same change or bug." + "reason": "A learning-rate scheduler feature request does not match a Wav2Vec2 alignment/time-offset issue." } ] }, @@ -18409,12 +18467,12 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "161909dd779045052e6bea88c394ea3879c4b6fa976f6bc0756fa77a1a9a713a", + "prepared_review_unit_hash": "36e9f630ab5dba8a921a46038c7cdcfcbb49c6afcf8220db72fe69c22d29a18e", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2479, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -18422,7 +18480,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2479, + "serialized_chars": 2478, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -18431,24 +18489,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These issues are only loosely related by Hugging Face model usage. #13244 is a tokenization mismatch bug, while #15980 is an authentication/error-message problem for private model downloads, and #16618 is a generic model-loading failure. They do not look like the same underlying defect.", - "confidence": 0.76, - "canonical_issue_reason": "#16618 is the broadest and most central user-facing failure in the cluster: it concerns inability to load a model, which is the closest shared theme among the items.", + "summary": "These issues are not duplicates: one is about tokenizer output mismatch, one about authentication/error handling when downloading private models, and one about a model-loading failure for bert-base-uncased. The only loose overlap is around model loading, but the concrete failure modes differ.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 16618 is the broadest of the three and the only one directly about model loading, but it is not a strong canonical match for the other two since their root problems are different.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the strongest representative issue because it has the most general symptom and highest discussion activity, making it the best anchor for this loose cluster.", + "best_issue_reason": "Issue 16618 is the best representative of the cluster only by being the most general/loading-related title; however, it should not be treated as a duplicate target for the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Tokenization differences versus TensorFlow code are a different bug class from failing to load 'bert-base-uncased'; no concrete shared code-path is evident." + "reason": "Tokenizer discrepancies for TAPAS and model-loading failures for bert-base-uncased are different bugs in different code paths." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Both mention model loading/download, but one is specifically a private-model login/error-message issue and the other is a generic inability to load a public model; this is too broad to treat as the same bug." + "reason": "A private-model authentication/download error is not the same underlying issue as a generic inability to load bert-base-uncased." } ] }, @@ -18478,7 +18536,7 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "90d4204ed9cbea65b1dbceac749fbea434fcbe1f6342ff45bc59b305f68f24e7", + "prepared_review_unit_hash": "372ee279eae1739978e3a99287d76ce7c84f558c7cf1d669dcbcde910f5d77b6", "component_budget": { "node_count": 2, "item_count": 2, @@ -18500,7 +18558,7 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two distinct community-contribution requests in different feature areas: Flash Attention 2 architecture support vs GGUF architecture support. They are not duplicates and should stay separate.", + "summary": "These are not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They share only a broad \u201csupport more architectures\u201d theme, not the same underlying change.", "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, @@ -18511,7 +18569,7 @@ "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different feature requests affecting different code paths and architectures; same broad theme of 'support for more architectures' is too generic to indicate the same underlying change." + "reason": "Different features in different code paths: Flash Attention 2 support vs GGUF support. Similar high-level goal, but not the same bug or change." } ] }, @@ -18570,184 +18628,184 @@ "issue:44370|issue:46226", "issue:46222|issue:46226" ], - "prepared_review_unit_hash": "06c10cdf32ce329438739a5dd74d0b9264a2e368d5bfffb874df251150c43fc9", + "prepared_review_unit_hash": "9ae707f4bf124ef02dc2759b13264d3b7905c769679c58580cce26046cc46daf", "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": 13526, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13521, - "estimated_input_tokens": 3381, - "estimated_eval_tokens": 7018 + "serialized_chars": 13526, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster splits into two clear groups: separate language-specific translation requests, and a smaller placeholder/template subgroup. I treat only the strongest same-title duplicates as merges and keep the language-specific issues separate.", - "confidence": 0.76, - "canonical_issue_reason": "43834 is the best canonical for the placeholder/template subgroup because it is the earliest generic template-style translation request and sits at the center of the near-identical placeholder cluster.", + "summary": "46222/46226 is a clear exact duplicate pair. The generic i18n template issues 43834/44365/44370 also look like duplicates of the same underlying request based on identical placeholder titles and near-identical similarity scores. The remaining language-coded translation issues do not have enough evidence here to safely collapse into those groups.", + "confidence": 0.69, + "canonical_issue_reason": "issue:46222 is the strongest canonical issue because it has the clearest exact-duplicate match (46226) plus inbound references, making it the most defensible representative in this cluster.", "canonical_pr_reason": null, - "best_issue_reason": "43834 is the best global issue to keep as the representative of the template-placeholder cluster; the other language-specific translation requests should remain separate.", + "best_issue_reason": "issue:46222 is the best global issue representative because it anchors the only explicit exact-duplicate pair and has some outside references/activity; the generic template issues are highly similar but less clearly established from the packet alone.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Turkish translation request vs generic template placeholder; same broad topic is not enough." + "reason": "Both are docs-translation issues, but the packet does not show they are the same underlying request." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Turkish translation request is distinct from the template placeholder issue." + "reason": "High template similarity alone is not enough to prove a duplicate here." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Concrete Turkish docs translation is not the same as the generic placeholder request." + "reason": "The titles differ materially and the evidence does not establish the same request." }, { "left": "issue:27088", "right": "issue:38435", "accept": false, - "reason": "Turkish and Romanian translation requests are separate language-specific issues." + "reason": "These are different language-coded translation requests; the packet does not support merging them." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Turkish and Spanish translation requests are different underlying requests." + "reason": "Turkish and Spanish translation requests are distinct and not shown to be the same issue." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Spanish translation request does not match the generic template placeholder issue." + "reason": "Similarity is too indirect to conclude these are duplicates." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Spanish request is separate from the template placeholder issue." + "reason": "The packet does not show the same underlying translation request." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Spanish and Romanian translation requests are separate issues." + "reason": "Spanish and Romanian translation requests are separate language-specific tasks." }, { "left": "issue:28936", "right": "issue:44370", "accept": false, - "reason": "Spanish request is distinct from the template placeholder issue." + "reason": "Shared docs/i18n wording is not enough to merge them as duplicates." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Romanian translation request and template placeholder issue are not the same change." + "reason": "The Romanian issue and the generic template issue are not clearly the same request from the available evidence." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Romanian request is separate from the template placeholder issue." + "reason": "The packet does not show that these are the same translation request." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Romanian docs translation is distinct from the generic template issue." + "reason": "The evidence is suggestive but still insufficient to treat them as the same issue." }, { "left": "issue:38435", "right": "issue:46222", "accept": false, - "reason": "Romanian and Thai-titled translation requests are different language-specific issues." + "reason": "Different titles and no direct evidence of a shared underlying request." }, { "left": "issue:38435", "right": "issue:46226", "accept": false, - "reason": "Romanian and Thai-titled translation requests are separate requests." + "reason": "Different titles and no direct evidence of a shared underlying request." }, { "left": "issue:43834", "right": "issue:44370", "accept": true, - "reason": "Same placeholder template request; titles and metadata are effectively the same duplicate family." + "reason": "Identical placeholder-style title and very high similarity strongly suggest the same underlying template issue." }, { "left": "issue:43834", "right": "issue:44365", "accept": true, - "reason": "Same placeholder template request with nearly identical metadata; likely duplicate reports." + "reason": "The titles are identical placeholders and the similarity score is near-perfect, consistent with a duplicate." }, { "left": "issue:43834", "right": "issue:46222", "accept": false, - "reason": "Template placeholder issue vs Thai-language request; different underlying requests." + "reason": "Despite high similarity, the available evidence is not enough to treat the generic template issue and the Thai issue as the same request." }, { "left": "issue:43834", "right": "issue:46226", "accept": false, - "reason": "Template placeholder issue vs Thai-language request; not the same report." + "reason": "Despite high similarity, the available evidence is not enough to treat the generic template issue and the Thai issue as the same request." }, { "left": "issue:44365", "right": "issue:44370", "accept": true, - "reason": "Same placeholder template issue; no meaningful distinction visible in the packet." + "reason": "Near-identical generic i18n template titles and extremely high similarity make this look like a duplicate." }, { "left": "issue:44365", "right": "issue:46222", "accept": false, - "reason": "Placeholder template issue vs Thai-language request; different issue family." + "reason": "No direct evidence that the generic template issue and the Thai issue are the same underlying request." }, { "left": "issue:44365", "right": "issue:46226", "accept": false, - "reason": "Placeholder template issue vs Thai-language request; different issue family." + "reason": "No direct evidence that the generic template issue and the Thai issue are the same underlying request." }, { "left": "issue:44370", "right": "issue:46222", "accept": false, - "reason": "Placeholder template issue vs Thai-language request; different issue family." + "reason": "The packet does not establish that the generic template issue and the Thai issue are duplicates." }, { "left": "issue:44370", "right": "issue:46226", "accept": false, - "reason": "Placeholder template issue vs Thai-language request; different issue family." + "reason": "The packet does not establish that the generic template issue and the Thai issue are duplicates." }, { "left": "issue:46222", "right": "issue:46226", "accept": true, - "reason": "Same Thai-language title and matching metadata strongly suggest a duplicate report of the same issue." + "reason": "Same exact Thai title and identical body length make this a clear duplicate pair." } ] }, "evaluator_result": { "accept": false, - "feedback": "The overall split between language-specific translation requests and the placeholder-template cluster is plausible, but the writeup overstates the evidence in a few places. In particular, the packet does not include bodies, so claims like \"titles and bodies appear effectively identical\" and \"near-identical body\" are not grounded. Also, issue 43834\u2019s title is not an exact match to 44365/44370; it is a template version, not the same title. Because the soft-edge accepts rely on stronger evidence than the packet provides, I can\u2019t accept this as fully conservative." + "feedback": "The 46222/46226 duplicate call is grounded, but the rest of the summary overreaches. The packet only shows identical generic i18n template titles for 43834/44365/44370 and very high similarity scores; it does not show the claimed language-specific differences. Treating all of those as clearly separate requests is not well supported, so several soft-edge false verdicts are too strong for the evidence provided." }, "evaluator_used": true, "retried": true, @@ -18829,274 +18887,280 @@ "issue:42503|issue:46653", "issue:45027|issue:46653" ], - "prepared_review_unit_hash": "0e021a128627ec30454fdb680cb8068df42e9737debfb8a4779db0dd33263d7c", + "prepared_review_unit_hash": "f6bb6e4f55e7b9ec19f79a19ce51c3366db7567871237c1ddee4071ccf3f4993", "component_budget": { "node_count": 18, "item_count": 18, "soft_pair_count": 40, - "serialized_chars": 24150, - "estimated_input_tokens": 6038, - "estimated_eval_tokens": 12332 + "serialized_chars": 24143, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 }, "budget": { "node_count": 18, "item_count": 18, "soft_pair_count": 40, - "serialized_chars": 24150, - "estimated_input_tokens": 6038, - "estimated_eval_tokens": 12332 + "serialized_chars": 24143, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is only loosely related by theme (model-support requests), but the items describe different models and different implementation targets. I found no true duplicates or mergeable PR-equivalents.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "This cluster is a mixed bag of unrelated model-addition requests (different model families, modalities, and features). They are superficially similar because many titles say \"Add model/support\", but they do not describe the same underlying bug or change, so they should not be deduplicated together.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue is a clear duplicate canonical for the cluster because the items request different models/features (e.g. EuroBert, RF-DETR, NeoBERT, MiniCPM3, Qwen3-TTS/ASR, etc.).", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "If forced to pick a representative, issue 45322 is one of the clearer, well-scoped model request reports with a concrete target and useful context; however, it is not a true canonical for the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different requests: EuroBert config addition vs DEIMv2 support; separate model families and code paths." + "reason": "EuroBert config support and DEIMv2 are unrelated model requests." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert config support and Qwen3-TTS support are unrelated model additions." + "reason": "EuroBert and Qwen3-TTS are different models and different modalities." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "EuroBert and RF-DETR are different model requests with no shared underlying change." + "reason": "EuroBert and RF-DETR are unrelated architecture additions." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are distinct models; not the same bug/change." + "reason": "RF-DETR and DEIMv2 are different detection model requests, not the same change." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "RF-DETR vision model support and LFM2.5 Audio support are unrelated." + "reason": "RF-DETR is a vision detection model; LFM2.5 Audio is an audio model request." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR and OpenAI weight-sparse transformer are separate requests." + "reason": "RF-DETR and OpenAI weight-sparse transformer are distinct model families and code paths." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR target different model families and implementation paths." + "reason": "RF-DETR and Qwen3-ASR are different tasks/models." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR support is unrelated to Qwen3-TTS support." + "reason": "RF-DETR and Qwen3-TTS are unrelated." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are different model additions; no single concrete fix." + "reason": "NeoBERT and Qwen3-ASR are different model support requests." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT support and circuit-sparsity transformer support are unrelated changes." + "reason": "NeoBERT and circuit-sparsity transformer are not the same underlying model change." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT and LFM2.5 Audio are distinct model requests." + "reason": "NeoBERT is a BERT-style LM; LFM2.5 Audio is an audio model." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are different models; not duplicates." + "reason": "RF-DETR and NeoBERT are unrelated model additions." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are separate model integrations." + "reason": "NeoBERT and ModernVBERT are different BERT-like models, not duplicates." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 architecture support and Qwen3-ASR support are different requests." + "reason": "MiniCPM3 architecture support and Qwen3-ASR are different model requests." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 and circuit-sparsity transformer changes do not share the same implementation target." + "reason": "MiniCPM3 support and weight-sparse transformer support are unrelated." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are unrelated model additions." + "reason": "MiniCPM3 and ModernVBERT are different model families." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 and Voxtral TTS are different model support requests." + "reason": "MiniCPM3 and Voxtral TTS support are different modalities/models." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are different model families with separate code paths." + "reason": "DEIMv2 and Qwen3-TTS are unrelated requests." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are unrelated support requests." + "reason": "ModernVBERT and Qwen3-ASR are different model additions." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are distinct model additions." + "reason": "ModernVBERT and Qwen3-TTS are not the same change." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT support and generic BERT-like RoPE support are not the same concrete change." + "reason": "ModernVBERT and BERT-like models with RoPE are related only loosely; they are not the same concrete request." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE support and Voxtral TTS support are unrelated." + "reason": "RoPE-enabled BERT-like models and Voxtral TTS are unrelated." }, { "left": "issue:42738", "right": "issue:46653", "accept": false, - "reason": "BERT-like RoPE support and a non-technical issue title are clearly not duplicates." + "reason": "BERT-like RoPE support and a profile/name issue have no overlap." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "TTS and ASR support requests for different Qwen3 modalities; not one underlying fix." + "reason": "Qwen3-TTS and Qwen3-ASR share a prefix but target different speech tasks and likely different implementations." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and Voxtral TTS are different model integrations." + "reason": "Qwen3-TTS and Voxtral-4B-TTS are different TTS model support requests." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "Qwen3-TTS support and DeepSeek-OCR2 support are unrelated requests." + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model requests." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS and weight-sparse transformer support are different code paths." + "reason": "Qwen3-TTS and weight-sparse transformer are different model changes." }, { "left": "issue:43671", "right": "issue:43909", "accept": false, - "reason": "Qwen3-TTS and LFM2.5 Audio are different audio-model requests." + "reason": "Qwen3-TTS and LFM2.5 Audio are different audio model requests." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio support are distinct model requests." + "reason": "Qwen3-ASR and LFM2.5 Audio are different speech/audio model requests." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR support and EUPE support are unrelated model additions." + "reason": "Qwen3-ASR and EUPE are different model requests with no shared code path shown." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "Qwen3-ASR and circuit-sparsity transformer support are not the same change." + "reason": "Qwen3-ASR and circuit-sparsity transformer are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different speech tasks/models." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are different model support requests." + "reason": "LFM2.5 Audio and SAM3-LiteText are different model domains." }, { "left": "issue:43909", "right": "issue:46645", "accept": false, - "reason": "LFM2.5 Audio support and Granite SWA+sink option are unrelated." + "reason": "LFM2.5 Audio support and Granite SWA+sink are different changes; one is model support, the other is an architectural option." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Weight-sparse transformer support and EUPE support are different implementations." + "reason": "OpenAI weight-sparse transformer and EUPE are unrelated model requests." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are different audio model requests." + "reason": "Qwen3-ASR and Voxtral-4B-TTS are different speech model requests." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are unrelated model additions." + "reason": "NeoBERT and Voxtral-4B-TTS are unrelated model additions." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR support and DeepSeek-OCR2 support are separate model requests." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model domains." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are different vision-language model requests." + "reason": "EUPE and TIPSv2 are different Meta/Google model requests with different purposes." }, { "left": "issue:42503", "right": "issue:46653", "accept": false, - "reason": "ModernVBERT support and a non-model/person issue are not duplicates." + "reason": "ModernVBERT models and a closed personal-name issue are unrelated." }, { "left": "issue:45027", "right": "issue:46653", "accept": false, - "reason": "Voxtral TTS support and a non-technical issue title are clearly unrelated." + "reason": "Voxtral TTS support and a personal-name issue are unrelated." } ] }, @@ -19126,7 +19190,7 @@ "soft_pairs": [ "issue:38740|issue:46443" ], - "prepared_review_unit_hash": "32af683614a4692bee049039917cde43e43ff393f35adb97b43e01c137bc6977", + "prepared_review_unit_hash": "2f975088240b738faec55763a42075a0601ee2668ab1de67f7b7fe7200faed28", "component_budget": { "node_count": 2, "item_count": 2, @@ -19148,18 +19212,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are unrelated: one is a docs request to add `pruna` as an optimization framework, the other is a Vision RoPE refactor. The soft similarity is a false positive; there is no duplicate relationship here.", + "summary": "The two issues are unrelated: one is a documentation request about adding `pruna` as an optimization framework, and the other is a Vision RoPE refactor. No duplicate relationship is supported.", "confidence": 0.99, - "canonical_issue_reason": null, + "canonical_issue_reason": "If a single issue must represent the cluster, issue 46443 is the better representative because it is a concrete code refactor rather than a docs request; however, the pair is not a duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46443 is the more substantive engineering issue and the better representative of the cluster, but it is not a duplicate of the docs request.", + "best_issue_reason": "Issue 46443 is the most suitable standalone issue because it describes an actual implementation refactor, whereas issue 38740 is only a documentation addition request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38740", "right": "issue:46443", "accept": false, - "reason": "Different topics and likely different code paths: documentation for an optimization framework vs a Vision RoPE refactor. No explicit linkage, shared targets, or file overlap." + "reason": "Different topics and artifact types: a docs request for adding `pruna` versus a Vision RoPE refactor. They do not describe the same underlying bug or change." } ] }, @@ -19189,12 +19253,12 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "89134da8c569617b2f514622b53946dd8c9bc67401c54ef4d6daf9e20686c70c", + "prepared_review_unit_hash": "e5075a69797bcde4ee5af32aaae787ffbd67a4c9b6c9b3eb53ebc515496789ff", "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 }, @@ -19202,7 +19266,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1479, + "serialized_chars": 1480, "estimated_input_tokens": 370, "estimated_eval_tokens": 996 }, @@ -19211,18 +19275,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues do not appear to describe the same underlying problem: one is about set decoding behavior, the other requests DeepSeek v3.2 model support. The soft similarity is too weak to treat them as duplicates.", - "confidence": 0.96, + "summary": "The two issues appear unrelated: one is about set block decoding mechanics, the other about adding DeepSeek v3.2 support. There is no evidence they describe the same underlying bug or change.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "If a representative issue is needed, \"DeepSeek v3.2 support\" is the more concrete feature request, but it is not a duplicate of the block decoding issue.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "Different scopes and outcomes: decoding implementation vs adding support for a specific model/version. No shared code path or explicit linkage indicating the same bug/change." + "reason": "Different problem domains with no shared files, targets, or explicit linkage; one concerns decoding behavior, the other model support. They do not look like the same underlying issue." } ] }, @@ -19252,7 +19316,7 @@ "soft_pairs": [ "issue:41306|issue:41826" ], - "prepared_review_unit_hash": "cf71513fc3de453ed94bf53e2ad0a6cef7abde2ec394ada8890410861001f454", + "prepared_review_unit_hash": "b1d12d6c466a4d62e69219ff8ab7dc5f941c8f88e3eba4f1f70e406df69a8fb8", "component_budget": { "node_count": 2, "item_count": 2, @@ -19274,18 +19338,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are only loosely related by the word \u201cTiled\u201d and do not describe the same bug or feature request. One is about exposing a `TiledFusedLogitsLoss` / FLCE flag in `from_pretrained()`, while the other requests integrating `TiledMLP` for reduced memory footprint.", + "summary": "These two issues are both about tiled/efficient model components, but they target different features and code paths: FLCE/LigerFusedLinearCrossEntropy flag handling in `from_pretrained()` versus integrating TiledMLP for reduced memory. They are not duplicates.", "confidence": 0.94, - "canonical_issue_reason": "Issue #41306 is the more concrete and actionable report: it names a specific integration point (`from_pretrained()`), a specific feature flag, and a specific fused loss implementation, making it a better anchor for this cluster.", + "canonical_issue_reason": "Issue 41306 is the more concrete and actionable report, focused on a specific loading/configuration gap for TiledFusedLogitsLoss in `from_pretrained()`.", "canonical_pr_reason": null, - "best_issue_reason": "#41306 is the best issue representative because it is narrowly scoped, implementation-oriented, and has higher discussion activity, suggesting clearer triage value than the broader TiledMLP memory-footprint request.", + "best_issue_reason": "Issue 41306 is the better representative because it names a precise component, API surface, and expected behavior, making it easier to triage and resolve than the broader TiledMLP integration request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41306", "right": "issue:41826", "accept": false, - "reason": "Different underlying changes: one concerns enabling FLCE/TiledFusedLogitsLoss via `from_pretrained()`, the other is about adding TiledMLP for memory savings. They share a subsystem name but not the same concrete problem or fix path." + "reason": "They concern different mechanisms and fixes: one is about enabling a loss-function flag during model loading, the other about adding TiledMLP for memory reduction. Same general theme, but not the same underlying bug/change." } ] }, @@ -19315,12 +19379,12 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "8d9fe6ccd9ea05c22c4a3f00437f886dc4f50c03c02c0f167810cb81c7bab0d7", + "prepared_review_unit_hash": "69d579f1bde9b2e53a2df7d800eca2806aebb6536ecec240321185786a5d5fc7", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1583, + "serialized_chars": 1584, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -19328,7 +19392,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1583, + "serialized_chars": 1584, "estimated_input_tokens": 396, "estimated_eval_tokens": 1048 }, @@ -19337,18 +19401,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two closed issues are both about Trainer customization, but they describe different problems: one is about `remove_unused_column` behavior, the other about FLOPS estimation for custom trainers. They should not be treated as duplicates.", - "confidence": 0.9, - "canonical_issue_reason": "No single canonical issue fits this cluster well; the issues concern separate Trainer code paths and different user-facing problems.", + "summary": "These two issues share only a loose theme around custom trainers, but they describe different Trainer behaviors: one is about `remove_unused_column` customizability, the other about FLOPS estimation. They should not be deduplicated.", + "confidence": 0.91, + "canonical_issue_reason": "issue #43598 is the broader, more central Trainer customizability request; issue #43600 is a separate FLOPS-estimation concern.", "canonical_pr_reason": null, - "best_issue_reason": "If forced to pick one representative, #43598 is the better cluster anchor because it is a more direct Trainer customization request and is broader in scope than the FLOPS-estimation-specific #43600.", + "best_issue_reason": "#43598 is the better representative issue because it focuses on a core Trainer customization behavior rather than a narrower metrics/estimation edge case.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Related by broad Trainer customization theme, but they target different behaviors (`remove_unused_column` customization vs FLOPS accounting for custom trainers) and are not the same underlying bug or change." + "reason": "Both mention custom trainers, but they target different code paths and user problems: unused-column removal vs FLOPS counting. They are not the same underlying bug or change." } ] }, @@ -19378,12 +19442,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "1b88eacb86d8acf2ad44340e2fd9da407f9a29707171736ab72cc64733d1f828", + "prepared_review_unit_hash": "efda7dcd3abfb8c99ad78b19bbac78b0a60ac9591dcefe8141a9d31d4f13cf17", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1610, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -19391,7 +19455,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1610, + "serialized_chars": 1611, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -19400,18 +19464,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related only at a very high level (both ask for Qwen model support), but they target different model families and different AutoModel heads, so they are not duplicates.", - "confidence": 0.92, - "canonical_issue_reason": "Issue 43684 is the broader, more foundational request: it asks for model registration in AutoModel/AutoModelForConditionalGeneration, which is a core missing integration point.", + "summary": "These are related at a very high level (both ask for new Qwen model support in auto classes), but they target different model families and different APIs, so they should not be treated as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue #43684 is the broader registration request and is the better anchor if one issue must represent the cluster, but it is still distinct from #44405 rather than a true duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43684 is the best representative issue because it addresses the general registration gap rather than a single task-specific head.", + "best_issue_reason": "#43684 is the more general and reusable request: it asks for AutoModel and AutoModelForConditionalGeneration registration for Qwen3-Omni, which is closer to a canonical tracking issue than the narrower Qwen3.5 sequence-classification request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different targets and scope: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 sequence-classification support. They are not the same underlying bug or change." + "reason": "Different underlying changes: one is registration for Qwen3-Omni in AutoModel/AutoModelForConditionalGeneration, the other is adding AutoModelForSequenceClassification support for Qwen3.5. Same broad subsystem, but not the same bug or mergeable fix." } ] }, @@ -19448,12 +19512,12 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "740ce29b34e46a0629cf53bce68f9ef820a29b801bfc9de5d78b10b86fd1f0a7", + "prepared_review_unit_hash": "4553bcd22152f2b27407144da2bb8ae8b173fd1686c5074d173f236e4ad7b94c", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4265, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -19461,7 +19525,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 6, - "serialized_chars": 4267, + "serialized_chars": 4265, "estimated_input_tokens": 1067, "estimated_eval_tokens": 2390 }, @@ -19470,54 +19534,54 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three closed issues share the exact title \"huggingface\" and look like duplicates of the same very terse report; issue 43835 (\"Fett\") appears unrelated. I\u2019d keep the oldest huggingface issue as the likely canonical issue.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 43845 is the oldest of the near-identical \"huggingface\" issues and is the best placeholder canonical issue for that duplicate cluster.", + "summary": "Three issues are effectively indistinguishable low-content reports titled \"huggingface \"; the \"Fett\" issue appears separate. Among the huggingface trio, the earliest report is the best canonical candidate.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface \" reports, so it is the best anchor for the duplicate cluster.", "canonical_pr_reason": null, - "best_issue_reason": "43845 is the best issue to represent the cluster because it is the earliest exact-title duplicate among the huggingface reports.", + "best_issue_reason": "Issue 43845 is the strongest global issue candidate because it is the first of the repeated identical-title reports and there is no richer signal to prefer another.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Different titles and no evidence they describe the same problem; 43835 looks unrelated." + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Different titles and no supporting detail tying them to the same underlying issue." + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Different titles and no supporting detail tying them to the same underlying issue." + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." }, { "left": "issue:43845", "right": "issue:43846", "accept": true, - "reason": "Same exact title and both are terse closed issues with very similar shape, so they plausibly duplicate the same report." + "reason": "Same title, near-identical low-content reports, and created one minute apart; likely the same underlying issue." }, { "left": "issue:43845", "right": "issue:43859", "accept": true, - "reason": "Same exact title and similarly minimal issue content; likely the same underlying report." + "reason": "Same title and similarly sparse reporting; likely duplicate instances of the same issue." }, { "left": "issue:43846", "right": "issue:43859", "accept": true, - "reason": "Same exact title and similarly minimal issue content; likely the same underlying report." + "reason": "Same title, close timing, and no distinguishing details; likely the same underlying issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: issue 43835 has a different title and the three \"huggingface\" issues are the only tight cluster, with 43845 being the earliest by timestamp. The soft-edge verdicts are conservative overall and do not overclaim beyond the available similarity evidence." + "feedback": "The summary is grounded in the packet: 43835 is the outlier by title, while 43845/43846/43859 are the only identical-title reports with similarly sparse bodies. The canonical choice of 43845 as the earliest of that trio is conservative and supported. The soft-edge accepts are also reasonable given the title match and low-content signals, with no unsupported leap beyond the available metadata." }, "evaluator_used": true, "retried": false, @@ -19555,7 +19619,7 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "d6493632431101b9dbb22a2a2c48b692ad27e55000ea18ad707503c8b221b210", + "prepared_review_unit_hash": "32482ae257a5a9107203e013c27e2634044454120aa2bcb0f171b26d6176e601", "component_budget": { "node_count": 5, "item_count": 5, @@ -19577,66 +19641,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All items are low-signal spam or generic placeholder issues with no substantive bodies, targets, or references, so there is no credible duplicate cluster to merge.", + "summary": "These are low-signal, likely unrelated placeholder/spam issues with no substantive evidence of a shared bug or feature request. No soft duplicates should be merged.", "confidence": 0.93, - "canonical_issue_reason": "If a representative must be chosen, issue 44418 is the least placeholder-like and has the most descriptive title; the rest are obvious spam/generic titles.", + "canonical_issue_reason": "issue:44371 is the earliest and most generic placeholder title, but it is not a strong canonical target because the cluster lacks a real shared problem statement.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44418 is the best representative purely by title quality, but it still lacks any concrete bug report content.", + "best_issue_reason": "issue:44371 is the least specific and earliest of the spam-like issues, so it is only the best placeholder representative, not a true duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are placeholder spam issues; no shared bug, code path, or actionable description to indicate the same underlying problem." + "reason": "Both are generic '' placeholders, but there is no underlying bug/change to unify." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Both are placeholder spam issues with no substantive content; similarity is only superficial." + "reason": "Same placeholder pattern, but these are not substantive duplicate reports." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "One is a spam placeholder and the other is a generic documentation-suggestion title; too little evidence of the same issue." + "reason": "Titles are different and there is no shared problem description beyond weak similarity." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "The titles are unrelated spam/generic text and there is no concrete shared bug report." + "reason": "No shared technical issue; 'Google LLC' is unrelated to the spam placeholder." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "Generic documentation-suggestion versus spam placeholder; no underlying defect or change is described." + "reason": "Documentation suggestion vs. spam placeholder: not the same underlying issue." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "Generic documentation-suggestion versus spam placeholder; not the same underlying issue." + "reason": "Documentation suggestion vs. spam placeholder: not a duplicate pair." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "Both are spam placeholders, but that alone is not enough to treat them as the same underlying bug or request." + "reason": "Identical spam titles, but they do not describe a concrete mergeable bug or change." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder and unrelated quoted-name title; no concrete evidence of a shared problem." + "reason": "No common issue content; one is spam, the other is an unrelated named topic." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "Spam placeholder versus unrelated quoted-name title; no substantive overlap." + "reason": "No common issue content; one is spam, the other is an unrelated named topic." } ] }, @@ -19666,7 +19730,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "1904f00edba13568725bfeb3383e37765647613ebab351c1921cefa60e0fba37", + "prepared_review_unit_hash": "f4da3f585eb1084379aaac9e6fe218cb5efac4ab97a2fbdcb84c44ea125472ef", "component_budget": { "node_count": 2, "item_count": 2, @@ -19688,24 +19752,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues have identical titles and a soft similarity signal, but the packet does not provide enough body or linkage evidence to confidently treat them as duplicates. They are likely related, but the duplicate claim should remain conservative.", - "confidence": 0.68, - "canonical_issue_reason": "If a single anchor is needed, #45145 is the better canonical candidate because it was created earlier and has an inbound reference. However, the evidence is not strong enough to state definitively that it is a duplicate of #45151.", + "summary": "The two issues look like a likely duplicate pair: they have the same title, similar body sizes, and moderate soft-similarity, but the evidence is not strong enough to call them near-identical with high confidence. No PRs are involved.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #45145 is the better canonical because it was created first and already has an inbound reference, so it is the more established thread.", "canonical_pr_reason": null, - "best_issue_reason": "#45145 is the best issue to represent the pair for triage because it is earlier and already referenced, but only as a provisional anchor rather than a confirmed duplicate.", + "best_issue_reason": "Issue #45145 is the strongest representative of the cluster for the same reason: earlier creation and existing reference activity.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": false, - "reason": "Identical titles and soft similarity alone are not enough to conclude these are duplicates without body text overlap or explicit linkage." + "reason": "Same title and moderate soft-similarity suggest a likely duplicate, but the evidence is not strong enough to confirm the same underlying issue. With only Jaccard 0.63, similar body length, and no explicit links or shared targets, this should stay conservative." } ] }, "evaluator_result": { "accept": false, - "feedback": "The packet supports that the titles are identical and the pair has soft similarity, but it does not provide enough evidence to claim the bodies are highly similar or that the issues are definitively duplicates. The soft-edge accept is too strong given only a soft_similarity signal and no body text or explicit linkage. Please keep the verdict more conservative." + "feedback": "The canonical-selection rationale is grounded (issue #45145 was created earlier and has an inbound reference), but the cluster summary and soft-edge verdict are a bit too strong for the evidence. We only have identical titles plus soft-similarity/Jaccard 0.63 and similar body lengths, which supports a likely duplicate but not confidently a \"near-identical\" or definitively same underlying issue. Please soften the wording and keep the verdict conservative." }, "evaluator_used": true, "retried": true, @@ -19786,40 +19850,40 @@ "soft_pairs": [ "issue:45672|issue:45674" ], - "prepared_review_unit_hash": "efb2a4e17b913885a0b7ff26ffde5cdba9995324cacadf269bcf398caa9cac7f", + "prepared_review_unit_hash": "ae5adbbf6ccb2b0d596567cf565d58b4aa0f02bfea6a29c0e6f288a609f988ee", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1672, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1672, - "estimated_input_tokens": 418, - "estimated_eval_tokens": 1092 + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two different 4-bit inference bugs: one is a Gemma4/audio-module uint8 torch.finfo() TypeError, the other is a BitsAndBytesConfig skip-modules regression that removes the default lm_head exclusion and triggers an AssertionError. They share a broad quantization context but not the same underlying failure.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 45674 is the better representative because it describes a specific, reproducible configuration regression in BitsAndBytesConfig with a clear assertion failure and actionable scope.", + "summary": "These are related to 4-bit inference regressions, but they describe different failure modes and likely different fixes: one is a torch.finfo()/uint8 path in Gemma4 audio modules, the other is a BitsAndBytesConfig skip-module interaction that drops the default lm_head exclusion and triggers an AssertionError.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 45674 is the strongest standalone issue: it names the offending API behavior, the impacted inference mode, and the likely root cause more concretely than the Gemma4-specific error report.", + "best_issue_reason": "Issue 45674 is the more general and clearly described 4-bit inference regression, but it is not a duplicate of 45672 because the root cause and failing code path differ.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Different bug classes and code paths: torch.finfo() TypeError in Gemma4 audio modules vs. lm_head exclusion being cleared by BitsAndBytesConfig. Shared 4-bit inference context is too broad to treat as the same issue." + "reason": "Different bugs: 45672 is a torch.finfo TypeError on uint8 weights in Gemma4 audio modules; 45674 is a BitsAndBytesConfig skip-module bug that removes the default lm_head exclusion and causes an AssertionError. Same broad area, not the same underlying defect." } ] }, @@ -19849,12 +19913,12 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "7be78913af3a9eff3989803df23e565070c70233caac61a0dff082260793870e", + "prepared_review_unit_hash": "e6c47a131ec34fc98fa7d01746135dd745bb49b4d6e9b001d9dc0db8b171c3a9", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1474, + "serialized_chars": 1476, "estimated_input_tokens": 369, "estimated_eval_tokens": 994 }, @@ -19862,7 +19926,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1474, + "serialized_chars": 1476, "estimated_input_tokens": 369, "estimated_eval_tokens": 994 }, @@ -19871,18 +19935,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues do not describe the same bug or change; they are just distinct name-and-date entries with no shared technical content.", - "confidence": 0.98, - "canonical_issue_reason": null, + "summary": "The cluster contains two unrelated closed issues with different person names and different dates; there is no evidence they describe the same underlying problem or request.", + "confidence": 0.97, + "canonical_issue_reason": "Neither issue has any body, references, or shared targets/filenames; the titles refer to different names and dates, so there is no clear canonical duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Both issues are equally sparse, but issue #45762 is marginally a better canonical placeholder only because it appears first; there is no substantive basis to prefer it.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "Different names and dates, no bug report text, no shared technical context, and no evidence they refer to the same underlying issue." + "reason": "The titles identify different people and different dates, and there is no supporting discussion or metadata tying them to the same bug or change." } ] }, @@ -19912,40 +19976,40 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "77751ffdb2171cd357a21a44e1ba424fadd9437abcbeeed88e14fdeed1f19275", + "prepared_review_unit_hash": "50d5e79d5ce959435080f868ebbc0b3a4c15a5c235cbcac1ede8082c0495605e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1481, - "estimated_input_tokens": 371, - "estimated_eval_tokens": 998 + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items are closed, contentless issues with unrelated personal-name titles; they do not appear to describe the same bug or request. This looks like two separate spam/placeholder issues rather than duplicates.", - "confidence": 0.95, - "canonical_issue_reason": null, + "summary": "These two closed issues do not look like duplicates or part of the same technical problem; they appear to be unrelated name/date entries.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue candidate: the issues are unrelated and do not describe the same bug or request.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Neither issue is a good representative for the cluster because there is no coherent shared technical topic.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45764", "right": "issue:45765", "accept": false, - "reason": "The titles are different names and dates with no shared technical problem, code path, or explicit linkage; not the same underlying issue." + "reason": "The titles are different personal-name/date strings with no shared technical context, so they do not appear to be the same underlying issue." } ] }, @@ -19975,40 +20039,40 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "063bd2a6620e0b7761117b42f1ed2f2ae995572d46fc00d42ef5ba70dad79f9b", + "prepared_review_unit_hash": "652ec25c436609ca5749939274f16b52ebc135640e20d57f9933be63eb929fd7", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1541, - "estimated_input_tokens": 386, - "estimated_eval_tokens": 1028 + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1541, - "estimated_input_tokens": 386, - "estimated_eval_tokens": 1028 + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one requests a new Qwen3_5 token-classification model class, while the other asks for FP8 training support. They target different features and code paths.", - "confidence": 0.98, - "canonical_issue_reason": "If a single representative is needed, #45810 is the more specific and concrete feature request; #46295 is a separate training-capability request, not the same underlying problem.", + "summary": "The two issues are not duplicates: one requests a Qwen3_5 token-classification model for value-model use, while the other asks for FP8 training support. They concern different features and code paths.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "#45810 is the better standalone issue because it is narrower and more actionable, but it should not be merged with #46295.", + "best_issue_reason": "issue #45810 is the more concrete and actionable request; #46295 is broader and more general.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different requests: adding a token-classification model class vs. enabling FP8 training support. They do not describe the same bug or change." + "reason": "Different underlying requests: model-class support for token classification vs FP8 training support. No shared bug/change target." } ] }, @@ -20048,12 +20112,12 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "6ac8a0fe449bb5b92daa7b25286b23c0c1442deb622be5f01ffdd6db132a76e7", + "prepared_review_unit_hash": "bd7ee2424c837fa074de051476e5f92cfba4b1b748ba0528bb977c7aec02e9ed", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5725, + "serialized_chars": 5728, "estimated_input_tokens": 1432, "estimated_eval_tokens": 3120 }, @@ -20061,7 +20125,7 @@ "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5725, + "serialized_chars": 5728, "estimated_input_tokens": 1432, "estimated_eval_tokens": 3120 }, @@ -20070,67 +20134,70 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "There are two exact duplicate issue pairs here: 45870/45871 and 45872/45873. Issue 45869 is related only by broad financial-data wording, not the same underlying request. No soft-edge pair should be merged across those topic boundaries.", - "confidence": 0.86, - "canonical_issue_reason": "45870 is a clean representative for the exact duplicate pair 45870/45871: same title and body size, with no evidence that it overlaps the separate 45872/45873 request or the 45869 dataset listing.", + "summary": "Two exact duplicate issue pairs stand out: 45870/45871 and 45872/45873. Issue 45869 is a separate dataset-promo post and is not the same underlying request. The example-request pair and the scraping-service pair are related only by theme ('financial time series') but describe different artifacts and should not be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue for the full set; the data splits into two separate exact-duplicate pairs (45870/45871 and 45872/45873) plus an unrelated dataset announcement (45869).", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the best issue to keep as the canonical representative because it is an exact duplicate of 45871 and is more narrowly tied to a single concrete example request than the broader, separate 45872/45873 service-oriented topic.", + "best_issue_reason": "If one issue must be chosen as the best representative, 45870 is a reasonable pick because it is the earlier half of an exact duplicate pair and has the clearest concrete example-request title.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Both are financial-data themed, but one is a dataset listing and the other is an example request; not the same underlying issue." + "reason": "Dataset/sample promo vs example/tutorial request; same broad finance theme, but not the same underlying issue." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Same broad domain only. A dataset bundle announcement is not a duplicate of an example/pattern request." + "reason": "Dataset/sample promo vs the duplicate example request; unrelated artifact types and goals." }, { "left": "issue:45870", "right": "issue:45871", - "accept": false, - "reason": "These look like the same issue textually, but the soft edge is not needed for acceptance here; as a duplicate triage pair it would normally be merged only if the bodies confirm identical content. With no body text, keep this rejected to avoid over-merging." + "accept": true, + "reason": "Exact same title and substantially the same body; clear duplicate pair." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Different concrete requests: one is a prediction example, the other adds a scraping service. Same broad financial topic, but not the same change." + "reason": "Both mention financial time series, but one is an example request and the other is a scraping-service request; different underlying change." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Different concrete requests; no evidence they describe the same underlying bug or change." + "reason": "Different concrete request: example/tutorial vs scraping service integration; not mergeable as one issue." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Prediction example vs scraping-service request are distinct artifacts despite shared domain wording." + "reason": "Same as 45870/45872: shared theme only, but distinct requested outcomes." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Same as 45871/45872: related subject area, but not the same underlying issue." + "reason": "Same as 45870/45873: not the same bug or change, just a related topic." }, { "left": "issue:45872", "right": "issue:45873", - "accept": false, - "reason": "They are near-exact textual duplicates, but without body confirmation we avoid auto-merging on title similarity alone." + "accept": true, + "reason": "Exact same title and substantially the same body; clear duplicate pair." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The two high-similarity pairs are treated as duplicates only where the evidence is strongest (45870/45871 and 45872/45873), and the cross-pair links are correctly rejected as related-only. 45869 is also reasonably separated as a different dataset-style post rather than the same request." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -20153,7 +20220,7 @@ "soft_pairs": [ "issue:46439|issue:46441" ], - "prepared_review_unit_hash": "cf2d44eeefa159a53f834400a2915fc2bd7097616a8cc4babf125c3795b31baa", + "prepared_review_unit_hash": "9d41aa86d52b0698d6f13a4744065c9f1fa0fcec45ac40579313dc41a23afb40", "component_budget": { "node_count": 2, "item_count": 2, @@ -20175,24 +20242,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both issues involve `cache_implementation=\"static\"` and `linear_attention`, but they describe different failure modes: one is a `KeyError: 'linear_attention'`, the other a shape-corruption `RuntimeError` during `early_initialization`. They look related at the subsystem level, but not clearly like the same underlying bug.", - "confidence": 0.85, - "canonical_issue_reason": "Issue 46439 is the better representative because it names the likely mechanism (`early_initialization`) and a concrete symptom pattern affecting linear-attention layers more broadly.", + "summary": "The two issues are related at the theme level (`cache_implementation=\"static\"` / `linear_attention`), but the packet does not establish a shared root cause. One reports wrong-shape corruption during `early_initialization`, the other a model-specific `KeyError` on Qwen3.5, so they should be clustered cautiously rather than treated as the same defect.", + "confidence": 0.43, + "canonical_issue_reason": "Issue #46439 is the better canonical issue because it is broader and more root-cause-oriented in its wording, covering the static-cache `early_initialization` path and the linear-attention breakage without being tied to one model.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46439 is the most general/root-cause-oriented report and is therefore the best single issue to anchor the cluster, even though the pair should not be deduplicated outright.", + "best_issue_reason": "Issue #46439 is the strongest representative for the cluster since it describes the more general failure mode, while #46441 reads as a narrower manifestation on a specific model and cache setting.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46439", "right": "issue:46441", "accept": false, - "reason": "Shared static-cache/linear-attention context is not enough: 46439 reports wrong tensor shapes and a `RuntimeError` from `early_initialization`, while 46441 reports a `KeyError` for `linear_attention` in Qwen3.5. These could be separate bugs in the same area rather than one fix." + "reason": "Reject as a same-bug match: the overlap is only the static-cache/linear_attention context, and the packet does not show a common root cause or a clearly shared fix path." } ] }, - "evaluator_result": null, - "evaluator_used": false, - "retried": false, + "evaluator_result": { + "accept": false, + "feedback": "The similarity is grounded only in the shared static-cache/linear_attention theme, but the packet does not establish a common root cause or a registry/early_initialization bug. The summary and canonicalization language go a bit too far beyond the titles, so the soft-edge accept should be more tentative rather than asserting the same underlying bug class." + }, + "evaluator_used": true, + "retried": true, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -20216,7 +20286,7 @@ "soft_pairs": [ "issue:46551|issue:46552" ], - "prepared_review_unit_hash": "01da7ff58561300cc537ddb5fa0fd1310e83d3b350d94b8993ca7106b4792b06", + "prepared_review_unit_hash": "20cab18958af5d5eab95ce91f3f947dc4ee4e6593fc8188b772d0553e3ba0be2", "component_budget": { "node_count": 2, "item_count": 2, @@ -20238,24 +20308,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are exact duplicates: same title and body, describing the same Wav2Vec2CTCTokenizer v5 word-delimiter special-token leak. The open issue is the better canonical candidate.", + "summary": "These are duplicate issues with identical titles and matching descriptions about Wav2Vec2CTCTokenizer v5 incorrectly treating the word delimiter as a special token. The open issue is the better canonical tracker.", "confidence": 0.99, - "canonical_issue_reason": "Issue 46552 is the open duplicate and has inbound references, so it is the better canonical tracker.", + "canonical_issue_reason": "Prefer issue #46552 because it is open and already has inbound references, making it the more active tracker for the same bug; #46551 appears to be the older closed duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46552 is the best issue to keep because it is still open and appears to be the actively referenced report.", + "best_issue_reason": "Issue #46552 is the best issue to keep because it is open, currently referenced, and describes the exact same problem as #46551.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46551", "right": "issue:46552", "accept": true, - "reason": "Same title, same body, same underlying tokenizer bug; these are clearly duplicate reports." + "reason": "Same title and same underlying tokenizer bug; the descriptions match and there is no meaningful distinction beyond open vs closed state." } ] }, "evaluator_result": { "accept": true, - "feedback": "The conclusion is grounded in the packet: both issues have the same title, identical body length, and maximum soft similarity, with 46552 open and carrying inbound references. The canonical choice is conservative and reasonable, and nothing materially overstates the evidence." + "feedback": "The summary is grounded in the packet: both issues have identical titles and matching body length, and issue #46552 is open with inbound references while #46551 is closed. The soft-edge verdict is conservative and supported by the evidence." }, "evaluator_used": true, "retried": false, @@ -20282,40 +20352,40 @@ "soft_pairs": [ "issue:46657|issue:46692" ], - "prepared_review_unit_hash": "22a1846b58e7a11a3e2b9799f48df882da024b7532ceadc930dd4a1bb1451287", + "prepared_review_unit_hash": "9f10787b81e226d70eccb0f94704ee017538d1b642809e16f5f5079fe0231c37", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1568, - "estimated_input_tokens": 392, - "estimated_eval_tokens": 1040 + "serialized_chars": 1569, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1568, - "estimated_input_tokens": 392, - "estimated_eval_tokens": 1040 + "serialized_chars": 1569, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues do not look like duplicates: one reports a tensor shape mismatch during object detection training, while the other reports gradual OOM during RT-DETR v2 training. They are different failure modes and likely different root causes.", - "confidence": 0.91, - "canonical_issue_reason": "Issue 46657 is the more concrete bug report and is centered on a distinct, reproducible training error (tensor size mismatch).", + "summary": "These two issues are not duplicates: one reports a tensor shape mismatch during object-detection training, while the other reports an out-of-memory failure after several epochs in RT-DETR v2. They describe different failure modes and likely different root causes.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits because the reports are about different symptoms and different code paths.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46657 is the better representative issue for this cluster because it describes a specific code-path failure rather than a broad resource symptom.", + "best_issue_reason": "Issue 46657 is the more general training-bug report, but it still does not subsume 46692; the latter is a distinct RT-DETR v2 OOM problem.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46657", "right": "issue:46692", "accept": false, - "reason": "Different symptoms and likely different root causes: shape mismatch vs out-of-memory after epochs. The similarity is only at the high-level training context, not the underlying bug." + "reason": "Different bugs: tensor size mismatch vs memory exhaustion after epochs. They are not the same underlying issue." } ] }, @@ -20345,12 +20415,12 @@ "soft_pairs": [ "issue:46822|issue:46823" ], - "prepared_review_unit_hash": "5fbf00bbe14092440530179fff2f6950434a5326b2de9651c7377c1717253639", + "prepared_review_unit_hash": "d12d19e1f71b18f102de502235631b8061e3191a5e1b1aef1924731b88e46266", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1671, + "serialized_chars": 1672, "estimated_input_tokens": 418, "estimated_eval_tokens": 1092 }, @@ -20358,7 +20428,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1671, + "serialized_chars": 1672, "estimated_input_tokens": 418, "estimated_eval_tokens": 1092 }, @@ -20367,18 +20437,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are about different model-loading failure modes in different code paths: ZeRO-3 zero.Init partitioning vs 4-bit device_map/max_memory dispatch. They are not duplicates.", - "confidence": 0.98, - "canonical_issue_reason": "No canonical issue for a duplicate cluster: the reports describe separate bugs with different triggers, symptoms, and subsystems.", + "summary": "These are two superficially similar but distinct bug reports about model loading/dispatch, not duplicates: one is a ZeRO-3 zero.Init partitioning/OOM problem for a composite multimodal submodule, the other is a 4-bit device_map/max_memory dispatch-to-CPU ValueError.", + "confidence": 0.93, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a true duplicate representative; if one must be chosen, 46823 is slightly more self-contained, but it is still a different bug from 46822.", + "best_issue_reason": "No single canonical issue stands out because the reports describe different code paths and failure modes; if one must be picked as the cluster representative, 46822 is slightly more concrete about the affected submodule and OOM symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46822", "right": "issue:46823", "accept": false, - "reason": "Different failure modes and mechanisms: ZeRO-3 composite submodule partitioning/OOM versus 4-bit device_map/max_memory incorrectly dispatching to CPU." + "reason": "Both concern loading/dispatch, but the underlying bugs differ: ZeRO-3 zero.Init fails to partition a composite submodule causing OOM, while 4-bit device_map='auto' with max_memory incorrectly sends modules to CPU and raises ValueError. They do not look like the same fixable defect." } ] }, @@ -20408,7 +20478,7 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "fc3e3c6fec1456b7e12828378998d0a3990a6d573d0ddb0c7044ae82104c9f72", + "prepared_review_unit_hash": "6d7039d86631c2ce089ba668a87c58747cf5e74dcaf552d61cf68118924cb2c1", "component_budget": { "node_count": 2, "item_count": 2, @@ -20430,18 +20500,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related VibeVoice workstreams, but they are not the same PR-level change. PR 40546 is the main VibeVoice implementation, while PR 42781 adds VibeVoice Realtime plus an acoustic tokenizer and broader API wiring, so they should not be merged as duplicates.", - "confidence": 0.94, + "summary": "These are related VibeVoice implementation PRs, but they target different scopes: the first adds the base VibeVoice model/docs, while the second adds VibeVoice Realtime plus an acoustic tokenizer and extra generation/auto-processing plumbing. They overlap in shared model files, but are not the same concrete change set.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 40546 is the earlier, non-draft core implementation with substantial review activity and broad model integration; it is the best representative of the base VibeVoice feature work.", + "canonical_pr_reason": "PR #42781 is the broader, newer VibeVoice deliverable and includes the base VibeVoice code plus realtime/tokenizer support, so it is the better representative PR for this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR 40546 is the most complete standalone VibeVoice implementation and is the clearest canonical artifact for this cluster.", + "best_pr_reason": "PR #42781 is the strongest representative of the VibeVoice work because it bundles the main model integration with the additional realtime path and supporting components; #40546 is a narrower precursor implementation.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "They share the VibeVoice subsystem, but 42781 is specifically about Realtime and acoustic-tokenizer support, with additional generation/auto-processing changes. This is broader and not the same concrete bug/change as 40546, so it should not be treated as a duplicate PR." + "reason": "Shared VibeVoice files alone are not enough: #40546 implements the base model, while #42781 adds realtime-specific functionality and an acoustic tokenizer with extra auto/generation changes. They are related but not the same fix and would not plausibly be merged as one PR." } ] }, @@ -20471,7 +20541,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "c615927a536cba2f157447a0234fc33f211f23a10aee1f65891689daa13e21d8", + "prepared_review_unit_hash": "4cbfc830f0de97c5b90d2ca4443a3a0053513ffd8c65a789e96f06cfe8ac8f15", "component_budget": { "node_count": 2, "item_count": 2, @@ -20493,18 +20563,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are unrelated feature additions in different model areas: TimesFM covariate forecasting vs. DINOv3 image-classification support. They do not describe the same bug or change path.", + "summary": "These two PRs are unrelated: one adds TimesFM forecasting with covariates, the other adds DINOv3 image-classification support. They do not share the same underlying bug or code change, so this is not a duplicate cluster.", "confidence": 0.99, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs target different models, different APIs, and different code paths, so they are not duplicates.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 41224 is the narrower and more self-contained change, but it is only a better representative of its own feature area; it does not subsume PR 40755.", + "best_pr_reason": "PR #41224 is the more self-contained feature addition, but it is still unrelated to PR #40755 and should not be treated as the canonical representative of a duplicate set.", "soft_edge_verdicts": [ { - "left": "pull_request:40755", - "right": "pull_request:41224", + "left": "[TimesFM] Add support for forecasting with covariates", + "right": "Add DINOv3ViTForImageClassification support", "accept": false, - "reason": "Reject: these PRs implement different features for different model families and would not plausibly be merged as one change." + "reason": "Different models, different code paths, and different feature areas; no evidence they fix the same underlying issue or could be merged into one PR." } ] }, @@ -20589,7 +20659,7 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-41855-16", + "cluster_id": "cluster-41855-15", "nodes": [ "pull_request:41855", "pull_request:42668", @@ -20605,7 +20675,6 @@ "pull_request:45655", "pull_request:46048", "pull_request:46267", - "pull_request:46744", "pull_request:46819" ], "soft_pairs": [ @@ -20626,196 +20695,187 @@ "pull_request:43995|pull_request:44827", "pull_request:44827|pull_request:45170", "pull_request:44827|pull_request:46267", - "pull_request:44827|pull_request:46744", "pull_request:44827|pull_request:46048", + "pull_request:44827|pull_request:46819", "pull_request:44827|pull_request:45213", "pull_request:43823|pull_request:45213", "pull_request:45655|pull_request:46819", "pull_request:44827|pull_request:45655", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "1b8afaf29435f53273d1f76a1be696bc42c2ce50ec6dc21f084edd03ca0b4f70", + "prepared_review_unit_hash": "ec35b27e7bdf8b56ce31f353a4206af9d28d2ba6fc9b69c29700eaef8cf16683", "component_budget": { - "node_count": 16, - "item_count": 16, + "node_count": 15, + "item_count": 15, "soft_pair_count": 24, - "serialized_chars": 42659, - "estimated_input_tokens": 10665, - "estimated_eval_tokens": 21586 + "serialized_chars": 40996, + "estimated_input_tokens": 10249, + "estimated_eval_tokens": 20754 }, "budget": { - "node_count": 16, - "item_count": 16, + "node_count": 15, + "item_count": 15, "soft_pair_count": 24, - "serialized_chars": 42659, - "estimated_input_tokens": 10665, - "estimated_eval_tokens": 21586 + "serialized_chars": 40996, + "estimated_input_tokens": 10249, + "estimated_eval_tokens": 20754 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly unrelated PRs spanning model additions, docs, tests, and assorted fixes. The only true duplicate-looking pair is the two fixed_cross_entropy patches (43251 and 43254), which both forward new kwargs into nn.functional.cross_entropy on the same code path.", - "confidence": 0.98, + "summary": "This cluster is mostly unrelated PRs, with one likely duplicate pair around `fixed_cross_entropy` kwargs handling. I would keep `pull_request:43251` as the canonical representative for that duplicate fix and reject the rest as separate changes.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43251 is the clearest canonical representative of the fixed_cross_entropy kwargs fix: it explicitly targets issue 43240 and forwards the supported cross_entropy arguments in the same function as 43254.", + "canonical_pr_reason": "`pull_request:43251` is the clearest representative of the only real duplicate pair in the set: it directly targets issue 43240 and fixes `fixed_cross_entropy` by forwarding the missing `cross_entropy` kwargs in `loss_utils.py`.", "best_issue_reason": null, - "best_pr_reason": "PR 43251 is the best overall representative for the overlapping loss-helper fix because it is tied to the tracked issue and presents the more complete implementation of the same change.", + "best_pr_reason": "`pull_request:43251` is the best PR anchor for the cluster because it is the most issue-linked and concrete fix among the near-duplicates; `43254` is essentially the same change but less directly anchored, while the remaining PRs are unrelated.", "soft_edge_verdicts": [ { "left": "pull_request:41855", "right": "pull_request:43823", "accept": false, - "reason": "Unrelated: Mistral tokenizer method additions vs a MobileLLM model introduction." + "reason": "Mistral tokenizer method additions vs adding MobileLLM; different components and no shared bug/change." }, { "left": "pull_request:41855", "right": "pull_request:43424", "accept": false, - "reason": "Unrelated: tokenizer API additions vs Executorch dynamic-shape export tests." + "reason": "Tokenizer API gap fix vs ExecuTorch dynamic-shape test; unrelated changes." }, { "left": "pull_request:42668", "right": "pull_request:43823", "accept": false, - "reason": "Unrelated: processor-from-pretrained robustness changes vs adding a new model." + "reason": "Processor auto-loading robustness vs a new model addition; not the same underlying change." }, { "left": "pull_request:43251", "right": "pull_request:46819", "accept": false, - "reason": "Different bugs: cross_entropy kwargs forwarding vs beam-search cache reordering." + "reason": "`fixed_cross_entropy` kwargs forwarding vs beam-search cache reordering; different code paths and bugs." }, { "left": "pull_request:43251", "right": "pull_request:43823", "accept": false, - "reason": "Different areas: loss helper kwargs vs MobileLLM model scaffolding." + "reason": "Loss utility fix and MobileLLM model add are unrelated." }, { "left": "pull_request:43251", "right": "pull_request:44827", "accept": false, - "reason": "Unrelated: loss helper API fix vs Mistral4 test/model work." + "reason": "Cross-entropy kwargs fix vs Mistral4 test/model work; unrelated." }, { "left": "pull_request:43424", "right": "pull_request:44827", "accept": false, - "reason": "Unrelated: Executorch export coverage vs Mistral4 tests/model fixes." + "reason": "ExecuTorch export test vs Mistral4 fixes; not the same underlying bug/change." }, { "left": "pull_request:43424", "right": "pull_request:43823", "accept": false, - "reason": "Unrelated: Executorch export tests vs MobileLLM model addition." + "reason": "Dynamic-shape export test vs model addition; unrelated." }, { "left": "pull_request:43823", "right": "pull_request:46819", "accept": false, - "reason": "Completely different changes: MobileLLM addition vs beam-search cache handling." + "reason": "MobileLLM model addition vs beam-search cache fix; no common code-path problem." }, { "left": "pull_request:43823", "right": "pull_request:45655", "accept": false, - "reason": "Unrelated: new MobileLLM model vs config-class resolution ordering fix." + "reason": "New model support vs `cls.config` resolution order fix; unrelated." }, { "left": "pull_request:43823", "right": "pull_request:43995", "accept": false, - "reason": "Unrelated: model addition vs Falcon output-capture refactor." + "reason": "New model addition vs Falcon output-collection refactor; different feature areas." }, { "left": "pull_request:43989", "right": "pull_request:46819", "accept": false, - "reason": "Unrelated: AutoVideoProcessor lookup fix vs beam-search cache logic." + "reason": "Auto video processor lookup fix vs beam-search cache handling; unrelated bugs." }, { "left": "pull_request:43995", "right": "pull_request:46819", "accept": false, - "reason": "Unrelated: Falcon forward/refactor changes vs generation beam-search fix." + "reason": "Falcon refactor vs beam-search cache reordering; not mergeable as one fix." }, { "left": "pull_request:43995", "right": "pull_request:45213", "accept": false, - "reason": "Unrelated: Falcon model refactor vs model-creation skill content." + "reason": "Model refactor vs model-creation skill content; unrelated." }, { "left": "pull_request:43995", "right": "pull_request:44827", "accept": false, - "reason": "Unrelated: Falcon refactor vs Mistral4 test/model changes." + "reason": "Falcon output-interface refactor vs Mistral4 test fixes; different subsystems." }, { "left": "pull_request:44827", "right": "pull_request:45170", "accept": false, - "reason": "Different bugs: Mistral4 test fixes vs layrnorm-to-layernorm renaming across CLIP-like models." + "reason": "Mistral4 test/model fixes vs `layrnorm` mapping rename; separate model families and fixes." }, { "left": "pull_request:44827", "right": "pull_request:46267", "accept": false, - "reason": "Unrelated: Mistral4 tests vs chat parsing/serve update." + "reason": "Mistral4 test fixes vs chat response parsing update; unrelated." }, { "left": "pull_request:44827", - "right": "pull_request:46744", + "right": "pull_request:46048", "accept": false, - "reason": "Unrelated: Mistral4 tests vs Apple Silicon docs update." + "reason": "Mistral4 test fixes vs TDT loss kernel addition; different features." }, { "left": "pull_request:44827", - "right": "pull_request:46048", + "right": "pull_request:46819", "accept": false, - "reason": "Unrelated: Mistral4 tests vs adding a TDT loss kernel." + "reason": "Mistral4 test fixes vs beam-search cache fix; unrelated." }, { "left": "pull_request:44827", "right": "pull_request:45213", "accept": false, - "reason": "Unrelated: Mistral4 test/model changes vs model-creation skill addition." + "reason": "Mistral4 fixes vs model-creation skill; completely different purpose." }, { "left": "pull_request:43823", "right": "pull_request:45213", "accept": false, - "reason": "Unrelated: MobileLLM model addition vs model-creation skill documentation/workflow." + "reason": "MobileLLM model addition vs AI skill/doc scaffolding; unrelated." }, { "left": "pull_request:45655", "right": "pull_request:46819", "accept": false, - "reason": "Unrelated: config resolution ordering fix vs beam-search cache handling." + "reason": "`cls.config` resolution order fix vs beam-search cache reordering; different code paths." }, { "left": "pull_request:44827", "right": "pull_request:45655", "accept": false, - "reason": "Unrelated: Mistral4 tests/model changes vs config resolution fix." - }, - { - "left": "pull_request:43251", - "right": "pull_request:43254", - "accept": true, - "reason": "Same underlying fix: both update fixed_cross_entropy to forward supported nn.functional.cross_entropy kwargs on the same code path, and could plausibly be merged into one PR." + "reason": "Mistral4 test/model work vs `cls.config` resolution fix; no underlying bug overlap." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet: 43251 and 43254 are the only clear same-function, same-issue-target pair, while the rest are mostly different model additions, docs, or unrelated fixes. The soft-edge rejections are generally conservative. One minor caveat: the phrase 'more complete implementation' for 43251 is a bit stronger than the packet strictly proves, but it does not materially undermine the overall assessment." - }, - "evaluator_used": true, + "evaluator_result": null, + "evaluator_used": false, "retried": false, - "accepted_nontrivial_soft_edge": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -20838,7 +20898,7 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "f14cdf9d68d652a89a5186f185164152997c182107c061dbdc73fd576795e771", + "prepared_review_unit_hash": "a031358726f6d92155050c1a671c98f8230885c97ead982f50abbadc9f3285dd", "component_budget": { "node_count": 2, "item_count": 2, @@ -20860,18 +20920,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are both GitHub Actions dependency bumps, but they target different action families and different maintenance goals: one is for Node 24 compatibility in core workflow actions, the other is a broader latest-version refresh including Docker build actions. They are not the same underlying change and should not be merged as duplicates.", - "confidence": 0.94, + "summary": "Both PRs are GitHub Actions dependency bump/maintenance changes, but they target different action families and different motivations, so they are not duplicates of the same underlying change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the cleaner canonical representative because it has a single, explicit purpose (Node 24 compatibility) and updates the core workflow actions in a more focused way.", + "canonical_pr_reason": "PR 42975 is the broader and more clearly scoped workflow maintenance change, upgrading multiple GitHub Actions across many workflows for Node 24 compatibility.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best representative overall because its scope is clearer and more specific; PR 42976 is a separate dependency refresh for Docker-related actions rather than the same fix.", + "best_pr_reason": "PR 42975 is the best representative if one PR must anchor the cluster because it covers the widest set of workflow-action updates and has a specific compatibility goal.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Do not merge: they update different action ecosystems and solve different problems. 42975 upgrades checkout/cache/upload-artifact for Node 24 compatibility, while 42976 bumps docker/build-push-action and other workflow actions to newer versions. Same broad maintenance theme, but not the same concrete change." + "reason": "They are both workflow dependency upgrades, but 42975 is about GitHub Actions version bumps for Node 24 compatibility, while 42976 upgrades Docker build actions to latest versions; different action types, different workflows, and not the same concrete bug or change." } ] }, @@ -20917,106 +20977,106 @@ "pull_request:45490|pull_request:46705", "pull_request:46562|pull_request:46658" ], - "prepared_review_unit_hash": "19e91320902d2706cdeb40c6aa8c6fa87593e03d80790d80633cf67e2c711a7c", + "prepared_review_unit_hash": "4c6a4c05fca73ef77ff365870dcf138ceab3f85e40e56d6b745f1f4a170502a7", "component_budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 12, - "serialized_chars": 24622, - "estimated_input_tokens": 6156, - "estimated_eval_tokens": 12568 + "serialized_chars": 24699, + "estimated_input_tokens": 6175, + "estimated_eval_tokens": 12606 }, "budget": { "node_count": 7, "item_count": 7, "soft_pair_count": 12, - "serialized_chars": 24622, - "estimated_input_tokens": 6156, - "estimated_eval_tokens": 12568 + "serialized_chars": 24699, + "estimated_input_tokens": 6175, + "estimated_eval_tokens": 12606 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated model-addition PRs that only overlap on shared docs/auto-registration scaffolding. The only likely duplicate-like pair is the Step 3.7 Flash integration PR and its companion WIP script PR.", - "confidence": 0.91, + "summary": "This cluster is heterogeneous: four separate model-addition PRs plus one distributed-config PR. The only mildly related pair is the Step3p7/Step_3_7_flash work, but they target different code paths and should not be merged as duplicates.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46562 is the clearest central artifact: it is the main Transformers integration for Step 3.7 Flash, and it has the strongest same-change link to 46658. The other PRs are separate model additions or an unrelated distributed-config change.", + "canonical_pr_reason": "No clear canonical PR: the items span unrelated model integrations and infra changes, so no single PR safely subsumes the others.", "best_issue_reason": null, - "best_pr_reason": "PR 46562 is the best representative of the only real subcluster here because it is the production model add rather than the accompanying WIP/demo repository changes in 46658.", + "best_pr_reason": "No PR is globally suitable as a duplicate representative; at best, one of the larger model-addition PRs could serve as a topic anchor, but none is a true duplicate match for the rest.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model families: ViT NEPA vs Molmo. Shared docs/auto-registry files are routine for new model PRs, not evidence of the same change." + "reason": "Different model families and code paths: ViT NEPA vs Molmo." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different model families: ViT NEPA vs Molmo2. Overlap is just shared Transformers registration/docs plumbing." + "reason": "Different model families: ViT NEPA vs Molmo2." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "Different model families: ViT NEPA vs CTSM. The shared files are generic model-registration updates." + "reason": "Different domains and implementations: vision model add vs time-series CTSM." }, { "left": "pull_request:42978", "right": "pull_request:46705", "accept": false, - "reason": "ViT NEPA model add is unrelated to the distributed configuration PR." + "reason": "Model addition PR vs distributed configuration infrastructure change." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are related by naming, but these are distinct model ports with different files and implementations; not the same concrete change." + "reason": "Related naming, but distinct models and separate implementations: Molmo vs Molmo2." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo vs CTSM are unrelated model additions; shared auto/docs files are not enough." + "reason": "Different architectures and tasks: Molmo vs CTSM." }, { "left": "pull_request:43448", "right": "pull_request:46705", "accept": false, - "reason": "Molmo model addition is unrelated to distributed config work." + "reason": "Different scope: model integration vs distributed config plumbing." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 vs CTSM are different models and codepaths; only generic registry/docs overlap." + "reason": "Different model families and tasks: Molmo2 vs CTSM." }, { "left": "pull_request:43451", "right": "pull_request:46705", "accept": false, - "reason": "Molmo2 model addition is unrelated to distributed config work." + "reason": "Different scope: model integration vs distributed config plumbing." }, { "left": "pull_request:45490", "right": "pull_request:46562", "accept": false, - "reason": "CTSM and Step 3.7 Flash are different models; overlap in auto-mappings/docs is broad subsystem reuse, not duplication." + "reason": "Both are model additions, but they target different products/code paths: CTSM vs Step3p7 Flash." }, { "left": "pull_request:45490", "right": "pull_request:46705", "accept": false, - "reason": "CTSM model addition is unrelated to distributed config work." + "reason": "Different scope: time-series model addition vs distributed config infrastructure." }, { "left": "pull_request:46562", "right": "pull_request:46658", - "accept": true, - "reason": "Both target Step 3.7 Flash and appear to describe the same underlying model port/reproduction effort; the WIP script repo in 46658 and the Transformers integration in 46562 are plausibly one combined change set." + "accept": false, + "reason": "Same broad family, but not the same concrete change: one adds the Step3p7 model, the other is a separate WIP Step 3 7 Flash script/mini-model effort." } ] }, @@ -21065,12 +21125,12 @@ "pull_request:46380|pull_request:46539", "pull_request:46380|pull_request:46579" ], - "prepared_review_unit_hash": "ab6661c66d8891873b91cadd777dfbf6f7394f449cb1e8ce577e3015b9e9da65", + "prepared_review_unit_hash": "01fb16fea64b647191074a0b5f38915ab1ab7ec01478748ca58814ca634308f5", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 14, - "serialized_chars": 23161, + "serialized_chars": 23162, "estimated_input_tokens": 5791, "estimated_eval_tokens": 11838 }, @@ -21078,7 +21138,7 @@ "node_count": 8, "item_count": 8, "soft_pair_count": 14, - "serialized_chars": 23161, + "serialized_chars": 23162, "estimated_input_tokens": 5791, "estimated_eval_tokens": 11838 }, @@ -21087,96 +21147,96 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is heterogeneous: Whisper test cleanup, tie_weights backward-compat, dummy test edits, FP8 dequant fixes, doctest cleanup, generation validation, CI resource patching, and typo fixes. I don\u2019t see any true duplicate pair.", - "confidence": 0.98, + "summary": "This cluster is a grab bag of unrelated PRs: Whisper tokenizer/tests, tie_weights compatibility, dummy vision test assertions, FP8 dequant work, doctest/typo cleanup, CI infra, and generation-config fixes. The overlaps are only superficial, so none of the soft edges look like true duplicates.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No clear canonical PR for duplicate-triage purposes; the cluster is heterogeneous. If one must be chosen as the best anchor, PR 46380 is the most substantive standalone runtime fix, but it is not a duplicate of the others.", "best_issue_reason": null, - "best_pr_reason": "PR 44815 is the strongest standalone bug-fix candidate because it targets a concrete dequantization path in model loading/FP8 code; however, it is not a duplicate of the other PRs here.", + "best_pr_reason": "PR 46380 is the strongest standalone bug-fix candidate: it changes generation config validation/runtime behavior in a concrete code path, unlike the mostly test/doc/typo-only PRs.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Whisper tokenizer/test fixes vs unrelated CLIP/ViT test assertion changes; different models and different bug surfaces." + "reason": "Whisper tokenizer/test fix vs. dummy assertion changes in CLIP/ViT tests; different bugs and code paths." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper decode/test fix vs FP8 dequant/model-loading fix; unrelated code paths." + "reason": "Whisper decode logic vs. FP8 dequant/core loading changes; unrelated subsystems." }, { "left": "pull_request:43291", "right": "pull_request:46380", "accept": false, - "reason": "Whisper test fix vs generation watermarking/validation and CI-related changes; not the same underlying bug." + "reason": "Whisper tokenizer fix vs. generation-config/Whisper validation fix; too broad and not the same underlying bug." }, { "left": "pull_request:43498", "right": "pull_request:46579", "accept": false, - "reason": "Backward-compat alias for tie_weights is a behavioral API change; typo cleanup is unrelated refactoring/docs hygiene." + "reason": "Backward-compatibility alias for tie_weights vs. broad typo cleanup; different change types and no shared bug." }, { "left": "pull_request:43665", "right": "pull_request:46539", "accept": false, - "reason": "Artificial failing assertions in model tests vs Phimoe CI integration/resource-limit patch; different intent and failure mode." + "reason": "Intentional failing test edits vs. Phimoe CI/integration-test infrastructure; not the same issue." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "Dummy test edits for CLIP/ViT vs FP8 dequant logic; no shared bug or change." + "reason": "Dummy test assertions vs. FP8 dequant fix; unrelated artifacts." }, { "left": "pull_request:43665", "right": "pull_request:46380", "accept": false, - "reason": "Dummy test edits vs generation config/watermarking fixes; unrelated." + "reason": "Dummy test assertions vs. generation-validation/runtime fix; no common concrete bug." }, { "left": "pull_request:43665", "right": "pull_request:46579", "accept": false, - "reason": "Dummy test edits vs broad code typo cleanup; unrelated." + "reason": "Dummy test assertions vs. typo/doc cleanup across many code files; not duplicate work." }, { "left": "pull_request:44815", "right": "pull_request:46539", "accept": false, - "reason": "FP8 dequant/model-loading fix vs CI runner memory workaround for Phimoe tests; different subsystems and issues." + "reason": "FP8 dequant/model-loading fix vs. Phimoe CI memory/device-map setup; different problems." }, { "left": "pull_request:44815", "right": "pull_request:46380", "accept": false, - "reason": "Both touch core_model_loading, but one is FP8 dequant behavior and the other is generation/watermarking validation; not the same concrete bug." + "reason": "Both touch some shared utilities, but one is FP8 dequantization and the other is generation-config validation; not the same bug." }, { "left": "pull_request:44815", "right": "pull_request:46579", "accept": false, - "reason": "Concrete FP8 dequant fix vs unrelated typo cleanup across code." + "reason": "Concrete dequant fix vs. typo cleanup PR; no overlap in underlying change." }, { "left": "pull_request:45114", "right": "pull_request:46579", "accept": false, - "reason": "Doctest/docstring cleanup vs code typo cleanup; they are both hygiene, but not the same change or bug." + "reason": "Doctest formatting updates vs. general typo cleanup in code; similar surface area only, not same change." }, { "left": "pull_request:46380", "right": "pull_request:46539", "accept": false, - "reason": "Generation configuration/Whisper validation fix vs Phimoe CI memory patch; different concrete problems." + "reason": "Generation config/Whisper runtime fix vs. Phimoe test-job infrastructure; unrelated." }, { "left": "pull_request:46380", "right": "pull_request:46579", "accept": false, - "reason": "Generation validation fix vs typo cleanup; no shared underlying bug." + "reason": "Generation validation/runtime fix vs. broad typo cleanup; not a duplicate or same bug family." } ] }, @@ -21209,52 +21269,52 @@ "pull_request:43488|pull_request:43532", "pull_request:43532|pull_request:46747" ], - "prepared_review_unit_hash": "b1fad96fedfd398a4a47e47bbb2d1f58928f49c454744865278428acdaab786f", + "prepared_review_unit_hash": "9b33a20f3e739f9530dfbd34118031fa2e0aea86169ce0518bd86ce63f0dd415", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6348, - "estimated_input_tokens": 1587, - "estimated_eval_tokens": 3430 + "serialized_chars": 6349, + "estimated_input_tokens": 1588, + "estimated_eval_tokens": 3432 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 6348, - "estimated_input_tokens": 1587, - "estimated_eval_tokens": 3430 + "serialized_chars": 6349, + "estimated_input_tokens": 1588, + "estimated_eval_tokens": 3432 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Three open '[don't merge]' test PRs with superficial title similarity, but they change different code paths and workflows. They are not duplicates of one underlying fix/change.", - "confidence": 0.95, + "summary": "All three pull requests are superficially similar as repo-maintenance/don't-merge changes, but they address different concrete areas: packaging/setup, CI comment automation, and PR security-gate workflow behavior. None look like duplicate fixes that could plausibly be merged into one PR.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "No canonical duplicate PR: the three PRs target unrelated changes and do not share the same underlying bug or code-path fix.", "best_issue_reason": null, - "best_pr_reason": null, + "best_pr_reason": "PR 46747 is the cleanest standalone representative because it is a single-file workflow change with one concrete behavior change, unlike the broader setup/workflow/test edits in the others.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:46747", "accept": false, - "reason": "Both are test PRs, but one edits setup/versioning and a Vit test while the other changes a PR security-gate workflow. The diffs are unrelated and would not be merged as one PR." + "reason": "Different concerns: setup/version/dependency and a Vit test tweak versus changing the PR security-gate workflow trigger. Not the same underlying change." }, { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "They only overlap on a Vit test filename; one is about package/version/dependency metadata, the other rewrites comment CI and related workflow/test utilities. Not the same underlying change." + "reason": "One changes packaging/dependencies and a test fixture; the other rewrites multiple GitHub workflow files and testing utilities. Different code paths and purposes." }, { "left": "pull_request:43532", "right": "pull_request:46747", "accept": false, - "reason": "Different workflow areas and different concrete changes: self-comment/self-scheduled/slack/test utility updates versus a single PR security-gate trigger change. Too broad and unrelated to treat as one fix." + "reason": "Both are workflow-related, but one handles self-comment CI/notification plumbing while the other changes the PR security-gate trigger model. Not a single fix." } ] }, @@ -21284,7 +21344,7 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "435adc5dd0a75ae6ae0c8d72078ec404e9af0d8f0da0475561ede680871315d8", + "prepared_review_unit_hash": "8d9f542cce616734d82cdff1a17380025eaf0f02f7353cba572a02f0bbd4864f", "component_budget": { "node_count": 2, "item_count": 2, @@ -21306,18 +21366,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related but distinct PRs: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. They share scaffolding and docs updates, but target different tasks and code paths, so they should not be merged as duplicates.", - "confidence": 0.95, + "summary": "These PRs are related but not duplicates: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. They share pipeline/docs scaffolding, but target different features and model/test surfaces.", + "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 43613 is the broader and more representative change set, covering the visual segmentation pipeline across multiple model integrations and tests; if one PR had to stand in for the cluster, this is the stronger candidate.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Different underlying features: concept segmentation vs visual segmentation. They share pipeline/docs boilerplate, but the task names, intended behavior, and model coverage differ, so they are not the same concrete change and would not be merged into one PR." + "reason": "Different underlying changes: concept segmentation vs visual segmentation are separate pipelines with different filenames, model integrations, and tests. The overlap is mostly shared infrastructure/docs, so they should not be merged as duplicates." } ] }, @@ -21347,7 +21407,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "dad9e7f8c4d41f6b27e7034e5e5d667ef8d456f0da810bf478870c0065433574", + "prepared_review_unit_hash": "3366c015551af7bcd4431574a82a521d67a4ade4d3904d199e2516808e8669c9", "component_budget": { "node_count": 2, "item_count": 2, @@ -21369,18 +21429,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs modify `trainer.py` to add customization hooks, but they address different Trainer behaviors: gradient-accumulation loss scaling vs. column/signature filtering. They are related by subsystem, not by the same underlying bug or change.", - "confidence": 0.92, + "summary": "These two PRs both touch `trainer.py` but address different trainer behaviors: gradient-accumulation loss scaling vs. signature-column filtering for dataset column pruning. They are not the same underlying change and should not be merged as duplicates.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43651 is the more concrete, self-contained trainer behavior change: it updates loss scaling in `training_step` and adds `_loss_is_scaled_for_ga` to control gradient accumulation handling. The other PR is a separate extensibility refactor for signature column filtering.", + "canonical_pr_reason": "No issue exists in this cluster; if one PR must be treated as the representative, `pull_request:43651` is the more self-contained code-path change, while `pull_request:43663` is an unrelated refactor of input-column filtering.", "best_issue_reason": null, - "best_pr_reason": "PR #43651 is the best representative of the cluster because it has a clearer functional purpose and a more direct code-path change, while PR #43663 is a different Trainer hook with a separate responsibility.", + "best_pr_reason": "`pull_request:43651` is the better representative because it fixes a specific training-step loss-scaling behavior with a focused property addition. `pull_request:43663` targets a different mechanism entirely (`_get_signature_columns`) and is not a duplicate of it.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Different code paths and behaviors: one changes loss scaling during training, the other changes how Trainer derives signature columns. Shared file alone is not enough to treat them as the same change." + "reason": "Different trainer code paths and different intended behavior: one changes gradient accumulation loss normalization, the other changes model signature inspection / column filtering. Shared filename alone is not enough to treat them as the same change." } ] }, @@ -21410,12 +21470,12 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "e990fc7afced5680c4b5489ffa7c0f6bae486193a2ff39145e4a6afd6dd33258", + "prepared_review_unit_hash": "5df3ba9feb0af633f67a7aed233fe1aaa9aee9ecaadf1a30bee7e334a082728e", "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 }, @@ -21423,7 +21483,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 3664, + "serialized_chars": 3663, "estimated_input_tokens": 916, "estimated_eval_tokens": 2088 }, @@ -21432,18 +21492,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs touch `src/transformers/cli/serve.py`, but they fix different `NameError`-style problems: PR 43656 is an annotations/typing cleanup with `from __future__ import annotations`, while PR 43842 specifically adds a `TYPE_CHECKING` import and forward reference for `TypeAdapter` when `pydantic` is absent. They are related but not the same underlying bug or mergeable into one duplicate fix.", - "confidence": 0.94, + "summary": "The two PRs are related by file and theme, but they fix different TypeAdapter-related failures in the CLI, so they should not be treated as duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR: these are separate fixes in the same file, each addressing a different code path and failure mode.", + "canonical_pr_reason": "PR 43842 is the narrower, more explicit fix for the runtime pydantic/TypeAdapter import problem and has a direct issue link; PR 43656 appears to address a separate annotation-evaluation NameError.", "best_issue_reason": null, - "best_pr_reason": "PR 43842 is the more direct fix for the reported `TypeAdapter` NameError when `pydantic` is not installed; PR 43656 is a separate typing/annotation change.", + "best_pr_reason": "If one PR must represent the cluster, 43842 is the better choice because it targets a concrete optional-dependency runtime failure with a specific fix, while 43656 is a different bug in the same file.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "Different root causes and fixes: one is future-annotations/typing cleanup, the other is a missing `TypeAdapter` import under `TYPE_CHECKING` for the no-pydantic case." + "reason": "Same file and similar wording, but different underlying problems: 43656 changes annotation handling to avoid a NameError from evaluated type hints, while 43842 makes TypeAdapter a TYPE_CHECKING-only import to avoid requiring pydantic at runtime. They are not the same fix and should not be merged as one PR." } ] }, @@ -21573,8 +21633,8 @@ "pull_request:44066|pull_request:44086", "pull_request:44018|pull_request:44068", "pull_request:44066|pull_request:44071", - "pull_request:44066|pull_request:44068", "pull_request:44018|pull_request:44066", + "pull_request:44066|pull_request:44068", "pull_request:44019|pull_request:44722", "pull_request:44086|pull_request:44722", "pull_request:44019|pull_request:44085", @@ -21588,8 +21648,8 @@ "pull_request:44018|pull_request:44086", "pull_request:44068|pull_request:44086", "pull_request:44071|pull_request:44085", - "pull_request:44059|pull_request:44086", "pull_request:44085|pull_request:44086", + "pull_request:44059|pull_request:44086", "pull_request:44098|pull_request:44101", "pull_request:44018|pull_request:44071", "pull_request:44059|pull_request:44722", @@ -21603,8 +21663,8 @@ "pull_request:44073|pull_request:44101", "pull_request:44073|pull_request:44161", "pull_request:44059|pull_request:44071", - "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:43997|pull_request:44017", @@ -21636,12 +21696,12 @@ "pull_request:44044|pull_request:44154", "pull_request:44024|pull_request:44076" ], - "prepared_review_unit_hash": "281fb1b2830168a70ea1e8e67e301b3064b0ad3c898d99aa58b395c6162f1b0f", + "prepared_review_unit_hash": "37ce14a4b46d1b66d03098235d7792529a93b26e6095582259477e1fe82e816b", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84504, + "serialized_chars": 84503, "estimated_input_tokens": 21126, "estimated_eval_tokens": 42508 }, @@ -21649,345 +21709,435 @@ "node_count": 28, "item_count": 28, "soft_pair_count": 72, - "serialized_chars": 76938, - "estimated_input_tokens": 19235, - "estimated_eval_tokens": 38726 + "serialized_chars": 76936, + "estimated_input_tokens": 19234, + "estimated_eval_tokens": 38724 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a theme cluster of unrelated model-by-model output-tracing refactors. Only a few pairs are true duplicates within the same code path, mainly the GPT-J/CodeGen and GPT-Neo subsets.", - "confidence": 0.82, + "summary": "Cluster is mostly a pile of model-specific output-tracing refactors; only the GPT-Neo and GPT-J/CodeGen subgroups look like true duplicates.", + "confidence": 0.73, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44722 is the strongest representative of the shared GPT-J/CodeGen output-tracing refactor: it targets the same concrete files and issue, and aligns with the duplicate subgroup best.", + "canonical_pr_reason": "44722 is the latest, most complete GPT-J/CodeGen output-tracing refactor and effectively supersedes the earlier GPT-J-focused variants.", "best_issue_reason": null, - "best_pr_reason": "PR 44722 is the clearest concrete output-tracing change in the cluster and best represents the duplicate GPT-J/CodeGen subgroup; the rest are separate model-specific refactors.", + "best_pr_reason": "44722 best represents the duplicate core because it covers the same GPT-J/CodeGen code path as the earlier matching PRs and is the broadest revision.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Same broad output-tracing theme, but different models/code paths (FNet/CVT vs GPT-J); not the same underlying change." + "reason": "Shared tracing theme only; different model code paths (CVT/FNet vs GPT-J)." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different model families and different implementation paths; only the refactor theme overlaps." + "reason": "Different model families and implementations; only the same umbrella tracing refactor theme." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both modify GPT-J output tracing in the same file and appear to be the same concrete decorator migration." + "reason": "Both are GPT-J output-tracing refactors on the same code path; one is a narrower subset of the other." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "Different backbones/models; shared issue target alone is not enough to make them duplicates." + "reason": "ResNet/RT-DETR vs EfficientNet are different code paths, despite the shared tracing migration theme." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet vs GPT-J/CodeGen are unrelated code paths despite similar refactor wording." + "reason": "Different model families; same standardized-output pattern is too broad to treat as a duplicate." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "Different model implementations; not the same bug or change." + "reason": "GPT-J/CodeGen changes are unrelated to EfficientNet's output tracing path." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa are unrelated; same refactor pattern only." + "reason": "MobileNetV2 and DeBERTaV2 are distinct implementations; only the refactor pattern overlaps." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "Different models and different code paths; not mergeable as one change." + "reason": "Different underlying models/code paths (GPT-J/CodeGen vs MGP-STR)." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Same GPT-Neo output-tracing refactor on the same file/tests; clearly the same underlying change." + "reason": "Both are GPT-Neo output-tracing refactors on the same model/code path." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J/CodeGen versus MPT; different code paths." + "reason": "GPT-J/CodeGen vs MPT; shared output-tracing infrastructure only." }, { - "left": "pull_request:44066", - "right": "pull_request:44068", + "left": "pull_request:44018", + "right": "pull_request:44066", "accept": false, - "reason": "GPT-J/CodeGen and GPT-Neo are separate model implementations, so this is only thematic overlap." + "reason": "GPT-Neo and GPT-J are different model paths; not the same concrete bug/change." }, { - "left": "pull_request:44018", - "right": "pull_request:44066", + "left": "pull_request:44066", + "right": "pull_request:44068", "accept": false, - "reason": "Different model families and code paths; not the same underlying fix." + "reason": "GPT-J/CodeGen and GPT-Neo are different models; same tooling, not same fix." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet refactor is unrelated to GPT-J/CodeGen output tracing." + "reason": "ResNet-related refactor does not fix the GPT-J/CodeGen code path." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J/CodeGen are unrelated changes." + "reason": "MGP-STR and GPT-J/CodeGen are unrelated code paths." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet and GPT-J are different code paths; only the general output-tracing theme matches." + "reason": "ResNet and GPT-J are different model families; only the broad tracing initiative matches." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated implementations." + "reason": "GPT-Neo and ResNet are unrelated code paths." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "Different model families; not the same bug or refactor." + "reason": "ResNet and MPT are different implementations; not the same underlying fix." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are different model code paths, so these are not duplicates." + "reason": "GPT-Neo and GPT-J/CodeGen are different model code paths." }, { "left": "pull_request:44068", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo versus Flaubert/XLM is only a shared refactor pattern, not the same change." + "reason": "GPT-Neo and Flaubert/XLM are unrelated fixes." }, { "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "Different model implementations and no shared concrete code path." + "reason": "GPT-Neo and Flaubert/XLM do not share the same concrete bug." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J/CodeGen are unrelated changes." + "reason": "MPT and GPT-J/CodeGen are different model paths." }, { "left": "pull_request:44018", "right": "pull_request:44085", "accept": false, - "reason": "GPT-Neo and GPT-J/CodeGen are separate refactors in different code paths." + "reason": "GPT-Neo vs GPT-J; same refactor style, different underlying model." }, { - "left": "pull_request:44059", + "left": "pull_request:44018", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 and MGP-STR are unrelated beyond the output-tracing theme." + "reason": "GPT-Neo and MGP-STR are unrelated code paths." }, { - "left": "pull_request:44085", + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are different model implementations." + }, + { + "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "Different model families and files; not the same underlying change." + "reason": "GPT2 and MGP-STR are unrelated concrete fixes." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert/XLM are unrelated code paths." + "reason": "ViLT and Flaubert/XLM are different model paths." }, { "left": "pull_request:44018", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo and MPT are different implementations; only the refactor style overlaps." + "reason": "GPT-Neo and MPT are different model code paths." }, { "left": "pull_request:44059", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 and GPT-J/CodeGen are different model code paths." + "reason": "GPT2 and GPT-J/CodeGen are different model paths; too broad to merge." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "Segformer and LongT5 are unrelated models; shared output-tracing language is not enough." + "reason": "SegFormer and LongT5 are unrelated implementations." }, { "left": "pull_request:44017", "right": "pull_request:44722", "accept": false, - "reason": "Different model families and concrete code paths." + "reason": "SegFormer and GPT-J/CodeGen are different model families." }, { "left": "pull_request:44017", "right": "pull_request:44073", "accept": false, - "reason": "Segformer and VisualBert are separate implementations; not the same bug or change." + "reason": "SegFormer and VisualBert are different concrete code paths." }, { "left": "pull_request:44068", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo and MPT are unrelated model code paths." + "reason": "GPT-Neo and MPT are unrelated model implementations." }, { "left": "pull_request:43997", "right": "pull_request:44073", "accept": false, - "reason": "RegNet and VisualBert are different backbones; only the refactor theme matches." + "reason": "RegNet and VisualBert are different code paths." }, { "left": "pull_request:44071", "right": "pull_request:44086", "accept": false, - "reason": "MPT and MGP-STR are unrelated changes." + "reason": "MPT and MGP-STR are unrelated model fixes." }, { - "left": "pull_request:44018", + "left": "pull_request:44017", "right": "pull_request:44071", "accept": false, - "reason": "GPT-Neo and MPT are different model implementations." + "reason": "SegFormer and MPT are different model families." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBert and Flaubert/XLM are different code paths." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "VisualBert and LongT5 are unrelated fixes." }, { "left": "pull_request:44059", "right": "pull_request:44071", "accept": false, - "reason": "GPT-2 and MPT are unrelated code paths." + "reason": "GPT2 and MPT are different underlying model paths." }, { - "left": "pull_request:44068", - "right": "pull_request:44086", + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", "accept": false, - "reason": "GPT-Neo and MGP-STR are separate refactors in different files/models." + "reason": "RegNet and SegFormer are different model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are unrelated concrete fixes." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT and ImageGPT are unrelated beyond the general output-tracing theme." + "reason": "MPT and ImageGPT are different model paths." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert/XLM are unrelated model implementations." + "reason": "TextNet and Flaubert/XLM are unrelated code paths." }, { "left": "pull_request:43997", "right": "pull_request:44018", "accept": false, - "reason": "RegNet and GPT-Neo are different model code paths." + "reason": "RegNet and GPT-Neo are different model families." }, { "left": "pull_request:44056", - "right": "pull_request:44101", + "right": "pull_request:44161", "accept": false, - "reason": "MPNet and Flaubert/XLM are unrelated refactors." + "reason": "MPNet and LongT5 are unrelated fixes." }, { - "left": "pull_request:44071", + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are different model paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet and LongT5 are unrelated implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44161", + "accept": false, + "reason": "GPT-Neo and LongT5 are different concrete code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are unrelated model fixes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different code paths." + }, + { + "left": "pull_request:44018", "right": "pull_request:44129", "accept": false, - "reason": "MPT and SpeechT5 are different model families." + "reason": "GPT-Neo and SpeechT5 are different model families." }, { - "left": "pull_request:43996", + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated concrete changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are different model paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different implementations." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT2 and GPT-Neo are different model code paths." + }, + { + "left": "pull_request:44101", "right": "pull_request:44161", "accept": false, - "reason": "FNet/CVT and LongT5 are unrelated code paths." + "reason": "Flaubert/XLM and LongT5 are unrelated fixes." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are different model families." }, { "left": "pull_request:43996", "right": "pull_request:44017", "accept": false, - "reason": "FNet/CVT and Segformer are separate model implementations." + "reason": "CVT/FNet and SegFormer are different code paths." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBert and DeBERTa are unrelated code paths." + "reason": "SqueezeBert and DeBERTaV2 are unrelated model fixes." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa and TextNet are different models; only the refactor pattern overlaps." + "reason": "DeBERTaV2 and TextNet are different underlying model paths." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J/CodeGen output-tracing decorator migration on the same concrete files; these could plausibly be merged into one PR." + "reason": "Same GPT-J/CodeGen output-tracing refactor; 44722 is the broader later revision." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both are GPT-J output-tracing refactors on the same file and appear to implement the same change, despite the title mismatch on 44085." + "reason": "Both target the same GPT-J output-tracing change; 44722 subsumes the narrower GPT-J-only variant." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are unrelated changes." + "reason": "GPT-J and SpeechT5 are different model code paths." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "FNet/CVT and ResNet/RT-DETR ResNet are unrelated model code paths." + "reason": "CVT/FNet and ResNet are different model implementations." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "FNet/CVT and SpeechT5 are unrelated implementations." + "reason": "CVT/FNet and SpeechT5 are unrelated fixes." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa and SpeechT5 are different code paths." + "reason": "DeBERTaV2 and SpeechT5 are different concrete code paths." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa and ViTS are unrelated model refactors." + "reason": "DeBERTaV2 and VITS are unrelated model fixes." }, { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "FocalNet and ImageGPT are different model families; only the output-tracing theme is shared." + "reason": "FocalNet and ImageGPT are different model paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays conservative. The accepted duplicate pairs are supported by shared files/targets and overlapping GPT-Neo/GPT-J code paths, while the rejected pairs mostly differ by model family and implementation path. The choice of PR 44722 as the canonical/representative GPT-J-CodeGen refactor is also supported by the diff and issue linkage." + "feedback": "The summary is conservative and consistent with the packet: it treats the same-model GPT-Neo and GPT-J/CodeGen PRs as duplicates, and rejects cross-model pairs rather than overclaiming mergeability. The canonical/best PR choices are grounded in the later, broader GPT-J/CodeGen revisions shown in the diffs." }, "evaluator_used": true, "retried": false, @@ -22022,12 +22172,12 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "60211b653b0659531337150322a79ecda8ca42cf8583787d5162bb7e0247f63a", + "prepared_review_unit_hash": "6d82f7eb02cb3a86fe8ffd329e80b7eae494296137f2838be26a613758818663", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84504, + "serialized_chars": 84503, "estimated_input_tokens": 21126, "estimated_eval_tokens": 42508 }, @@ -22035,7 +22185,7 @@ "node_count": 6, "item_count": 6, "soft_pair_count": 5, - "serialized_chars": 13066, + "serialized_chars": 13067, "estimated_input_tokens": 3267, "estimated_eval_tokens": 6790 }, @@ -22044,42 +22194,42 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all model-specific refactors to standardized output tracing/capture, but they touch different architectures and code paths, so they should not be merged as duplicates.", - "confidence": 0.91, + "summary": "All items are model-specific PRs that standardize output tracing/capture across different architectures. They share a refactoring theme, but they do not appear to fix the same concrete code path or to be mergeable into one PR as duplicates.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44129 is the best anchor because it is explicitly tied to tracking issue 43979 and represents a concrete, self-contained output-capture refactor.", + "canonical_pr_reason": "PR 44129 is the strongest cluster representative: it has the only explicit issue target, a focused model-specific refactor, and the clearest connection to the shared output-tracing migration theme.", "best_issue_reason": null, - "best_pr_reason": "PR 44129 is the strongest representative of the cluster: it has the clearest issue linkage and a substantive implementation change, while the others are similarly themed but model-specific.", + "best_pr_reason": "PR 44129 is the best standalone PR in the set because it is issue-linked and cleanly exemplifies the standardized output-capture refactor pattern.", "soft_edge_verdicts": [ { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Both are output-tracing refactors, but for different models (VisualBert vs TextNet) and different files/code paths; same pattern, not the same change." + "reason": "Both are output-tracing refactors, but they touch different models and code paths (VisualBert vs TextNet); same pattern, not the same change." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "MPNet and VisualBert are unrelated model implementations; they share a refactor theme but not a common underlying bug or mergeable change." + "reason": "MPNet and VisualBert are separate model implementations; this is a shared refactor style, not a duplicate bug or one mergeable PR." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "MPNet vs TextNet is different model code with different tracing hooks; this is a broad refactor similarity only." + "reason": "Different model targets and different implementation details; they are related by framework migration only, not by the same underlying fix." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are separate model paths; the diffs both migrate output handling, but they do not fix the same concrete issue." + "reason": "RegNet and SqueezeBert each migrate output tracing independently; no evidence they address the same concrete bug or code path." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "SqueezeBert and SpeechT5 are different architectures and modules; the shared output-capture refactor theme is too broad to treat as one PR." + "reason": "These are separate architecture migrations with different affected layers and outputs; too broad to treat as one duplicate PR." } ] }, @@ -22159,274 +22309,274 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "9857042405c6889b802c089ce0d5b9c3d26ba7d38acc1fc51727ea25f646358f", + "prepared_review_unit_hash": "555d84ea5f1f3e8611428eb74fe2a7ad8d2cb6a1b335bbe20a62ed17580810d6", "component_budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "budget": { "node_count": 13, "item_count": 13, "soft_pair_count": 40, - "serialized_chars": 40170, - "estimated_input_tokens": 10043, - "estimated_eval_tokens": 20342 + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are all separate model-specific output-tracing refactors, not duplicates of one underlying bug or change. No issue artifacts are present; the PRs share a theme but target different architectures and code paths.", - "confidence": 0.96, + "summary": "These PRs are thematically similar refactors around output tracing / `can_return_tuple`, but they target different model families and code paths. I do not see true duplicates or a single underlying bug/change shared across the set.", + "confidence": 0.87, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44003 is the strongest representative of the cluster: it has the broadest concrete tracing refactor and spans the closely related mamba/falcon_mamba code paths with explicit output-capturing plumbing.", + "canonical_pr_reason": "PR 44003 is the broadest representative of the cluster: it applies the output-tracing refactor across two related Mamba-family files and introduces the shared capture plumbing, making it the most central example of this theme.", "best_issue_reason": null, - "best_pr_reason": "PR 44003 is the best single exemplar here because it is the most substantial and concrete output-tracing change among otherwise unrelated model-specific refactors.", + "best_pr_reason": "PR 44003 is the best single representative in this set because it has the widest and clearest shared implementation change among closely related models, while the others are more isolated per-model refactors.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "timm_backbone and UperNet are different model families; same tracing refactor pattern only." + "reason": "Same refactor pattern, but timm_backbone and upernet are different model paths with separate forward behavior; not the same underlying change." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "timm_backbone and vision_text_dual_encoder use different forward paths and architectures." + "reason": "Both are output-tracing refactors, but they touch unrelated model families and code paths." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "timm_backbone and UnivNet are unrelated code paths; only the refactor style matches." + "reason": "Different architectures and separate forward implementations; shared theme is too broad to treat as one change." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "timm_backbone is unrelated to mamba/falcon_mamba; no shared underlying fix." + "reason": "timm_backbone refactor is unrelated to the Mamba/Falcon-Mamba changes in 44003." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "timm_backbone and RWKV are distinct architectures; same output-tracing theme is too broad." + "reason": "Different model families; no shared concrete bug or code path." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "MobileNetV1 and UnivNet are different models and domains; not the same change." + "reason": "MobileNetV1 and UnivNet are unrelated implementations; same refactor style does not make them duplicates." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "MobileNetV1 and UperNet touch different model code paths and output types." + "reason": "Separate models and separate forward signatures; only the general tracing refactor matches." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "MobileNetV1 and DepthAnything are separate architectures; shared decorator plumbing is not enough." + "reason": "MobileNetV1 and Depth Anything are different code paths; not one mergeable fix." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "MobileNetV1 and VisionEncoderDecoder are unrelated model paths, not one bug." + "reason": "Different subsystems with no shared concrete behavior change beyond output handling cleanup." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "MobileNetV1 and SpeechEncoderDecoder are different encoder-decoder families with distinct fixes." + "reason": "MobileNetV1 and Speech-Encoder-Decoder are unrelated; this is only a broad refactor similarity." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "vision_text_dual_encoder and UnivNet are unrelated models; same refactor theme only." + "reason": "Vision-text dual encoder and UnivNet are unrelated models; no common underlying bug/change." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "vision_text_dual_encoder and UperNet are different tasks and forward paths." + "reason": "These are separate model refactors in different code paths, not the same fix." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "vision_text_dual_encoder and mamba/falcon_mamba do not fix the same code-path problem." + "reason": "Shared tracing pattern, but CLIP-like vision-text and Mamba-family code are unrelated." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "vision_text_dual_encoder and DepthAnything are separate model families; no shared concrete bug." + "reason": "Different model families and forward logic; do not look mergeable into one PR." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "UnivNet and UperNet are unrelated audio vs vision models." + "reason": "UnivNet and UperNet are unrelated; same output-cleanup theme is too broad." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "UnivNet and mamba/falcon_mamba are different code paths and architectures." + "reason": "Different model families and distinct code paths; not a duplicate change." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "UnivNet and DepthAnything are unrelated model implementations." - }, - { - "left": "pull_request:44001", - "right": "pull_request:44004", - "accept": false, - "reason": "UnivNet and CodeGen are distinct model families; same output-handling theme is too broad." + "reason": "UnivNet and Depth Anything are separate implementations with no shared concrete bug." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "UperNet and RWKV are different architectures and output contracts." + "reason": "UperNet and RWKV are unrelated models; only the refactor pattern overlaps." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "UperNet and mamba/falcon_mamba do not share the same underlying fix." + "reason": "UperNet and Mamba-family code are different subsystems; not the same underlying change." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "mamba/falcon_mamba and SuperPoint are unrelated code paths; refactor style matches only." + "reason": "Mamba-family refactor and SuperPoint refactor are separate model-specific changes." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "mamba/falcon_mamba and CodeGen are different models with different output tracing needs." + "reason": "Both are output-signature refactors, but Mamba and CodeGen target different implementation details and are not one fix." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "timm_backbone and CodeGen are unrelated architectures; not mergeable as one PR." + "reason": "timm_backbone and CodeGen are unrelated; shared refactor language is insufficient." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "UnivNet and CodeGen are different domains and forward signatures." + "reason": "UnivNet and CodeGen are different models with distinct code paths." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "UperNet and CodeGen are separate model paths; no shared concrete bug." + "reason": "UperNet and CodeGen are unrelated beyond broad tracing cleanup." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder and CodeGen are not the same underlying change." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "DepthAnything and VisionEncoderDecoder are distinct architectures and changes." + "reason": "Depth Anything and Vision-Encoder-Decoder are separate model families; no shared concrete bug." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "DepthAnything and SpeechEncoderDecoder are different model families." + "reason": "Depth Anything and Speech-Encoder-Decoder are unrelated code paths." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "DepthAnything and SuperPoint are unrelated vision models with different fixes." + "reason": "These are different model-specific refactors; no single fix to merge." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "DepthAnything and RWKV are unrelated architectures; same refactor pattern only." + "reason": "Depth Anything and RWKV are unrelated models and implementation paths." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "DepthAnything and DPR are separate model code paths." + "reason": "Depth Anything and DPR are separate code paths; not a duplicate." }, { "left": "pull_request:44026", - "right": "pull_request:44027", + "right": "pull_request:44029", "accept": false, - "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are related only at a high level, but this is not one concrete change." + "reason": "Vision-Encoder-Decoder and RWKV are unrelated; only the output-tracing theme overlaps." }, { "left": "pull_request:44026", - "right": "pull_request:44028", + "right": "pull_request:44027", "accept": false, - "reason": "VisionEncoderDecoder and SuperPoint are unrelated implementations." + "reason": "Vision-Encoder-Decoder and Speech-Encoder-Decoder are different implementations and not the same change." }, { "left": "pull_request:44026", - "right": "pull_request:44029", + "right": "pull_request:44028", "accept": false, - "reason": "VisionEncoderDecoder and RWKV are different model families and output contracts." + "reason": "Vision-Encoder-Decoder and SuperPoint are unrelated model families." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "VisionEncoderDecoder and DPR are distinct code paths; not the same bug fix." + "reason": "Different models and different forward paths; no duplicate fix." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "SpeechEncoderDecoder and SuperPoint are unrelated model changes." + "reason": "Speech-Encoder-Decoder and SuperPoint are unrelated code paths." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "SpeechEncoderDecoder and RWKV are different architectures and outputs." + "reason": "Different model families; same refactor pattern is not enough." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "SpeechEncoderDecoder and DPR are separate model implementations." + "reason": "Speech-Encoder-Decoder and DPR are separate implementations with no shared bug." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "SuperPoint and RWKV are unrelated code paths and cannot be merged as one PR." + "reason": "SuperPoint and RWKV are unrelated; not mergeable into one PR." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "SuperPoint and DPR are different model families; same refactor theme is insufficient." + "reason": "SuperPoint and DPR are different model paths and not the same underlying fix." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "RWKV and DPR are unrelated model families with different output tracing behavior." + "reason": "RWKV and DPR are distinct models; only the generic tracing refactor overlaps." } ] }, @@ -22456,7 +22606,7 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "1bf6e10508b31e950934d6956e3594c47320d7cf200c9c3adee081805e58ba3d", + "prepared_review_unit_hash": "9000232f9688616c747369ab36366cb5c5644264577d0ce6ed53860dc3d148c4", "component_budget": { "node_count": 2, "item_count": 2, @@ -22478,18 +22628,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate model-implementation PRs that only overlap on shared Transformers scaffolding (docs, auto registries, package init). They target different models and code paths, so they should not be treated as duplicates.", - "confidence": 0.96, + "summary": "Two unrelated model-implementation PRs: one adds PenguinVL, the other adds Voxtral TTS. They share boilerplate auto-registry and docs paths, but target different models, codepaths, and test suites.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45401 is the better cluster representative because it is tied to an explicit tracking issue and is narrowly scoped to a single Voxtral TTS integration.", + "canonical_pr_reason": "PR #45401 is the better canonical PR for this cluster because it has an explicit issue target (45027) and is a focused, self-contained model addition.", "best_issue_reason": null, - "best_pr_reason": "PR 45401 is the best standalone PR in this cluster: it has an explicit issue target, a focused title, and a self-contained implementation for one model family.", + "best_pr_reason": "PR #45401 is the best PR to represent the cluster only in the sense that it is the most clearly scoped and traceable to a specific task; however, it is not a duplicate of #44662.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Reject: PenguinVL and Voxtral TTS are different models with different implementation surfaces; the shared files are just common registration/docs boilerplate, not the same underlying bug or mergeable change." + "reason": "Reject: these are different model additions (PenguinVL vs Voxtral TTS) with different implementations, configs, processors, and tests. Shared auto-registration/docs edits are broad framework boilerplate, not evidence of the same underlying change." } ] }, @@ -22519,12 +22669,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "0475c81e0a3b854d2d6486e9c8af0f0c9553bc33cc23574a6c226ccfcdeff436", + "prepared_review_unit_hash": "af8c378707a694dc81f26f9bdda697272192e6c698201bbe7fb60a3c7a3ac348", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4406, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -22532,7 +22682,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4407, + "serialized_chars": 4406, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -22541,18 +22691,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are about fusion-related functionality in Transformers, but they implement different layers and concrete mechanisms: #44979 adds a generic Module Fusion API (`module_fusion.py`), while #45363 adds `KernelConfig`-driven n-to-1 kernel fusion and integration changes. They are thematically close but not the same underlying change.", - "confidence": 0.92, + "summary": "These two pull requests are related by the broader idea of module/kernel fusion, but they implement different features and code paths: one adds a generic Module Fusion API, while the other adds n-to-1 kernel fusion through KernelConfig and model-loading integration. They are not duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45363 is the better canonical PR because it is the more complete, user-facing implementation of a specific fusion workflow (`KernelConfig`-based n-to-1 kernel fusion) with integrations and example usage.", + "canonical_pr_reason": "PR 45363 is the better canonical representative because it is the more user-facing end-to-end feature and touches the actual kernel-fusion integration path, whereas PR 44979 is a lower-level API foundation.", "best_issue_reason": null, - "best_pr_reason": "PR #45363 is the strongest representative of the cluster: it spans the main runtime path, related integration code, and an end-to-end example, whereas #44979 is a separate generic API proposal.", + "best_pr_reason": "PR 45363 is the best representative of the cluster because it covers the concrete fusion workflow with config, integration, and example usage; PR 44979 is more of an enabling API layer.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "Do not merge: they share the broad fusion theme, but #44979 defines a generic module fusion API and #45363 implements a different kernel-fusion configuration path. They do not appear to fix the same concrete code-path problem and would not plausibly be one PR." + "reason": "Related theme, but not the same change: 44979 introduces a generic Module Fusion API, while 45363 implements KernelConfig-driven n-to-1 kernel fusion and model-loading integration. They do not appear mergeable into one duplicate PR." } ] }, @@ -22582,7 +22732,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "40db47fbcd6d0ccd294084a4049999415e00d99774a1150225d0bab86a8bf5c5", + "prepared_review_unit_hash": "f9bc5de197ebddec69077274427d25e21db06edef5ef2d0065f838185b46be3b", "component_budget": { "node_count": 2, "item_count": 2, @@ -22604,18 +22754,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same concrete bug in `_patch_mistral_regex`: an `AttributeError` caused by using `tokenizer.backend_tokenizer.pre_tokenizer` instead of `tokenizer.pre_tokenizer`. The later PR is the more complete version because it adds tests, but the two changes are effectively the same fix.", - "confidence": 0.98, + "summary": "Both PRs implement the same Mistral tokenizer fix for the same AttributeError in `_patch_mistral_regex`, changing the same lines in the same file; 45317 adds regression tests on top of the identical code fix.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45317 is the better canonical PR because it fixes the same bug as #45086 and includes test coverage, making it the more complete representation of the change.", + "canonical_pr_reason": "pull_request:45317 is the most complete version of the shared fix: it applies the same code-path correction as 45086 and adds test coverage.", "best_issue_reason": null, - "best_pr_reason": "PR #45317 is the strongest representative of the cluster: same code-path bug, same target issue, and it adds regression tests while keeping the fix narrowly scoped.", + "best_pr_reason": "pull_request:45317 is the best representative because it contains the same bug fix plus regression tests, making it the stronger standalone change.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying AttributeError fix in the same function and file; the second PR is an expanded version with tests, so they are merge-equivalent for deduplication purposes." + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`, same target file and code change; 45317 just extends it with tests." } ] }, @@ -22647,12 +22797,12 @@ "pull_request:45115|pull_request:45569", "pull_request:45569|pull_request:46565" ], - "prepared_review_unit_hash": "e4c1e22f9a6db83503d8b8640803f447b068dafc78b044c1c70f3f49c7c6ac53", + "prepared_review_unit_hash": "c16742028f2a3fbc00a007d53da26f40d4c39bf3b4d5a71b164f69347fe2f575", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 9123, + "serialized_chars": 9122, "estimated_input_tokens": 2281, "estimated_eval_tokens": 4818 }, @@ -22660,7 +22810,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 9123, + "serialized_chars": 9122, "estimated_input_tokens": 2281, "estimated_eval_tokens": 4818 }, @@ -22669,24 +22819,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are all Nemotron-related, but they are not the same underlying change. PR 45569 is the main NemotronH family refactor, PR 45115 is a narrower conversion-mapping/inheritance cleanup, and PR 46565 is a separate Nemotron3.5 ASR feature addition.", - "confidence": 0.89, + "summary": "All three PRs are in the Nemotron family, but they target different changes: a NemotronH conversion/refactor, a NemotronH dense/sparse split and docs, and a separate Nemotron3.5 ASR addition. They are related contextually, but not duplicate fixes or mergeable as one PR.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45569 is the best cluster representative because it is the broadest and most central NemotronH refactor, touching the core model/config/auto-registration pieces rather than just a conversion shim or an unrelated model family.", + "canonical_pr_reason": "PR #45569 is the best representative of the cluster because it is the broadest NemotronH code change and the most central to the family split/refactor, touching the core model, auto-mapping, docs, and tests.", "best_issue_reason": null, - "best_pr_reason": "PR 45569 is the strongest standalone PR in this cluster: it covers the main NemotronH restructuring and includes the related dense/sparse split, making it the most representative artifact for the NemotronH work.", + "best_pr_reason": "PR #45569 is the strongest standalone PR for this cluster: it spans the main NemotronH implementation work rather than only a conversion mapping tweak (#45115) or an unrelated ASR model addition (#46565).", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "Related theme, but not the same concrete change. 45115 focuses on conversion-mapping renames for NemotronH compatibility, while 45569 is a broader model split/refactor with docs, auto mappings, and new dense/sparse packages." + "reason": "Both are NemotronH-related, but #45115 is a conversion-mapping/inheritance refactor while #45569 is the larger architecture split into NemotronH dense/sparse variants. They do not look like the same concrete bug or change." }, { "left": "pull_request:45569", "right": "pull_request:46565", "accept": false, - "reason": "Different model and different code path. 45569 is about NemotronH architecture/package refactoring; 46565 adds Nemotron3.5 ASR streaming with separate model, processor, generation, and test files." + "reason": "These address different models and code paths: NemotronH dense/sparse refactoring versus Nemotron3.5 ASR streaming support. Shared family naming is not enough to treat them as duplicates." } ] }, @@ -22716,7 +22866,7 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "05c07f894591bf9b1e8633a8ad43b44d9b9f3b6f65a1076a3cd9171d78fffb9b", + "prepared_review_unit_hash": "d72820a195f8f27352df4dd6017f465aded3b618cbc6e9ec06f058265f5d36a9", "component_budget": { "node_count": 2, "item_count": 2, @@ -22738,18 +22888,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs cover the same heterogeneous-transformer feature area, but they are not duplicates: one is a broader end-to-end feature PR (config + modeling), while the other is a narrower config-only slice. The overlap is substantial, but the scope differs enough that they look like complementary/iterative work rather than the same change.", - "confidence": 0.84, + "summary": "These two PRs are closely related parts of the same heterogeneous-model feature work. PR #45332 is the broader, more complete change set, while PR #45333 is a narrower config-only slice that overlaps heavily with it. The overlap is enough to treat them as the same change family for triage.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45332 is the better canonical representative because it covers the broader heterogeneous-model feature set, including both per-layer configuration and modeling utilities, while 45333 is limited to configuration support.", + "canonical_pr_reason": "#45332 is the better canonical PR because it covers the full heterogeneous-model feature, including per-layer config plus modeling/runtime changes, whereas #45333 is a narrower subset focused on config support only.", "best_issue_reason": null, - "best_pr_reason": "PR 45332 is the strongest overall artifact in the cluster: it has the wider implementation scope, more affected files, and appears to be the more complete feature integration.", + "best_pr_reason": "#45332 is the best representative of the cluster because it subsumes the shared config work and adds the broader modeling integration, making it the most complete instance of the underlying feature.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", - "accept": false, - "reason": "They overlap on heterogeneous configuration code and docs, but 45332 adds substantial modeling support beyond 45333. This looks like a superset/subset or split feature work, not the same concrete PR-level change." + "accept": true, + "reason": "Same underlying heterogeneous configuration feature; #45333 is a subset of the broader implementation in #45332, with substantial overlap in config/docs/tests and no conflicting scope." } ] }, @@ -22779,40 +22929,40 @@ "soft_pairs": [ "pull_request:45346|pull_request:46818" ], - "prepared_review_unit_hash": "c0190ed2a1a216d0bc2d9f6c144cad3a7a718d5f0071c29517cdd0ef974f09f7", + "prepared_review_unit_hash": "f66b95f83c9e0c960f83091dc23bcc8b96b4951ae1d813928e306d50d3eeb342", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5624, - "estimated_input_tokens": 1406, - "estimated_eval_tokens": 3068 + "serialized_chars": 5591, + "estimated_input_tokens": 1398, + "estimated_eval_tokens": 3052 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5624, - "estimated_input_tokens": 1406, - "estimated_eval_tokens": 3068 + "serialized_chars": 5591, + "estimated_input_tokens": 1398, + "estimated_eval_tokens": 3052 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs are not duplicates: one fixes MoE router logits being softmaxed twice in Mixtral/Qwen2-MoE, while the other fixes DeepGEMM FP8 scale handling and EP router contracts across many different models. They target different code paths and different failure modes.", - "confidence": 0.98, + "summary": "These two pull requests are not duplicates: one is a narrow regression test for MoE router logits being raw logits instead of softmaxed probabilities, while the other is a broad expert-parallel/DeepGEMM change about router contract handling and FP8 scale format across many models.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46818 is the more substantial standalone fix: it addresses a concrete runtime correctness issue in EP/FP8 and spans the actual implementation paths, not just regression tests.", + "canonical_pr_reason": "PR 46818 is the better canonical representative because it contains the substantive implementation changes and spans the actual code paths involved, whereas PR 45346 is a small test-focused regression fix for a different bug.", "best_issue_reason": null, - "best_pr_reason": "PR 46818 is the best representative of a real code-path fix because it contains the broader implementation changes and addresses an execution-time bug rather than only adding test coverage.", + "best_pr_reason": "PR 46818 is the best overall representative of the cluster because it is the larger code change and more central architectural fix; PR 45346 is too narrow and addresses a distinct MoE router-logits regression.", "soft_edge_verdicts": [ { "left": "pull_request:45346", "right": "pull_request:46818", "accept": false, - "reason": "Different underlying bugs: 45346 is about router logits being softmaxed twice in MoE models, while 46818 is about DeepGEMM/FP8 scale-format handling and EP router contracts. They do not appear mergeable into one PR." + "reason": "Different underlying bugs: 45346 tests for double-softmaxed router logits in Mixtral/Qwen2 MoE, while 46818 changes EP router contracts and FP8/DeepGEMM scale handling across many models. They do not look mergeable into one PR." } ] }, @@ -22842,12 +22992,12 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "66e00163a41d04226f0e9a08ab05e14b136a55f843f57469879a8de82dc256f4", + "prepared_review_unit_hash": "d090635122b2ae75c0ea70f4041755a82a229a55b06d5ded368a2697f71ccb2f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5514, + "serialized_chars": 5513, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -22855,7 +23005,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5514, + "serialized_chars": 5513, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -22864,18 +23014,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs are typing-related, but they address different scopes: #45415 is a broad type-checking sweep across many Transformers modules, while #45425 is a narrower typing cleanup centered on PEFT/modeling utilities and shared typing helpers. They are related in theme, not the same underlying change.", - "confidence": 0.91, + "summary": "Two open PRs in the same typing cleanup area, but they address different slices of the codebase and are not the same concrete fix/change.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45415 is the broader umbrella change and best represents the cluster\u2019s main theme of expanding type-check coverage across the codebase.", + "canonical_pr_reason": "PR 45415 is the broader, more central type-checking sweep across many core `src/transformers/*.py` files, so it best represents the main work in this cluster.", "best_issue_reason": null, - "best_pr_reason": "PR #45415 is the most representative single PR here because it spans the larger cross-module typing rollout; #45425 looks like a narrower follow-on/refinement rather than the same concrete fix.", + "best_pr_reason": "PR 45415 is the best single representative PR because it covers the widest set of typing changes and touches core runtime modules, whereas PR 45425 is a narrower follow-up typing cleanup centered around `_typing.py`/`modeling_utils` integration.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Both are typing-focused, but they fix different concrete problems and touch different code paths. #45415 adds typing support across many source files; #45425 adds specific typing protocols/aliases for PEFT, flash attention, and modeling utilities. They are not mergeable as one PR for the same bug/change." + "reason": "Both are typing-related and share some files, but the diffs target different concrete changes: 45415 is a broad type-checking pass over many modules, while 45425 adds typing protocol aliases/helpers and related adjustments. They are related cleanup work, not the same underlying bug or mergeable single PR." } ] }, @@ -22905,7 +23055,7 @@ "soft_pairs": [ "pull_request:45513|pull_request:46084" ], - "prepared_review_unit_hash": "82415c2fd358493d97b9a10ea39e342a4414e8b545f0a37393f5014d5527953a", + "prepared_review_unit_hash": "8155fdec04f5cb941da3a77331c3d4f98ba87e1f4dbed3bb77d40d4e6eee7ac0", "component_budget": { "node_count": 2, "item_count": 2, @@ -22927,18 +23077,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two PRs with superficially similar cache/seq_len>1 fixes, but they address different model families and different code paths, so they are not duplicates.", - "confidence": 0.95, + "summary": "Both PRs fix cache-related multi-token decode bugs, but they affect different model families and code paths (Qwen3.5 linear attention vs. Mamba2 CUDA kernels), so they are not duplicates and should not be merged into one PR.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46084 is the clearest representative: it has an explicit issue target and a narrowly scoped Mamba2 cache-path fix.", + "canonical_pr_reason": "PR #46084 is the clearest representative: it has an explicit issue target (#46032), a narrowly scoped fix for the seq_len>1 cache path in Mamba2, and accompanying tests.", "best_issue_reason": null, - "best_pr_reason": "PR 46084 is the best single PR to surface because it is concrete, self-contained, and tied to a specific tracked bug (#46032).", + "best_pr_reason": "PR #46084 is the best single PR to keep as the cluster representative because it is tightly scoped, explicitly linked to an issue, and addresses one concrete cached-forward bug path.", "soft_edge_verdicts": [ { "left": "pull_request:45513", "right": "pull_request:46084", "accept": false, - "reason": "Reject: both involve cached forward handling, but PR 45513 is about Qwen3.5/Qwen3-next linear attention multi-token cached continuation, while PR 46084 fixes Mamba2 CUDA kernel cache behavior for seq_len > 1. Different underlying bugs and not plausibly one merged PR." + "reason": "Both concern cached multi-token forwarding, but they fix different architectures and implementations, so they are not the same underlying change." } ] }, @@ -22968,7 +23118,7 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "6ec7dd66801e62894a8afe37717e2f10936f62138c01e868b08c12efdf607e29", + "prepared_review_unit_hash": "7e56a455d57f3a01088ed44223a30e38638c9bef8b199fe07dfccf834b13ff63", "component_budget": { "node_count": 2, "item_count": 2, @@ -22990,18 +23140,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two PRs both touch expert parallelism loading, but they fix different bugs in different code paths: one targets DeepSpeed ZeRO-3/accelerate config serialization and EP detection, the other targets FSDP2/post-shard parameter wrapping to prevent expert weights from being overwritten. They are related, but not duplicates.", - "confidence": 0.97, + "summary": "Both PRs are EP-related, but they fix different failure modes in different loading paths: #45548 focuses on preserving distributed_config/has_ep for DeepSpeed ZeRO-3 loading via accelerate launch, while #45662 fixes expert weights being overwritten during FSDP2 loading/broadcast. They are adjacent, not duplicates.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "#45662 is the stronger representative of the cluster because it addresses a concrete corruption bug in the expert-loading path and includes the broader set of code changes tied to that failure mode.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the better representative of the cluster because it addresses a concrete, narrowly-scoped EP loading corruption bug with explicit loader/runtime changes across multiple files.", + "best_pr_reason": "#45662 is the best single PR to represent the cluster because it is the most specific, technically grounded fix and the more substantial end-to-end change around EP loading behavior.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Same broad EP area, but different concrete failures and fixes: 45548 is about preserving distributed_config/has_ep for accelerate + ZeRO-3 loading, while 45662 prevents rank-0 overwrite via post-shard wrapping for FSDP2. Not the same underlying bug and not a single mergeable PR." + "reason": "Different underlying bugs and code paths: #45548 preserves distributed config state for ZeRO-3/accelerate launch, whereas #45662 prevents experts from being overwritten during FSDP2 loading. Shared EP terminology and one common file are not enough to treat them as the same change." } ] }, @@ -23031,7 +23181,7 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "fa5cc9674480896c3a456ff0c5ed3a7d953b42b3d0b37b4e0050e742ef76880e", + "prepared_review_unit_hash": "a9fd8bfc920f0594f6a6b8dae31d8ae951837c857dbcb21f497ed75d7a5338ba", "component_budget": { "node_count": 2, "item_count": 2, @@ -23053,18 +23203,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs address the same xdist-related captured_info collision in testing_utils; 45645 appears to be the more complete follow-up, adding CI log aggregation for multiple files.", - "confidence": 0.91, + "summary": "Both PRs address the same xdist collision around captured_info/debug logging in `testing_utils`, with 45645 expanding the fix to CI artifact collection as well. 45645 is the better canonical PR because it is the more complete end-to-end change.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the broader, more complete fix: it updates testing_utils and the CI workflow to handle worker-specific captured_info files, whereas 45639 only fixes the writer side.", + "canonical_pr_reason": "PR 45645 is the broader, more complete fix: it includes the xdist-safe output handling in `src/transformers/testing_utils.py` and also updates the workflow to read multiple `captured_info*.txt` files, preserving CI debug logs. That makes it the best representative of the change set.", "best_issue_reason": null, - "best_pr_reason": "45645 best represents the cluster because it covers the same bug and includes the extra CI log-preservation changes needed to use the new per-worker files.", + "best_pr_reason": "PR 45645 is the strongest single PR for this cluster because it covers both the underlying collision fix and the downstream CI log collection behavior. PR 45639 appears to be the narrower precursor/subset focused only on making the write path xdist-safe.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Both PRs fix the same underlying xdist collision for captured_info artifacts; 45645 is essentially a superset of 45639 and could plausibly be merged as one change." + "reason": "Same underlying xdist collision/captured_info bug in `testing_utils`; 45645 looks like a superset that adds the workflow-side handling needed to preserve logs, so they could plausibly be merged as one change." } ] }, @@ -23094,7 +23244,7 @@ "soft_pairs": [ "pull_request:45720|pull_request:46259" ], - "prepared_review_unit_hash": "31ab4a3f114405540ee7b3d09b589ecaaf933cb3bff386c8a33f6154b3151fb5", + "prepared_review_unit_hash": "da8025eed121ea1f7dd0ebd60011307245cfb48d7051e91d0d914154cb0f8eea", "component_budget": { "node_count": 2, "item_count": 2, @@ -23116,18 +23266,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two unrelated pull requests were grouped only because they both touch many model files and are linter-related. PR 45720 implements TRF003 by adding `can_return_tuple` / removing `return_dict` in forward methods; PR 46259 implements TRF018 by changing weight initialization logic. They should not be merged as duplicates.", - "confidence": 0.97, + "summary": "These are two different linter-rule PRs in the same general area, but they address unrelated changes: TRF003 adds a return-tuple decorator to model forward methods, while TRF018 changes weight-initialization logic across model classes. They are not duplicates and should not be merged together.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45720 is the cleaner canonical representative because it focuses on one concrete code-pattern change across model forwards, with a clear, narrow rule scope (TRF003).", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 45720 is the best representative of the cluster since it is a more self-contained, specific linter-driven change and is easier to use as the anchor for this grouping than the broader initialization-rule PR.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:45720", "right": "pull_request:46259", "accept": false, - "reason": "Different linter rules and different code paths: one changes forward return handling (TRF003), the other changes model weight initialization (TRF018). Shared model filenames are incidental, not the same underlying fix." + "reason": "Different underlying changes and code paths: 45720 is about forward/return_dict handling via can_return_tuple, while 46259 is about _init_weights initialization behavior. Similar linter framing and overlapping model files are not enough to treat them as the same PR." } ] }, @@ -23223,7 +23373,7 @@ "soft_pairs": [ "pull_request:45932|pull_request:46100" ], - "prepared_review_unit_hash": "7595cab697726dbd1cd78a49e1620a977be9122fc8695f8850c223bc0a530e41", + "prepared_review_unit_hash": "6be53d2cde71f79b2bea5956de30138a64ee1a9c3fad034ad2e359a67dd710de", "component_budget": { "node_count": 2, "item_count": 2, @@ -23245,18 +23395,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two unrelated PRs in the same broad repository area, but they do not appear to fix the same concrete bug or implement the same change. One is a model/config and CI-related sweep; the other is a test migration adding kernel/model tests.", + "summary": "The two pull requests are not duplicates: one is a broad model/code-path and CI-secret change set, while the other is a test migration with new kernel/model tests. They should remain separate.", "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45932 is the broader code-change PR and is the better cluster representative because it contains the main implementation/config edits across model files and workflow plumbing, whereas #46100 is primarily a test migration.", + "canonical_pr_reason": "PR 45932 is the more substantial representative PR: it changes model code and workflow configuration, whereas 46100 is primarily a test migration and test additions.", "best_issue_reason": null, - "best_pr_reason": "PR #45932 is the strongest single representative of the cluster because it has substantive production-code/config changes; #46100 is mostly test scaffolding and less representative of a core fix/change.", + "best_pr_reason": "PR 45932 is the best single representative of this cluster because it is the broader functional/code change; 46100 is focused on test infrastructure and does not cover the same underlying change.", "soft_edge_verdicts": [ { "left": "pull_request:45932", "right": "pull_request:46100", "accept": false, - "reason": "Reject: these PRs touch different concrete concerns. #45932 changes CI/workflow plus several model/config files, while #46100 adds new kernel tests and test migration files. They do not look like the same bug fix or the same mergeable change." + "reason": "They do not appear to fix the same concrete bug or implement the same change. 45932 modifies model implementation/mappings and CI workflow secrets, while 46100 adds kernel tests and a DDP test config for test migration. The overlap is only broad testing/model areas, not one mergeable code-path fix." } ] }, @@ -23286,7 +23436,7 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "7001a17f3524860f237ad60edab4120737395908fa9497bd242302b085f70eb7", + "prepared_review_unit_hash": "8ba0719448ac7deea64b2bf7c14be2bbb9bf9b9eaf34ee5720f4b98ba38295f4", "component_budget": { "node_count": 2, "item_count": 2, @@ -23308,18 +23458,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs overlap on distributed code, but they are not the same change: one is a focused EP/Trainer integration PR and the other is a broad distributed branch base spanning many unrelated areas.", - "confidence": 0.91, + "summary": "These are not duplicate PRs: one is a focused EP + Trainer integration change, while the other is a broad distributed branch base touching many unrelated areas. The overlap in distributed files looks incidental to the shared branch rather than the same concrete fix.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46126 is the most specific, self-contained change in the cluster: EP + Trainer integration with targeted distributed-file edits. PR 46269 is a much broader base branch touching CI, FSDP, generation, integrations, and model code.", + "canonical_pr_reason": "PR #46126 is the more canonical change because it is narrowly scoped and centers on a single feature/fix set (EP + Trainer integration around DistributedConfig).", "best_issue_reason": null, - "best_pr_reason": "PR 46126 is the best representative PR because it has a clear concrete purpose and tighter scope, making it more suitable as the cluster canonical than the broad umbrella branch.", + "best_pr_reason": "PR #46126 is the best representative artifact: it is coherent, specific, and directly describes the underlying change, whereas #46269 is too broad and base-like to serve as a good canonical PR.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Shared distributed files are not enough here. 46126 is a focused EP/Trainer integration change, while 46269 is a wide distributed base branch with many unrelated modifications; they do not look like the same underlying PR." + "reason": "The overlap is only on broad distributed files; #46126 implements EP/Trainer integration, while #46269 is a large distributed branch base spanning CI and multiple subsystems. They do not look like the same concrete code-path fix and would not plausibly merge into one PR." } ] }, @@ -23385,12 +23535,12 @@ "pull_request:46504|pull_request:46563", "pull_request:46504|pull_request:46716" ], - "prepared_review_unit_hash": "3e6eec50827d83c638c55397bb184eca5d8d9c5fe8a7c797a3d05ad62cc6bb5c", + "prepared_review_unit_hash": "62102420565375cbce5f562fac75c3482f70d5787129751ae2fa28053f97e30c", "component_budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 29, - "serialized_chars": 25918, + "serialized_chars": 25920, "estimated_input_tokens": 6480, "estimated_eval_tokens": 13216 }, @@ -23398,7 +23548,7 @@ "node_count": 10, "item_count": 10, "soft_pair_count": 29, - "serialized_chars": 25918, + "serialized_chars": 25920, "estimated_input_tokens": 6480, "estimated_eval_tokens": 13216 }, @@ -23407,186 +23557,186 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Mostly unrelated workflow-maintenance PRs. There is a small security-gate subgroup in .github/workflows/pr-ci-security-gate.yml, but the diffs are distinct follow-ups rather than the same fix, so no soft merges are justified.", - "confidence": 0.96, + "summary": "The cluster is a set of unrelated workflow-maintenance PRs. The only apparent mini-cluster is pr-ci-security-gate.yml, but those PRs make different security/permission changes, so they are not duplicates. No soft edges should be merged.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46554 is the most substantive representative of the security-gate subgroup, with the core functional change; the others are separate hardening/comment-only follow-ups.", + "canonical_pr_reason": "No single true canonical PR across the whole cluster; pull_request:46554 is the closest representative because it is the most substantial and central pr-ci-security-gate hardening change, but the surrounding PRs are separate follow-up edits rather than the same fix.", "best_issue_reason": null, - "best_pr_reason": "PR 46554 is the best overall representative of the cluster because it contains the largest concrete workflow behavior change, while the rest are independent workflow tweaks.", + "best_pr_reason": "pull_request:46554 is the best representative of the tightest subcluster by breadth and connectivity, though it is not a true duplicate hub because the other pr-ci-security-gate PRs change different behaviors.", "soft_edge_verdicts": [ { "left": "pull_request:46138", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows and different fixes: self-comment CI stdin handling vs PR CI security-gate logic." + "reason": "Different workflow and bug: self-comment-ci switches PR comment handling to stdin, while 46554 hardens pr-ci-security-gate. Not the same underlying change." }, { "left": "pull_request:46138", "right": "pull_request:46558", "accept": false, - "reason": "Unrelated workflow updates; one edits comment parsing, the other changes PR CI gate permissions/approval flow." + "reason": "Different workflow and code path; one fixes comment piping, the other changes security-gate permissions/approval flow." }, { "left": "pull_request:46138", "right": "pull_request:46563", "accept": false, - "reason": "Self-comment CI command piping and a comment-only note in pr-ci-security-gate are not the same underlying change." + "reason": "Different workflow maintenance edits in different code paths; not the same bug or change." }, { "left": "pull_request:46138", "right": "pull_request:46716", "accept": false, - "reason": "Different workflows and unrelated changes: stdin handling vs secrets inheritance removal." + "reason": "Self-comment-ci stdin fix is unrelated to the pr-ci-security-gate secrets-scope change." }, { "left": "pull_request:46138", "right": "pull_request:46504", "accept": false, - "reason": "Build CI docker image trigger handling is unrelated to self-comment CI message piping." + "reason": "Different workflows and fixes: comment stdin handling versus COMMIT_MESSAGE quoting in build-ci-docker-images." }, { "left": "pull_request:46432", "right": "pull_request:46503", "accept": false, - "reason": "SSH runner hardening and benchmark dispatch enablement are separate workflow changes." + "reason": "SSH-runner hardening is unrelated to benchmark_v2 dispatch-trigger changes." }, { "left": "pull_request:46432", "right": "pull_request:46501", "accept": false, - "reason": "SSH runner env/monitor changes are unrelated to anti-slop permissions placement." + "reason": "Different workflow files and unrelated changes; ssh-runner hardening is not the same as anti-slop permission reshuffling." }, { "left": "pull_request:46432", "right": "pull_request:46502", "accept": false, - "reason": "SSH runner changes do not address the assign-reviewers checkout ref behavior." + "reason": "Different workflow and fix: ssh-runner secret/memory-monitor hardening versus assign-reviewers checkout ref adjustment." }, { "left": "pull_request:46432", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows and different problems: SSH runner security cleanup vs PR CI security-gate behavior." + "reason": "Different workflow and bug; ssh-runner changes are unrelated to pr-ci-security-gate behavior." }, { "left": "pull_request:46432", "right": "pull_request:46558", "accept": false, - "reason": "SSH runner hardening is unrelated to PR CI security-gate permissions and approval gating." + "reason": "Different workflow and concern: ssh-runner security/hardening versus pr-ci-security-gate permission tweaks." }, { "left": "pull_request:46501", "right": "pull_request:46503", "accept": false, - "reason": "Anti-slop permission scoping and benchmark workflow dispatch support are unrelated changes." + "reason": "Different workflow files and different logic; anti-slop permission change is not the benchmark_v2 trigger update." }, { "left": "pull_request:46501", "right": "pull_request:46502", "accept": false, - "reason": "Different workflows and different fixes; no shared code path." + "reason": "Different workflows and unrelated fixes; not the same underlying change." }, { "left": "pull_request:46501", "right": "pull_request:46504", "accept": false, - "reason": "Anti-slop permissions update is unrelated to build-ci-docker-images commit-message handling." + "reason": "anti-slop.yml permission adjustment and build-ci-docker-images message quoting are separate changes." }, { "left": "pull_request:46501", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and different behaviors; not the same underlying bug or change." + "reason": "Different workflow files and unrelated concerns; not a duplicate." }, { "left": "pull_request:46502", "right": "pull_request:46503", "accept": false, - "reason": "Assign-reviewers checkout ref and benchmark workflow dispatch are unrelated." + "reason": "assign-reviewers checkout ref fix and benchmark_v2 dispatch support are different workflow changes." }, { "left": "pull_request:46502", "right": "pull_request:46504", "accept": false, - "reason": "Assign-reviewers base-ref checkout is unrelated to build-ci-docker-images trigger parsing." + "reason": "Different workflows and fixes; no shared concrete bug." }, { "left": "pull_request:46502", "right": "pull_request:46554", "accept": false, - "reason": "No overlap in workflow behavior: reviewer assignment vs PR CI security checks." + "reason": "Different workflow files and unrelated behavior changes." }, { "left": "pull_request:46503", "right": "pull_request:46504", "accept": false, - "reason": "Benchmark workflow dispatch support and docker-image tag trigger parsing are separate changes." + "reason": "Benchmark dispatch trigger and build-ci-docker-images grep/quoting fix are unrelated." }, { "left": "pull_request:46503", "right": "pull_request:46554", "accept": false, - "reason": "Different workflows and different functionality; not a duplicate fix." + "reason": "Different workflow and code path; benchmark_v2 dispatch support is not pr-ci-security-gate hardening." }, { "left": "pull_request:46504", "right": "pull_request:46554", "accept": false, - "reason": "Build-ci docker image trigger parsing is unrelated to PR CI security-gate logic." + "reason": "Different workflows and fixes; build-ci-docker-images COMMIT_MESSAGE handling is unrelated to pr-ci-security-gate changes." }, { "left": "pull_request:46504", "right": "pull_request:46558", "accept": false, - "reason": "Different workflows and different fixes; not the same code-path problem." + "reason": "Different workflow files and different behavior; not the same underlying bug/change." }, { "left": "pull_request:46554", "right": "pull_request:46558", "accept": false, - "reason": "Both touch pr-ci-security-gate.yml, but one adds checkout/diff-based security checks while the other tightens permissions and approval gating; separate follow-up hardening, not the same fix." + "reason": "Same file, but different concrete changes: 46554 adds checkout and diff-based checks, while 46558 changes permissions and approval gating." }, { "left": "pull_request:46554", "right": "pull_request:46563", "accept": false, - "reason": "Functional security-gate changes versus a comment-only safety note are not the same underlying change." + "reason": "Same workflow file, but distinct changes: security checks versus reusable-workflow review note/comments." }, { "left": "pull_request:46554", "right": "pull_request:46716", "accept": false, - "reason": "Security-gate logic changes and secrets:{} on the reusable workflow call are distinct fixes." + "reason": "Same file, but different fixes: added security checks versus changing secrets inheritance to an empty map." }, { "left": "pull_request:46558", "right": "pull_request:46563", "accept": false, - "reason": "Permissions/approval gating and a documentation comment do not fix the same bug." + "reason": "Same workflow, but one changes permissions/if-guard and the other adds documentation about the reusable workflow. Not the same change." }, { "left": "pull_request:46558", "right": "pull_request:46716", "accept": false, - "reason": "Different hardening steps in the same workflow file, but not the same concrete change." + "reason": "Same file, but distinct security-gate adjustments: permission/approval logic versus secrets scoping." }, { "left": "pull_request:46563", "right": "pull_request:46716", "accept": false, - "reason": "Comment-only reassurance and secrets inheritance removal are unrelated changes." + "reason": "Same workflow file, but different concrete edits: explanatory comments versus secrets inheritance change." }, { "left": "pull_request:46504", "right": "pull_request:46563", "accept": false, - "reason": "Build-ci docker image trigger parsing is unrelated to a pr-ci-security-gate explanatory comment." + "reason": "Different workflows and unrelated fixes; build-ci-docker-images message handling is not the pr-ci-security-gate note/comment change." }, { "left": "pull_request:46504", "right": "pull_request:46716", "accept": false, - "reason": "Different workflows and different fixes; no shared underlying bug." + "reason": "Different workflows and code paths; no shared underlying bug or change." } ] }, @@ -23616,7 +23766,7 @@ "soft_pairs": [ "pull_request:46152|pull_request:46207" ], - "prepared_review_unit_hash": "68bcc048c33d3e0a610fd0f5cb37edf0c450d9a09ac06fd7a03306e161b47727", + "prepared_review_unit_hash": "b9e92902365bd7141da431b6fa238c57bb046543f8757e31e5561db4e86457e4", "component_budget": { "node_count": 2, "item_count": 2, @@ -23638,18 +23788,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are thematically similar refactors around source inspection, but they fix different code paths in different modules and are not duplicates.", - "confidence": 0.93, + "summary": "Both PRs replace brittle source-file open() usage with safer source inspection, but they fix different code paths: import_utils resource resolution vs modeling_utils attention-implementation detection.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the changes are in separate helpers (`import_utils._resolve_traversable` vs `modeling_utils._can_set_attn_implementation`) and address different source-reading call sites.", + "canonical_pr_reason": "No canonical PR: the changes are thematically similar but target different modules and different behaviors, so they are not interchangeable duplicates.", "best_issue_reason": null, - "best_pr_reason": "PR 46207 is the stronger standalone change because it has a narrower, well-defined target and includes tests, but it still does not subsume PR 46152.", + "best_pr_reason": "PR 46207 is the cleaner representative of the theme because it addresses one concrete source-reading heuristic and includes tests, but it should remain separate from 46152.", "soft_edge_verdicts": [ { "left": "pull_request:46152", "right": "pull_request:46207", "accept": false, - "reason": "Same general pattern (replacing file opens for source access), but different modules, different functions, and different underlying behavior; not the same concrete fix." + "reason": "Shared 'avoid open()' theme only; one is resource-path traversal in import_utils, the other is source introspection in modeling_utils. Different bugs and code paths." } ] }, @@ -23742,7 +23892,7 @@ "soft_pairs": [ "pull_request:46310|pull_request:46546" ], - "prepared_review_unit_hash": "11b1e9287aed76823b5fee46fc7ffcac56256f6f9621259557011933829b6603", + "prepared_review_unit_hash": "44cd8039da96fc3c1b6692d3daf1d927165a2ef4025832317d84adde1f4d5a40", "component_budget": { "node_count": 2, "item_count": 2, @@ -23764,18 +23914,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs address the same off-by-one right-window boundary in the sliding-window mask helper for Moonshine streaming; the later PR is a superset with tests and an additional Gemma4 fix.", - "confidence": 0.89, + "summary": "Both PRs fix the same Moonshine streaming sliding-window boundary bug (`-dist <= right_window_size` vs `<`) in the same code path; PR #46546 is the broader, more complete version because it includes the Moonshine fix plus tests and a parallel Gemma4 boundary fix.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46546 is the stronger representative: it contains the same Moonshine streaming boundary fix as #46310, adds test coverage, and reflects the more complete patch.", + "canonical_pr_reason": "PR #46546 is the better canonical representative: it fixes the Moonshine streaming boundary bug, adds tests, and is the more complete patch even though it also includes an additional Gemma4 change.", "best_issue_reason": null, - "best_pr_reason": "PR #46546 is the best standalone candidate because it includes the Moonshine fix plus validation tests and is the more fully developed patch, even though it also applies the same boundary correction to Gemma4.", + "best_pr_reason": "PR #46546 is the strongest overall candidate because it addresses the same concrete Moonshine streaming bug as #46310 and includes test coverage, making it the more complete fix.", "soft_edge_verdicts": [ { "left": "pull_request:46310", "right": "pull_request:46546", "accept": true, - "reason": "Both PRs fix the same Moonshine streaming right-window off-by-one (`<` to `<=`) in the same mask helper. #46546 is a superset with tests, so they plausibly represent the same underlying bug fix." + "reason": "Both PRs change the same Moonshine streaming right-window boundary check in the same function family; #46546 is a superset that could plausibly absorb #46310." } ] }, @@ -23805,7 +23955,7 @@ "soft_pairs": [ "pull_request:46376|pull_request:46417" ], - "prepared_review_unit_hash": "75021654cbf0a7465310db97f312fa6222a9cf3d1f343888279010fd6f45bb88", + "prepared_review_unit_hash": "1ec740b1c5682b5aa57701693d50a8baed39ea95d271efbd895ce89e01e39c7b", "component_budget": { "node_count": 2, "item_count": 2, @@ -23827,18 +23977,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests are unrelated: one adds multimodal support to continuous batching internals, while the other adds a new HunYuan VL model with docs, configs, processors, and tests. They do not target the same code path or change the same feature set.", + "summary": "The two pull requests are unrelated: one adds multimodal continuous batching infrastructure, while the other adds the HunYuan VL model package, docs, and tests. They should not be clustered as duplicates.", "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 46417 is the more complete, non-draft, self-contained feature addition and better represents the cluster if a single PR had to be surfaced, though it is not a duplicate of PR 46376.", + "best_pr_reason": "PR 46417 is the more self-contained and complete change, with a coherent model addition plus docs/tests. PR 46376 is a draft infrastructure enhancement and not a duplicate of it.", "soft_edge_verdicts": [ { "left": "pull_request:46376", "right": "pull_request:46417", "accept": false, - "reason": "Different underlying changes: 46376 modifies continuous batching to support multimodal caching/scheduling, while 46417 adds a new HunYuan VL model stack, docs, auto-mappings, and tests. No shared concrete bug or mergeable common code path." + "reason": "Different underlying changes and code paths: continuous batching/multimodal runtime support vs. a new HunYuan VL model integration. No shared filenames or concrete fix target, so they are not mergeable as one PR." } ] }, @@ -23868,7 +24018,7 @@ "soft_pairs": [ "pull_request:46574|pull_request:46580" ], - "prepared_review_unit_hash": "e88adbb80c676c11dcb11df04e0351fbdec8b7762f4574cfed1df86eba95cfb4", + "prepared_review_unit_hash": "fe627fd5c8f3ace331bb4121eafdb455e1d7d9ab708453999acf6c8604bc6e73", "component_budget": { "node_count": 2, "item_count": 2, @@ -23890,18 +24040,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are not duplicates: PR #46574 fixes CodeLlama decode stripping of real leading spaces, while PR #46580 fixes DeepSeek-Coder/DeepSeek-LLM loading through the wrong tokenizer class in auto-tokenization.", + "summary": "These are two unrelated pull requests: one fixes CodeLlama decode stripping behavior, the other fixes DeepSeek-Coder/DeepSeek-LLM tokenizer-class selection in AutoTokenizer. They do not describe the same underlying bug or change.", "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: they fix different bugs in different tokenizer code paths and should not be merged.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "No single best PR for deduping; neither can represent the other because the underlying problems and affected models differ.", + "best_pr_reason": "PR #46574 is the cleaner representative if one must be picked: it is a focused single-bug tokenizer fix with a narrow code path and explicit issue target.", "soft_edge_verdicts": [ { "left": "pull_request:46574", "right": "pull_request:46580", "accept": false, - "reason": "Different root causes and code paths: one changes CodeLlama decoding behavior, the other changes auto-tokenizer class selection for DeepSeek checkpoints." + "reason": "Different models, different failure modes, and different code paths: CodeLlama decode stripping vs DeepSeek checkpoint-to-tokenizer-class routing. They are not plausibly mergeable into one PR." } ] }, @@ -23931,12 +24081,12 @@ "soft_pairs": [ "pull_request:46603|pull_request:46604" ], - "prepared_review_unit_hash": "0a668f55f8655bdf21ff2defc2982315554401da4e8652817b8d942cb47753f3", + "prepared_review_unit_hash": "7bb1685d8c8f84a2cf7367398f51bbe9d3a1a0da9684fd7271d67d0f8ada431f", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5496, + "serialized_chars": 5495, "estimated_input_tokens": 1374, "estimated_eval_tokens": 3004 }, @@ -23944,7 +24094,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5496, + "serialized_chars": 5495, "estimated_input_tokens": 1374, "estimated_eval_tokens": 3004 }, @@ -23953,18 +24103,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs share Mistral tokenizer infrastructure but target different changes: one refactors the existing slow-tokenizer converter into the Mistral integration package, while the other adds tekken tokenizer detection, conversion, and save utilities. They are related but not the same underlying change.", + "summary": "Both PRs touch the Mistral tokenizer stack and share many files, but they address different changes: 46603 is a refactor moving MistralConverter into the integrations package, while 46604 adds tekken tokenizer detection/conversion/save utilities. They are related but not duplicates.", "confidence": 0.94, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 46604 is the more substantive end-state feature PR for Mistral tokenizer support; PR 46603 is a narrower packaging/refactor change that does not cover the tekken detection/conversion work.", "best_issue_reason": null, - "best_pr_reason": "PR #46604 is the more substantial and self-contained Mistral change: it adds new tokenizer detection/conversion/save behavior and associated tests, so it is the better representative artifact for this cluster.", + "best_pr_reason": "PR 46604 best represents the main functional change in the cluster because it introduces new tokenizer behavior and utilities, whereas 46603 is only a code move/refactor.", "soft_edge_verdicts": [ { "left": "pull_request:46603", "right": "pull_request:46604", "accept": false, - "reason": "Reject: these are distinct Mistral work items. #46603 is a code move/refactor for MistralConverter; #46604 introduces new tekken tokenizer functionality and utilities. Shared files alone are not enough to treat them as the same fix/change." + "reason": "Shared Mistral files reflect the same subsystem, but the PRs fix/add different concrete behaviors and would not be merged as one duplicate change." } ] }, @@ -23998,12 +24148,12 @@ "pull_request:46741|pull_request:46773", "pull_request:46741|pull_request:46831" ], - "prepared_review_unit_hash": "9f3accbc8b1634cd8ff0d05dce0e29e39f35769ef9c173f61f6d53685196b8ed", + "prepared_review_unit_hash": "55a74069bc9ac1db2e56ba33f3470a06e10154c9a663ea740c755ea0c5100afd", "component_budget": { "node_count": 4, "item_count": 4, "soft_pair_count": 3, - "serialized_chars": 10051, + "serialized_chars": 10052, "estimated_input_tokens": 2513, "estimated_eval_tokens": 5282 }, @@ -24011,7 +24161,7 @@ "node_count": 4, "item_count": 4, "soft_pair_count": 3, - "serialized_chars": 10051, + "serialized_chars": 10052, "estimated_input_tokens": 2513, "estimated_eval_tokens": 5282 }, @@ -24020,30 +24170,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These four PRs are related only by broad cache/model-forward logic, but they fix different bugs in different model families and code paths. None of the soft edges should be merged as duplicates.", - "confidence": 0.96, + "summary": "These four PRs are all about model-runtime bugs, but they target different code paths and failure modes: cached decoding in SSM mixers, padding-mask handling, quantized cache construction, and CPU-offload hook behavior. They are not duplicates of one another.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46773 is the clearest standalone bug fix: it has a precise failure mode (padding mask skipped for batch size 1) and a straightforward, broadly applicable fix across linear-attention models.", + "canonical_pr_reason": "PR 46741 is the best representative of the set because it fixes a narrow, concrete decoding bug with a focused model-specific patch and test coverage; the others address unrelated runtime issues in different subsystems.", "best_issue_reason": null, - "best_pr_reason": "PR #46773 is the strongest overall candidate to keep as the representative PR because it addresses a concrete, reproducible bug with broad model impact and a focused code change.", + "best_pr_reason": "PR 46741 is the strongest standalone fix: it targets a specific cached single-token decoding path and is comparatively self-contained, while the other PRs are broader or solve different problems entirely.", "soft_edge_verdicts": [ { "left": "pull_request:46741", "right": "pull_request:46781", "accept": false, - "reason": "Different bugs: #46741 fixes Mamba2/Zamba2/Nemotron-H cached single-token/speculative decoding state handling; #46781 fixes QuantizedCache initialization for hybrid SSM+attention models." + "reason": "Different bugs: 46741 fixes cached speculative/chunked decoding in Zamba2/Nemotron-H mixers, while 46781 fixes QuantizedCache construction for hybrid SSM+attention models. Separate code paths and not mergeable as one PR." }, { "left": "pull_request:46741", "right": "pull_request:46773", "accept": false, - "reason": "#46741 is about cache/state handling in Mamba2-style mixers during decoding; #46773 is about applying padding masks for batch size 1 in linear-attention models. Separate failure modes." + "reason": "Different issues: 46741 is about single-token cached decode logic in Mamba2-style mixers; 46773 changes padding-mask application when batch size is 1. Same broad family, but not the same underlying bug." }, { "left": "pull_request:46741", "right": "pull_request:46831", "accept": false, - "reason": "#46741 targets Mamba2/Zamba2/Nemotron-H precomputed-state decoding, while #46831 fixes CPU-offload hook handling in Qwen3.5/Qwen3-Next. Different model families and code paths." + "reason": "Different runtime problem: 46741 addresses cached decode state handling, whereas 46831 fixes garbage generation under device_map CPU offload by triggering a conv1d hook. Not the same code-path defect." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index 0791809118c278fe131781ffbe6b0a327577223c..38b547040618634b05c39f4202e4df9dccdfd887 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260623t060018z", + "analysis_id": "hybrid-model-20260623t120018z", "archived_artifacts": { - "hybrid": "snapshots/20260623T060018Z/analysis-runs/hybrid-model-20260623t060018z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260623T060018Z/analysis-runs/hybrid-model-20260623t060018z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-06-23T06:11:45Z", + "published_at": "2026-06-23T12:12:35Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260623T060018Z", + "snapshot_id": "20260623T120018Z", "variant": "hybrid" } diff --git a/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..e29ed71f2ca652936cb6883bcc2c1c5ececab5bc --- /dev/null +++ b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json @@ -0,0 +1,2627 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260623T120018Z", + "generated_at": "2026-06-23T12:12:33Z", + "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-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": 41116, + "issue_numbers": [ + 41115 + ], + "pr_numbers": [ + 41116, + 45613 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 41116, + "right_pr_number": 45613, + "code_similarity": 0.792, + "size_similarity": 0.893, + "file_overlap": 0.692, + "area_overlap": 0.882, + "patch_similarity": 0.648, + "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": [ + [ + 737, + 744 + ] + ], + "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": [ + [ + 301, + 307 + ], + [ + 746, + 752 + ], + [ + 1366, + 1372 + ] + ], + "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, + 150 + ] + ], + "right_ranges": [ + [ + 1, + 126 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 575 + ] + ], + "right_ranges": [ + [ + 1, + 522 + ] + ] + }, + { + "filename": "src/transformers/models/minicpm3/modular_minicpm3.py", + "left_ranges": [ + [ + 1, + 354 + ] + ], + "right_ranges": [ + [ + 1, + 342 + ] + ] + }, + { + "filename": "tests/models/minicpm3/test_modeling_minicpm3.py", + "left_ranges": [ + [ + 1, + 135 + ] + ], + "right_ranges": [ + [ + 1, + 136 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 + ], + "pr_numbers": [ + 43251, + 43254 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, + "shared_filenames": [ + "src/transformers/loss/loss_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/loss/loss_utils.py", + "left_ranges": [ + [ + 30, + 50 + ] + ], + "right_ranges": [ + [ + 30, + 43 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45647-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45647.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 45647, + "canonical_pr_number": 45738, + "issue_numbers": [ + 45647 + ], + "pr_numbers": [ + 45737, + 45738 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 45737, + "right_pr_number": 45738, + "code_similarity": 0.526, + "size_similarity": 0.11, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.024, + "shared_filenames": [ + "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/musicgen_melody/modeling_musicgen_melody.py", + "left_ranges": [ + [ + 491, + 497 + ] + ], + "right_ranges": [ + [ + 491, + 497 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46159-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46159.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 46159, + "canonical_pr_number": 46174, + "issue_numbers": [ + 46159 + ], + "pr_numbers": [ + 46174, + 46780 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 46174, + "right_pr_number": 46780, + "code_similarity": 0.841, + "size_similarity": 0.87, + "file_overlap": 1.0, + "area_overlap": 0.767, + "patch_similarity": 0.657, + "shared_filenames": [ + "src/transformers/models/aimv2/modeling_aimv2.py", + "src/transformers/models/d_fine/modeling_d_fine.py", + "src/transformers/models/deimv2/modeling_deimv2.py", + "src/transformers/models/pp_doclayout_v2/modeling_pp_doclayout_v2.py", + "src/transformers/models/pp_doclayout_v3/modeling_pp_doclayout_v3.py", + "src/transformers/models/rt_detr/modeling_rt_detr.py", + "src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py", + "src/transformers/models/vit_mae/modeling_vit_mae.py", + "src/transformers/models/vit_mae/modular_vit_mae.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/aimv2/modeling_aimv2.py", + "left_ranges": [ + [ + 134, + 141 + ], + [ + 144, + 154 + ], + [ + 157, + 165 + ] + ], + "right_ranges": [ + [ + 134, + 140 + ], + [ + 143, + 153 + ], + [ + 156, + 162 + ] + ] + }, + { + "filename": "src/transformers/models/d_fine/modeling_d_fine.py", + "left_ranges": [ + [ + 631, + 638 + ], + [ + 641, + 651 + ], + [ + 654, + 662 + ] + ], + "right_ranges": [ + [ + 631, + 637 + ], + [ + 640, + 650 + ], + [ + 653, + 659 + ] + ] + }, + { + "filename": "src/transformers/models/deimv2/modeling_deimv2.py", + "left_ranges": [ + [ + 703, + 710 + ], + [ + 713, + 723 + ], + [ + 726, + 734 + ] + ], + "right_ranges": [ + [ + 703, + 709 + ], + [ + 712, + 722 + ], + [ + 725, + 731 + ] + ] + }, + { + "filename": "src/transformers/models/pp_doclayout_v2/modeling_pp_doclayout_v2.py", + "left_ranges": [ + [ + 1551, + 1558 + ], + [ + 1561, + 1571 + ], + [ + 1574, + 1582 + ] + ], + "right_ranges": [ + [ + 1551, + 1557 + ], + [ + 1560, + 1570 + ], + [ + 1573, + 1579 + ] + ] + }, + { + "filename": "src/transformers/models/pp_doclayout_v3/modeling_pp_doclayout_v3.py", + "left_ranges": [ + [ + 804, + 811 + ], + [ + 814, + 824 + ], + [ + 827, + 835 + ] + ], + "right_ranges": [ + [ + 804, + 810 + ], + [ + 813, + 823 + ], + [ + 826, + 832 + ] + ] + }, + { + "filename": "src/transformers/models/rt_detr/modeling_rt_detr.py", + "left_ranges": [ + [ + 847, + 854 + ], + [ + 857, + 867 + ], + [ + 870, + 878 + ] + ], + "right_ranges": [ + [ + 847, + 853 + ], + [ + 856, + 866 + ], + [ + 869, + 875 + ] + ] + }, + { + "filename": "src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py", + "left_ranges": [ + [ + 976, + 983 + ], + [ + 986, + 996 + ], + [ + 999, + 1007 + ] + ], + "right_ranges": [ + [ + 976, + 982 + ], + [ + 985, + 995 + ], + [ + 998, + 1004 + ] + ] + }, + { + "filename": "src/transformers/models/vit_mae/modeling_vit_mae.py", + "left_ranges": [ + [ + 152, + 159 + ], + [ + 162, + 172 + ], + [ + 175, + 183 + ] + ], + "right_ranges": [ + [ + 152, + 158 + ], + [ + 161, + 171 + ], + [ + 174, + 180 + ] + ] + }, + { + "filename": "src/transformers/models/vit_mae/modular_vit_mae.py", + "left_ranges": [ + [ + 62, + 69 + ], + [ + 72, + 82 + ], + [ + 85, + 93 + ] + ], + "right_ranges": [ + [ + 62, + 68 + ], + [ + 71, + 81 + ], + [ + 84, + 90 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45561-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45561, + "canonical_pr_number": 45645, + "issue_numbers": [ + 45561 + ], + "pr_numbers": [ + 45639, + 45645 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45639, + "right_pr_number": 45645, + "code_similarity": 0.721, + "size_similarity": 0.973, + "file_overlap": 0.75, + "area_overlap": 0.563, + "patch_similarity": 0.696, + "shared_filenames": [ + "src/transformers/testing_utils.py", + "tests/utils/test_testing_utils.py", + "utils/notification_service.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/testing_utils.py", + "left_ranges": [ + [ + 3611, + 3618 + ], + [ + 3835, + 3869 + ] + ], + "right_ranges": [ + [ + 3525, + 3558 + ], + [ + 3782, + 3788 + ] + ] + }, + { + "filename": "tests/utils/test_testing_utils.py", + "left_ranges": [ + [ + 1, + 144 + ] + ], + "right_ranges": [ + [ + 1, + 114 + ] + ] + }, + { + "filename": "utils/notification_service.py", + "left_ranges": [ + [ + 935, + 967 + ] + ], + "right_ranges": [ + [ + 935, + 967 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-46305-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46305.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46305, + "canonical_pr_number": 46546, + "issue_numbers": [ + 46305 + ], + "pr_numbers": [ + 46310, + 46546 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46310, + "right_pr_number": 46546, + "code_similarity": 0.566, + "size_similarity": 0.444, + "file_overlap": 0.5, + "area_overlap": 0.867, + "patch_similarity": 0.16, + "shared_filenames": [ + "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/moonshine_streaming/modeling_moonshine_streaming.py", + "left_ranges": [ + [ + 356, + 362 + ] + ], + "right_ranges": [ + [ + 355, + 361 + ] + ] + }, + { + "filename": "src/transformers/models/moonshine_streaming/modular_moonshine_streaming.py", + "left_ranges": [ + [ + 269, + 275 + ] + ], + "right_ranges": [ + [ + 269, + 275 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-45081-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #45081.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 45081, + "canonical_pr_number": 45317, + "issue_numbers": [ + 45081 + ], + "pr_numbers": [ + 45086, + 45317 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45086, + "right_pr_number": 45317, + "code_similarity": 0.589, + "size_similarity": 0.34, + "file_overlap": 0.5, + "area_overlap": 1.0, + "patch_similarity": 0.136, + "shared_filenames": [ + "src/transformers/tokenization_utils_tokenizers.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/tokenization_utils_tokenizers.py", + "left_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ], + "right_ranges": [ + [ + 1360, + 1370 + ], + [ + 1374, + 1380 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-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-45332-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 45333, + "issue_numbers": [], + "pr_numbers": [ + 45332, + 45333 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 45332, + "right_pr_number": 45333, + "code_similarity": 0.61, + "size_similarity": 0.344, + "file_overlap": 0.389, + "area_overlap": 0.988, + "patch_similarity": 0.525, + "shared_filenames": [ + "docs/source/en/_toctree.yml", + "docs/source/en/heterogeneous_configurations.md", + "src/transformers/configuration_utils.py", + "src/transformers/integrations/heterogeneity/__init__.py", + "src/transformers/integrations/heterogeneity/configuration_utils.py", + "tests/heterogeneity/__init__.py", + "tests/heterogeneity/test_configuration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/_toctree.yml", + "left_ranges": [ + [ + 16, + 23 + ] + ], + "right_ranges": [ + [ + 16, + 23 + ] + ] + }, + { + "filename": "docs/source/en/heterogeneous_configurations.md", + "left_ranges": [ + [ + 1, + 206 + ] + ], + "right_ranges": [ + [ + 1, + 206 + ] + ] + }, + { + "filename": "src/transformers/configuration_utils.py", + "left_ranges": [ + [ + 30, + 36 + ], + [ + 123, + 129 + ], + [ + 188, + 195 + ], + [ + 302, + 311 + ], + [ + 316, + 324 + ], + [ + 443, + 456 + ], + [ + 859, + 867 + ], + [ + 965, + 973 + ], + [ + 1026, + 1034 + ], + [ + 1076, + 1084 + ] + ], + "right_ranges": [ + [ + 30, + 36 + ], + [ + 123, + 129 + ], + [ + 188, + 195 + ], + [ + 302, + 311 + ], + [ + 316, + 324 + ], + [ + 443, + 456 + ], + [ + 859, + 867 + ], + [ + 965, + 973 + ], + [ + 1026, + 1034 + ], + [ + 1076, + 1084 + ] + ] + }, + { + "filename": "src/transformers/integrations/heterogeneity/__init__.py", + "left_ranges": [ + [ + 1, + 34 + ] + ], + "right_ranges": [ + [ + 1, + 21 + ] + ] + }, + { + "filename": "src/transformers/integrations/heterogeneity/configuration_utils.py", + "left_ranges": [ + [ + 1, + 389 + ] + ], + "right_ranges": [ + [ + 1, + 389 + ] + ] + }, + { + "filename": "tests/heterogeneity/test_configuration_utils.py", + "left_ranges": [ + [ + 1, + 323 + ] + ], + "right_ranges": [ + [ + 1, + 323 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [], + "duplicate_prs": [ + { + "cluster_id": "cluster-41115-3", + "canonical_pr_number": 41116, + "duplicate_pr_numbers": [ + 45613 + ], + "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 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-45332-2", + "canonical_pr_number": 45333, + "duplicate_pr_numbers": [ + 45332 + ], + "target_issue_number": null, + "reason": "PRs in cluster-45332-2 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." + }, + { + "cluster_id": "cluster-46159-3", + "canonical_pr_number": 46174, + "duplicate_pr_numbers": [ + 46780 + ], + "target_issue_number": 46159, + "reason": "PRs in cluster-46159-3 are treated as duplicates because they converge on issue #46159 with closing_reference evidence." + }, + { + "cluster_id": "cluster-46305-3", + "canonical_pr_number": 46546, + "duplicate_pr_numbers": [ + 46310 + ], + "target_issue_number": 46305, + "reason": "PRs in cluster-46305-3 are treated as duplicates because they converge on issue #46305 with closing_reference, shared_issue_target, soft_similarity evidence." + } + ], + "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/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..670a6031d54db93624e29c7999483f43aec21c62 --- /dev/null +++ b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,24208 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260623T120018Z", + "generated_at": "2026-06-23T12:12:33Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "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:43425", + "issue:43541", + "issue:43596", + "issue:43638", + "issue:43716", + "issue:43827", + "issue:43828", + "issue:43975", + "issue:44292", + "issue:44322", + "issue:44457", + "issue:44509", + "issue:44512", + "issue:44560", + "issue:44805", + "issue:44841", + "issue:44918", + "issue:45081", + "issue:45161", + "issue:45237", + "issue:45464", + "issue:45507", + "issue:45685", + "issue:45800", + "issue:45823", + "issue:45859", + "issue:45874", + "issue:45901", + "issue:45902", + "issue:46015", + "issue:46165", + "issue:46294", + "issue:46468", + "issue:46515", + "issue:46661" + ], + "soft_pairs": [ + "issue:44322|issue:45464", + "issue:45237|issue:45685", + "issue:43716|issue:45237", + "issue:43828|issue:45237", + "issue:45237|issue:46468", + "issue:45237|issue:45507", + "issue:30277|issue:35545", + "issue:35545|issue:42915", + "issue:35545|issue:41720", + "issue:45901|issue:46015", + "issue: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:46165|issue:46468", + "issue:44560|issue:46165", + "issue:43596|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:44841|issue:46661", + "issue:24643|issue:30277", + "issue:34928|issue:44805", + "issue:45859|issue:45902", + "issue:43638|issue:45161", + "issue:34928|issue:36331", + "issue:43596|issue:45161", + "issue:43716|issue:45507", + "issue:44292|issue:45237", + "issue:45874|issue:46294", + "issue:39290|issue:45081", + "issue:45823|issue:45874", + "issue:45800|issue:46515", + "issue:43425|issue:44292", + "issue:34928|issue:41720", + "issue:43828|issue:45161", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:43716|issue:46468", + "issue:43716|issue:45685", + "issue:44292|issue:44918", + "issue:45507|issue:45685", + "issue:43716|issue:44292", + "issue:44918|issue:45161", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43828|issue:45685", + "issue:43039|issue:43828", + "issue:44918|issue:45685", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:45237|issue:46015", + "issue:43039|issue:44805" + ], + "prepared_review_unit_hash": "7cbe1529059cde0fc240a6be98d67feeed44e5d7939bb7ed8bd66c735ab64e48", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 62, + "serialized_chars": 45724, + "estimated_input_tokens": 11431, + "estimated_eval_tokens": 23118 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly noisy and cross-model/topic. I accept only one true duplicate family: the DeepSpeed ZeRO-3/Bert index-0 out-of-bounds reports. The Qwen3_5Moe rotary-embedding pair is related but distinct (CPU-offload incompatibility vs uninitialized memory after meta\u2192to_empty(cuda)), so it should not be merged.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43596 is the clearest representative of the only confirmed duplicate family here: the ZeRO-3/Bert size-0/index-0 out-of-bounds failure shared with issue:43638.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43596 is the best anchor for triage because it captures the core ZeRO-3/Bert failure mode and has the strongest true duplicate match in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different model/config bug vs streaming chat/completions failure; no shared code path." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Different backend/hardware failures: AMD GPU problem vs MPS histogram-on-Int error." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Unrelated symptoms: Mistral-3 image dtype mismatch vs GPT-OSS failing on AMD GPUs." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Both are runtime dtype/backend issues, but on different models and paths." + }, + { + "left": "issue:45237", + "right": "issue:46468", + "accept": false, + "reason": "Generic GPU failure vs broken official example; not the same bug." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "Different models and failures: AMD GPU incompatibility vs invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch is unrelated to ModernBERT ONNX export." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ModernBERT ONNX export and Qwen3Moe FP8 failure are unrelated bugs." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Export-time ONNX issue vs Qwen3 auto device-mapping CUDA assert; different failures." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Table QA crash and outdated RAG docs are unrelated." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs cleanup items, but they refer to different removed references." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ModernBERT ONNX export vs GenerationMixin import breakage are unrelated." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "TRl SFT embedding unpacking failure vs resume-training with remote code are different code paths." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask-shape IndexError and Gemma4 device-map mismatch are not the same bug." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both mention index/shape errors, but the failure modes and contexts differ too much." + }, + { + "left": "issue:43638", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3 Bert loading error vs remote-code resume issue are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "ZeRO-3/Bert size-0 IndexError is different from a mask/tensor shape mismatch." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Mask shape mismatch is unrelated to Qwen3.5 video token/feature mismatch." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Different problems: generic mask index error vs GPT-OSS TP-only failure." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Resume-training with remote code is unrelated to a broken official Nemotron example." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Video StopIteration issue and remote-code training resume issue are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "ZeRO-3/Bert loading error is unrelated to remote-code resume training." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice_preset and ModernBERT ONNX export are unrelated." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Different failures and models: Jamba collective mismatch vs Bert ZeRO-3 index error." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same underlying ZeRO-3/Bert index-0 out-of-bounds failure; the titles differ mainly by model specificity." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy call path is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "DeepSeek detokenization and LoRA merge/save mismatch are different bugs." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen-3 NVFP4 runtime error and LoRA merge/save inconsistency are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:46661", + "accept": false, + "reason": "Processor failure for Voxtral is unrelated to remote-code model_class/config_class errors." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "DeepSpeed training weight-shape error and collective mismatch are different failures." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "FSDP activation-checkpointing size issue is unrelated to mask-shape IndexError." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Related Qwen3_5Moe rotary-embedding reports, but one is CPU-offload incompatibility and the other is uninitialized memory after meta\u2192to_empty(cuda); not the same failure mode." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Bert ZeRO-3 loading error is unrelated to GPT-OSS tensor-parallel failure." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "FSDP activation-checkpointing size mismatch is unrelated to CustomTrainer compute_loss signature change." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "ZeRO-3 Bert loading error is unrelated to GPT-OSS TP-only failure." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Mistral-3 image dtype mismatch and GraniteMoEHybrid invalid method are different issues." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Qwen-3 NVFP4 runtime error is unrelated to the AMD GPU failure for GPT-OSS." + }, + { + "left": "issue:45874", + "right": "issue:46294", + "accept": false, + "reason": "Gemma4 reverse embedding lookup and Qwen3.5 video token mismatch are unrelated." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 vLLM attribute error and Mistral tokenizer regex crash are unrelated." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Both are Gemma4 issues, but one is device-map mismatch and the other is expensive reverse embedding lookup." + }, + { + "left": "issue:45800", + "right": "issue:46515", + "accept": false, + "reason": "Both mention torch 2.4.1 incompatibility, but the failure details are too broad to establish one bug." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Torch version incompatibility and Qwen-3 NVFP4 runtime error are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "FSDP activation-checkpointing size issue is unrelated to Qwen3 auto device-mapping CUDA assert." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Different model/backend paths: autocast dtype mismatch vs MPS histogram-on-Int error." + }, + { + "left": "issue:44918", + "right": "issue:45685", + "accept": false, + "reason": "TRl SFT embedding issue is unrelated to the MPS histogram error." + }, + { + "left": "issue:30277", + "right": "issue:36331", + "accept": false, + "reason": "DeepSpeed collective mismatch and trainer compute_loss signature change are unrelated." + }, + { + "left": "issue:34928", + "right": "issue:36010", + "accept": false, + "reason": "Activation-checkpointing/FSDP size mismatch is unrelated to GenerationMixin import failure." + }, + { + "left": "issue:45237", + "right": "issue:46015", + "accept": false, + "reason": "GPT-OSS GPU issue and outdated RAG examples are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger Kernel cross-entropy issue is unrelated to the mask/tensor shape IndexError." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The writeup is mostly conservative, but it overstates at least one soft-edge pair: issue:45859 vs issue:45902 is treated as a true duplicate even though the packet only shows related Qwen3_5Moe rotary-embedding wording, not a clearly same failure mode. The two titles point to different concrete problems (CPU-offload incompatibility vs uninitialized memory after meta\u2192to_empty(cuda) materialization), and the packet marks the pair as non-deterministic. Because the summary promotes that pair as one of the only strong duplicate signals, it is not fully grounded and should be softened/rejected." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 2, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:36010", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:43039", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43541", + "issue:43572", + "issue:43596", + "issue:43638", + "issue:43792", + "issue:43828", + "issue:43854", + "issue:43866", + "issue:43872", + "issue:43901", + "issue:43975", + "issue:44292", + "issue:44512", + "issue:44560", + "issue:44589", + "issue:44841", + "issue:44863", + "issue:45020", + "issue:45070", + "issue:45084", + "issue:45237", + "issue:45362", + "issue:45399", + "issue:45464", + "issue:45685", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45901", + "issue:45920", + "issue:46015", + "issue:46294", + "issue:46304", + "issue:46402", + "issue:46661", + "issue:46772" + ], + "soft_pairs": [ + "issue:43866|issue:46402", + "issue:45800|issue:46304", + "issue:43296|issue:46304", + "issue:41762|issue:43596", + "issue:39290|issue:41720", + "issue:43901|issue:44512", + "issue:44589|issue:46304", + "issue:45237|issue:45901", + "issue:43039|issue:44560", + "issue:45823|issue:46294", + "issue:43975|issue:46015", + "issue:44841|issue:45823", + "issue:43866|issue:46772", + "issue:44863|issue:46015", + "issue:45020|issue:46304", + "issue:45812|issue:45920", + "issue:43866|issue:44863", + "issue:43872|issue:46304", + "issue:44292|issue:46015", + "issue:39290|issue:43531", + "issue:41093|issue:43638", + "issue:43792|issue:45901", + "issue:43854|issue:43866", + "issue:43425|issue:45070", + "issue:43975|issue:45901", + "issue:43792|issue:45070", + "issue:43596|issue:45800", + "issue:45362|issue:45464", + "issue:44292|issue:45901", + "issue:39290|issue:45823", + "issue:45464|issue:45685", + "issue:36010|issue:39290", + "issue:43541|issue:45685", + "issue:43541|issue:43828", + "issue:44841|issue:45084", + "issue:43404|issue:43866", + "issue:41093|issue:41762", + "issue:45399|issue:46661", + "issue:43572|issue:45070", + "issue:45800|issue:45823", + "issue:43296|issue:43366" + ], + "prepared_review_unit_hash": "3c5c01df1ba5f95e30263a259877d5f1a9ad7821cb3a53fd0ef4a3f0f3441b3f", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 41, + "serialized_chars": 39201, + "estimated_input_tokens": 9801, + "estimated_eval_tokens": 19858 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are clearly unrelated issue reports. The only strong duplicate candidate is the AutoTokenizer v5 regression pair, where 45920 is the broader, better canonical issue.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 45920 is the broadest and clearest report of the tokenizer-ID regression, covering multiple affected model families and subsuming the narrower Granite-specific report.", + "canonical_pr_reason": null, + "best_issue_reason": "45920 best represents the underlying bug because it states the shared failure mode across several models, making it the most useful canonical issue for triage.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43866", + "right": "issue:46402", + "accept": false, + "reason": "Different model-loading failures in unrelated models; no shared bug path is evident." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both mention version incompatibility, but they describe different breakages and code paths." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "Unrelated failures: vLLM/PaddleOCR-VL loading vs distributed/PyTorch support." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": false, + "reason": "Both are index errors under ZeRO-3, but the affected models and likely failure sites differ." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 config attribute error and Qwen3 device-map CUDA assert are unrelated." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs complaints, but they concern different documentation regressions." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Float8 storage lookup error is unrelated to the USE_DISTRIBUTED=0 support regression." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "Different subsystems and symptoms; no common underlying bug is apparent." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Liger Kernel cross-entropy dispatch and Qwen3-VL video StopIteration are unrelated." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Device-mapping mismatch and video-token/template mismatch are different model-input bugs." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Tokenizer detokenization bug vs outdated RAG docs; not the same issue." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different problems." + }, + { + "left": "issue:43866", + "right": "issue:46772", + "accept": false, + "reason": "Corrupted checkpoint vs NVFP4 compatibility are distinct causes." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "NemotronH checkpoint loading failure is unrelated to RAG documentation issues." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Remote-code model breakage and USE_DISTRIBUTED=0 breakage are separate regressions." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Same underlying AutoTokenizer v5 regression causing wrong token IDs across multiple model families; 45920 is the broader report." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Different checkpoint/loading failures for different models." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "bitsandbytes Int8Params constructor mismatch is unrelated to distributed support." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "NVFP4 model runtime error vs documentation issue; no shared code path." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Both mention sliding-window/Qwen3, but they target different models and failures." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Both are shape/index errors, but one is a mask mismatch and the other is ZeRO-3/Bert initialization." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper runtime failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Model test loading failure and checkpoint corruption are not the same bug." + }, + { + "left": "issue:43425", + "right": "issue:45070", + "accept": false, + "reason": "Torch-version incompatibility and pydantic PretrainedConfig regression are different breakages." + }, + { + "left": "issue:43975", + "right": "issue:45901", + "accept": false, + "reason": "Tokenizer detokenization bug and table QA crash do not match." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper loading/runtime issue is unrelated to the pydantic model regression." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "Different ZeRO-3/indexing issue versus a Torch/Transformers compatibility report." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Same model family, but different API surfaces and failure modes; not enough to treat as one bug." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Chat/completions streaming failure and MPS histogram kernel error are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "ImportError for GenerationMixin and Gemma3 config attribute error are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Grouped_mm/dynamo tracing issue and MPS histogram Int error are different code paths." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Both are precision-related MoE errors, but the concrete failures and triggers differ." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure and template-node compilation error are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Weight-tying bug and checkpoint corruption are distinct issues." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Different index-error signatures with different model/loading contexts." + }, + { + "left": "issue:45399", + "right": "issue:46661", + "accept": false, + "reason": "Flash-attn fallback logic and remote-code config_class failure are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Specific StableLmConfig field regression vs broader pydantic PretrainedConfig regression." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/Transformers incompatibility and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "PaddleOCR-VL loading in vLLM and GGUF gpt-oss support are different model-support requests." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted duplicate pair, issue:45812 \u2194 issue:45920, is well supported by the shared AutoTokenizer v5 wrong-token-ID regression, and choosing 45920 as the broader canonical issue is reasonable." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 3, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:39186", + "issue:40990", + "issue:41093", + "issue:42915", + "issue:43054", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43873", + "issue:43975", + "issue:43994", + "issue:44291", + "issue:44292", + "issue:44360", + "issue:44387", + "issue:44479", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44589", + "issue:44805", + "issue:44863", + "issue:44918", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45161", + "issue:45183", + "issue:45399", + "issue:45464", + "issue:45507", + "issue:45632", + "issue:45698", + "issue:45759", + "issue:45800", + "issue:45902", + "issue:46294", + "issue:46463", + "issue:46489", + "issue:46491", + "issue:46640", + "issue:46661" + ], + "soft_pairs": [ + "issue:44387|issue:45005", + "issue:44805|issue:44918", + "issue:43541|issue:45161", + "issue:43828|issue:45464", + "issue:43366|issue:45464", + "issue:44291|issue:45070", + "issue:43531|issue:45070", + "issue:43541|issue:44589", + "issue:43716|issue:46294", + "issue:45020|issue:45800", + "issue:41093|issue:43596", + "issue:43054|issue:43994", + "issue:44479|issue:46294", + "issue:45759|issue:45902", + "issue:46640|issue:46661", + "issue:43716|issue:44918", + "issue:39186|issue:44918", + "issue:40990|issue:44960", + "issue:43645|issue:45399", + "issue:45632|issue:45698", + "issue:43404|issue:45005", + "issue:42915|issue:45399", + "issue:44530|issue:45005", + "issue:43873|issue:45005", + "issue:44360|issue:44512", + "issue:44863|issue:45005", + "issue:44492|issue:44512", + "issue:45183|issue:46463", + "issue:44918|issue:45507", + "issue:46489|issue:46491", + "issue:43975|issue:44292", + "issue:44292|issue:45507" + ], + "prepared_review_unit_hash": "b617567ab8c6db251ac478ed507ff6481abdc2412e916db043131796b066fc99", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "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": "Most pairs are not duplicates; the clear duplicate-looking groups are SigLIP2 output regression, Qwen3.5 video-input regression, remote-code cache/custom-module loading, and generic input-validation error-message bugs.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45698 is the best umbrella issue here: it describes the user-facing failure ('from_pretrained loads wrong custom module after save_pretrained'), while issue:45632 is the narrower cache-collision cause.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45698 is the strongest canonical issue candidate among the set because it is broad, actively discussed, and cleanly subsumes the more specific leaf-directory cache collision report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Both involve quantization/tied-weights themes, but they describe different concrete failures on different models/code paths." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Both are Qwen-related runtime errors, but the mask-shape mismatch and the embedding-unpacking failure are different bugs." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Different failures: torch dynamo grouped_mm tracing vs tensor-parallel support on GPT-OSS MoE." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Different model families and symptoms: autocast dtype mismatch vs streaming chat/completions failure." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "gpt-oss GGUF architecture support is unrelated to the Qwen3.5 streaming inference bug." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "One is an init_empty_weights/_is_hf_initialized constructor issue; the other is a pydantic config-field regression." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE sliding-window handling is unrelated to the pydantic PretrainedConfig field breakage." + }, + { + "left": "issue:43541", + "right": "issue:44589", + "accept": false, + "reason": "Different concrete errors and code paths: grouped_mm tracing vs missing Float8 storage object." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Both are multimodal input mismatches, but image preprocessor/model dtype mismatch is not the same bug as video-token/template mismatch." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "General remote_code breakage is too broad to merge with a torch/transformers version incompatibility report." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are indexing/runtime errors, but the mask-length mismatch and deepspeed zero3 out-of-bounds issue are different." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 underlying regression: one reports degraded text embeddings, the other nonsensical AutoModel/pipeline outputs on the same model family." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": true, + "reason": "46294 looks like a specific reproduction of the broader Qwen3.5 video-input regression described in 44479." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "Different multimodal/Qwen issues: config unwrapping from_config vs uninitialized rotary embedding memory after to_empty." + }, + { + "left": "issue:46640", + "right": "issue:46661", + "accept": false, + "reason": "Flash Attention 2 behavior and remote-code config_class lookup are unrelated failures." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Different multimodal bugs: dtype mismatch in preprocessing vs TRL SFT unpacking failure." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP 2-D weight error is unrelated to Qwen3.5 embedding unpacking." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "Perplexity quality regression and GLM5 support are not the same underlying bug." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Custom models in notebooks vs flash-attn2 fallback blocking are different issues." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": true, + "reason": "Same underlying remote-code cache/custom-module bug; 45632 is the narrower leaf-directory collision cause, 45698 is the broader symptom." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Both mention tied weights, but they concern different model families and user flows; not enough evidence of the same concrete bug." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "FineGrainedFP8Config failure and flash-attn2 fallback gating are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash on Qwen3.5 is a different code path from tied-weights regressions." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Offloading/quantization behavior is broader and different from the specific tied-weights regression reported in 45005." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "DSA indexer ReLU discussion and docs mentioning the removed run command are unrelated." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is not the same bug as tied weights in translation models." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A typo in cache strategies is unrelated to stale docs about the removed run command." + }, + { + "left": "issue:45183", + "right": "issue:46463", + "accept": true, + "reason": "46463 appears to be a specific T5-family manifestation of the broader input-validation XOR error-message bug described in 45183." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Qwen3.5 embedding unpacking under TRL and GraniteMoEHybrid invalid-method calls are different failures." + }, + { + "left": "issue:46489", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and different round-trip behavior." + }, + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "DeepSeek detokenization and Qwen NVFP4 runtime failure are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall: the canonical issue choice is well supported by the titles, and the accepted soft pairs are conservative duplicates-by-subsumption rather than overclaims. The rejected pairs also look appropriately distinct based on the packet." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 4, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:24643", + "issue:28530", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:41553", + "issue:41762", + "issue:42491", + "issue:42915", + "issue:42947", + "issue:43054", + "issue:43493", + "issue:43643", + "issue:43644", + "issue:43782", + "issue:43824", + "issue:43828", + "issue:43856", + "issue:43872", + "issue:43881", + "issue:43883", + "issue:43950", + "issue:43975", + "issue:44368", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44589", + "issue:44661", + "issue:44910", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45356", + "issue:45399", + "issue:45520", + "issue:45663", + "issue:45685", + "issue:45701", + "issue:45823", + "issue:45859", + "issue:45902", + "issue:45920", + "issue:46489" + ], + "soft_pairs": [ + "issue:43950|issue:44534", + "issue:39692|issue:43054", + "issue:43644|issue:43950", + "issue:36331|issue:41762", + "issue:43872|issue:44589", + "issue:43828|issue:44928", + "issue:45920|issue:46489", + "issue:43054|issue:45701", + "issue:38175|issue:43054", + "issue:43950|issue:45902", + "issue:41553|issue:43883", + "issue:44910|issue:45859", + "issue:43054|issue:43493", + "issue:42947|issue:44387", + "issue:45663|issue:45902", + "issue:41553|issue:43881", + "issue:44403|issue:44451", + "issue:43975|issue:45520", + "issue:44403|issue:44488", + "issue:24643|issue:28530", + "issue:42947|issue:43856", + "issue:43643|issue:45020", + "issue:41762|issue:42915", + "issue:44991|issue:45399", + "issue:43824|issue:43872", + "issue:43881|issue:43883", + "issue:43782|issue:44960", + "issue:45685|issue:45823", + "issue:45020|issue:45823", + "issue:45081|issue:45399", + "issue:44661|issue:45020", + "issue:42491|issue:44368", + "issue:43828|issue:45823", + "issue:45020|issue:45356" + ], + "prepared_review_unit_hash": "e369832d8de57252858f7e6f005983745f14cec818765bba8534a131d1ef4dfc", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 38201, + "estimated_input_tokens": 9551, + "estimated_eval_tokens": 19358 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated Transformers issues, but there is a clear duplicate set around the v5 regression that corrupts non-persistent buffers during `from_pretrained()`/materialization. Most other pairs are only loosely related by model family or subsystem and should stay separate.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43950 is the best umbrella report for the non-persistent buffer corruption regression: it names the concrete API (`from_pretrained()`), the affected behavior (`register_buffer(persistent=False)`), and the version regression. It subsumes the near-identical reports in 43644 and 44534, plus the more specific manifestation in 45902.", + "canonical_pr_reason": null, + "best_issue_reason": "43950 is the strongest representative issue because it is the most concrete and actionable description of the shared bug, while still broad enough to cover the duplicate phrasing variants around the same buffer-corruption regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying v5 bug: non-persistent buffers getting filled/corrupted with junk." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "One is a broken docs example; the other is a model-quality/embedding regression. Same model family, not the same bug." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Near-identical report of the same non-persistent buffer corruption regression." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Different failures: trainer API incompatibility vs DeepSpeed ZeRO-3 load-time index error." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Both are compatibility errors, but they involve different storage/backends and different code paths." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different models and different symptoms: dtype mismatch under autocast vs gradient explosion from 3D position_ids/SDPA fallback." + }, + { + "left": "issue:45920", + "right": "issue:46489", + "accept": false, + "reason": "Both concern tokenizer regressions, but 45920 is a broad multi-model token-ID issue while 46489 is a specific DeepSeek-Coder detokenization problem tied to a prior fix gap." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "Comparing SigLIP2 embedding quality to SigLIP is not the same as a generic tokenizer/version-change report." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both are about SigLIP2, but one reports zero probabilities and the other reports worse text embeddings; not the same concrete bug." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": true, + "reason": "Specific manifestation of the same non-persistent buffer/materialization corruption bug." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Bad AutoTokenizer error message for Voxtral vs missing attribute on MolmoForCausalLM are unrelated failures." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Flash-Attention illegal memory access from 3D position_ids is unrelated to CPU offload incompatibility in Qwen3_5MoeTextRotaryEmbedding." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "Both mention SigLIP2, but one is an output-quality complaint and the other is a HF-vs-JAX implementation discrepancy; too broad to treat as the same bug." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "Different issues: ineffective gradient checkpointing with LoRA vs increased reserved memory/OOM under int4 quantization." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "Gemma4 FSDP2 KeyError is a different code-path problem from non-persistent buffer materialization corruption." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Different model-loading failures with no shared concrete code path." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "Generic loading noise vs a specific model load failure; not the same bug." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Tokenizer detokenization regression vs missing `flash_attn` import on Python 3.13 are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Generic loading noise vs a specific tokenizer/model load failure for `cjvt/sleng-bert`." + }, + { + "left": "issue:24643", + "right": "issue:28530", + "accept": false, + "reason": "DeepSpeed training shape error and early-stopping metric handling are unrelated." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "PEFT LoRA checkpointing behavior is not the same issue as Qwen3 MoE memory usage." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "Both involve `remote_code`, but one is missing fields in `AutoConfig` while the other is a broader model-loading regression; not clearly the same bug." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different models and different failure modes: Gemma3 DeepSpeed load error vs Qwen3 MoE LoRA incompatibility across branches." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading failure for a specific model vs flash-attn2 fallback gating; unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Importing a missing model class and a bitsandbytes constructor incompatibility are different problems." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "Different load-time failures on different models and code paths." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Qwen3VL weight-only load error and a GLM5 issue are not the same bug." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram dtype limitation and Gemma4 device mismatch are unrelated." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Broad `remote_code` breakage is not the same as a Gemma4 PLE device-placement bug." + }, + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer regex crash and flash-attn2 fallback checks are separate problems." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "`TOKENIZER_MAPPING_NAMES` add-new-model-like failure is unrelated to the broader `remote_code` regression report." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Flash-attn crash from packed-sequence misinterpretation is unrelated to CPU offload incompatibility." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The umbrella issue choice is supported by the packet, and the accepted pairs are conservative duplicates of the same non-persistent-buffer/materialization regression. The remaining rejections stay safely on the cautious side." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 5, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36331", + "issue:41093", + "issue:41553", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43596", + "issue:43638", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43872", + "issue:43940", + "issue:43950", + "issue:44220", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44560", + "issue:44805", + "issue:44863", + "issue:44928", + "issue:44991", + "issue:45020", + "issue:45305", + "issue:45313", + "issue:45325", + "issue:45488", + "issue:45588", + "issue:45663", + "issue:45715", + "issue:45759", + "issue:45823", + "issue:45902", + "issue:45923", + "issue:46315", + "issue:46402", + "issue:46489", + "issue:46515", + "issue:46661" + ], + "soft_pairs": [ + "issue:45715|issue:46315", + "issue:45488|issue:46489", + "issue:43827|issue:44509", + "issue:45759|issue:46661", + "issue:43404|issue:43950", + "issue:43872|issue:44291", + "issue:42915|issue:43872", + "issue:45020|issue:46661", + "issue:43792|issue:44220", + "issue:41553|issue:44991", + "issue:43257|issue:45305", + "issue:36331|issue:41093", + "issue:42947|issue:45923", + "issue:43940|issue:45313", + "issue:43854|issue:44863", + "issue:41553|issue:44488", + "issue:44928|issue:45588", + "issue:41093|issue:44805", + "issue:44479|issue:44560", + "issue:24643|issue:30064", + "issue:44530|issue:44863", + "issue:44863|issue:46402", + "issue:43366|issue:43828", + "issue:43596|issue:46515", + "issue:43366|issue:43531", + "issue:45325|issue:45902", + "issue:43638|issue:46515", + "issue:45663|issue:45823" + ], + "prepared_review_unit_hash": "bb2445bbd3eabef2ca35d498442d15b50d4a71fbf8f9a7fe83310bf6a64a19fb", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "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 mostly a mix of unrelated Transformers bug reports, with a few clear duplicate pairs around docs, `_is_hf_initialized` loading regressions, tokenizer regressions, and a video-input regression.", + "confidence": 0.61, + "canonical_issue_reason": "Issue 45020 is the broadest umbrella here: a general recent-version/`remote_code` compatibility regression that best represents the cluster\u2019s recurring loader breakages.", + "canonical_pr_reason": null, + "best_issue_reason": "45020 is the strongest anchor issue because it is the most general and can subsume several of the more specific model-loading failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer performance complaints, but they hit different APIs (`convert_ids_to_tokens` vs `convert_tokens_to_ids`) and don\u2019t look like the same underlying bug." + }, + { + "left": "issue:45488", + "right": "issue:46489", + "accept": true, + "reason": "Both point to the same DeepSeek/Llama tokenizer regression in v5, where tokenizer handling changes produce incorrect tokenization/output for DeepSeek-family models." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "These are the same docs regression: v5 pipeline-task removals are still referenced in summarization/translation docs." + }, + { + "left": "issue:45759", + "right": "issue:46661", + "accept": false, + "reason": "Both involve model construction, but one is about unwrapping `text_config` for composite Qwen configs while the other is a remote-code `config_class` attribute error; different bugs." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "One is a tied-weight bug in a specific multimodal model, the other is a broad `from_pretrained` non-persistent-buffer regression; no shared code-path evidence." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Both center on the same `_is_hf_initialized` argument regression during model initialization/loading, causing TypeErrors in related loading paths." + }, + { + "left": "issue:42915", + "right": "issue:43872", + "accept": false, + "reason": "These are different failures: FineGrainedFP8Config on Qwen3Moe versus an `_is_hf_initialized` bitsandbytes/init bug." + }, + { + "left": "issue:45020", + "right": "issue:46661", + "accept": false, + "reason": "`45020` is a broad umbrella for remote_code breakage, but `46661` is a specific `config_class` attribute error; not enough evidence they are the same concrete bug." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "The Whisper model failure and the `_torch_extract_fbank_features()` issue are related only by audio processing; the titles don\u2019t show the same exact bug." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer-load problems on different models; one is a bad error message, the other is a failed tokenizer load regression." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "Both mention DeepSpeed and Qwen, but one is about weight conversion/loading and the other about gradient averaging; distinct code paths." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "Unexpected `compute_loss` kwarg and a mask/tensor shape mismatch are unrelated runtime failures." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing not taking effect with PEFT LoRA is a different bug from a missing `supports_gradient_checkpointing` flag on a remote-code model." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are ZeRO-3 weight-loading failures in Qwen models, but they affect different model families with different missing-weight symptoms." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-loading failures for different architectures; no shared concrete bug is evident." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "A tokenizer error message issue for Voxtral is unrelated to loading `cjvt/sleng-bert`." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Catastrophic NaN/SDPA-fallback behavior in RLHF is unrelated to a flash-attention crash on sink-less models." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": true, + "reason": "These are the same generic mask-vs-indexed-tensor shape mismatch error pattern, likely the same off-by-one or alignment bug." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": true, + "reason": "Both describe the same v5.3.0 Qwen video-input regression family; `44560` looks like a specific manifestation of the broader video-input breakage in `44479`." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "Training with DeepSpeed and void-segmentation-map processing are unrelated issues." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "PagedAttentionCache/linear_attention and NemotronH checkpoint loading are unrelated." + }, + { + "left": "issue:44863", + "right": "issue:46402", + "accept": false, + "reason": "Different models and different loading failures; no same bug signal." + }, + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "GGUF gpt-oss support and a Phi autocast dtype mismatch are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:46515", + "accept": false, + "reason": "Zero-3 BertModel index error and torch 2.4.1 incompatibility are not the same concrete issue." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "GGUF gpt-oss support is unrelated to a Qwen3-MoE sliding_window bug." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "These are different positional-encoding bugs in different Qwen variants; same general area, but not the same issue." + }, + { + "left": "issue:43638", + "right": "issue:46515", + "accept": false, + "reason": "A Zero-3/non-pretrained BertModel index error is unrelated to a torch-version incompatibility." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Gemma4 FSDP2 shared_kv_states issues and device_map mismatches are different forward-pass bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is broadly grounded in the packet, and the conservative false rejects look fine. The accepted soft pairs are mostly defensible from the titles alone: the docs issues, the `_is_hf_initialized` loading regressions, the mask-shape errors, and the Qwen video-input regression family all look related. The tokenizer pair is a little broader, but still stays within a plausible shared regression family rather than asserting an exact duplicate." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 6, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43720", + "issue:43824", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:44960", + "issue:44977", + "issue:45030", + "issue:45276", + "issue:45584", + "issue:45685", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46459", + "issue:46489", + "issue:46512", + "issue:46515" + ], + "soft_pairs": [ + "issue:43366|issue:44589", + "issue:44155|issue:45584", + "issue:43716|issue:44560", + "issue:43824|issue:43957", + "issue:43950|issue:45834", + "issue:42915|issue:43957", + "issue:44560|issue:45685", + "issue:43645|issue:43646", + "issue:43550|issue:44910", + "issue:44683|issue:46515", + "issue:44560|issue:44918", + "issue:42617|issue:43366", + "issue:36010|issue:42915", + "issue:45812|issue:46489", + "issue:45823|issue:46515", + "issue:44661|issue:45823", + "issue:44560|issue:44805", + "issue:38175|issue:43994", + "issue:43632|issue:45834", + "issue:44877|issue:45030", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:43317|issue:44756", + "issue:44661|issue:44805", + "issue:42491|issue:43720", + "issue:43644|issue:44534", + "issue:38175|issue:39692", + "issue:41720|issue:42915", + "issue:42915|issue:44756", + "issue:42491|issue:45276", + "issue:44451|issue:44488", + "issue:44960|issue:44977", + "issue:46459|issue:46512" + ], + "prepared_review_unit_hash": "ba5cdb299e2c788a40fce4dcb32acaedeca8855c0433a0d9bef468c65d8032f5", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37477, + "estimated_input_tokens": 9370, + "estimated_eval_tokens": 18996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous issue set with a few clear duplicate pairs. The strongest duplicates are the custom-model initialization regression, the SigLIP2 bad-output reports, and the non-persistent-buffer corruption reports; most other soft pairs are different bugs.", + "confidence": 0.74, + "canonical_issue_reason": "issue:43644 is the best representative issue: it describes a broad Transformers 5.0 regression with clear duplicate coverage from 44534, and it\u2019s more general than the more model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest global issue choice because it captures a concrete, broadly applicable v5 regression and has an obvious duplicate, making it a good cluster anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "Different problems: GGUF architecture support vs a Float8 storage lookup error." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "AudioFlamingo3 batching leak and Whisper empty-transcription failure are unrelated code paths." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Model/processor dtype mismatch is unrelated to a Qwen3-vl video StopIteration error." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Missing import for a specific class and meta-device loading failures are different issues." + }, + { + "left": "issue:43950", + "right": "issue:45834", + "accept": false, + "reason": "Non-persistent-buffer corruption is unrelated to Kosmos2.5 long-OCR index errors." + }, + { + "left": "issue:42915", + "right": "issue:43957", + "accept": false, + "reason": "Qwen3Moe FP8 failure and meta-device loading breakage are different bugs." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Video embedding StopIteration and an MPS Int histogram error are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "Same underlying custom-model initialization regression in Transformers 5; one report is just notebook-specific." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different model and different failure mode: torch.compile/SDPA vs FlashAttention misreading 3D position_ids." + }, + { + "left": "issue:44683", + "right": "issue:46515", + "accept": false, + "reason": "Both mention torch versions, but they describe different incompatibilities and different failure paths." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Qwen3-vl video inference StopIteration and TRL SFT embedding unpacking are not the same bug." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to GGUF gpt-oss architecture support." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Importing GenerationMixin and Qwen3Moe FP8 loading are unrelated." + }, + { + "left": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but for different model families and likely different fixes." + }, + { + "left": "issue:45823", + "right": "issue:46515", + "accept": false, + "reason": "Gemma4 device-mapping forward mismatch is unrelated to a general torch 2.4.1 incompatibility report." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping constraint and Gemma4 PLE device mismatch are different problems." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different failures: video StopIteration vs tensor/mask shape mismatch." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 base model regression with incorrect/zero outputs; 43994 is the broader symptom report." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "_is_hf_initialized flag breakage and Kosmos2.5 OCR indexing are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both involve config validation, but the affected models and concrete failures differ too much to be the same bug." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Deepspeed zero3 Bert indexing and add-new-model-like tokenizer mapping failure are different code paths." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Zero3 Bert init IndexError is unrelated to the tokenizer-mapping issue." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "GPU+CPU offload loading failure and Strix Halo mmap OOM are different issues." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Qwen3_moe LoRA compatibility across versions and BitNet packed-weight loading are unrelated." + }, + { + "left": "issue:43644", + "right": "issue:44534", + "accept": true, + "reason": "Same non-persistent-buffer regression in Transformers 5; the titles describe the same underlying bug." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": false, + "reason": "Both touch SigLIP2, but one is bad model output while the other is a docs example with multiple distinct errors." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Qwen3 device-map cudaErrorAssert and Qwen3Moe FP8 failure are different problems." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3Moe FP8 failure is unrelated to mmap-related OOM avoidance." + }, + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Different models and different symptoms: LoRA incompatibility vs resize_token_embeddings not propagating." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-load failures, but the models and the likely root cause are not specific enough to confirm a duplicate." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "GLM5 and Qwen3.5 flash-attention failures are separate model-specific issues." + }, + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "One is dtype being ignored during checkpoint loading; the other is config.dtype staying stale after from_config." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted duplicate pairs are well supported by highly similar titles and descriptions, and the rejected pairs avoid obvious overreach. The canonical issue choice of 43644 is also defensible as the broad v5 non-persistent-buffer regression with a clear duplicate in 44534." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 7, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42947", + "issue:43257", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44079", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44402", + "issue:44488", + "issue:44756", + "issue:44910", + "issue:44960", + "issue:44991", + "issue:45276", + "issue:45292", + "issue:45313", + "issue:45335", + "issue:45357", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45507", + "issue:45588", + "issue:45663", + "issue:45704", + "issue:45759", + "issue:45812", + "issue:46050", + "issue:46468", + "issue:46829" + ], + "soft_pairs": [ + "issue:45507|issue:46468", + "issue:45313|issue:45357", + "issue:44488|issue:44991", + "issue:43644|issue:43645", + "issue:43856|issue:43873", + "issue:43856|issue:44387", + "issue:39692|issue:43994", + "issue:42491|issue:45292", + "issue:43257|issue:43854", + "issue:43854|issue:43931", + "issue:43708|issue:44075", + "issue:42491|issue:43931", + "issue:44960|issue:46050", + "issue:45488|issue:45812", + "issue:44164|issue:44291", + "issue:45362|issue:45406", + "issue:45276|issue:45335", + "issue:44910|issue:45588", + "issue:44402|issue:44488", + "issue:43425|issue:43957", + "issue:43873|issue:44387", + "issue:44079|issue:46829", + "issue:41720|issue:44155", + "issue:39692|issue:44960", + "issue:45663|issue:45759", + "issue:42947|issue:45704", + "issue:43404|issue:43454", + "issue:44291|issue:44756", + "issue:36010|issue:39404", + "issue:44387|issue:44756" + ], + "prepared_review_unit_hash": "c0ed4cd6e9bd0f3d09618d975515d8c215818e3c8fda5aa411d2124250be2e1a", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36817, + "estimated_input_tokens": 9205, + "estimated_eval_tokens": 18666 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous set of Transformers issues across unrelated models/subsystems. I found one likely duplicate pair around `resize_token_embeddings`; the rest look like distinct bugs and should not be merged.", + "confidence": 0.83, + "canonical_issue_reason": "Issue 45276 is the broader of the only clear duplicate pair: it covers `embed_tokens_per_layer` and `output_embeddings`, while 45335 is the narrower T5Gemma variant.", + "canonical_pr_reason": null, + "best_issue_reason": "45276 is the best representative issue for this cluster because it is the broadest, most general-looking ticket among the few genuinely overlapping reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45507", + "right": "issue:46468", + "accept": false, + "reason": "Different models and failure modes; one is a model-method bug, the other a broken example." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Both involve Qwen3.5, but one is ZeRO-3 loading and the other is save_pretrained key serialization; different bugs." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading complaints, but for different models and likely different root causes." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5 regressions, but one is non-persistent buffer initialization and the other is Jupyter custom model initialization." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Different problems: Qwen3 MoE training memory usage vs quantization offloading behavior." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Both mention memory/OOM, but one is MoE training and the other is int4 quantization reserved-memory growth." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Related SigLIP2 area, but one is a docs example with multiple errors and the other is incorrect outputs from AutoModel/pipeline." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "Different models and different symptoms; one is LoRA/hf5 compatibility, the other is resize_token_embeddings output handling." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Different Qwen3 MoE loading paths: accelerate+deepspeed conversion vs unit-test model loading failure." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Both are loading errors, but for different models and different mismatch conditions." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint-resume math is unrelated to SGD optimizer arguments." + }, + { + "left": "issue:42491", + "right": "issue:43931", + "accept": false, + "reason": "No shared bug: LoRA checkpoint compatibility vs Qwen3-VL shape mismatch on load." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "Different generations of the same family, but one is a generic GLM5 issue and the other is a quantized save/load regression." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer-related regressions, but one is a LlamaTokenizer pre-tokenizer override and the other is Granite AutoTokenizer ID mapping." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different save/load paths: extra_state handling vs init_empty_weights unexpected argument." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Both are Qwen3.5/Gemma4 adjacent, but one is chat crashes and the other is processor attribute access in serve." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": true, + "reason": "Same API regression class: `resize_token_embeddings` fails to propagate to model-specific embedding/output weights; these look like the same underlying bug across Gemma4 and T5Gemma." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both are flash-attention crashes, but the root causes differ: Qwen3.5 packed sequence handling vs sink-less `s_aux=None` handling." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "One is tokenizer vocab-size mismatch, the other is a separate model-loading failure." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Torch version incompatibility is unrelated to meta-device loading regressions." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Quantization offloading behavior and int4 reserved-memory OOM are related broadly, but not the same concrete bug." + }, + { + "left": "issue:44079", + "right": "issue:46829", + "accept": false, + "reason": "ModelOutput key assignment bug is unrelated to VideoPrism output-field wiring." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different issues: Qwen3 auto device mapping crash vs AudioFlamingo3 batched inference leakage." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "SigLIP2 docs/model mismatch is unrelated to a GLM5 issue." + }, + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different model families and different failures: FSDP2 attention state vs AutoModelForCausalLM config unwrapping." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Different bugs: PEFT gradient checkpointing vs T5 using apex.FusedRMSNorm memory leak." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Very similar symptom, but different model classes; not enough evidence they share one concrete code path." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "init_empty_weights argument handling is unrelated to mmap OOM on Strix Halo." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to Whisper pipeline `return_language` behavior." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Different resource issues: int4 quantization reserved memory vs mmap-induced OOM on Strix Halo." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted 45276/45335 pair is plausibly the same resize_token_embeddings propagation bug across closely related model implementations, and the other soft-edge rejections do not overclaim shared root cause. No mergeability issue 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": 24, + "review_unit_index": 8, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:39404", + "issue:42796", + "issue:42947", + "issue:43335", + "issue:43381", + "issue:43425", + "issue:43611", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43716", + "issue:43828", + "issue:43854", + "issue:43957", + "issue:44155", + "issue:44164", + "issue:44292", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44530", + "issue:44589", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:45084", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45685", + "issue:45698", + "issue:45701", + "issue:45812", + "issue:45823", + "issue:45859", + "issue:46315", + "issue:46326", + "issue:46468", + "issue:46491", + "issue:46665" + ], + "soft_pairs": [ + "issue:44292|issue:44912", + "issue:45663|issue:45859", + "issue:42947|issue:43381", + "issue:43957|issue:44756", + "issue:44155|issue:45823", + "issue:43335|issue:45663", + "issue:44863|issue:46468", + "issue:45276|issue:45292", + "issue:45685|issue:46468", + "issue:44479|issue:44877", + "issue:43381|issue:44928", + "issue:44164|issue:44756", + "issue:45684|issue:45698", + "issue:45588|issue:45663", + "issue:43854|issue:45216", + "issue:42796|issue:46326", + "issue:45292|issue:45335", + "issue:43828|issue:43957", + "issue:43646|issue:43708", + "issue:44402|issue:44451", + "issue:43645|issue:44164", + "issue:44589|issue:44756", + "issue:43425|issue:43828", + "issue:45812|issue:46315", + "issue:46315|issue:46491", + "issue:39404|issue:45701", + "issue:43611|issue:43645", + "issue:44530|issue:45084", + "issue:39404|issue:46665", + "issue:44683|issue:45823", + "issue:43425|issue:45823", + "issue:43425|issue:43716", + "issue:44589|issue:45084", + "issue:44479|issue:44805" + ], + "prepared_review_unit_hash": "76afb22914413a2272209d49bc88157d153d4d8081d2880c74721d140eed64cf", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37872, + "estimated_input_tokens": 9468, + "estimated_eval_tokens": 19192 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of unrelated regressions across tokenizers, custom-model loading, quantization, device placement, and training behavior. The only clearly duplicate-looking edges are the resize_token_embeddings issues, which appear to be the same propagation bug in different model variants.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45698 is the broadest and most central custom-model loading regression in the set, with the most discussion and inbound references, and it overlaps with several nearby save/from_pretrained issues.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45698 is the best anchor for triage because it describes a concrete, user-facing from_pretrained/save_pretrained failure and has the strongest evidence of ongoing impact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Same general quantization/loading area, but the models and failure modes differ; not the same underlying bug." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Different models and different code paths: FSDP attention state rebuild vs CPU offload rotary embedding compatibility." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is about ineffective checkpointing with LoRA and the other about eval-mode restrictions." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device loading and mmap OOM are separate issues with different triggers and fixes." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "AudioFlamingo3 batched inference leak and Gemma4 device mismatch are unrelated failures." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "Sparse-layer construction bug and Gemma4 FSDP attention KeyError are not the same bug." + }, + { + "left": "issue:44863", + "right": "issue:46468", + "accept": false, + "reason": "Both mention Nemotron, but one is checkpoint loading and the other is a broken example; no shared concrete failure." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Both describe the same resize_token_embeddings propagation bug affecting output embeddings; one is a Gemma4-specific manifestation." + }, + { + "left": "issue:45685", + "right": "issue:46468", + "accept": false, + "reason": "MPS histogram/int error and a broken Nemotron example are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Video-input regression for Qwen VL models and strict granite_speech config loading are different issues." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction is unrelated to the RLHF NaN explosion from 3D position_ids/SDPA fallback." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "extra_state save/load handling and Strix Halo mmap OOM are separate problems." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Both involve custom model save/load, but one is about read-only file permissions and the other is about loading the wrong module." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash attention sink-less crash and Gemma4 FSDP KeyError are distinct attention-path bugs." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "GLM-4.7-Flash loading failure and Qwen3.5 save_pretrained regression are not the same underlying issue." + }, + { + "left": "issue:42796", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer BPE-marker handling and continue_final_message behavior are different tokenizer bugs." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Both are the same resize_token_embeddings bug: the embedding/output weights are not updated consistently across model variants." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE and meta-device loading failures are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization in notebooks and Trainer resume max_steps miscalculation are separate issues." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab-size mismatch for one model and inability to load another model are too weakly connected to treat as the same bug." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Notebook custom model initialization and extra_state save/load handling do not look like one bug." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Missing Float8 storage and mmap OOM are unrelated runtime/environment failures." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch 2.10 incompatibility is broader and not the same as the specific autocast dtype mismatch." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "One is correctness (wrong token IDs), the other is performance (slow convert_tokens_to_ids); not the same bug." + }, + { + "left": "issue:46315", + "right": "issue:46491", + "accept": false, + "reason": "Tokenizer performance regression and CodeLlama leading-space round-trip regression are different issues." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language regression and tokenization changes are not the same underlying bug." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both are custom-model related, but base_model_prefix loading and notebook initialization failures are different failure modes." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and non-template-node compilation error are unrelated." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "Whisper pipeline behavior and from_pretrained inside Python classes are not the same issue." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Compiled flex_attention torch-version failure and Gemma4 device mismatch are separate code paths." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as a Gemma4 device-map forward mismatch." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "General torch-version incompatibility and a model/preprocessor dtype mismatch are unrelated." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup failure and compilation of non-template nodes are different problems." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Qwen video regression and mask/tensor shape IndexError do not appear to be the same bug." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The anchor choice for issue:45698 is reasonable given its discussion activity and direct save/from_pretrained scope, and the only accepted soft-edge pairs are the resize_token_embeddings cases, which do look plausibly like the same underlying bug across model variants. The rejected pairs are generally justified as distinct failure modes." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 9, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:32090", + "issue:36032", + "issue:39404", + "issue:42796", + "issue:42915", + "issue:43299", + "issue:43404", + "issue:43425", + "issue:43452", + "issue:43541", + "issue:43596", + "issue:43611", + "issue:43643", + "issue:43644", + "issue:43716", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44451", + "issue:44462", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44661", + "issue:44671", + "issue:44683", + "issue:44737", + "issue:44756", + "issue:44898", + "issue:44912", + "issue:44918", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45759", + "issue:45859", + "issue:46050", + "issue:46285", + "issue:46326", + "issue:46402", + "issue:46489", + "issue:46821" + ], + "soft_pairs": [ + "issue:43906|issue:44530", + "issue:43901|issue:44509", + "issue:43994|issue:44079", + "issue:44671|issue:44991", + "issue:43425|issue:44918", + "issue:44265|issue:44898", + "issue:36032|issue:43452", + "issue:39404|issue:44912", + "issue:44164|issue:45759", + "issue:46050|issue:46326", + "issue:43906|issue:46326", + "issue:43299|issue:43931", + "issue:44991|issue:45081", + "issue:43856|issue:44756", + "issue:43596|issue:44683", + "issue:44164|issue:46821", + "issue:44488|issue:45081", + "issue:44488|issue:44987", + "issue:43611|issue:43644", + "issue:44462|issue:46489", + "issue:42796|issue:43404", + "issue:43716|issue:45084", + "issue:36032|issue:42915", + "issue:46285|issue:46402", + "issue:32090|issue:39404", + "issue:44737|issue:45859", + "issue:44661|issue:44683", + "issue:43643|issue:44530", + "issue:43541|issue:45084", + "issue:42915|issue:44530", + "issue:44451|issue:44987" + ], + "prepared_review_unit_hash": "8494029e0a0d359a7e5c3ff22d812a517dc2a7afa589fb4fce1f1c3fa3a3aac7", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37071, + "estimated_input_tokens": 9268, + "estimated_eval_tokens": 18792 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The items are heterogeneous: tokenizer/load regressions, model-shape/config bugs, compilation/tracing failures, docs issues, and resource/OOM reports. None of the soft pairs look like the same underlying bug or a PR-worthy merge target, so I reject all soft edges.", + "confidence": 0.94, + "canonical_issue_reason": "No single issue cleanly represents this set; the entries span several unrelated regressions. If forced to pick a broad loading-regression umbrella, 43611 is the closest, but it does not subsume most of the others.", + "canonical_pr_reason": null, + "best_issue_reason": "There isn\u2019t a strong global issue representative here; 43611 is the broadest loading-regression candidate, but it is still not a safe canonical for the whole collection.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "Different problems: a reproduction of a prior bug vs a Qwen3.5 paged-attention cache crash." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but they cover different pipeline docs regressions and different removed/changed tasks." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Different bug classes: bad model/pipeline outputs vs ModelOutput key assignment when a key was previously None." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "Different models and root causes: CamemBERT masked-LM regression vs tokenizer loading failure for EMBEDDIA/est-roberta." + }, + { + "left": "issue:43425", + "right": "issue:44918", + "accept": false, + "reason": "Torch version incompatibility is not the same as a TRL trainer embedding-unpack failure." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "Different code paths: torch.export/export with torch_compilable_check vs Perceiver interpolation image-classification failure." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer method conflict is unrelated to gguf_file handling in AutoTokenizer/AutoModel loading." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback behavior." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "save/from_pretrained extra_state handling is a different issue from from_config text_config unwrapping." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized model save/load breakage is unrelated to mistral_common continue_final_message behavior." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "A reproduction issue and a mistral_common tokenizer bug are not the same underlying defect." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both involve Qwen3-VL MoE loading, but one is a version regression and the other is a weight-shape mismatch; root causes are not clearly the same." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer failures on different model families; no shared concrete bug path." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3 MoE training memory inefficiency is not the same as disabling mmap on Strix Halo to avoid OOM." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "Zero3/BertModel indexing failure is unrelated to compiled flex_attention on newer torch." + }, + { + "left": "issue:44164", + "right": "issue:46821", + "accept": false, + "reason": "extra_state save/load handling does not match a TimesFM window_size AttributeError." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Loading sleng-bert vs mistral regex patching are different tokenizer bugs." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Different model loading failures with different symptoms and likely different fixes." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "base_model_prefix loading breakage is not the same as non-persistent buffers being filled with junk." + }, + { + "left": "issue:44462", + "right": "issue:46489", + "accept": false, + "reason": "Ignoring tokenizer.json from repo is distinct from DeepSeek-Coder v1 producing incorrect output after a prior fix gap." + }, + { + "left": "issue:42796", + "right": "issue:43404", + "accept": false, + "reason": "Ministral BPE-marker handling and Mistral3 lm_head weight tying are different model/tokenizer bugs." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Image preprocessor/model dtype mismatch is unrelated to a template-node compilation error." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer method conflict and Qwen3Moe FineGrainedFP8Config failure are unrelated." + }, + { + "left": "issue:46285", + "right": "issue:46402", + "accept": false, + "reason": "CLIP structure inconsistency is not the same as Sam3Model text encoder weights missing on load." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error is unrelated to Whisper pipeline return_language behavior." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "Both mention device/offload concerns, but they are different code paths and different failures." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Model-addition mapping failure is unrelated to flex_attention compilation on torch>=2.9." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "trust_remote_code config field loss is not the same as a PagedAttentionCache linear_attention crash." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Torch dynamo tracing of grouped_mm in Mixtral is different from a non-template-node compile error." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3Moe FP8Config failure and Qwen3.5 linear_attention cache crash are different bugs." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Different model-loading failures for different repositories and no shared concrete 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": 24, + "review_unit_index": 10, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28530", + "issue:34567", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:42175", + "issue:42915", + "issue:43262", + "issue:43317", + "issue:43377", + "issue:43404", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43582", + "issue:43632", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43742", + "issue:43746", + "issue:43819", + "issue:43883", + "issue:43994", + "issue:44062", + "issue:44206", + "issue:44220", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44488", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44811", + "issue:44849", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45325", + "issue:45507", + "issue:45584", + "issue:45987", + "issue:46178", + "issue:46326" + ], + "soft_pairs": [ + "issue:40990|issue:43421", + "issue:43582|issue:43632", + "issue:43644|issue:43646", + "issue:34567|issue:40444", + "issue:44811|issue:45584", + "issue:43883|issue:44488", + "issue:42175|issue:46178", + "issue:44206|issue:44220", + "issue:40444|issue:46326", + "issue:36032|issue:43645", + "issue:43452|issue:43482", + "issue:43317|issue:43482", + "issue:45084|issue:45507", + "issue:44987|issue:44991", + "issue:43262|issue:43377", + "issue:43746|issue:46326", + "issue:43883|issue:44991", + "issue:42915|issue:45084", + "issue:44403|issue:45081", + "issue:44464|issue:45987", + "issue:43646|issue:44534", + "issue:43645|issue:44534", + "issue:44849|issue:45325", + "issue:43377|issue:43819", + "issue:43404|issue:44466", + "issue:43994|issue:44610", + "issue:44987|issue:45081", + "issue:44451|issue:45081", + "issue:42915|issue:44291", + "issue:28530|issue:34567", + "issue:36032|issue:43742", + "issue:42915|issue:44589", + "issue:44062|issue:44987" + ], + "prepared_review_unit_hash": "561f43a490aa8fe7a3f3e78ddee583e7808deb40e6e0dc2253af76f08263e0e8", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37705, + "estimated_input_tokens": 9427, + "estimated_eval_tokens": 19110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous. It contains several unrelated transformers v5 regressions, tokenizer/model-loading failures, and audio/vision edge cases, but the soft pairs do not look like true duplicates of the same underlying bug.", + "confidence": 0.34, + "canonical_issue_reason": "issue:43632 is the broadest and most central v5 regression here, centered on broken model initialization/state handling, so it is the best umbrella representative among the issues.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43632 is the clearest representative of the cluster\u2019s main theme: transformers v5 initialization/state regressions, rather than a narrow model- or tokenizer-specific failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "Unrelated symptoms: perplexity on gpt-oss vs runtime special-token/post-processor mutation." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "Different failures: Apple Silicon allocator warmup TypeError vs `_is_hf_initialized` flag regression." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 init-related, but one is junk non-persistent buffers and the other is custom model notebook initialization; not the same bug." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "Trainer token accounting bug vs Qwen2.5-VL iterable dataset with multiple images; different subsystems." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Both touch Whisper/tokenization, but one ignores `skip_special_tokens` and the other fails on empty transcription after `align_special_tokens`." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Different model-loading problems: missing `all_tied_weights_keys` vs failing to load a specific BERT model." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Packaging/backends mismatch for `.[torch]` vs an unsupported backend mapping entry for `tensorflow_text`." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Related audio area, but different code paths and errors: unsupported `center` arg vs `_torch_extract_fbank_features()` issue." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Different multimodal/tokenizer bugs: iterable dataset multi-image prompt handling vs `continue_final_message` on mistral_common tokenizers." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "Tokenizer method conflict at load time vs custom model initialization in Jupyter; not the same underlying issue." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "Both involve GGUF loading, but one is a general `gguf_file` API break and the other is a specific Qwen2.5-GGUF v5 load failure." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Dequantized device-map offload failure is a different path from Qwen2.5-GGUF loading failure." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Completely different errors: template compilation TypeError vs invalid method call in GraniteMoEHybrid." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Both are load regressions in v5, but for different models and likely different tokenizer/model code paths." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio processor sampling-rate defaulting and MIMI batch padding-mask behavior are different bugs." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "PEFT/local-checkpoint loading for GraniteSpeech is unrelated to mistral_common `continue_final_message` handling." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Different failures: tied-weights attribute missing vs tokenizer loading for EMBEDDIA/est-roberta." + }, + { + "left": "issue:42915", + "right": "issue:45084", + "accept": false, + "reason": "FineGrainedFP8Config failure and non-template node compilation error are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Generic load noise vs a specific Mistral tokenizer regex patch crash; not the same bug." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Different generation/cache issues: chunked compiled-forward inconsistency vs `get_seq_length()` returning the wrong type." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Custom model init failure and non-persistent buffer junk are different regressions, even if both are v5-related." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Notebook custom model initialization and non-persistent buffer serialization are distinct issues." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Qwen3.5 hidden-state output bug vs Qwen2.5-VL rope/position-id scaling bug; different models and paths." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "Batched-vs-single padding-mask mismatch in MIMI is unrelated to DAC from_latents / missing STE behavior." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "One is missing weight tying in a model class; the other is inconsistent serialization of tied weights. Related theme, but not the same concrete bug." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Siglip2 nonsensical outputs and OmDet-Turbo 640px-vs-224px preprocessing mismatch are different issues." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer/model load failures: physical-intelligence/fast vs Mistral regex patch crash." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Model loading failure for ScandiBERT is unrelated to Mistral tokenizer regex patching." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FP8 config failure and `_is_hf_initialized` argument regression are separate v5 breakages." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Early stopping/eval metric mismatch vs trainer token counter not updating; different trainer features." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer class method collision and MobileLLM key error are unrelated loading failures." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "FP8 configuration failure and float8 storage lookup failure are different low-level regressions." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken keyword collision vs loading failure for a specific model; not the same code-path." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 11, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:36032", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42796", + "issue:43089", + "issue:43296", + "issue:43322", + "issue:43381", + "issue:43425", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43582", + "issue:43611", + "issue:43632", + "issue:43746", + "issue:43792", + "issue:44079", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44479", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44683", + "issue:44756", + "issue:44811", + "issue:44877", + "issue:44991", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45538", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:45987", + "issue:46315", + "issue:46491", + "issue:46710", + "issue:46821" + ], + "soft_pairs": [ + "issue:44265|issue:44610", + "issue:43089|issue:45987", + "issue:43632|issue:46821", + "issue:44479|issue:45325", + "issue:43322|issue:43540", + "issue:43531|issue:44877", + "issue:36032|issue:41720", + "issue:45216|issue:45357", + "issue:44991|issue:46315", + "issue:43425|issue:44403", + "issue:43296|issue:43531", + "issue:44220|issue:44683", + "issue:43611|issue:44534", + "issue:40444|issue:45325", + "issue:45812|issue:46491", + "issue:44403|issue:44991", + "issue:43632|issue:44756", + "issue:43792|issue:44589", + "issue:43381|issue:45923", + "issue:42796|issue:43746", + "issue:44291|issue:44589", + "issue:33666|issue:40444", + "issue:40990|issue:44811", + "issue:42796|issue:46710", + "issue:43482|issue:43525", + "issue:35707|issue:40444", + "issue:44079|issue:45663", + "issue:41553|issue:43582", + "issue:43482|issue:45538", + "issue:43632|issue:44534" + ], + "prepared_review_unit_hash": "bdf952b5e2ebff877b7c2d689015adb079ed7267d6df85b535a484ae05f28ad9", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36501, + "estimated_input_tokens": 9126, + "estimated_eval_tokens": 18508 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated issues. The only clear duplicate-like pair is the Qwen3.5 save_pretrained regression/follow-up; the rest share at most a broad subsystem or model family, which is not enough to merge.", + "confidence": 0.87, + "canonical_issue_reason": "issue:45216 is the broader, earlier regression report for the Qwen3.5 save_pretrained bug, while issue:45357 looks like a narrower follow-up about one concrete manifestation (visual encoder keys).", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the strongest cluster anchor because it is the broad upstream regression report and the only issue with a clearly related follow-up in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "Different bugs: torch.export/torch_compilable_check vs OmDet-Turbo processor image-size mismatch." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Both touch generation/cache behavior, but one is a performance overhead report and the other is a cache-length contract bug." + }, + { + "left": "issue:43632", + "right": "issue:46821", + "accept": false, + "reason": "Unrelated: a Transformers v5 init flag regression vs a TimesFM window_size AttributeError." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Both involve Qwen VL media handling, but one is a broad video regression and the other is a specific still-image position_id scaling bug." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "Different models and failure modes: Llava Next loading segfault vs Qwen3OmniMoe video ValueError." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "No same code path: Qwen3-MoE sliding_window behavior vs granite_speech config strictness." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer method conflict vs Qwen3 auto device mapping CUDA assert; no shared bug." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Same Qwen3.5 save_pretrained regression; 45357 is a narrower follow-up about incorrect visual encoder keys." + }, + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Both are tokenizer-related, but one is a load failure for a specific model and the other is a performance regression in convert_tokens_to_ids." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Different topics: Torch 2.10 compatibility vs noisy loading output." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "Different model families and bugs: PaddleOCR-VL loading in vLLM vs Qwen3-MoE sliding_window." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Unrelated audio feature extraction bug vs compiled flex_attention failure on newer torch." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Different regressions: base_model_prefix loading vs non-persistent buffer corruption." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "Both involve Qwen2.5-VL, but one is IterableDataset/multi-image finetuning and the other is rope/position_id scaling." + }, + { + "left": "issue:45812", + "right": "issue:46491", + "accept": false, + "reason": "Different tokenizers and symptoms: Granite token IDs regression vs CodeLlama leading-space round-trip issue." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Noise during loading is not the same bug as EMBEDDIA/est-roberta tokenizer loading failure." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "Different problem domains: HF init flag regression vs Strix Halo mmap OOM workaround." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper model loading/runtime issue vs Float8 storage lookup error; no concrete shared code path." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "Related to gradient checkpointing in name only: eval-mode restriction vs missing supports_gradient_checkpointing flag." + }, + { + "left": "issue:42796", + "right": "issue:43746", + "accept": false, + "reason": "Tokenizer BPE marker handling is unrelated to GraniteSpeech PEFT checkpoint loading." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized TypeError is a different failure than missing Float8 storage." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Both are Qwen VL training issues, but one is multi-GPU training and the other is multi-image iterable dataset failure." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "Completely different subsystems: GPT-OSS perplexity vs Whisper batch_decode skip_special_tokens behavior." + }, + { + "left": "issue:42796", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer marker handling for Ministral is unrelated to DeepSeek accuracy regression." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "Different loading failures: Qwen2.5-GGUF with transformers v5 vs missing pad_token_id on Llama4Config." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Different generation/training paths: progressive generation with inputs_embeds/past_key_values vs Qwen2.5-VL finetuning on multi-image datasets." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment bug is unrelated to Gemma-4 FSDP2 shared_kv_states KeyError." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad AutoTokenizer error message vs Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "Different tokenizer/model issues: Qwen2.5-GGUF loading vs CLIPTokenizer max length constant." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "Both are Transformers v5 regressions, but they affect different internals and symptoms." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted soft pair, issue:45216 and issue:45357, is plausibly a same-bug regression/follow-up based on their titles. The other rejections are generally well-justified by distinct models, subsystems, or failure modes. Minor caution: wording like \"upstream\" is a bit interpretive, but not enough to invalidate the cluster judgment." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 12, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:40444", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43421", + "issue:43454", + "issue:43482", + "issue:43493", + "issue:43531", + "issue:43606", + "issue:43611", + "issue:43644", + "issue:43646", + "issue:43701", + "issue:43792", + "issue:43818", + "issue:43856", + "issue:43873", + "issue:43957", + "issue:43994", + "issue:44079", + "issue:44265", + "issue:44291", + "issue:44462", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44568", + "issue:44829", + "issue:44877", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45491", + "issue:45542", + "issue:45701", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46402", + "issue:46821" + ], + "soft_pairs": [ + "issue:43317|issue:43856", + "issue:43531|issue:44291", + "issue:43421|issue:44568", + "issue:42175|issue:45542", + "issue:41720|issue:45081", + "issue:44079|issue:44534", + "issue:44462|issue:45812", + "issue:43296|issue:46821", + "issue:44829|issue:45479", + "issue:43818|issue:46402", + "issue:43377|issue:45491", + "issue:43611|issue:43646", + "issue:42175|issue:43296", + "issue:43644|issue:44079", + "issue:43818|issue:45357", + "issue:43792|issue:43957", + "issue:44265|issue:45072", + "issue:45325|issue:45381", + "issue:43493|issue:43994", + "issue:45325|issue:45834", + "issue:40444|issue:45381", + "issue:43454|issue:44466", + "issue:42915|issue:43482", + "issue:44291|issue:44877", + "issue:43701|issue:45325", + "issue:43873|issue:45701", + "issue:42796|issue:45381", + "issue:43606|issue:44265", + "issue:44479|issue:45823" + ], + "prepared_review_unit_hash": "0a31a62378b8194b8d57c1e7b8ec2dba3da1c5395e121af4673b4fbd0d8fb148", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36674, + "estimated_input_tokens": 9169, + "estimated_eval_tokens": 18594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous issue cluster with a few clear duplicate-style subpairs around Qwen visual position handling and SigLIP2 tokenizer/model behavior; the rest are different regressions in unrelated subsystems.", + "confidence": 0.66, + "canonical_issue_reason": "Issue 44479 is the broadest and most representative report of the Qwen2.5-VL/Qwen3* video-input regression family, capturing the same visual position-id bug line as the narrower Qwen visual-position issues.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44479 is the best single representative because it describes the regression most generally and across multiple affected Qwen model variants.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Different problems: CPU/GPU offload loading failure vs MoE training memory usage." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Different code paths: sliding_window/Qwen3-MoE behavior vs init_empty_weights/_is_hf_initialized loading error." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Both tokenizer-related, but one is runtime post-processor updates and the other is add_special_tokens/BOS-EOS handling; not the same bug." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Both concern TF-related packaging, but the reported failures are different extras/dependency setups." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Unrelated failures: Qwen3 auto device-map CUDA assert vs Mistral regex patch AttributeError." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "Different issues in different internals: ModelOutput key assignment vs non-persistent buffer initialization." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Tokenizer loading vs Granite token-ID regression; related area, but not clearly the same underlying bug." + }, + { + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "Different models and failures: PaddleOCR-VL load in vLLM vs TimesFM window_size AttributeError." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Both cause degenerate training, but via distinct mechanisms and model families." + }, + { + "left": "issue:43818", + "right": "issue:46402", + "accept": false, + "reason": "Different model bugs: Video-LLaVA tower/weight sharing vs Sam3Model text encoder loading." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Both are batching/masking issues, but they affect different models and different attention/padding failure modes." + }, + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers v5 regressions, but one is about base_model_prefix loading and the other custom model initialization." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "Packaging/backend issue vs model-loading failure in vLLM; not the same bug." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "Non-persistent buffer corruption vs ModelOutput key assignment bug." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Different visual-model bugs: missing temporal attention in Video-LLaVA vs incorrect visual encoder keys in save_pretrained." + }, + { + "left": "issue:43792", + "right": "issue:43957", + "accept": false, + "reason": "Different loading failures: Whisper runtime issue vs torch.device('meta') loading breakage." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "Different problems: torch.export/torch_compilable_check vs bf16 dtype mismatch in inference." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": true, + "reason": "Same Qwen2.5-VL vision-position regression family in 5.3.0; both point to get_rope_index/position-id handling." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 bug family: HF implementation disagrees with the original JAX behavior and yields bad outputs for the same model." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Both are vision/ocr-adjacent, but one is Qwen2.5-VL position-id scaling and the other is a Kosmos2.5 long-input index error." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Both mention Qwen2.5-VL, but one is finetuning with multiple images per prompt and the other is a video-input position-id regression." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "Related to lm_head/tied weights, but one is a weight-tying bug and the other is a serialization inconsistency." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Unrelated model-specific failures: Qwen3MoE FP8 config vs Qwen2.5-GGUF loading on v5." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "Different failures: init_empty_weights argument mismatch vs strict config rejecting granite_speech." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "Checkpoint resume key mismatch is unrelated to Qwen2.5-VL position-id computation." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Offloading/quantization issue vs tokenizer-version-dependent tokenization change." + }, + { + "left": "issue:42796", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer BPE-marker handling is unrelated to Qwen2.5-VL vision position ids." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU-offload device mismatch vs torch.export/compilable-check export failure." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Video-input regression in Qwen models vs Gemma4 PLE device-mapping mismatch; different symptoms and code paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering summary is grounded in the packet and stays conservative. The two accepted soft pairs (Qwen position-id handling and SigLIP2 behavior) are supported by the issue titles, and the rejected pairs generally involve clearly different model families or failure modes. The choice of issue 44479 as the broad representative of the Qwen video/position-id regression family is reasonable from the titles provided." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 13, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:36331", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42796", + "issue:42915", + "issue:42947", + "issue:43257", + "issue:43262", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43454", + "issue:43526", + "issue:43550", + "issue:43572", + "issue:43606", + "issue:43720", + "issue:43746", + "issue:43749", + "issue:43819", + "issue:43828", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44186", + "issue:44265", + "issue:44488", + "issue:44534", + "issue:44568", + "issue:44610", + "issue:44898", + "issue:44991", + "issue:45072", + "issue:45584", + "issue:45759", + "issue:45823", + "issue:45920", + "issue:45923", + "issue:46050", + "issue:46209", + "issue:46305", + "issue:46315", + "issue:46821" + ], + "soft_pairs": [ + "issue:40990|issue:45584", + "issue:43262|issue:43746", + "issue:43828|issue:46821", + "issue:43881|issue:44488", + "issue:35707|issue:41720", + "issue:34634|issue:35707", + "issue:43454|issue:43550", + "issue:43606|issue:43819", + "issue:39692|issue:40444", + "issue:44186|issue:44265", + "issue:41720|issue:45823", + "issue:43377|issue:44610", + "issue:46305|issue:46315", + "issue:43257|issue:45923", + "issue:44568|issue:45920", + "issue:45823|issue:46821", + "issue:43454|issue:45072", + "issue:42796|issue:46209", + "issue:43317|issue:43873", + "issue:43377|issue:44898", + "issue:43454|issue:43720", + "issue:40990|issue:42947", + "issue:42617|issue:42915", + "issue:43749|issue:44534", + "issue:43572|issue:45759", + "issue:43526|issue:43883", + "issue:43322|issue:46050", + "issue:41720|issue:43322", + "issue:43881|issue:44991", + "issue:36331|issue:42617" + ], + "prepared_review_unit_hash": "d536694e56253d2139145f2b399c81fafb99f27718e9dc972441cb5be46f9f3a", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36810, + "estimated_input_tokens": 9203, + "estimated_eval_tokens": 18662 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly heterogeneous issue reports across unrelated models and subsystems. I do not see a safe cluster-wide duplicate canonical; none of the soft-edge pairs look mergeable into one concrete fix.", + "confidence": 0.88, + "canonical_issue_reason": "No safe canonical issue for the whole set: the reports span unrelated bugs in tokenization, loading, device mapping, training, audio, and export.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to name a closest representative, issue 43873 is the broadest loading/offloading report, but it still does not represent the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "Different models and bug surfaces: perplexity/eval behavior vs Whisper empty-transcription generation after token alignment." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Both touch audio, but one is a sampling-rate default bug and the other is a PEFT/local checkpoint loading failure." + }, + { + "left": "issue:43828", + "right": "issue:46821", + "accept": false, + "reason": "Unrelated failures: autocast dtype mismatch in Phi-MoE vs an AttributeError for TimesFM window_size." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Both are model-loading failures, but they involve different models and likely different root causes." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Generation with inputs_embeds/past_key_values is unrelated to a device-map CUDA assert on Qwen3." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "Bark voice_preset handling and progressive generation with past key values are different code paths." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "Weight tying in AyaVisionConfig is unrelated to Bamba SDPA/torch.compile failures." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "Both are audio-model issues, but one is CPU offload device mismatch and the other is DAC latent reconstruction mismatch." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "Docs example errors for SigLIP2 do not match Qwen2.5-VL finetuning with multi-image iterable datasets." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer crash in LayoutLMv2 is not the same as torch.export failing on torch_compilable_check models." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Both involve device_map=auto, but the failures differ: CUDA assert on Qwen3 vs device mismatch during Gemma4 forward." + }, + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "Missing padding-mask support in MIMI encoder is unrelated to OmDet-Turbo processor/model input-size mismatch." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Streaming mask off-by-one and tokenizer convert_tokens_to_ids slowdown are different bugs." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "Qwen3 MOE weight conversion under accelerate/deepspeed is unrelated to a missing supports_gradient_checkpointing flag." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "Specific BOS/EOS handling in one tokenizer is not the same as broad AutoTokenizer ID regressions across many models." + }, + { + "left": "issue:45823", + "right": "issue:46821", + "accept": false, + "reason": "Gemma4 device-mismatch forward failure and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "LM head weight tying and bfloat16 dtype mismatches in inference are different problems." + }, + { + "left": "issue:42796", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are unrelated." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "They share offloading/quantization wording, but the repros and failure modes differ enough that they should not be merged." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Padding-mask batch consistency and Perceiver non-default resolution failures are different model/processor bugs." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "AyaVision weight tying and BitNet packed-weight unpacking during accelerate loading are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Perplexity on a benchmark and gradient-checkpointing behavior with PEFT LoRA are unrelated." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "3d_parallel script execution and Qwen3MoE FP8 loading are different code paths." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "Both mention loading/regression, but FSDP CPU RAM efficient loading and junk non-persistent buffers are not the same defect." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to unwrapping text_config for composite Qwen multimodal configs." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "BEiT image-processor label reduction and Molmo tied-weights attribute errors are unrelated." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Llava Next load crashes and sinq save/load quantization breakage are different failures." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Both are model-loading issues, but one is a device-map CUDA assert and the other is a segmentation fault in Llava Next loading." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Different tokenizer/model-loading failures for different repositories; no shared concrete bug." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Trainer compute_loss signature mismatch is unrelated to the 3d_parallel script 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": 24, + "review_unit_index": 14, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:26413", + "issue:28530", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43065", + "issue:43295", + "issue:43317", + "issue:43404", + "issue:43450", + "issue:43531", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43873", + "issue:44008", + "issue:44186", + "issue:44387", + "issue:44423", + "issue:44466", + "issue:44610", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45632", + "issue:45684", + "issue:45685", + "issue:45701", + "issue:45704", + "issue:45823", + "issue:46402", + "issue:46665", + "issue:46821" + ], + "soft_pairs": [ + "issue:44387|issue:45704", + "issue:41720|issue:44977", + "issue:36331|issue:42915", + "issue:44466|issue:45357", + "issue:44877|issue:44991", + "issue:43450|issue:45701", + "issue:40990|issue:45701", + "issue:43065|issue:46402", + "issue:43873|issue:44756", + "issue:41720|issue:42617", + "issue:44186|issue:45072", + "issue:26413|issue:28530", + "issue:43550|issue:43720", + "issue:43577|issue:43708", + "issue:45685|issue:46821", + "issue:43295|issue:45701", + "issue:43550|issue:43606", + "issue:43646|issue:46665", + "issue:44186|issue:44610", + "issue:42915|issue:43317", + "issue:44610|issue:44898", + "issue:43404|issue:45357", + "issue:44008|issue:45823", + "issue:36010|issue:42617", + "issue:44423|issue:45406", + "issue:45632|issue:45684", + "issue:41720|issue:45161", + "issue:43531|issue:43572", + "issue:28530|issue:33666" + ], + "prepared_review_unit_hash": "ec30d31d55ee7d0569890e6ab100da31cc2abba2312670390ca259e4f02c4306", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35902, + "estimated_input_tokens": 8976, + "estimated_eval_tokens": 18208 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs share only broad themes (same model family, same API area, or same symptom class) but not the same concrete bug. I would not merge any of the proposed soft edges.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45701 is the broadest, most hub-like issue here (generic version-regression/tokenization wording, multiple inbound references), so it is the best umbrella representative for the noisy cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45701 is the most representative issue by scope and connectivity, but only as a loose thematic hub rather than a true duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Both mention memory pressure, but one is int4 quantization/OOM while the other is a T5 RMSNorm memory leak; different code paths and fixes." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Both are Qwen3-related, but one is a device-mapping cuda assert and the other is a flash-attention generation issue; not the same bug." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "`compute_loss(num_items_in_batch)` is a trainer API compatibility issue; Qwen3MoE + FineGrainedFP8Config is a different model/config failure." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both concern saving/loading model weights, but one is tied `lm_head` serialization and the other is incorrect visual encoder keys; distinct failures." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Both involve loading/serialization regressions in v5, but one is strict config handling and the other is tokenizer loading for a specific model; not the same defect." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Video processor batch-shape handling is unrelated to the tokenization-version regression described in 45701." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "High perplexity on gpt-oss is an evaluation/model-behavior problem, not the same underlying tokenization change bug." + }, + { + "left": "issue:43065", + "right": "issue:46402", + "accept": false, + "reason": "Both are Sam3-related, but one is a dummy Conv2d concern and the other is missing text-encoder weight loading; different code paths." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Quantization/offload behavior and Strix Halo mmap OOM are both memory-related but arise from different mechanisms and fixes." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Qwen3 auto-device-map cuda assert and failure to run `3d_parallel.py` are not the same underlying issue." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer crashes on NER/batched padding and bfloat16 dtype mismatches are unrelated preprocessing vs precision bugs." + }, + { + "left": "issue:26413", + "right": "issue:28530", + "accept": false, + "reason": "Checkpoint resumption exhausting the epoch iterator and early-stopping metric lookup are separate trainer behaviors." + }, + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "One is torch.compile + SDPA on Bamba, the other is packed-weight handling during accelerate loading; different failure modes." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "BLIP2 dtype propagation and resume-from-checkpoint max_steps calculation are unrelated bugs." + }, + { + "left": "issue:45685", + "right": "issue:46821", + "accept": false, + "reason": "A MoE MPS histogram dtype error and a TimesFM window_size AttributeError are unrelated model-specific failures." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "Processor/tokenizer API regression and a broader tokenization-version complaint are related only superficially, not the same concrete bug." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Both touch device/runtime behavior, but one is compile+SDPA and the other is CPU-offload device mismatch; not mergeable as one fix." + }, + { + "left": "issue:43646", + "right": "issue:46665", + "accept": false, + "reason": "Custom model initialization breakage and inconsistent `from_pretrained` behavior inside classes are separate API issues." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "LayoutLMv2 tokenizer NER/padding crashes and OmDet-Turbo image-size mismatch are both processor issues but not the same bug." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "FineGrainedFP8Config failure and dequantized model loading with cpu offload are different quantization/offload problems." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both involve image sizing, but one is an OmDet processor/model expectation mismatch and the other is Perceiver interpolation behavior; distinct issues." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Tied `lm_head` weights in Mistral3 and incorrect visual encoder keys in Qwen3.5 save_pretrained are different serialization bugs." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3n variable-name collision in forward and Gemma4 PLE device mismatch with `device_map=auto` are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "GenerationMixin import regression and failure to run `3d_parallel.py` are unrelated issues." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "`continuous-batching` multimodal crash and Gemma4Processor missing `_tokenizer` are both serve-related but different code paths and fixes." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Trust-remote-code cache path collision and read-only permission propagation are unrelated save/load filesystem issues." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Both mention Qwen/GPT-OSS/MoE style model families, but one is an A800 device-map crash and the other is tensor-parallel support; different problems." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Sliding-window handling for Qwen3-MoE and missing `pad_token_idx` in StableLmConfig are separate config/model bugs." + }, + { + "left": "issue:28530", + "right": "issue:33666", + "accept": false, + "reason": "Early-stopping metric lookup and Qwen2-VL multi-GPU training are not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 15, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28530", + "issue:29159", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36032", + "issue:36331", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:43316", + "issue:43322", + "issue:43329", + "issue:43388", + "issue:43425", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43653", + "issue:43716", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:43881", + "issue:44186", + "issue:44361", + "issue:44387", + "issue:44492", + "issue:44509", + "issue:44534", + "issue:44589", + "issue:44671", + "issue:44743", + "issue:44913", + "issue:45059", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:46291", + "issue:46665", + "issue:46821" + ], + "soft_pairs": [ + "issue:44671|issue:44743", + "issue:42915|issue:46291", + "issue:36331|issue:41720", + "issue:43526|issue:43881", + "issue:44186|issue:44361", + "issue:43425|issue:46821", + "issue:43716|issue:46821", + "issue:43316|issue:44913", + "issue:42915|issue:43322", + "issue:28530|issue:30064", + "issue:33357|issue:35707", + "issue:43540|issue:43782", + "issue:30064|issue:34634", + "issue:43550|issue:43653", + "issue:43873|issue:45704", + "issue:29159|issue:36032", + "issue:44387|issue:45701", + "issue:43606|issue:43720", + "issue:44492|issue:44509", + "issue:44534|issue:46665", + "issue:43531|issue:45759", + "issue:41720|issue:43746", + "issue:42490|issue:43329", + "issue:43388|issue:45059", + "issue:43606|issue:43653", + "issue:45406|issue:45464", + "issue:34567|issue:35707", + "issue:43531|issue:46821", + "issue:34634|issue:41720", + "issue:43531|issue:44589" + ], + "prepared_review_unit_hash": "7bf1362ce6d44d9833b3ee15a705c132dcb681b606ca5b83a168aaed600c4cd0", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36607, + "estimated_input_tokens": 9152, + "estimated_eval_tokens": 18560 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Highly heterogeneous cluster: it mixes unrelated tokenizer, config, multimodal, training, quantization, serving, and docs issues. None of the suggested soft pairs look like the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue stands out; the set is too heterogeneous to have a meaningful duplicate anchor.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global best issue; even the broadest titles here describe different failure modes in different subsystems.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "Different models and code paths: CamemBERT masked-LM predictions vs Qwen3.5 recurrent state reset with cache." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and pipelines import failure from kernels version/revision requirements are unrelated." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "Trainer compute_loss signature mismatch is unrelated to Qwen3 auto device mapping cudaErrorAssert." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug and glm-4v-9b loading failure are different subsystems and symptoms." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer bugs, but LayoutLMv2 NER/padding crashes and MLukeAttributeError are different concrete defects." + }, + { + "left": "issue:43425", + "right": "issue:46821", + "accept": false, + "reason": "Torch 2.10 incompatibility is a version-support problem, not the TimesFM window_size AttributeError." + }, + { + "left": "issue:43716", + "right": "issue:46821", + "accept": false, + "reason": "Image preprocessor dtype mismatch and TimesFM window_size AttributeError are unrelated issues." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy and GPTNeoX rotary_pct reload regression involve different config fields and behaviors." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Qwen3Moe fine-grained FP8 config failure and Llava Next segmentation fault are not the same bug." + }, + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Early stopping/eval metric lookup has nothing to do with void segmentation map processing." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values are different runtime problems." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Video-input ValueError in Qwen3OmniMoe and Qwen3VL weight_only load failure are different code paths." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Void segmentation map processing and BarkProcessor voice_preset handling are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "torch.compile SDPA failure in Bamba-9B-v2 is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization/offloading behavior and T5 apex FusedRMSNorm memory leak are distinct bugs." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Slow/fast tokenizer inconsistency is different from the T5Tokenizer add_special_tokens naming conflict." + }, + { + "left": "issue:44387", + "right": "issue:45701", + "accept": false, + "reason": "Int4 quantization OOM and version-dependent tokenization changes are not the same underlying issue." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "CPU offload device mismatch for Bark differs from BitNet packed-weights unpacking during accelerate loading." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A typo in cache strategies is unrelated to stale docs mentioning removed pipeline tasks." + }, + { + "left": "issue:44534", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffers filled with junk and from_pretrained behavior inside Python classes are different loader semantics." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and AutoModelForCausalLM config unwrapping are separate Qwen-related bugs." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 auto device mapping cudaErrorAssert is unrelated to GraniteSpeech PEFT adapter loading." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Trailing-slash local model import warning and undefined variables in multimodal token counting are different issues." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "gather_for_metrics dropping tuple labels is unrelated to SAM3 PCS text/bounding-box behavior." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "CPU offload device mismatch and mask token special-token registration are different bugs." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to Qwen3.5 streaming chat/completions failure." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "TrainerState token counters not updating and progressive generation cache behavior are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:46821", + "accept": false, + "reason": "Both mention windowing, but Qwen3-MoE sliding_window and TimesFM window_size AttributeError are different models and bugs." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "BarkProcessor voice_preset failure is unrelated to Qwen3 device-mapping cudaErrorAssert." + }, + { + "left": "issue:43531", + "right": "issue:44589", + "accept": false, + "reason": "Qwen3-MoE sliding_window bug and missing Float8 storage object 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": 24, + "review_unit_index": 16, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:43381", + "issue:43421", + "issue:43644", + "issue:43653", + "issue:43746", + "issue:43782", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44062", + "issue:44077", + "issue:44186", + "issue:44368", + "issue:44451", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44625", + "issue:44671", + "issue:44843", + "issue:44898", + "issue:44918", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45072", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:45923", + "issue:46032", + "issue:46050", + "issue:46402", + "issue:46463", + "issue:46655", + "issue:46665", + "issue:46693" + ], + "soft_pairs": [ + "issue:44451|issue:44991", + "issue:45759|issue:45859", + "issue:45357|issue:46402", + "issue:44077|issue:44625", + "issue:43644|issue:46665", + "issue:44062|issue:44451", + "issue:44898|issue:45072", + "issue:44987|issue:45042", + "issue:44625|issue:45479", + "issue:43381|issue:43844", + "issue:39692|issue:44610", + "issue:44368|issue:45357", + "issue:46655|issue:46693", + "issue:41720|issue:45704", + "issue:43746|issue:46050", + "issue:43421|issue:43927", + "issue:45704|issue:45923", + "issue:43381|issue:45704", + "issue:44918|issue:44936", + "issue:42886|issue:44843", + "issue:44514|issue:44545", + "issue:45759|issue:45923", + "issue:46032|issue:46463", + "issue:39692|issue:43873", + "issue:43746|issue:43873", + "issue:43994|issue:44671", + "issue:43421|issue:43653", + "issue:30064|issue:36010", + "issue:43421|issue:44186", + "issue:43782|issue:45381" + ], + "prepared_review_unit_hash": "83b52acba4548dd8206f45733b47441b136ddc09978c6a46d6633763d37e2711", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36365, + "estimated_input_tokens": 9092, + "estimated_eval_tokens": 18440 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous Transformers issues with only two clear duplicate pairs: the offline tokenizer-cache regression and the Qwen2_5_VL batched padding crash. The rest are similar only at a broad subsystem level, not the same underlying bug.", + "confidence": 0.87, + "canonical_issue_reason": "Issue 43421 is the best cluster anchor among the issues: it is the broadest tokenizer-centric report and sits near several other special-token/tokenizer behavior reports, making it the most representative issue in this set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43421 is the strongest overall issue candidate because it captures a concrete recurring tokenizer state/post-processor problem and is broader than the more model-specific tokenizer reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading failures, but they involve different models and likely different root causes." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Both mention Qwen 3.5/3.6, but one is a config unwrap issue and the other is CPU offload compatibility; not the same bug." + }, + { + "left": "issue:45357", + "right": "issue:46402", + "accept": false, + "reason": "Different model families and different loading failures; no shared concrete code path." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "One is about post_init being optional, the other about num_labels propagation; unrelated behaviors." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer corruption and from_pretrained behavior in Python classes are unrelated." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "A tokenizer AddedToken constructor error is not the same as a model-specific load failure." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Different subsystems and failure modes: image resolution handling vs dtype mismatch." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "Loading a specific model and PIL backend torchvision requirements are separate bugs." + }, + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "One is config propagation, the other is loss calculation degeneracy; different code paths." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Gradient checkpointing in eval mode and a ZeRO-3 gradient blow-up are distinct issues." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Both are vision-related, but one is a doc/example mismatch and quantization failure while the other is a processor size mismatch." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "A warning about tie_word_embeddings is unrelated to saving incorrect visual encoder keys." + }, + { + "left": "issue:46655", + "right": "issue:46693", + "accept": false, + "reason": "Incorrect video padding calculation and a FA2 performance regression are different problems." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "CUDA assert during Qwen3 device mapping is unrelated to a T5 Apex RMSNorm memory leak." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "Local PEFT checkpoint loading and sinq save/load breakage are different serialization problems." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Both involve tokenizer metadata, but one is runtime special-token updates and the other is save/load losing custom token IDs." + }, + { + "left": "issue:45704", + "right": "issue:45923", + "accept": false, + "reason": "A memory leak in T5 norm code and a missing gradient_checkpointing flag are not the same bug." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Eval-mode gradient checkpointing behavior does not match a T5 Apex memory leak." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "Trainer embedding unpacking failure and evaluate-after-train failure are different trainer issues." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": true, + "reason": "Both describe the same offline tokenizer-loading regression; 44843 identifies the unconditional model_info() call in _patch_mistral_regex as the root cause." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "These are the same Qwen2_5_VLProcessor.apply_chat_template crash on batched input with padding=False." + }, + { + "left": "issue:45759", + "right": "issue:45923", + "accept": false, + "reason": "Different Qwen model issues: from_config text_config unwrapping vs missing supports_gradient_checkpointing." + }, + { + "left": "issue:46032", + "right": "issue:46463", + "accept": false, + "reason": "A silent wrong-result cache bug and an error-message bug are not the same underlying change." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 doc/example issues and quantization offloading behavior are not the same bug." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "PEFT checkpoint loading and quantized offloading misbehavior are separate problems." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different models and failure modes: nonsensical pipeline outputs vs masked LM predictions." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Both concern tokenizer special tokens, but one is runtime post-processor updating and the other is a BigBirdTokenizer registration/decode bug." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "ImageProcessor segmentation-map handling and a GenerationMixin ImportError are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Special-token post-processor updates and LayoutLMv2 NER/padding crashes are different tokenizer bugs." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Qwen3VL weight_only loading and Qwen2.5-VL video position IDs are separate multimodal issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet: it correctly characterizes the set as mostly heterogeneous and identifies the two strongest duplicate pairs (42886/44843 and 44514/44545). The remaining soft-edge rejections are conservative and do not overclaim shared root cause from the titles alone. I don\u2019t see an evidence-overstatement problem here." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 17, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:30064", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43295", + "issue:43335", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43611", + "issue:43618", + "issue:43643", + "issue:43653", + "issue:43697", + "issue:43720", + "issue:43756", + "issue:43819", + "issue:43927", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44442", + "issue:44448", + "issue:44485", + "issue:44488", + "issue:44509", + "issue:44661", + "issue:44671", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44991", + "issue:45072", + "issue:45081", + "issue:45084", + "issue:45356", + "issue:45381", + "issue:45834", + "issue:45907", + "issue:45987", + "issue:46294", + "issue:46665", + "issue:46693", + "issue:46829", + "issue:46832" + ], + "soft_pairs": [ + "issue:46829|issue:46832", + "issue:45081|issue:45356", + "issue:45987|issue:46829", + "issue:45356|issue:45907", + "issue:43720|issue:43819", + "issue:43653|issue:43720", + "issue:43596|issue:44220", + "issue:43295|issue:45834", + "issue:43550|issue:45072", + "issue:43550|issue:43927", + "issue:44062|issue:44488", + "issue:43540|issue:46294", + "issue:45381|issue:46693", + "issue:44220|issue:44661", + "issue:43531|issue:43643", + "issue:43611|issue:46665", + "issue:43618|issue:43697", + "issue:43653|issue:43819", + "issue:44186|issue:44898", + "issue:30064|issue:36331", + "issue:41720|issue:44991", + "issue:43335|issue:43756", + "issue:43643|issue:44877", + "issue:43550|issue:43819", + "issue:43540|issue:44008", + "issue:44821|issue:45084", + "issue:44843|issue:45081", + "issue:39692|issue:45834", + "issue:44485|issue:44509", + "issue:44186|issue:44442", + "issue:44448|issue:44671" + ], + "prepared_review_unit_hash": "a4dba96b6eb8c870385ade848c2ba953e93bc2fd5c4ca26a7369842b7c4e9c7d", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37253, + "estimated_input_tokens": 9314, + "estimated_eval_tokens": 18884 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose topical neighborhood, not a true duplicate cluster: it mixes unrelated bugs across vision/video models, tokenizers, configs, and training/runtime paths. No soft edge is strong enough to merge, so all candidate links are rejected.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue is clearly justified because the cluster is heterogeneous; at most it loosely centers on tokenizer-regex regressions, but most items are unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick one representative issue, 45081 is the clearest concrete regression in the tokenizer-regex code path (_patch_mistral_regex crash with a reproducible failure).", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46829", + "right": "issue:46832", + "accept": false, + "reason": "Different model families and failures: VideoPrism output field mismatch vs RT-DETR feature-level shape error." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both touch fix_mistral_regex, but one is a crash in _patch_mistral_regex and the other is a Kimi tokenizer regression with codec/warning issues; not clearly the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache seq length contract bug is unrelated to a video classification output field mix-up." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression and repo_files revision forwarding are unrelated changes." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "Different subsystems and failures: accelerate loading of packed BitNet weights vs DAC from_latents/STE mismatch." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight loading are unrelated tokenizer/model-loading bugs." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed zero3/BertModel init error is unrelated to _torch_extract_fbank_features() behavior." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Processor/tokenizer regression in v4.57.5 is unrelated to Kosmos2.5 long-OCR indexing." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "Bamba torch.compile SDPA failure is unrelated to SwitchTransformers/TimmWrapperModel dtype mismatches." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Torch.compile SDPA failure and DiaConfig save/load token-ID loss are different bugs." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken duplicate special-arg error is unrelated to loading a specific model/tokenizer checkpoint." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and Qwen3.5 video apply_chat_template token/feature mismatch are different code paths." + }, + { + "left": "issue:45381", + "right": "issue:46693", + "accept": false, + "reason": "Qwen2.5-VL video position IDs and FA2 long-context performance regression are unrelated." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio fbank feature extraction issue and add-new-model-like TOKENIZER_MAPPING_NAMES failure are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to AutoConfig trust_remote_code missing fields." + }, + { + "left": "issue:43611", + "right": "issue:46665", + "accept": false, + "reason": "base_model_prefix loading breakage and from_pretrained behavior inside Python classes are different loader regressions." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "Missing CLIPOutput attentions field and RTDetrV2 output drift are unrelated model-output regressions." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "Tokenizer special-token registration and DAC forward/from_latents mismatch are unrelated." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash and Perceiver non-default-res image classification failure are different bugs." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Image Processor void segmentation map handling and CustomTrainer.compute_loss signature mismatch are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert and est-roberta tokenizer loading failure do not share a concrete code path." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation and Smollm3 RoPE layer-dropping are different configuration/model-definition bugs." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "AutoConfig trust_remote_code missing fields and granite_speech strict-config loading are distinct config-loading regressions." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba torch.compile SDPA failure and DAC latent conversion mismatch are unrelated model/runtime issues." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and Gemma3n variable-name collision are unrelated." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading and TypeError compiling non-template nodes are unrelated." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Both involve _patch_mistral_regex, but one is offline-mode model_info() access and the other is a backend_tokenizer attribute crash; not clearly the same underlying bug." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "SigLIP2 docs/example errors and Kosmos2.5 OCR indexing are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:44509", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion and removed pipeline-task docs are different change requests." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2Tokenizer NER/padding crash and AutoTokenizer failing to load FastSpeech2ConformerTokenizer are unrelated." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Pegasus v4/v5 output differences and CamemBERT masked-LM prediction regression are different model-specific 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": 24, + "review_unit_index": 18, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:34634", + "issue:36010", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43720", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43825", + "issue:43827", + "issue:43950", + "issue:43994", + "issue:44008", + "issue:44291", + "issue:44479", + "issue:44561", + "issue:44610", + "issue:44805", + "issue:44811", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45183", + "issue:45381", + "issue:45479", + "issue:45584", + "issue:45685", + "issue:45834", + "issue:45987", + "issue:46032", + "issue:46291", + "issue:46294", + "issue:46463", + "issue:46512", + "issue:46665" + ], + "soft_pairs": [ + "issue:43644|issue:43749", + "issue:45183|issue:45479", + "issue:43322|issue:45685", + "issue:34634|issue:36010", + "issue:43606|issue:45072", + "issue:45584|issue:45834", + "issue:43716|issue:44008", + "issue:44479|issue:45834", + "issue:43388|issue:46032", + "issue:43950|issue:45987", + "issue:44991|issue:46291", + "issue:43582|issue:46512", + "issue:43577|issue:43818", + "issue:43825|issue:43994", + "issue:43782|issue:44977", + "issue:46294|issue:46463", + "issue:43421|issue:44811", + "issue:44561|issue:45020", + "issue:44805|issue:44936", + "issue:43540|issue:45381", + "issue:44610|issue:45072", + "issue:43720|issue:45072", + "issue:43825|issue:43827", + "issue:28218|issue:29159", + "issue:43749|issue:44291", + "issue:43645|issue:46665", + "issue:43577|issue:43582" + ], + "prepared_review_unit_hash": "25988cda0876a91002c474f381645b8c7fb1e0b49b42d5071e50cdff0a71b581", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35657, + "estimated_input_tokens": 8915, + "estimated_eval_tokens": 18086 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly noisy and spans many unrelated regressions. The only strong duplicate-looking pair is the tokenizer added-token spacing vs. slow/fast tokenizer inconsistency report.", + "confidence": 0.61, + "canonical_issue_reason": "issue:45020 is the broadest umbrella report in the set for recent-version remote_code regressions, making it the best general anchor for this mixed cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the most general, high-level issue here; the rest are mostly narrow, model-specific regressions or unrelated bug reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Different loading paths and failures: non-persistent buffer corruption vs. FSDP CPU RAM efficient loading." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "One is input_ids/inputs_embeds validation messaging; the other is a classification-loss bug for num_labels=1." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated models and code paths: Llava Next load segfault vs. an MPS histogram/int support error." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "BarkProcessor voice_preset behavior is unrelated to the GenerationMixin import failure." + }, + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Different bug classes: CPU offload device mismatch vs. bfloat16 dtype mismatch in other models." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty transcription handling is unrelated to Kosmos2.5 long OCR indexing." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "Both are multimodal, but they describe different failures in different models and code paths." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Qwen video regression and Kosmos2.5 OCR indexing are not the same underlying bug." + }, + { + "left": "issue:43388", + "right": "issue:46032", + "accept": false, + "reason": "Gather-for-metrics label trimming is unrelated to Mamba2 cache misuse with seq_len > 1." + }, + { + "left": "issue:43950", + "right": "issue:45987", + "accept": false, + "reason": "Non-persistent buffer corruption and StaticCache return-type mismatch are different regressions." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading failure and kernels version/revision import failure are distinct issues." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon warmup TypeError is unrelated to stale config dtype in from_config()." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "BLIP2 dtype mismatch and Video-LLaVA architecture/weight-sharing issues are not the same bug." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "Pipeline v5 error-message wording and SigLIP2 model output corruption are unrelated." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "from_pretrained weight_only loading error is unrelated to flash-attention generation problems." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen3.5 video template mismatch and T5 decoder-input validation messaging are different code paths." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Runtime post-processor updates and Whisper batch_decode skip_special_tokens handling are separate tokenizer/decoder issues." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both mention remote_code breakage, but one is specifically about is_torch_fx_available removal while the other is a broad umbrella report." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Mask shape IndexError and trainer.evaluate() after train() are not the same failure." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Video input ValueError in Qwen3OmniMoe is unrelated to video position-id regression in Qwen2.5-VL." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "OmDet-Turbo image-size mismatch and bfloat16 dtype mismatches are different problems." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Packed-weight loading failure in BitNet is unrelated to bfloat16 dtype mismatch bugs." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "A misleading runtime error and stale documentation references are related thematically but not the same artifact." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": true, + "reason": "Both describe tokenizer behavior divergence around added tokens / spacing, with the slow-vs-fast inconsistency matching the extra-space symptom." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "FSDP CPU RAM loading and init_empty_weights/_is_hf_initialized argument handling are different initialization/loading bugs." + }, + { + "left": "issue:43645", + "right": "issue:46665", + "accept": false, + "reason": "Jupyter custom-model initialization failure is not the same as from_pretrained-in-class inconsistent behavior." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype handling and Apple Silicon caching allocator warmup are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge decisions stay conservative overall. The one accepted pair is plausible from the titles, and the rejections do not overclaim duplicate status." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 19, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:36331", + "issue:42175", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43582", + "issue:43638", + "issue:43645", + "issue:43653", + "issue:43697", + "issue:43742", + "issue:43782", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44060", + "issue:44368", + "issue:44554", + "issue:44589", + "issue:44671", + "issue:44857", + "issue:44912", + "issue:44936", + "issue:44991", + "issue:45003", + "issue:45020", + "issue:45042", + "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", + "issue:46697", + "issue:46728" + ], + "soft_pairs": [ + "issue:44671|issue:45701", + "issue:42175|issue:45081", + "issue:45987|issue:46192", + "issue:34634|issue:36331", + "issue:45003|issue:46285", + "issue:46697|issue:46728", + "issue:43653|issue:44912", + "issue:43262|issue:43819", + "issue:43582|issue:44857", + "issue:43782|issue:45325", + "issue:45081|issue:45406", + "issue:44589|issue:45042", + "issue:45372|issue:46326", + "issue:43742|issue:45020", + "issue:43550|issue:44912", + "issue:45084|issue:45538", + "issue:43329|issue:45600", + "issue:43526|issue:43697", + "issue:44060|issue:44368", + "issue:43452|issue:44991", + "issue:43526|issue:44554", + "issue:43257|issue:43329", + "issue:43906|issue:46209", + "issue:36010|issue:36331", + "issue:43819|issue:43927", + "issue:36010|issue:43645", + "issue:43531|issue:45042", + "issue:43638|issue:44936" + ], + "prepared_review_unit_hash": "a6cde3af40de2c8a143dbad6ca6ab94e5c6b605b2d4063bdb4b298e634706937", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36422, + "estimated_input_tokens": 9106, + "estimated_eval_tokens": 18468 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is mostly heterogeneous, and the soft edges are broad-topic false positives rather than true duplicates. I would not merge any of these pairs; no single issue/PR stands out as a valid canonical representative for the whole group.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "Different bugs: one is a CamemBERT masked-LM regression, the other is a general tokenization change across versions." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Unrelated subsystems: pip backend dependency packaging vs a Mistral tokenizer regex patch crash." + }, + { + "left": "issue:45987", + "right": "issue:46192", + "accept": false, + "reason": "Different API surfaces: cache sequence length return type vs chat-template kwargs handling." + }, + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Unrelated models/code paths: Bark voice preset vs trainer compute_loss signature regression." + }, + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "Both touch model internals, but the bugs are different: sys.modules access vs CLIP text model structure inconsistency." + }, + { + "left": "issue:46697", + "right": "issue:46728", + "accept": false, + "reason": "Both are image-processor issues, but they affect different models and different counting/padding logic." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "Tokenizer special-token registration bug is unrelated to MXFP4 quantization loading fallback." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sampling-rate default bug is unrelated to DAC latent/forward mismatch." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "Different platform and failure mode: Apple Silicon allocator TypeError vs CUDA AMP loss crash." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Both are Qwen VL-related, but one is a load-time weight_only error and the other is a rope/indexing bug." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Similar processor/tokenizer loading failures, but different models and different missing attributes." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Completely unrelated: Float8 storage lookup error vs PIL backend torchvision dependency regression." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Both involve mistral_common, but one is an import break in Gemma 4 loading and the other is a continue_final_message tokenizer bug." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "KeyError on a specific model is not the same underlying issue as remote_code breakage in newer versions." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Different concrete failures: torch.compile+SDPA on Bamba vs MXFP4 quantization fallback on git-oss-20b." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Different functionality entirely: template compilation error vs CLIPTokenizer max length constant." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Both affect model plumbing, but one is multimodal token counting and the other is stale auto-mapping references." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "Vision-model issues, but not the same bug: label reduction in BEiT fast processor vs RTDetrV2 output drift." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention Qwen tied-weights warnings, but one is an incorrect tying bug and the other is a user-facing warning in a different model/version." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer loading failures, but they concern different tokenizer configs and different failure paths." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "Unrelated vision processor label handling vs an MPS attention correctness issue." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Different Qwen3-MoE problems: weight conversion/loading under accelerate+deepspeed vs undefined multimodal-token helper usage." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "No shared bug: isolated reproduction of an older issue vs FP8 inference failure on B200." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "Different regressions: missing GenerationMixin import vs trainer compute_loss signature change." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "Different model/config bugs: DAC latent reconstruction mismatch vs DiaConfig losing custom token IDs after save/load." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "ImportError for GenerationMixin is unrelated to notebook custom-model initialization failures." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Different code paths: Qwen3-MoE sliding-window logic vs PIL backend processor dependency handling." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Training-time Zero3/empty-tensor index crash is not the same as evaluate-after-train state 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": 24, + "review_unit_index": 20, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28282", + "issue:29159", + "issue:29870", + "issue:34567", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:42175", + "issue:42490", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43450", + "issue:43540", + "issue:43606", + "issue:43643", + "issue:43650", + "issue:43696", + "issue:43720", + "issue:43742", + "issue:43746", + "issue:43749", + "issue:43854", + "issue:43906", + "issue:43927", + "issue:43931", + "issue:44291", + "issue:44387", + "issue:44479", + "issue:44610", + "issue:44617", + "issue:44756", + "issue:44811", + "issue:44821", + "issue:44857", + "issue:44912", + "issue:44977", + "issue:45042", + "issue:45459", + "issue:45759", + "issue:45803", + "issue:45859", + "issue:46291", + "issue:46421", + "issue:46424", + "issue:46701", + "issue:46832" + ], + "soft_pairs": [ + "issue:42175|issue:46291", + "issue:43720|issue:44610", + "issue:42490|issue:43257", + "issue:39692|issue:44912", + "issue:38175|issue:44610", + "issue:43746|issue:44811", + "issue:43643|issue:45759", + "issue:43854|issue:44977", + "issue:44387|issue:44617", + "issue:43257|issue:43696", + "issue:43262|issue:44811", + "issue:46421|issue:46424", + "issue:43257|issue:43931", + "issue:43329|issue:44479", + "issue:42915|issue:43742", + "issue:45459|issue:45803", + "issue:43329|issue:43450", + "issue:43749|issue:44756", + "issue:43650|issue:43906", + "issue:44291|issue:45042", + "issue:42915|issue:44821", + "issue:34567|issue:40990", + "issue:43606|issue:45859", + "issue:43540|issue:46701", + "issue:44857|issue:46832", + "issue:43927|issue:44912", + "issue:29159|issue:34567", + "issue:28282|issue:29870" + ], + "prepared_review_unit_hash": "9b946c827a9e9da402f9d31288b3665049576879820e5614ca18dfe3dc468ad2", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36200, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: loading/configuration, quantization, multimodal/video, tokenizer, and training-state issues all mixed together. I don\u2019t see any soft pair that clearly describes the same underlying bug or change.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 43749 is the broadest infrastructure/loading bug in the set and has the strongest cluster-central feel (high discussion and inbound references), so it is the least narrow anchor even though the cluster is not a true duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43749 is the best overall representative of the cluster\u2019s loading/runtime-failure theme; it is broader and more central than the more model-specific or narrowly scoped reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "One is about missing TensorFlow backend support from a torch-only install; the other is an import failure caused by the kernels package version/revision requirement. Different dependency problems." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "Accelerate loading with packed weights is unrelated to a processor/model input-size mismatch in OmDet-Turbo." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "A trailing-slash local-path import warning is not the same bug as Qwen3 MoE weights failing to convert under accelerate + deepspeed." + }, + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Documentation example errors and an MXFP4 quantization fallback bug are different problems in different code paths." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "Unexpected zero probabilities in SigLIP2 is unrelated to OmDet-Turbo\u2019s processor/output-size mismatch." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "Local checkpoint loading with PEFT adapters is a different failure mode from Whisper batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Both involve config loading, but one loses fields with trust_remote_code while the other fails to unwrap text_config for composite multimodal configs; not the same bug." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "A unit-test model load failure for GLM-4.7 is not the same as a flash-attention generation regression in Qwen3.5." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Int4 CUDA memory growth under Transformers 5.x is a different issue from Sam3Video\u2019s out-of-memory failure." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3 MOE weight-conversion failure during loading is unrelated to a plain CUDA OOM on GPT-oss-20b." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Defaulting audio chat templates to 16kHz is a different processor bug from Whisper batch_decode skipping special-token handling." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Both are cache-related, but one is about lazy initialization causing recompilation and the other about max_cache_length only applying to Quantized Cache; not the same concrete bug." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Accelerate/deepspeed weight conversion failure and Qwen3-VL weight-shape mismatch are distinct model-loading errors." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "Both touch video regressions, but one is a specific undefined-helper bug in multimodal token counting while the other is a broader v5.3.0 generation regression across several Qwen video models." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and a MobileLLM key error are unrelated loading bugs." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both are exception-handling problems, but they affect different functions and different failure modes: protobuf decode errors vs bare except swallowing interrupts in Fuyu tensor conversion." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined variables in video token counting is not the same as batched video processor shape corruption." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "FSDP CPU RAM-efficient loading and disabling mmap on Strix Halo are different memory/loading issues with different root causes." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "One item is just a placeholder/nonsense title, the other is an isolated reproduction report; they are not the same bug." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "The unexpected _is_hf_initialized argument error is unrelated to PIL image processors incorrectly requiring torchvision." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "A Qwen3Moe FP8 configuration bug is unrelated to failing to load AutoImageProcessor from a URL." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counting and GPT-oss perplexity are different issues in different subsystems." + }, + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both mention CPU offload, but they involve different models and different code paths: device mismatch in bark-small versus rotary embedding incompatibility in Qwen3_5Moe." + }, + { + "left": "issue:43540", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3OmniMoe video input validation failure and DiffusionGemma pipeline failure are unrelated." + }, + { + "left": "issue:44857", + "right": "issue:46832", + "accept": false, + "reason": "A float16 AMP crash in LwDetrImageLoss is not the same as an RTDetr forward TypeError when feature-level counts do not match backbone outputs." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig save/load token-ID loss and MXFP4 quantization fallback are separate model/config bugs." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistencies are unrelated to TrainerState\u2019s token counter not updating." + }, + { + "left": "issue:28282", + "right": "issue:29870", + "accept": false, + "reason": "Missing PyTorch during AutoModel import is a dependency issue, while the Bart generate crash is an unbound decoder_attention_mask 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": 24, + "review_unit_index": 21, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:24643", + "issue:27957", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43317", + "issue:43404", + "issue:43425", + "issue:43452", + "issue:43454", + "issue:43525", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43742", + "issue:43749", + "issue:43824", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44560", + "issue:44568", + "issue:44821", + "issue:44849", + "issue:44936", + "issue:44987", + "issue:44991", + "issue:45127", + "issue:45137", + "issue:45356", + "issue:45357", + "issue:45405", + "issue:45488", + "issue:45538", + "issue:45701", + "issue:45907", + "issue:46291", + "issue:46512", + "issue:46693", + "issue:46821" + ], + "soft_pairs": [ + "issue:43525|issue:46291", + "issue:43454|issue:43577", + "issue:44568|issue:45356", + "issue:44987|issue:45538", + "issue:28282|issue:34689", + "issue:43425|issue:46291", + "issue:42915|issue:45538", + "issue:43697|issue:45357", + "issue:33357|issue:34567", + "issue:43404|issue:45127", + "issue:42175|issue:43824", + "issue:45137|issue:45405", + "issue:28282|issue:29127", + "issue:43742|issue:44291", + "issue:43317|issue:45538", + "issue:44936|issue:46821", + "issue:44479|issue:46693", + "issue:44062|issue:44991", + "issue:44849|issue:45701", + "issue:43577|issue:46512", + "issue:42915|issue:43749", + "issue:43452|issue:43582", + "issue:41720|issue:43540", + "issue:27957|issue:28282", + "issue:24643|issue:29159", + "issue:43540|issue:44560", + "issue:44987|issue:45907", + "issue:42175|issue:44488", + "issue:43540|issue:44479", + "issue:44291|issue:44821", + "issue:44568|issue:45488" + ], + "prepared_review_unit_hash": "190711a3fb18f07370d5a33f50dc3a759722f0cf752b29dd7b003c4160bdcd6b", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36968, + "estimated_input_tokens": 9242, + "estimated_eval_tokens": 18740 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are a heterogeneous mix of unrelated bug reports; the soft pairs are mostly only loosely similar by subsystem or symptom, not the same underlying fix. I found no duplicate pair that is strong enough to merge, and there are no PRs in the set.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Different failures: a missing `pad_token_id` config attribute versus a kernels package API/version requirement." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "Both touch model loading internals, but one is about untied `lm_head` weights while the other is a dtype-casting bug in BLIP2." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Tokenizer-related, but different concrete regressions and model families; not the same bug path." + }, + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "A model load failure for one repo is not the same as CLIP tokenizer `model_max_length` behavior." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "One is missing PyTorch in the environment; the other is a specific model-loading regression in a newer Transformers release." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "Torch compatibility and kernels revision forwarding are unrelated dependency issues." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "FineGrainedFP8/Qwen3Moe loading is unrelated to CLIP tokenizer length settings." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "Different models and different bugs: output differences in RTDetrV2 versus incorrect save_pretrained keys in Qwen3.5." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "MacOS bus error during CLIP model load is unrelated to TrainerState token counting." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both involve tied weights, but the concrete problems differ: missing tie in Mistral3 versus LoRA merge/extended-vocab collapse." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Packaging/backend inclusion issue versus missing exported Qwen2_5_VL class import." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error and a PEFT minimum-version bump are unrelated." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "Environment import failure versus LayoutLMv3 error-message clarity." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "A key error while loading one model is not the same as the `_is_hf_initialized` argument regression." + }, + { + "left": "issue:43317", + "right": "issue:45538", + "accept": false, + "reason": "Device-map/offload loading failure is unrelated to CLIP tokenizer max-length behavior." + }, + { + "left": "issue:44936", + "right": "issue:46821", + "accept": false, + "reason": "Trainer lifecycle bug versus a TimesFM `window_size` attribute error." + }, + { + "left": "issue:44479", + "right": "issue:46693", + "accept": false, + "reason": "Both are regressions, but one is a video-input bug and the other is a long-context FA2 performance regression." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer load failures, but they appear to stem from different concrete code paths and model-specific issues." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "One is a Qwen3.5 hidden-states bug; the other is a broader tokenization-version change." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "Model weight dtype mismatch and stale `config.dtype` are related themes but not the same underlying defect." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "Qwen3Moe FP8 loading and FSDP CPU RAM efficient loading are separate training/loading problems." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "gguf/AutoModel loading behavior is unrelated to Apple Silicon allocator warmup type handling." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Different model families and failures: CUDA assert during device mapping versus video input processing error." + }, + { + "left": "issue:27957", + "right": "issue:28282", + "accept": false, + "reason": "Flash Attention 2 on XLM-Roberta is unrelated to the PyTorch import error." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "DeepSpeed weight-shape runtime error and tokenizer inconsistency are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Both mention video inputs, but the model families and failure modes differ enough that they do not look like one fix." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "Model loading failure versus Hugging Face Hub revision forwarding in `list_repo_files`." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Backend install packaging issue versus a model-specific loading failure." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both involve tokenizer regressions, but one is about `add_special_tokens` behavior and the other about a tokenizer.json pre-tokenizer override." + } + ] + }, + "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": 24, + "review_unit_index": 22, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:24643", + "issue:29870", + "issue:34567", + "issue:36331", + "issue:38175", + "issue:43257", + "issue:43452", + "issue:43526", + "issue:43531", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43716", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:44062", + "issue:44479", + "issue:44488", + "issue:44568", + "issue:44589", + "issue:44625", + "issue:44737", + "issue:44792", + "issue:44821", + "issue:44877", + "issue:44898", + "issue:44964", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45479", + "issue:45588", + "issue:45600", + "issue:45823", + "issue:45920", + "issue:46399", + "issue:46463", + "issue:46468", + "issue:46491", + "issue:46693", + "issue:46701", + "issue:46821", + "issue:46829" + ], + "soft_pairs": [ + "issue:38175|issue:44898", + "issue:44479|issue:45381", + "issue:44792|issue:44964", + "issue:43257|issue:43756", + "issue:34567|issue:36331", + "issue:44062|issue:45081", + "issue:43582|issue:46821", + "issue:43742|issue:44589", + "issue:45042|issue:45081", + "issue:43716|issue:46701", + "issue:44737|issue:46463", + "issue:43756|issue:46399", + "issue:43756|issue:45325", + "issue:44568|issue:46491", + "issue:43531|issue:44821", + "issue:43906|issue:46468", + "issue:45405|issue:45600", + "issue:43577|issue:45072", + "issue:45823|issue:46463", + "issue:24643|issue:29870", + "issue:43906|issue:46693", + "issue:43697|issue:46829", + "issue:43452|issue:44488", + "issue:44737|issue:45588", + "issue:43526|issue:45479", + "issue:44625|issue:44877", + "issue:45357|issue:45920" + ], + "prepared_review_unit_hash": "7b45b69a4ccffd62ab991a9de362f7c821df4c2df03e3d5fdf01c8b68cc73169", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36162, + "estimated_input_tokens": 9041, + "estimated_eval_tokens": 18338 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are distinct model-specific regressions or unrelated loader/trainer bugs, so none of the soft edges are true duplicates. The best issue representative is the broad, high-impact tokenizer regression in issue 45920.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45920 is the broadest, most central regression here: it impacts AutoTokenizer across multiple model families, has clear reproduction, and reflects the kind of high-signal user-facing breakage the cluster otherwise lacks.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 has the strongest combination of breadth, impact, and documentation quality among the issues in this noisy cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Both are image-model inference bugs, but they hit different architectures and different failure modes: SigLIP2 zero probabilities vs Perceiver non-default-resolution failure." + }, + { + "left": "issue:44479", + "right": "issue:45381", + "accept": false, + "reason": "Same general Qwen VL/video area, but not the same concrete bug: one is a broad v5.3.0 regression across several models, the other is a specific qwen2.5-vl vision_position_ids issue." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Different model families and paths: Janus image generation test failure versus Phi-4 multimodal loading failure." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Unrelated bugs: Qwen3 MoE weight conversion with accelerate/deepspeed versus SmolLM3 RoPE-layer configuration mismatch." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Trainer token-count tracking not updating is unrelated to CustomTrainer.compute_loss() receiving an unexpected keyword argument." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer AddedToken keyword handling and Mistral regex patching fail in different code paths with different root causes." + }, + { + "left": "issue:43582", + "right": "issue:46821", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and TimesFM window_size AttributeError are unrelated runtime issues." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Loading facebook/MobileLLM-125M key errors and Float8 storage lookup failures are distinct loader/runtime problems." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "One is an image-processor torchvision dependency regression; the other is a tokenizer regex patch crash." + }, + { + "left": "issue:43716", + "right": "issue:46701", + "accept": false, + "reason": "Different symptoms and code paths: BLIP2 dtype mismatch versus DiffusionGemma image-text-to-text pipeline failure." + }, + { + "left": "issue:44737", + "right": "issue:46463", + "accept": false, + "reason": "XLNet CPU tensor creation bug and T5 decoder-input validation error message are not the same underlying issue." + }, + { + "left": "issue:43756", + "right": "issue:46399", + "accept": false, + "reason": "SmolLM3 RoPE-layer loss and Qwen3.5 kernel-registration crash are unrelated model implementation bugs." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Both involve Qwen-family rotary/position behavior, but one is about SmolLM3 layer dropping and the other is a Qwen2.5-VL still-image position-id scaling bug." + }, + { + "left": "issue:44568", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but one is BOS/EOS insertion behavior and the other is code-lama leading-space round-trip loss." + }, + { + "left": "issue:43531", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior and AutoImageProcessor URL loading are unrelated." + }, + { + "left": "issue:43906", + "right": "issue:46468", + "accept": false, + "reason": "An isolated reproduction of an old issue and a broken Nemotron example are not the same concrete defect." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "PEFT version constraint problems are unrelated to stale auto_mappings references to removed configs." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both mention dtype, but they affect different model families and different inference paths." + }, + { + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Gemma4 device-map mismatch and T5 decoder validation messaging are unrelated bugs." + }, + { + "left": "issue:24643", + "right": "issue:29870", + "accept": false, + "reason": "DeepSpeed training weight-shape error and BART generate() unbound-local mask error are different failures." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "The first is a reproduction of another issue; the second is a performance regression in long-context generation with FA2." + }, + { + "left": "issue:43697", + "right": "issue:46829", + "accept": false, + "reason": "RTDetrV2 output changes in v5 and VideoPrism hidden_states misassignment are distinct model-output bugs." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "Both are load-time problems, but one is a gguf_file integration breakage and the other is a specific model load failure; not the same root cause." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU arange placement and flash_attention sink-less crash are unrelated implementation bugs." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BEiT label reduction and sequence-classification zero-loss behavior are different issues, despite both touching classification preprocessing/training." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3.5 num_labels propagation and strict-config loading failures are separate config-handling bugs." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "Qwen3.5 save_pretrained visual-key corruption and AutoTokenizer wrong token IDs are different code paths 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": 24, + "review_unit_index": 23, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28282", + "issue:29159", + "issue:32090", + "issue:33357", + "issue:38175", + "issue:41762", + "issue:42898", + "issue:43295", + "issue:43452", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43761", + "issue:43819", + "issue:43994", + "issue:44361", + "issue:44442", + "issue:44466", + "issue:44493", + "issue:44568", + "issue:44811", + "issue:44843", + "issue:44898", + "issue:44913", + "issue:44936", + "issue:45005", + "issue:45200", + "issue:45216", + "issue:45356", + "issue:45372", + "issue:45375", + "issue:45542", + "issue:46050", + "issue:46121", + "issue:46170", + "issue:46178", + "issue:46463", + "issue:46491", + "issue:46693", + "issue:46701", + "issue:46829" + ], + "soft_pairs": [ + "issue:43540|issue:46693", + "issue:44913|issue:46121", + "issue:28282|issue:32090", + "issue:43697|issue:45216", + "issue:43720|issue:44898", + "issue:43761|issue:46829", + "issue:43643|issue:45375", + "issue:45200|issue:46463", + "issue:44466|issue:45005", + "issue:41762|issue:44936", + "issue:44361|issue:44442", + "issue:44493|issue:46050", + "issue:43653|issue:44361", + "issue:43819|issue:44811", + "issue:43688|issue:43749", + "issue:43525|issue:46701", + "issue:38175|issue:43295", + "issue:43452|issue:44843", + "issue:43994|issue:44898", + "issue:45372|issue:46170", + "issue:43531|issue:43742", + "issue:45542|issue:46178", + "issue:42898|issue:46491", + "issue:43295|issue:45356", + "issue:29159|issue:33357", + "issue:43653|issue:44568" + ], + "prepared_review_unit_hash": "8ca91472a8b5224c5f02bc28cdfebfbe04398250b1e9427ae695a6514435b4eb", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35916, + "estimated_input_tokens": 8979, + "estimated_eval_tokens": 18214 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most items are unrelated one-off issue reports across different models/subsystems. The only soft pair that plausibly describes the same underlying bug is the v5 tied-weights serialization regression: issue 44466 and issue 45005.", + "confidence": 0.58, + "canonical_issue_reason": "Issue 45005 is the best representative because it is broader and more detailed: it describes the tied-weights regression in v5 across translation models, which subsumes the narrower lm_head.weight/device-specific serialization problem in issue 44466.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45005 is the strongest overall issue candidate for the cluster because it most clearly captures the shared tied-weights serialization bug and appears to be the broader root report.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "Different problems: video input processing error in Qwen3OmniMoe vs a FA2 long-context performance regression." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "Both mention rope-related code, but one is config reload losing rotary_pct while the other is a TypeError in rope validation for JSON-loaded lists." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "Unrelated failures: missing PyTorch import for AutoModel vs a Trainer GPU broadcast TypeError on NoneType." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "Different model/regression bugs: RTDetrV2 output drift vs Qwen3.5 save_pretrained checkpoint corruption." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "Different code paths: BitNet packed-weight unpacking during accelerate loading vs Perceiver positional interpolation on non-default image size." + }, + { + "left": "issue:43761", + "right": "issue:46829", + "accept": false, + "reason": "Both are output-field issues in vision models, but they affect different models and different incorrect fields; not the same underlying bug." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "Both involve config fields, but one is trust_remote_code returning incomplete configs while the other is a specific Qwen vision config field dropped by strict validation." + }, + { + "left": "issue:45200", + "right": "issue:46463", + "accept": false, + "reason": "Unrelated: Gemma 4 text-only token defaults vs an incorrect decoder-input validation message for T5." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": true, + "reason": "Both describe the same tied-weights/save_pretrained regression in v5, centered on lm_head.weight serialization and translation models." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Different failures: DeepSpeed ZeRO-3 loading IndexError for Gemma3 vs trainer.evaluate() failing after training." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Both are tokenizer-loading problems, but they involve different tokenizer classes and distinct failure modes." + }, + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "Both mention unexpected keys/quantization, but one is a generic position-id key issue and the other is sinq quantization save/load corruption." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "Different issues: BigBird mask token registration/empty decode vs MLukeTokenizer AttributeError on tasks." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "Unrelated: DAC latent reconstruction mismatch vs Whisper processor.batch_decode ignoring skip_special_tokens." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Different subsystems: auxiliary-loss normalization in MoE models vs broken FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43525", + "right": "issue:46701", + "accept": false, + "reason": "Different model families and failures: Llama4Config missing pad_token_id vs DiffusionGemma image-text pipeline breakage." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Both are regressions, but one is SigLIP2 zero-probabilities output quality and the other is CLIPVisionModel hidden_states handling." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both involve AutoTokenizer.from_pretrained, but one is gguf_file breakage for tokenizer/model loading and the other is an offline-mode mistral regex hub call." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Different vision-model bugs: nonsensical SigLIP2 outputs vs Perceiver resize/interpolation failure." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Unrelated dependency/import problems: Gemma 4 processor loading needs ReasoningEffort vs Windows ROCm torch.distributed import/type issues." + }, + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Different model-specific bugs: Qwen3-MoE sliding_window behavior vs a key error loading MobileLLM-125M." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Both involve backend detection, but one is tensorboard-without-TensorFlow backend handling and the other is a missing tensorflow_text backend mapping entry." + }, + { + "left": "issue:42898", + "right": "issue:46491", + "accept": false, + "reason": "Related only at a broad tokenizer level; one is a global clean_up_tokenization_spaces semantic change, the other is a CodeLlama leading-space round-trip regression." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Different regressions: CLIPVisionModel hidden_states output vs Kimi-K2.5 tokenizer codec handling and mistral regex warning behavior." + }, + { + "left": "issue:29159", + "right": "issue:33357", + "accept": false, + "reason": "Unrelated: slow-vs-fast tokenizer inconsistency vs a MacOS bus error with a community CLIP model." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both are special-token/tokenizer bugs, but they affect different tokenizers and different behaviors; not the same underlying issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is conservative and grounded in the packet. The only clearly related soft pair is 44466/45005, and the rest of the pairwise rejections are appropriately cautious given the issue titles. 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": 24, + "review_unit_index": 24, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:32090", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:41720", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43452", + "issue:43504", + "issue:43550", + "issue:43577", + "issue:43606", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43720", + "issue:43749", + "issue:43756", + "issue:43782", + "issue:43927", + "issue:44361", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44484", + "issue:44554", + "issue:44561", + "issue:44610", + "issue:44625", + "issue:44704", + "issue:44734", + "issue:44743", + "issue:44821", + "issue:44843", + "issue:44977", + "issue:44991", + "issue:45440", + "issue:45491", + "issue:45701", + "issue:45704", + "issue:46002", + "issue:46050", + "issue:46402", + "issue:46612", + "issue:46693", + "issue:46821" + ], + "soft_pairs": [ + "issue:43646|issue:43749", + "issue:44464|issue:44743", + "issue:43452|issue:43577", + "issue:43756|issue:46402", + "issue:43688|issue:45704", + "issue:43653|issue:43927", + "issue:43688|issue:45701", + "issue:44843|issue:44991", + "issue:44821|issue:46821", + "issue:44479|issue:44625", + "issue:44484|issue:44704", + "issue:45491|issue:46002", + "issue:43550|issue:44361", + "issue:43688|issue:45440", + "issue:33666|issue:36010", + "issue:44734|issue:46612", + "issue:32090|issue:33666", + "issue:44464|issue:44977", + "issue:44466|issue:46050", + "issue:41720|issue:44464", + "issue:44977|issue:46693", + "issue:43504|issue:43550", + "issue:43504|issue:43720", + "issue:43577|issue:44554", + "issue:43643|issue:44561", + "issue:43720|issue:43927", + "issue:33666|issue:36331", + "issue:43504|issue:44610", + "issue:43606|issue:43927", + "issue:43278|issue:43688", + "issue:43782|issue:46693", + "issue:43262|issue:43295" + ], + "prepared_review_unit_hash": "ffd7ee935324fc01dfea8518bfa027c9d84294fe8f1f84b84591f72a6e05cfe3", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37395, + "estimated_input_tokens": 9349, + "estimated_eval_tokens": 18954 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated regressions, loading bugs, and model-specific issues. None of the soft pairs look like the same underlying bug/change, so I would not merge any of them as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "If a single representative is needed, issue 43646 is the broadest cross-cutting regression report in the set, but the cluster is too heterogeneous to justify a true canonical duplicate target.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43646 is the best overall representative because it describes a wide Transformers 5.0 breakage affecting custom model initialization, which is broader than the mostly model-specific reports here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "Different failure domains: custom model initialization vs FSDP CPU-RAM-efficient loading." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Both involve generation/caching, but one is chunked generation with compiled forward and the other is recurrent-state reset in modular Qwen3.5." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "Tokenizer/model loading via gguf_file is unrelated to Blip2 dtype propagation." + }, + { + "left": "issue:43756", + "right": "issue:46402", + "accept": false, + "reason": "Different models and bugs: RoPE layer dropping vs missing text encoder weights on load." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Aux-loss normalization bug is unrelated to T5 using apex RMSNorm and leaking memory." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Tokenizer special-token registration issue is not the same as config token IDs being lost on save/load." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux-loss normalization is unrelated to tokenizer behavior changes across versions." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Offline hub call in tokenizer patching is different from a model-specific tokenizer load failure." + }, + { + "left": "issue:44821", + "right": "issue:46821", + "accept": false, + "reason": "Image processor URL loading and TimesFM window_size AttributeError are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Video-input regression in Qwen variants is different from num_labels propagation in Qwen3.5 config." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to AutoProcessor kwargs forwarding." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Both mention NaNs, but one is Gemma3 sliding-window attention and the other is classification-head init/checkpointing." + }, + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "torch.compile/SDPA failure is unrelated to MLukeTokenizer AttributeError." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Aux-loss normalization and DeepseekV3MoE implementation divergence are different problems." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU Qwen2-VL training and GenerationMixin import failure are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46612", + "accept": false, + "reason": "Both involve generation internals, but the bugs are different: KV-cache indexing crash vs skipped beam-search cache reorder." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Generic Trainer broadcast error is not the same as Qwen2-VL multi-GPU training behavior." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency is distinct from Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Device-dependent lm_head serialization is not the same as sinq quantization save/load breakage." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "cudaErrorAssert on auto device mapping is unrelated to compiled-forward generation drift." + }, + { + "left": "issue:44977", + "right": "issue:46693", + "accept": false, + "reason": "Correctness failure in Qwen3.5 flash-attention is not the same as a performance regression in FA2 long-context generation." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Legacy-field model loading bug is unrelated to Bamba torch.compile SDPA failure." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Legacy-field load failure is different from packed-weights handling during accelerate loading." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "Blip2 dtype propagation and MPS attention correctness are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "Missing fields from trust_remote_code config loading is different from removal of is_torch_fx_available breaking remote-code models." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Accelerate packed-weight loading and DiaConfig token-ID persistence are unrelated." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "Qwen2-VL multi-GPU training and compute_loss signature mismatch are different trainer issues." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Beit preset loading bug is unrelated to OmDet-Turbo processor/model input-size mismatch." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU-offload device mismatch does not match DiaConfig token-ID loss." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Embedding dtype drift between train/eval is unrelated to auxiliary-loss normalization." + }, + { + "left": "issue:43782", + "right": "issue:46693", + "accept": false, + "reason": "weight_only loading error for Qwen3VL is unrelated to FA2 performance regression." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Audio sampling-rate default bug and processor.tokenizer regression are different processor 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": 24, + "review_unit_index": 25, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28218", + "issue:33357", + "issue:33666", + "issue:39692", + "issue:42796", + "issue:43278", + "issue:43295", + "issue:43452", + "issue:43493", + "issue:43525", + "issue:43572", + "issue:43646", + "issue:43701", + "issue:43720", + "issue:43824", + "issue:43825", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44361", + "issue:44451", + "issue:44466", + "issue:44488", + "issue:44514", + "issue:44554", + "issue:44568", + "issue:44589", + "issue:44625", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:44964", + "issue:44977", + "issue:45072", + "issue:45290", + "issue:45357", + "issue:45405", + "issue:45584", + "issue:45812", + "issue:45823", + "issue:46050", + "issue:46192", + "issue:46209", + "issue:46365", + "issue:46489", + "issue:46710", + "issue:46821", + "issue:46832" + ], + "soft_pairs": [ + "issue:39692|issue:43493", + "issue:43720|issue:44361", + "issue:44514|issue:45290", + "issue:44964|issue:45405", + "issue:44554|issue:46365", + "issue:43525|issue:43572", + "issue:43493|issue:43825", + "issue:44977|issue:46209", + "issue:44625|issue:45584", + "issue:43452|issue:44451", + "issue:45823|issue:46050", + "issue:28218|issue:33357", + "issue:43994|issue:46192", + "issue:42796|issue:44977", + "issue:44849|issue:46209", + "issue:43278|issue:43844", + "issue:39692|issue:43295", + "issue:43493|issue:43844", + "issue:33357|issue:33666", + "issue:43646|issue:44466", + "issue:46209|issue:46710", + "issue:44568|issue:46489", + "issue:44849|issue:45357", + "issue:43278|issue:43701", + "issue:44792|issue:45405", + "issue:43927|issue:44361", + "issue:45072|issue:46832", + "issue:43824|issue:44589", + "issue:46821|issue:46832", + "issue:44568|issue:45812", + "issue:44488|issue:44843" + ], + "prepared_review_unit_hash": "bfdc4d8c5d4a412964245fdae1e741a36a53b7e2e54ecdc47b5177dd9e56224b", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37067, + "estimated_input_tokens": 9267, + "estimated_eval_tokens": 18790 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: it contains several unrelated tokenizer/config/load regressions, mostly around Transformers v5+, but none of the proposed soft pairs look like the same underlying bug. I would use the Granite tokenizer regression as the closest representative, but with low confidence.", + "confidence": 0.66, + "canonical_issue_reason": "Issue 45812 is the most concrete and central tokenizer regression here: a silent v4\u2192v5 change causing wrong token IDs across Granite models, which matches the tokenizer-focused theme better than the broader or model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "45812 is the best anchor for this cluster because it is broad, user-visible, and directly about tokenizer output correctness rather than a one-off model-specific failure.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Different problems: documentation/example errors and quantization failure vs an implementation discrepancy with JAX." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "Unrelated subsystems: BitNet accelerate-loading packed weights vs an MLukeTokenizer AttributeError." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both involve apply_chat_template crashes, but the triggers and code paths differ (batched padding=False vs tool-call assistant messages without content)." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Model load failure vs dependency version bump; not the same underlying bug." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "MPS attention correctness issue and Trainer import failure are unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Same general class of missing config attributes after v5, but different models, attributes, and failure modes." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and a pipeline error-message regression are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/precision failures: flash-attention generation vs FP8 inference on B200." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Different features and symptoms: num_labels propagation vs Whisper empty transcription generation failure." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "Both are load-time failures, but they affect different models and likely different root causes." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Different bugs: device-map mismatch in Gemma4 forward vs sinq save/load quantization breakage." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer spacing bug and a MacOS bus error in CLIP are not the same issue." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "Wrong SigLIP2 outputs and silently ignored return_tensors are unrelated." + }, + { + "left": "issue:42796", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation failure are different bugs." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 hidden-state output bug is unrelated to FP8 inference on B200." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Evaluate-time dtype change and ZeRO-3 gradient blow-up are not the same root cause." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Different regressions: SigLIP2 doc example errors vs processor.tokenizer/custom image-tokenizer access breakage." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "Different model/runtime issues: SigLIP2 discrepancy vs ZeRO-3 gradient anomaly." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS bus error in CLIP and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Generic custom model initialization breakage vs lm_head.weight serialization inconsistency are distinct problems." + }, + { + "left": "issue:46209", + "right": "issue:46710", + "accept": false, + "reason": "FP8 inference failure and an accuracy regression after upgrade are not the same bug." + }, + { + "left": "issue:44568", + "right": "issue:46489", + "accept": false, + "reason": "Different tokenizer regressions affecting different models and behaviors; not a shared code-path bug." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Hidden-states bug and visual-encoder key serialization bug are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Dtype mismatch in eval vs resume_from_checkpoint key mismatch are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image-generation test failure and a PEFT version bump are unrelated." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "DiaConfig token-id persistence bug and MLukeTokenizer AttributeError are different failures." + }, + { + "left": "issue:45072", + "right": "issue:46832", + "accept": false, + "reason": "BF16 dtype mismatch issues are unrelated to RT-DETR feature-levels shape/type error." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "ImportError for a model class and missing Float8 storage type are unrelated." + }, + { + "left": "issue:46821", + "right": "issue:46832", + "accept": false, + "reason": "TimesFM window_size AttributeError and RT-DETR forward TypeError are distinct bugs." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different tokenizers and failure modes; not the same underlying bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 26, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:29942", + "issue:30333", + "issue:32090", + "issue:34567", + "issue:42490", + "issue:43257", + "issue:43299", + "issue:43334", + "issue:43504", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43653", + "issue:43696", + "issue:43844", + "issue:43873", + "issue:43937", + "issue:43994", + "issue:44112", + "issue:44291", + "issue:44464", + "issue:44493", + "issue:44589", + "issue:44869", + "issue:44964", + "issue:44977", + "issue:45084", + "issue:45200", + "issue:45229", + "issue:45245", + "issue:45405", + "issue:45479", + "issue:45520", + "issue:45584", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46050", + "issue:46285", + "issue:46424", + "issue:46468", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46829", + "issue:46832" + ], + "soft_pairs": [ + "issue:44589|issue:45520", + "issue:43575|issue:45405", + "issue:43257|issue:43299", + "issue:43504|issue:43606", + "issue:43606|issue:44112", + "issue:43334|issue:45759", + "issue:45245|issue:46710", + "issue:43334|issue:43572", + "issue:43572|issue:44291", + "issue:44493|issue:46285", + "issue:44464|issue:46424", + "issue:43504|issue:43653", + "issue:43937|issue:46701", + "issue:44464|issue:45584", + "issue:45200|issue:45229", + "issue:46468|issue:46832", + "issue:43844|issue:44964", + "issue:44869|issue:46752", + "issue:45229|issue:45823", + "issue:43873|issue:46050", + "issue:29942|issue:30333", + "issue:43994|issue:44493", + "issue:45588|issue:46050", + "issue:29159|issue:32090", + "issue:42490|issue:43696", + "issue:44977|issue:46701", + "issue:45084|issue:45520", + "issue:28218|issue:34567", + "issue:45479|issue:46829" + ], + "prepared_review_unit_hash": "9168df70f907084be56dd0c4d501da7185fb4c32fa6041ce973c2f6b4e2262a8", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36410, + "estimated_input_tokens": 9103, + "estimated_eval_tokens": 18462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft edge looks like a true duplicate; the cluster is a broad false-positive mix of tokenizer, loading/config, quantization, and model-specific regression reports. I\u2019d treat it as heterogeneous rather than one underlying bug cluster.", + "confidence": 0.87, + "canonical_issue_reason": "If a representative issue is needed, #44291 is the broadest loader-regression report in the set (generic model loading failure under transformers >= 5.0), but this cluster is too heterogeneous to treat as a clean duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "#44291 is the best overall issue representative because it captures a concrete, broadly applicable loading regression; still, it is not a true canonical anchor for the whole mixed cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Different failures: Float8 storage lookup vs flash_attn import_utils on Python 3.13; unrelated code paths." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "OOM on Qwen2 tensor-parallel load vs a PEFT version bump; not the same bug." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both are Qwen-family loading regressions, but one is accelerate/deepspeed weight conversion and the other is Qwen3VL MOE config loading; different concrete paths." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field load failure vs Bark CPU-offload device mismatch; unrelated." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "One is a real model/offload mismatch, the other a stale GraniteSpeech CI test; same wording, different underlying issue." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Both touch multimodal Qwen config handling, but pad_token_id missing and text_config unwrapping are distinct loader bugs." + }, + { + "left": "issue:45245", + "right": "issue:46710", + "accept": false, + "reason": "Runtime category-limit error vs accuracy regression after upgrade; unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Both mention missing token fields after 5.0, but they affect different configs/models and different attributes." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Both are 5.0-era loading regressions, but StableLm config metadata and init_empty_weights arg handling are different code paths." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "Unexpected position-id key vs CLIP text-model structure change; unrelated." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Compiled-forward chunked generation inconsistency vs cache max-length semantics; not the same bug." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "BEiT legacy preset loading vs BigBirdTokenizer special-token registration; unrelated." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "GenerationConfig validation failure vs image-text pipeline output issue; different problems." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward generation inconsistency vs Whisper empty-transcription failure; unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Same model family, but one is input defaulting for fine-tuning and the other is multi-GPU inference OOM." + }, + { + "left": "issue:46468", + "right": "issue:46832", + "accept": false, + "reason": "Nemotron example breakage vs RT-DETR forward type error; unrelated." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient behavior vs Phi-4 multimodal loading failure; unrelated." + }, + { + "left": "issue:44869", + "right": "issue:46752", + "accept": false, + "reason": "Both are input-edge-case crashes, but one is Whisper timestamp decoding and the other is empty chat-template handling; different APIs." + }, + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4 inference issues, but OOM and device-map mismatch are different failure modes." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Quantization/offloading theme overlaps, but offload behavior and sinq save/load persistence are distinct bugs." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "FlashAttention CI failures vs MLflow reporting on failed training; unrelated." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "Siglip2 nonsensical outputs vs unexpected position-id key regression; unrelated." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "Flash-attention sink-less-model crash vs sinq quantized model save/load breakage; unrelated." + }, + { + "left": "issue:29159", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer fast/slow inconsistency vs GPU broadcast TypeError; unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "Trailing-slash import warning vs GPT-oss OOM; unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3.5 flash-attention generation issue vs DiffusionGemma image-text pipeline bug; unrelated." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Template compiler error vs flash_attn import KeyError on Python 3.13; unrelated." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer adds extra space vs TrainerState token counter not updating; unrelated." + }, + { + "left": "issue:45479", + "right": "issue:46829", + "accept": false, + "reason": "Degenerate zero-loss classification vs VideoPrism output field mix-up; 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": 24, + "review_unit_index": 27, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:15354", + "issue:29870", + "issue:29942", + "issue:37428", + "issue:38175", + "issue:39692", + "issue:41720", + "issue:42490", + "issue:43065", + "issue:43232", + "issue:43295", + "issue:43450", + "issue:43493", + "issue:43540", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43653", + "issue:43701", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43957", + "issue:44112", + "issue:44186", + "issue:44423", + "issue:44554", + "issue:44610", + "issue:44734", + "issue:44849", + "issue:44977", + "issue:44987", + "issue:45072", + "issue:45198", + "issue:45200", + "issue:45356", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46121", + "issue:46178", + "issue:46192", + "issue:46369", + "issue:46710", + "issue:46821" + ], + "soft_pairs": [ + "issue:43065|issue:46369", + "issue:44423|issue:44734", + "issue:44554|issue:45446", + "issue:15354|issue:38175", + "issue:43493|issue:43701", + "issue:41720|issue:46710", + "issue:45406|issue:46121", + "issue:44977|issue:46710", + "issue:45072|issue:45198", + "issue:37428|issue:43957", + "issue:43295|issue:43450", + "issue:45588|issue:45759", + "issue:43540|issue:45200", + "issue:44186|issue:46192", + "issue:43295|issue:43873", + "issue:43572|issue:46821", + "issue:45200|issue:45823", + "issue:45823|issue:46192", + "issue:43295|issue:44610", + "issue:29870|issue:29942", + "issue:45520|issue:46178", + "issue:43572|issue:44987", + "issue:43232|issue:44849", + "issue:43550|issue:43844", + "issue:43450|issue:43575", + "issue:44554|issue:45356", + "issue:37428|issue:43824", + "issue:42490|issue:43493", + "issue:39692|issue:46192", + "issue:38175|issue:43493", + "issue:43653|issue:44112" + ], + "prepared_review_unit_hash": "4237fb223061a2fdc125d614f7d83e8b266b5890eaab1fc5f5c3a97e05348241", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37226, + "estimated_input_tokens": 9307, + "estimated_eval_tokens": 18870 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The candidate pairs are mostly broad similarity matches by subsystem or model family, but they describe different bugs, regressions, or code paths. I don\u2019t see any true duplicate pairs here, so no soft edge should be accepted.", + "confidence": 0.92, + "canonical_issue_reason": "No clear canonical issue: the soft pairs are all false positives with different triggers, symptoms, or code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must be singled out, issue:45200 is the most concrete and actionable standalone report, but it does not act as a true duplicate hub for the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3-related, but one is about a dummy Conv2d in the decoder and the other is memory overflow in a litetext model." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both hit transformers serve, but one crashes on multimodal continuous batching and the other on KV-cache continuation indexing." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "Different problems: MPS attention correctness vs a PyTorch version check for AuxRequest import in flex_attention." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "Unrelated failures: TorchScript export of a ViT model vs zero probabilities in SigLIP2 inference." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 model discrepancy and a checkpoint resume key mismatch are unrelated bugs." + }, + { + "left": "issue:41720", + "right": "issue:46710", + "accept": false, + "reason": "Different failure modes: CUDA assert with auto device mapping vs accuracy regression after a version upgrade." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Gemma4 processor/tokenizer crash and rope-parameter JSON list handling are separate code paths." + }, + { + "left": "issue:44977", + "right": "issue:46710", + "accept": false, + "reason": "Flash-attention generation bug and a broad accuracy regression are not the same defect." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "One is a dtype mismatch in bfloat16 inference; the other is Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Import error for a flash-attention helper and meta-device loading breakage are distinct issues." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer access in custom code and batched video processor shape mismatch are different regressions." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Flash-attention sink handling and Qwen multimodal config unwrapping do not describe the same bug." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video-input processing in Qwen3OmniMoe is unrelated to Gemma4 mm_token_type_ids defaults." + }, + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "Tokenizer crashes on NER/padding vs return_tensors being ignored are separate tokenizer behaviors." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Custom processor-tokenizer regression and quantization offloading failure are not the same underlying problem." + }, + { + "left": "issue:43572", + "right": "issue:46821", + "accept": false, + "reason": "Both mention config arguments, but one is StableLm pad_token_idx and the other is TimesFM window_size." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4-related, but one is token-type defaults for text-only fine-tuning and the other is a device_map forward mismatch." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "Gemma4 device placement and a generic return_tensors handling bug are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Processor/tokenizer access regression and OmDet-Turbo input-size mismatch are different code paths." + }, + { + "left": "issue:29870", + "right": "issue:29942", + "accept": false, + "reason": "BART generation UnboundLocalError and Flash Attention 2 test failures are not the same bug." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "Different import/backend failures: flash_attn on Python 3.13 vs tensorflow_text backend mapping." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "StableLm config field loss and Qwen3.5 flash-attention generation issues are unrelated." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Generation kwargs sync handling and output_hidden_states behavior in Qwen3.5 are separate bugs." + }, + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "torch.compile/SDPA failure and ZeRO-3 gradient growth are different training issues." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "Video processor batch-shape bug and Qwen2-57B-A14B-Instruct OOM are not duplicates." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "MPS attention correctness and Kimi-K2.5 tokenizer regression are unrelated." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Different import errors: flash-attention utility import vs Qwen2.5-VL class import." + }, + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "Trailing-slash model-path warning and SigLIP2 HF/JAX discrepancy are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 documentation example errors and return_tensors handling are different problems." + }, + { + "left": "issue:38175", + "right": "issue:43493", + "accept": false, + "reason": "Both are SigLIP2-related, but one reports zero probabilities while the other reports implementation divergence vs JAX." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Tokenizer special-token registration and a stale CI device override test 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": 24, + "review_unit_index": 28, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28218", + "issue:35141", + "issue:41720", + "issue:42915", + "issue:43262", + "issue:43278", + "issue:43334", + "issue:43335", + "issue:43388", + "issue:43450", + "issue:43452", + "issue:43525", + "issue:43526", + "issue:43550", + "issue:43819", + "issue:43824", + "issue:43827", + "issue:43856", + "issue:43901", + "issue:43950", + "issue:43975", + "issue:44112", + "issue:44206", + "issue:44442", + "issue:44466", + "issue:44479", + "issue:44521", + "issue:44561", + "issue:44610", + "issue:44792", + "issue:44857", + "issue:44898", + "issue:45083", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45325", + "issue:45381", + "issue:45412", + "issue:45563", + "issue:45584", + "issue:45593", + "issue:46192", + "issue:46209", + "issue:46234", + "issue:46285", + "issue:46512" + ], + "soft_pairs": [ + "issue:44466|issue:45127", + "issue:44521|issue:45290", + "issue:43278|issue:43856", + "issue:45381|issue:46209", + "issue:28218|issue:35141", + "issue:42915|issue:43824", + "issue:43950|issue:46285", + "issue:43824|issue:43975", + "issue:43452|issue:46512", + "issue:43334|issue:43525", + "issue:43262|issue:43450", + "issue:44442|issue:45198", + "issue:44792|issue:45563", + "issue:44898|issue:45412", + "issue:44479|issue:46192", + "issue:43819|issue:44112", + "issue:45200|issue:45325", + "issue:43827|issue:43901", + "issue:43388|issue:43526", + "issue:44561|issue:46234", + "issue:43335|issue:43550", + "issue:41720|issue:46209", + "issue:43819|issue:45083", + "issue:44857|issue:45593", + "issue:44610|issue:44857", + "issue:44206|issue:45584" + ], + "prepared_review_unit_hash": "0eb28aace78bfeb96f05ac53c118aadcaf458b5783d1a5436d96e7c87e8daefe", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35889, + "estimated_input_tokens": 8973, + "estimated_eval_tokens": 18202 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster does not look like a duplicate set. It is a grab bag of unrelated transformer bugs, regressions, and docs issues that only share broad API or subsystem keywords.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both concern tied weights, but one is device-dependent lm_head serialization and the other is LoRA merge collapse with extended vocabulary; different concrete bugs." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both involve apply_chat_template, but one is all-zero assistant masks for multimodal input and the other is a crash on tool-call messages; not the same failure." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "Embedding dtype drift in evaluate and Qwen3 MoE memory usage are unrelated issues." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "Qwen2.5-VL vision position ids bug and FP8 inference failure on B200 are different model/runtime problems." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer spacing after added tokens is unrelated to embedding reinitialization after resizing." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "Qwen3Moe FP8 failure and Qwen2.5-VL import error are different code paths and symptoms." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to CLIP text model structural inconsistency." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Missing import for a VL class and incorrect detokenization are separate bugs." + }, + { + "left": "issue:43452", + "right": "issue:46512", + "accept": false, + "reason": "gguf_file loading failures and stale config dtype after from_config are unrelated." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": false, + "reason": "Both mention missing pad_token_id, but they affect different model configs and loading paths; not the same regression." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "Last-batch label dropping in gather_for_metrics is unrelated to BeitImageProcessorFast reducing labels to one element." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate default and batched video shape handling are different processor bugs." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "Tokenizer loading failure and save_pretrained/tokenization failure in Wav2Vec2 are not the same issue." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "Janus image generation test failure and stale paged-generate warning are unrelated." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver interpolation failure and RT-DETR memory release bug are different models and failure modes." + }, + { + "left": "issue:44479", + "right": "issue:46192", + "accept": false, + "reason": "Qwen video input regression and return_tensors being ignored are unrelated API bugs." + }, + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC forward/from_latents mismatch and a stale GraniteSpeech CI test do not describe the same bug." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Gemma 4 mm_token_type_ids defaulting and Qwen2.5-VL rope indexing are different multimodal bugs." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs issues, but they cover different pipeline docs and different stale references." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile SDPA failure are unrelated model/runtime issues." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "Auto device mapping cudaErrorAssert and FP8 inference on B200 are different hardware/runtime failures." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC latent mismatch and qwen3_omni_moe feature-length helper behavior are unrelated." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "AMP CUDA crash in LwDetrImageLoss is unrelated to D-FINE not using auxiliary losses." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "OmDet-Turbo output size mismatch and LwDetrImageLoss AMP crash are different bugs." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Unsupported center arg in LasrFeatureExtractor and Whisper empty-transcription failure are unrelated." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Both are missing import_utils helpers, but they involve different removed/absent functions and different breakages; not one 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": 24, + "review_unit_index": 29, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28282", + "issue:29127", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:42490", + "issue:43278", + "issue:43335", + "issue:43425", + "issue:43504", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43720", + "issue:43825", + "issue:43854", + "issue:43931", + "issue:44112", + "issue:44351", + "issue:44448", + "issue:44514", + "issue:44521", + "issue:44743", + "issue:44855", + "issue:45083", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45290", + "issue:45412", + "issue:45520", + "issue:45588", + "issue:45593", + "issue:45759", + "issue:45859", + "issue:46170", + "issue:46285", + "issue:46326", + "issue:46431", + "issue:46512", + "issue:46566", + "issue:46821" + ], + "soft_pairs": [ + "issue:46512|issue:46821", + "issue:45759|issue:46285", + "issue:43606|issue:45083", + "issue:45520|issue:46170", + "issue:38175|issue:45412", + "issue:37428|issue:43425", + "issue:43335|issue:43653", + "issue:43550|issue:44112", + "issue:29942|issue:32090", + "issue:28282|issue:33357", + "issue:43611|issue:43931", + "issue:44448|issue:44743", + "issue:43335|issue:43504", + "issue:44855|issue:45083", + "issue:43278|issue:43825", + "issue:43720|issue:44112", + "issue:29127|issue:32090", + "issue:43335|issue:45083", + "issue:44514|issue:44521", + "issue:43335|issue:45593", + "issue:37428|issue:44351", + "issue:43335|issue:44112", + "issue:45290|issue:46326", + "issue:45588|issue:45859", + "issue:46431|issue:46566", + "issue:34689|issue:35141", + "issue:45127|issue:45245", + "issue:42490|issue:43854", + "issue:45200|issue:45588" + ], + "prepared_review_unit_hash": "7c7dec748ebb22b47ca2980dae8bdec94424a46f6a536d209ec7e8af1c1c4d7b", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 36179, + "estimated_input_tokens": 9045, + "estimated_eval_tokens": 18346 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a loose similarity bucket of unrelated Transformers issues. The soft edges mostly connect superficially similar bug reports (imports, loading, chat templates, offload, config, CI) rather than the same underlying defect, so none should be merged.", + "confidence": 0.79, + "canonical_issue_reason": "No single canonical issue: the tickets span many distinct subsystems and failure modes, and the closest pairs are only title-level similar.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global representative stands out; the cluster is heterogeneous and not centered on one bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46512", + "right": "issue:46821", + "accept": false, + "reason": "Different bugs in different models: stale config dtype vs TimesFM window_size AttributeError." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Both concern multimodal model/config structure, but they hit different code paths and failure modes." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "Unrelated regressions: CPU offload device mismatch vs a helper length-calculation bug." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "Both are import/dependency issues, but the platform, module, and root cause differ." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Different subsystems entirely: zero probabilities in Siglip2 vs RT-DETR memory cleanup." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Version-compatibility problems, but one is a missing internal symbol import and the other is Torch 2.10 incompatibility." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Different model/tokenizer bugs: sparse layer construction vs special-token decode behavior." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/runtime issues, but they affect different models and code paths." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention test failure vs Trainer GPU broadcast NoneType error are unrelated bugs." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "Missing PyTorch import and MacOS bus error on CLIP are different failures." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "Different model-loading breakages: base_model_prefix handling vs weight-shape mismatch." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Different regression types: output drift across versions vs recurrent-state reset in modular Qwen code." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "Separate bugs in different models; no shared underlying defect." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Python 3.13 JIT parsing issue vs qwen3_omni_moe helper behavior are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "Embedding dtype persistence and pipeline task messaging are different issues." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "Accelerate loading of packed weights vs a stale CI device-override test are unrelated." + }, + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "Different errors in different components: LayoutLMv3 validation messaging vs Trainer broadcast handling." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "No shared root cause; these are distinct model-specific bugs." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both are chat-template related, but one is batched padding behavior and the other is assistant-mask generation for multimodal inputs." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "Different models and code paths; no same underlying bug." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they involve different symbols and regressions." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both touch chat templating, but they fail in different branches and with different inputs." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Distinct failures: flash_attention integration on s_aux=None vs CPU-offload incompatibility in Qwen3_5Moe." + }, + { + "left": "issue:46431", + "right": "issue:46566", + "accept": false, + "reason": "Missing chat template support vs meta-device loading are unrelated." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Different model-loading problems: one breaks Llama 3.2 Vision loading, the other reinitializes embeddings after resize." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA merge collapse and category-count overflow are unrelated bugs." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "Local-path slash warning and GLM-4.7-Flash unit-test loading failure are different issues." + }, + { + "left": "issue:45200", + "right": "issue:45588", + "accept": false, + "reason": "Gemma token-type defaults and flash_attention s_aux handling are unrelated code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 30, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:26413", + "issue:27957", + "issue:28218", + "issue:29870", + "issue:29942", + "issue:31501", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:38175", + "issue:39692", + "issue:40444", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43701", + "issue:43819", + "issue:44206", + "issue:44360", + "issue:44448", + "issue:44493", + "issue:44561", + "issue:44617", + "issue:44779", + "issue:44857", + "issue:44977", + "issue:45072", + "issue:45083", + "issue:45229", + "issue:45290", + "issue:45310", + "issue:45381", + "issue:45459", + "issue:45478", + "issue:45823", + "issue:45920", + "issue:46192", + "issue:46234", + "issue:46697" + ], + "soft_pairs": [ + "issue:45310|issue:45478", + "issue:43819|issue:44493", + "issue:44448|issue:44779", + "issue:43606|issue:44857", + "issue:39692|issue:43526", + "issue:26413|issue:31501", + "issue:28218|issue:32090", + "issue:44617|issue:46697", + "issue:45459|issue:46234", + "issue:39692|issue:44493", + "issue:44206|issue:45823", + "issue:43701|issue:45083", + "issue:44857|issue:45072", + "issue:32090|issue:33357", + "issue:42490|issue:43575", + "issue:43550|issue:45083", + "issue:44360|issue:44561", + "issue:38175|issue:44493", + "issue:27957|issue:34689", + "issue:29942|issue:35141", + "issue:29870|issue:32090", + "issue:44448|issue:44977", + "issue:43278|issue:43329", + "issue:45920|issue:46192", + "issue:40444|issue:43526", + "issue:45290|issue:45381", + "issue:10000|issue:15354", + "issue:44977|issue:45229", + "issue:43278|issue:43381" + ], + "prepared_review_unit_hash": "77ae8d35f31b02d8c467ed3ba2f76c39322aef0ff7a9f43ad992e58f4536deb8", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35775, + "estimated_input_tokens": 8944, + "estimated_eval_tokens": 18144 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous: the only clear duplicate pair is the two identical Qwen3.5 MoE from_pretrained reports (45310/45478). The rest are broad text-similarity matches across unrelated bugs. No PRs are present.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45310 is the earliest of the only exact duplicate pair, and it cleanly represents the Qwen3.5 MoE from_pretrained regression also reported in 45478.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45920 is the strongest standalone issue in the set to use as a representative: it describes a broad v5 tokenizer regression affecting multiple named models and has the most cross-references.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Same title and same Qwen3.5 MoE from_pretrained failure; these look like the same underlying bug." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "Different code paths: DAC.from_latents mismatch vs position-id key warnings across many models." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Both are v5 regressions, but one is Pegasus generation output drift and the other is DeepSeek tokenizer IDs; not the same bug." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "Both involve runtime/device issues, but Bark CPU offload mismatch and LwDetr AMP crash are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 documentation example errors and BeitImageProcessorFast reduce_labels behavior are unrelated." + }, + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "Checkpoint resume epoch_iterator failure and FineWeb training not starting are different training issues." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer added-space bug and Trainer GPU broadcast NoneType error are unrelated." + }, + { + "left": "issue:44617", + "right": "issue:46697", + "accept": false, + "reason": "CUDA OOM in Sam3Video and over-padding in Swin2SR image processing are different bugs." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Both touch import/utils, but hidden tokenizer decode errors vs a missing availability helper are different problems." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "Documentation/quantization example issue vs position-id key warnings; no shared bug." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "LasrFeatureExtractor center-arg crash and Gemma4 device-map mismatch are unrelated." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "Resume-from-checkpoint key mismatch and qwen3_omni_moe feature-length behavior are different code paths." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "Both are dtype/runtime mismatches, but they affect different models and mechanisms; not the same fix." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer NoneType broadcast error and MacOS CLIP bus error are unrelated failures." + }, + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "Trailing-slash local model import warning and tensor-parallel OOM are unrelated." + }, + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Bamba torch.compile/SDPA failure and qwen3_omni_moe feature-length issue are different bugs." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA ReLU discussion and removal of is_torch_fx_available breaking trust_remote_code are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 zero-probability issue and position-id key warnings are different regressions." + }, + { + "left": "issue:27957", + "right": "issue:34689", + "accept": false, + "reason": "Both mention model loading/attention, but XLMRoberta Flash Attention 2 and Llama 3.2 Vision loading are distinct issues." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention tests failing and embedding reinitialization on resize are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:32090", + "accept": false, + "reason": "BART generate decoder_attention_mask bug and GPU broadcast NoneType error are different failures." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Both are generation regressions, but Pegasus v4/v5 output drift and Qwen3.5 flash-attention generation failure are not the same bug." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype drift in evaluate and undefined video-token helper variables are unrelated." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Both are tokenizer/API regressions in v5, but wrong token IDs across models vs return_tensors being ignored are different underlying bugs." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Qwen2.5-VL iterable dataset finetuning failure and BeitImageProcessorFast reduce_labels bug are unrelated." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call crash and qwen2.5-vl video vision_position_ids bug are different multimodal issues." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel issue and GeneratorExp/torch.jit.script export failure are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Qwen3.5 flash-attention generation bug and Gemma4 multi-GPU OOM are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "BF16-to-FP32 dtype change in evaluation and gradient checkpointing in eval mode are different behaviors." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only clearly supported duplicate pair is 45310/45478 with identical titles, and the remaining soft-edge rejections do not overstate similarity. No PRs are present, so the PR-related fields are appropriately null." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 31, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:25251", + "issue:29127", + "issue:29870", + "issue:32090", + "issue:33290", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:37428", + "issue:39692", + "issue:41720", + "issue:42175", + "issue:42490", + "issue:43278", + "issue:43381", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43701", + "issue:43873", + "issue:43927", + "issue:43976", + "issue:44186", + "issue:44361", + "issue:44448", + "issue:44466", + "issue:44561", + "issue:44568", + "issue:44743", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:45198", + "issue:45229", + "issue:45290", + "issue:45372", + "issue:45440", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45698", + "issue:45803", + "issue:45823", + "issue:46121", + "issue:46399", + "issue:46655" + ], + "soft_pairs": [ + "issue:44855|issue:46399", + "issue:29870|issue:33290", + "issue:45290|issue:46655", + "issue:44857|issue:45588", + "issue:43927|issue:45698", + "issue:37428|issue:42175", + "issue:32090|issue:33290", + "issue:44561|issue:44779", + "issue:43873|issue:45229", + "issue:25251|issue:32090", + "issue:44448|issue:44466", + "issue:43976|issue:45440", + "issue:42490|issue:43873", + "issue:44568|issue:45491", + "issue:45588|issue:45823", + "issue:43278|issue:43575", + "issue:41720|issue:45229", + "issue:43381|issue:43701", + "issue:44977|issue:45563", + "issue:44871|issue:44977", + "issue:33357|issue:39692", + "issue:29127|issue:34567", + "issue:45198|issue:45803", + "issue:43653|issue:44743", + "issue:44361|issue:45198", + "issue:44186|issue:45198", + "issue:45372|issue:45588", + "issue:43611|issue:46121", + "issue:29870|issue:34689" + ], + "prepared_review_unit_hash": "f3c29caf80f3b2d23c98b385813afa525fbef9511ca04dbdd8b27818e10e635f", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36740, + "estimated_input_tokens": 9185, + "estimated_eval_tokens": 18626 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a heterogeneous mix of unrelated Transformers bugs. The soft-similarity pairs overlap lexically on broad areas like generation, serialization, tokenizers, or memory, but they do not appear to describe the same concrete defect, so none should be merged as duplicates.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 is the strongest issue candidate by discussion/inbound activity and broad infrastructure relevance, but it is still only one bug report in a largely unrelated set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Both are import/registration failures, but they affect different model code paths (DebertaV2 vs Qwen3.5) and different root causes." + }, + { + "left": "issue:29870", + "right": "issue:33290", + "accept": false, + "reason": "Bart generate() UnboundLocalError and DeepSpeed/Adafactor OOM are unrelated training/runtime problems." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Chat template tool-call crash and SmolVLM video padding bug are different APIs and failure modes." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "One is an AMP/CUDA loss crash, the other is a flash-attention integration AttributeError on sink-less models." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load behavior, but one loses config token IDs and the other loads the wrong custom module; different underlying bugs." + }, + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "Flash-attention import error and missing TensorFlow backend in a torch install are unrelated." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer broadcast NoneType and Adafactor OOM are different failure classes with no shared code-path." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "A removed utility breaking trust_remote_code models is unrelated to DeepSeek tokenizer output regressions." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization/offloading behavior and Gemma4 multi-GPU OOM are both memory-adjacent, but not the same bug or fix." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k nesting bug and Trainer GPU broadcast error are unrelated." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Both are v5 regressions, but one is generation output drift and the other is serialization of lm_head.weight; different concrete issues." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility breakage and DeepSeekV3MoE divergence are not the same defect." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "Trailing-slash import warning and quantization/offloading failure are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer special-token behavior and Gemma3 NaN embeddings from sliding-window attention are different problems." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "Flash-attention sinkless-model crash and Gemma4 device_map mismatch are distinct forward-path bugs." + }, + { + "left": "issue:43278", + "right": "issue:43575", + "accept": false, + "reason": "Training/eval dtype drift and Qwen2 tensor-parallel OOM are unrelated." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both involve large-model inference failures, but one is cudaErrorAssert on auto device mapping and the other is plain OOM; not the same root cause." + }, + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Gradient checkpointing in eval mode and resume_from_checkpoint key mismatch do not share a code-path." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Flash-attention generation failure and a stale num_return_sequences warning are not the same bug." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "Gemma eos_token_id config mismatch and Qwen3.5 flash-attention generation failure are unrelated model-specific issues." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "MacOS bus error on a CLIP model and SigLIP2 doc example errors are different scenarios and failure types." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error-message clarity and TrainerState token counter not updating are unrelated." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 save/tokenization failure and a bare-except in Fuyu tensor conversion are different bugs." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated tokenizer vs model-state issues." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "Both touch tokenization, but MLuke AttributeError and Wav2Vec2 save_pretrained/tokenization failure are not the same defect." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 NER/padding crash and Wav2Vec2 save/tokenization failure are different tokenization pathways." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma 4 processor import breakage and flash-attention sinkless crash are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "base_model_prefix loading regression and rope-validation TypeError involve different loading/validation code." + }, + { + "left": "issue:29870", + "right": "issue:34689", + "accept": false, + "reason": "Bart generate() UnboundLocalError and Llama 3.2 Vision model-loading breakage are different model-specific regressions." + } + ] + }, + "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": 24, + "review_unit_index": 32, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:27957", + "issue:29127", + "issue:29942", + "issue:30064", + "issue:33290", + "issue:33357", + "issue:35141", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43066", + "issue:43381", + "issue:43412", + "issue:43421", + "issue:43526", + "issue:43575", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43824", + "issue:43931", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44279", + "issue:44387", + "issue:44442", + "issue:44462", + "issue:44479", + "issue:45072", + "issue:45198", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45712", + "issue:45834", + "issue:46369", + "issue:46693" + ], + "soft_pairs": [ + "issue:42915|issue:42994", + "issue:43412|issue:45588", + "issue:44206|issue:44479", + "issue:45372|issue:45406", + "issue:43756|issue:45440", + "issue:29942|issue:30064", + "issue:43653|issue:44442", + "issue:45491|issue:46693", + "issue:43065|issue:45412", + "issue:29127|issue:33357", + "issue:45245|issue:45563", + "issue:43065|issue:43994", + "issue:45459|issue:45712", + "issue:44265|issue:45198", + "issue:43575|issue:43701", + "issue:43066|issue:44462", + "issue:45372|issue:45459", + "issue:45440|issue:45593", + "issue:29127|issue:35141", + "issue:42490|issue:43381", + "issue:43994|issue:45412", + "issue:45412|issue:46369", + "issue:43421|issue:45491", + "issue:43575|issue:44387", + "issue:44442|issue:45072", + "issue:42915|issue:43931", + "issue:43526|issue:45834", + "issue:43526|issue:46693", + "issue:27957|issue:29127", + "issue:43526|issue:44186", + "issue:43824|issue:44279", + "issue:27957|issue:33290" + ], + "prepared_review_unit_hash": "bd6c6f1a3424914df0f3692c45f33c29ccd44910e2a5197951ed8e0f6c347e4d", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36569, + "estimated_input_tokens": 9143, + "estimated_eval_tokens": 18542 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The soft pairs look like broad-topic false positives rather than true duplicates: they share surface themes such as tokenizers, quantization, Flash Attention, or memory, but each points to a different model, code path, or failure mode. No soft edge is strong enough to merge into one duplicate cluster.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45491 is the clearest standalone, user-facing regression with a tightly scoped reproducer; the rest of the items are mostly unrelated false-positive matches.", + "canonical_pr_reason": null, + "best_issue_reason": "45491 is the strongest issue to anchor on: it has a concrete failure mode, specific triggering conditions, and clear impact.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Different phases and bugs: FP8 config failure versus quantized model saving failure." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "ExecuTorch export crash and flash-attention integration crash are separate code paths." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Feature extractor arg regression is unrelated to the Qwen VL video-input regression." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "One is a dependency import error; the other is a missing Gemma4Processor attribute in serve." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Different model-specific implementation bugs: RoPE layer dropping versus DeepSeekV3MoE divergence." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "FA2 test failures and void segmentation-map processing are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "Tokenizer special-token registration bug does not match a FastSpeech2Conformer tokenizer load failure." + }, + { + "left": "issue:45491", + "right": "issue:46693", + "accept": false, + "reason": "Both mention FA2/long context, but one is NaN embeddings and the other is a performance regression." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy Conv2d placeholder code is not the same as an RT-DETR memory leak." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "Error-message clarity for LayoutLMv3 is unrelated to a CLIP bus error on macOS." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-count runtime error and stale generate warning are different problems." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "A dummy class cleanup issue does not match SigLIP2 nonsensical outputs." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Protobuf error masking and leftover dummy classes are separate maintenance issues." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Qwen TP OOM and resume_from_checkpoint key mismatch do not share the same bug." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type and ignoring tokenizer.json are distinct tokenizer-loading issues." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma 4 dependency import failure is not the same as protobuf exception handling." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepSeekV3MoE divergence and D-FINE auxiliary-loss behavior are unrelated model bugs." + }, + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "Message clarity for one error is not the same as embedding reinitialization logic." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "Path-trailing-slash import warning and gradient-checkpointing eval-mode restriction are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SigLIP2 bad outputs and RT-DETR memory leak do not share a concrete code path." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "Memory leak on deletion and memory overflow during sam3-litetext usage are different failures." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer post-processor sync bug is unrelated to Gemma3 NaN embeddings." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Model-specific TP OOM and int4 reserved-memory regression are not the same bug." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer load failure and bfloat16 dtype mismatch are different subsystems." + }, + { + "left": "issue:27957", + "right": "issue:29127", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to LayoutLMv3 error handling." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "Missing Qwen2.5-VL import symbol and a generic dependency issue are not the same concrete problem." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug and Kosmos2.5 OCR index error are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:46693", + "accept": false, + "reason": "Label-reduction bug and FA2 long-context performance regression are unrelated." + }, + { + "left": "issue:27957", + "right": "issue:33290", + "accept": false, + "reason": "XLMRoberta FA2 support and deepspeed/adafactor OOM are different failures." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 33, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:27957", + "issue:29870", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:42175", + "issue:42490", + "issue:43262", + "issue:43316", + "issue:43334", + "issue:43381", + "issue:43493", + "issue:43525", + "issue:43531", + "issue:43646", + "issue:43720", + "issue:43723", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43994", + "issue:44062", + "issue:44077", + "issue:44206", + "issue:44442", + "issue:44479", + "issue:44514", + "issue:44743", + "issue:44811", + "issue:44898", + "issue:44977", + "issue:45003", + "issue:45245", + "issue:45250", + "issue:45325", + "issue:45381", + "issue:45440", + "issue:45479", + "issue:45588", + "issue:45600", + "issue:45859", + "issue:45907", + "issue:46121", + "issue:46234", + "issue:46369", + "issue:46399", + "issue:46640", + "issue:46697" + ], + "soft_pairs": [ + "issue:46121|issue:46399", + "issue:43994|issue:46369", + "issue:43334|issue:43531", + "issue:33290|issue:34689", + "issue:45250|issue:46640", + "issue:33290|issue:35141", + "issue:45588|issue:45600", + "issue:43381|issue:43873", + "issue:43761|issue:45859", + "issue:44514|issue:45381", + "issue:43646|issue:44077", + "issue:44977|issue:45245", + "issue:43262|issue:44206", + "issue:44743|issue:44811", + "issue:44442|issue:44898", + "issue:42490|issue:44743", + "issue:45003|issue:45440", + "issue:43720|issue:44442", + "issue:42175|issue:46234", + "issue:43316|issue:43525", + "issue:27957|issue:29870", + "issue:43723|issue:45479", + "issue:43493|issue:43827", + "issue:44077|issue:44479", + "issue:45325|issue:46697", + "issue:44062|issue:45907" + ], + "prepared_review_unit_hash": "d508085053ccfefd6b03f4fbfabe7d7ebe449f2da56e71140feb1bbed09f0ec7", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35565, + "estimated_input_tokens": 8892, + "estimated_eval_tokens": 18040 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is noisy and mostly heterogeneous; nearly all soft pairs are clearly unrelated. The only plausible duplicate-looking pair is the two very generic Flash Attention 2 reports, but the titles are too vague to confidently merge.", + "confidence": 0.58, + "canonical_issue_reason": "issue:46640 is the broadest and most current flash-attention-related report in the set, so it is the best umbrella candidate if any Flash Attention 2 duplicates are consolidated.", + "canonical_pr_reason": null, + "best_issue_reason": "Same as canonical: it is the most generic issue in the only apparent shared theme, and there is no better representative with clearer scope.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "Different bugs: rope-validation TypeError vs a Qwen3.5 kernel registration crash." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "Unrelated models and symptoms: SigLIP2 output quality vs Sam3 LiteText memory overflow." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Both are Qwen3-family, but one is a missing pad_token_id attribute and the other is a sliding_window behavior bug; not the same issue." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Different subsystems entirely: DeepSpeed/Adafactor OOM vs model loading for Llama 3.2 Vision." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Both are very generic Flash Attention 2 reports, but the available evidence is too thin to confirm the same underlying bug." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "OOM in optimizer training is unrelated to token embedding resize/post_init reinitialization." + }, + { + "left": "issue:45588", + "right": "issue:45600", + "accept": false, + "reason": "One is a flash_attention integration crash on s_aux=None; the other is an auto-mappings CI break from removed configs." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient-checkpointing eval-mode behavior is unrelated to quantization/offloading." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "Different model paths and failures: CLIPVision hidden_states regression vs Qwen3_5Moe CPU offload incompatibility." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Both concern Qwen2.5-VL, but one is batched chat-template padding behavior and the other is video position-id scaling; different bugs." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "A general custom model init regression is not the same as patchtsmixer's post_init allowance issue." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3.5 flash-attention generation failure is unrelated to the categories > 2^24 runtime error." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting is unrelated to LastFeatureExtractor's unsupported center argument." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Different bugs in different areas: recurrent-state reset vs Whisper batch_decode skipping special tokens." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer loading failure and Perceiver image-resolution failure are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "Local-path import warnings do not match the Qwen recurrent-state cache bug." + }, + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "Unsafe sys.modules access is unrelated to DeepSeekV3MoE divergence from remote implementation." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet accelerate-loading/unpacked-weights failure is unrelated to tokenizer loading." + }, + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "Packaging backend inclusion and a missing import_utils helper are different issues." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both are config attribute discrepancies, but they concern different model families and different missing fields." + }, + { + "left": "issue:27957", + "right": "issue:29870", + "accept": false, + "reason": "XLM-Roberta Flash Attention 2 support and BART decoder_attention_mask assignment are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:45479", + "accept": false, + "reason": "Tokenizer loading in v5 is unrelated to the degenerate zero-loss classification bug." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation mismatch and docs still referencing pipeline() are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "PatchTSMixer post_init policy and Qwen VL video-input regression are unrelated." + }, + { + "left": "issue:45325", + "right": "issue:46697", + "accept": false, + "reason": "Qwen2.5-VL temporal position-id scaling and Swin2SR over-padding are unrelated." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "AddedToken special-argument handling is unrelated to list_repo_files revision forwarding." + } + ] + }, + "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": 24, + "review_unit_index": 34, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:27957", + "issue:28282", + "issue:32090", + "issue:42617", + "issue:42907", + "issue:43232", + "issue:43295", + "issue:43316", + "issue:43388", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43643", + "issue:43701", + "issue:43716", + "issue:43742", + "issue:43749", + "issue:43761", + "issue:43825", + "issue:43873", + "issue:43931", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44442", + "issue:44464", + "issue:44496", + "issue:44534", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44987", + "issue:45137", + "issue:45216", + "issue:45584", + "issue:45698", + "issue:45759", + "issue:45803", + "issue:46178", + "issue:46234", + "issue:46291", + "issue:46421", + "issue:46506", + "issue:46821" + ], + "soft_pairs": [ + "issue:43873|issue:44871", + "issue:43761|issue:44079", + "issue:43643|issue:44913", + "issue:44855|issue:45137", + "issue:43749|issue:43931", + "issue:43316|issue:45759", + "issue:43761|issue:44188", + "issue:44987|issue:46178", + "issue:43232|issue:43388", + "issue:44190|issue:44496", + "issue:43931|issue:45216", + "issue:42907|issue:45698", + "issue:44442|issue:45803", + "issue:43701|issue:46506", + "issue:42617|issue:43575", + "issue:27957|issue:32090", + "issue:43575|issue:43742", + "issue:43550|issue:43701", + "issue:25251|issue:28282", + "issue:43606|issue:43701", + "issue:44464|issue:46421", + "issue:46234|issue:46291", + "issue:16998|issue:22355", + "issue:43295|issue:43825", + "issue:43295|issue:44871", + "issue:44871|issue:45584", + "issue:43316|issue:43716", + "issue:43316|issue:46821", + "issue:44077|issue:44534" + ], + "prepared_review_unit_hash": "fc69c4ba0d71bceda346f6009aa817f5ff56613f1678fc22bc9d4fdb95a9d571", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36313, + "estimated_input_tokens": 9079, + "estimated_eval_tokens": 18414 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft edges are mostly superficial topical overlaps, not the same underlying bug or change. I would not merge any of the listed pairs as duplicates. #43873 is the most active issue, but it is still too broad to serve as a real canonical duplicate hub.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue: the items span unrelated bugs; issue:43873 is only the most active thread, not a true duplicate hub.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43873 has the strongest engagement (discussion/inbound refs) and is the closest thing to a hub, though it is still too broad to represent the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Different problems: offloading/quantization handling vs Gemma-3 eos_token_id config inconsistency." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVision hidden_states regression and ModelOutput key assignment are unrelated code paths." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "AutoConfig trust_remote_code missing fields is unrelated to GPTNeoXConfig rotary_pct reload behavior." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "DebertaV2 import/script parsing on Python 3.13 is unrelated to DeepSpeed ZeRO3 deque errors." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Qwen3-VL weight-shape mismatch are different loading failures." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both involve multimodal/composite configs, but the concrete bugs and models differ." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "Hidden_states regression in CLIPVision is unrelated to attention-kernel divergence under torch.compile." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "Physical-intelligence/fast loading failure is unrelated to the tensorflow_text backend-mapping error." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation sync_gpus kwargs bug is unrelated to gather_for_metrics truncating tuple labels." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset loading in an example script is unrelated to an unrecognized model/config loading error." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Both touch Qwen3-VL/Qwen3.5, but one is a load-time shape mismatch and the other is a save_pretrained regression." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Dequantized model saving failure and wrong custom-module loading are separate save/load bugs." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "AutoTokenizer loading failure and a bare except in FuyuBatchFeature are unrelated." + }, + { + "left": "issue:43701", + "right": "issue:46506", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is unrelated to CPU/CUDA divergence in PPDocLayoutV3." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py runtime failure and tp-load OOM on Qwen2 are different issues." + }, + { + "left": "issue:27957", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 support for XLMRoberta is unrelated to Trainer _gpu_broadcast_one NoneType errors." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Qwen2 tensor-parallel OOM and MobileLLM config key error are different loading bugs." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile SDPA failure in Bamba is unrelated to resume_from_checkpoint key mismatches." + }, + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "pipeline top_k nesting behavior is unrelated to missing PyTorch import errors." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch and checkpoint key mismatch are different failure modes." + }, + { + "left": "issue:44464", + "right": "issue:46421", + "accept": false, + "reason": "Chunked-generation output inconsistency and lazy cache recompilation are related broadly to generation, but not the same concrete bug." + }, + { + "left": "issue:46234", + "right": "issue:46291", + "accept": false, + "reason": "Missing import_utils symbol and kernels backend version/revision failure are unrelated import-time issues." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "DeBERTa model_max_length question and transformers.onnx missing-module error are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Processor/tokenizer regression and pipeline translation-task messaging are separate v5 issues." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Processor/tokenizer regression is unrelated to Gemma-3 eos_token_id inconsistency." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma eos_token config mismatch and Whisper empty-transcription failure are different tokenization bugs." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy is unrelated to Mistral-3 image-preprocessor dtype mismatch." + }, + { + "left": "issue:43316", + "right": "issue:46821", + "accept": false, + "reason": "Gemma3TextConfig API discrepancy is unrelated to TimesFM window_size AttributeError." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "patchtsmixer post_init policy and non-persistent buffer junk initialization are different model/config 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": 24, + "review_unit_index": 35, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:41950", + "issue:43065", + "issue:43352", + "issue:43388", + "issue:43404", + "issue:43441", + "issue:43452", + "issue:43475", + "issue:43504", + "issue:43525", + "issue:43531", + "issue:43540", + "issue:43572", + "issue:43646", + "issue:43650", + "issue:43673", + "issue:43688", + "issue:43761", + "issue:44162", + "issue:44190", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44617", + "issue:44743", + "issue:44792", + "issue:44871", + "issue:44913", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45542", + "issue:45600", + "issue:45632", + "issue:45698", + "issue:45803", + "issue:46050", + "issue:46170", + "issue:46178", + "issue:46202", + "issue:46234", + "issue:46365", + "issue:46369", + "issue:46424", + "issue:46640", + "issue:46832" + ], + "soft_pairs": [ + "issue:44871|issue:46050", + "issue:43646|issue:44913", + "issue:44162|issue:44792", + "issue:45803|issue:46365", + "issue:43452|issue:44496", + "issue:43404|issue:43441", + "issue:45803|issue:46170", + "issue:45632|issue:46424", + "issue:44610|issue:46369", + "issue:43761|issue:46832", + "issue:43650|issue:46202", + "issue:43572|issue:46178", + "issue:43540|issue:44514", + "issue:45542|issue:46234", + "issue:43065|issue:44617", + "issue:43504|issue:45698", + "issue:43388|issue:43688", + "issue:44792|issue:44977", + "issue:44792|issue:45600", + "issue:43673|issue:44743", + "issue:43352|issue:46640", + "issue:45127|issue:45200", + "issue:41950|issue:43475", + "issue:43525|issue:43531", + "issue:44190|issue:44964", + "issue:44534|issue:44913" + ], + "prepared_review_unit_hash": "c768a69374aa9753a34cbfeefbdf8fab8ea34c052fdeb437afaa859fb8b76b65", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35389, + "estimated_input_tokens": 8848, + "estimated_eval_tokens": 17952 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a collection of distinct Transformers bug reports with only superficial thematic overlap (loading/saving, config regressions, flash-attn, multimodal/video, cache behavior). I would not merge any of the soft pairs as duplicates. No PRs are present.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45698 is the broadest and most active report here: it describes a general from_pretrained/save_pretrained custom-module loading regression, has strong discussion/inbound references, and is the best representative if one issue must anchor the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45698 is the strongest standalone issue to keep: it is open, well-discussed, and framed at the right abstraction level for a reusable loading/serialization bug rather than a model-specific symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "Both concern generation/model config after v5 changes, but one is an eos_token_id mismatch and the other is quantized-model roundtrip corruption; different underlying bugs." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Both are config/regression reports, but one is a broad v5 custom-model initialization breakage while the other is a specific GPTNeoX rotary_pct reload bug." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 being broken and janus image-generation test failure are unrelated model-specific failures, not the same bug." + }, + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "One is a bare-except bug in Fuyu tensor conversion; the other is an import failure when datasets is installed. Different code paths and failure modes." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "Both are loading-related, but one is about gguf_file handling in from_pretrained and the other is an unrecognized model/config.json problem for a specific model repository." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Both mention model support constraints, but one is a missing weight tie in Mistral3 and the other is a FlashAttention compatibility issue for Ministral-3." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "KeyboardInterrupt-swallowing in Fuyu tensor conversion and a Windows/ROCm import dependency issue are unrelated." + }, + { + "left": "issue:45632", + "right": "issue:46424", + "accept": false, + "reason": "Cache-path collision for local models and max_cache_length only applying to Quantized Cache involve different cache subsystems and different bugs." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Processor output size mismatch and Sam3-lite-text memory overflow are separate SAM/vision-model issues with different root causes." + }, + { + "left": "issue:43761", + "right": "issue:46832", + "accept": false, + "reason": "CLIPVision hidden_states regression and RT-DETR feature-level mismatch are unrelated architecture-specific failures." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "A generic 'ADD THE DATA' issue is not the same as Gemma pipeline roundtrip failure on the underscore token; no shared bug semantics." + }, + { + "left": "issue:43572", + "right": "issue:46178", + "accept": false, + "reason": "StableLm pad_token_idx regression and a tensorflow_text backend mapping error are different config/backend issues." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and Qwen2_5_VLProcessor batched chat-template crash are both multimodal, but they fail in different code paths." + }, + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "TensorBoard-without-TensorFlow backend error and missing is_falcon_mamba_ssm_available helper are unrelated dependency/API issues." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "A dummy Conv2d in Sam3PixelDecoder and Sam3Video CUDA OOM are not the same bug; one is architectural placeholder code, the other is resource exhaustion." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Beit legacy-field preset loading failure and wrong custom-module loading after save_pretrained are both loading problems, but they affect different mechanisms and code paths." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Batch-label dropping in gather_for_metrics and auxiliary-loss normalization are both training-related, but they are separate bugs in different subsystems." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Janus image-generation test failure and Qwen3.5 flash-attention generation issues are distinct model/runtime failures." + }, + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Janus test failure and auto_mappings.py referencing removed Sam3LiteText configs are unrelated; one is a runtime test bug, the other is CI/import mapping breakage." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Generation cache missing during chunked_prefill and recurrent states resetting in modular_qwen3_5.py both involve state/caching, but they are different mechanisms and code paths." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "NemotronH FlashAttention 2 support gap and a generic Flash Attention 2 issue are too vague and model-specific to treat as the same bug." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA/extended-vocab collapse with tied embeddings and Gemma-4 mm_token_type_ids defaulting are different model-training problems." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Video-classification pipeline looking for image processors and Sam3VisionEncoderOutput missing fpn_position_embeddings are separate video/model API bugs." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Missing pad_token_id in Llama4Config and sliding_window issues in Qwen3-MoE are both config-related, but they are not the same defect." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading in an example script and loading microsoft/Phi-4-multimodal-instruct with latest transformers are unrelated failures." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffers filled with junk in v5 and GPTNeoX rotary_pct silently reverting are both regressions, but they affect different internals." + } + ] + }, + "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": 24, + "review_unit_index": 36, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:29870", + "issue:35141", + "issue:39692", + "issue:41720", + "issue:42490", + "issue:43065", + "issue:43335", + "issue:43550", + "issue:43644", + "issue:43673", + "issue:43827", + "issue:43931", + "issue:43994", + "issue:44060", + "issue:44077", + "issue:44112", + "issue:44190", + "issue:44464", + "issue:44479", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44855", + "issue:44869", + "issue:44877", + "issue:44913", + "issue:44933", + "issue:45127", + "issue:45245", + "issue:45357", + "issue:45362", + "issue:45381", + "issue:45406", + "issue:45563", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45632", + "issue:46082", + "issue:46369", + "issue:46421", + "issue:46617", + "issue:46655", + "issue:46697", + "issue:46832" + ], + "soft_pairs": [ + "issue:44734|issue:45406", + "issue:35141|issue:42490", + "issue:46082|issue:46697", + "issue:44869|issue:46832", + "issue:45632|issue:46617", + "issue:44855|issue:45588", + "issue:43673|issue:46421", + "issue:43931|issue:44060", + "issue:43827|issue:43994", + "issue:29870|issue:35141", + "issue:44617|issue:45584", + "issue:45563|issue:45584", + "issue:43644|issue:44913", + "issue:44933|issue:45362", + "issue:25251|issue:27957", + "issue:43335|issue:44855", + "issue:44479|issue:44734", + "issue:45381|issue:46369", + "issue:43550|issue:44190", + "issue:44077|issue:44877", + "issue:41720|issue:45245", + "issue:43065|issue:44610", + "issue:43335|issue:44190", + "issue:44464|issue:45563", + "issue:45127|issue:45357", + "issue:39692|issue:44190", + "issue:45381|issue:46655", + "issue:44112|issue:45600" + ], + "prepared_review_unit_hash": "7639a545e4c9da51cf13be95e93ab7c8ed6c8370c0a5e5f3318a94cb198ca7d5", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35624, + "estimated_input_tokens": 8906, + "estimated_eval_tokens": 18068 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No soft-edge pair looks like the same underlying bug/change; these issues span unrelated subsystems (generation, model configs, docs, vision/video, caching, and CI).", + "confidence": 0.94, + "canonical_issue_reason": "issue:43644 is the broadest, most infrastructure-level report here (a v5.0.0 regression affecting non-persistent buffers across models), so it is the best single issue to anchor the cluster even though the cluster is not a true duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43644 is the strongest representative issue because it is cross-cutting, concrete, and has downstream impact beyond one model family; the rest are narrower, model-specific, or documentation/CI-only reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "Different failure modes and code paths: KV-cache continuation tensor indexing in serve vs. Gemma4Processor missing _tokenizer." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "One is output embedding reinit after resize; the other is a local-path trailing-slash import warning. No shared bug." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "LlamaConfig head_dim validation and Swin2SR image over-padding are unrelated." + }, + { + "left": "issue:44869", + "right": "issue:46832", + "accept": false, + "reason": "Whisper timestamp decode crash and RT-DETR backbone feature-level TypeError are distinct model bugs." + }, + { + "left": "issue:45632", + "right": "issue:46617", + "accept": false, + "reason": "Both involve trust_remote_code/cache, but one is leaf-directory cache collision and the other is FileNotFoundError from cache loading; not the same root cause." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python 3.13 JIT parsing/IndentationError is unrelated to flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:43673", + "right": "issue:46421", + "accept": false, + "reason": "Both mention chunked prefill/cache, but one is missing cache in v5.0.0 while the other is lazy cache init causing recompilation." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-Next tied-weights warning are different issues in different model families." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Documentation references removed pipeline APIs; the other is a SigLIP2 inference correctness problem." + }, + { + "left": "issue:29870", + "right": "issue:35141", + "accept": false, + "reason": "Bart generate() decoder_attention_mask bug is unrelated to token-embedding reinitialization after resize." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "Sam3Video OOM is not the same as Whisper empty-transcription/alignment crash." + }, + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "A stale warning about num_return_sequences is a different bug from Whisper failing on empty transcription after align_special_tokens." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Generic v5 buffer corruption and GPTNeoX rotary_pct reload regression are unrelated config/runtime bugs." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Nonexistent import from image_utils and Qwen3.5 chat crash are different problems." + }, + { + "left": "issue:25251", + "right": "issue:27957", + "accept": false, + "reason": "Pipeline top_k nesting bug and XLMRoberta Flash Attention 2 support are unrelated." + }, + { + "left": "issue:43335", + "right": "issue:44855", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation on zero layers has nothing to do with DebertaV2Model import parsing on Python 3.13." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Qwen VL video input regression and serve KV-cache continuation crash affect different paths and symptoms." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Wrong vision_position_ids in Qwen2.5-VL is unrelated to Sam3-LiteText memory overflow." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba-9B-v2 is unrelated to loading a local dataset in the image-classification script." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "PatchTSMixer post_init policy and granite_speech config loading are different config-validation issues." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to the category-count limit RuntimeError." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder and OmDet-Turbo processor size mismatch are distinct model/processor bugs." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers sparse-layer bug is unrelated to dataset loading in the no_trainer example." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Chunked generation inconsistency with compiled forward and a stale warning about num_return_sequences are not the same bug." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "LoRA merge collapse with extended vocabulary and incorrect visual-encoder keys on save_pretrained are different regressions." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 docs/model mismatch and local dataset loading failure are unrelated." + }, + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Qwen2.5-VL video position IDs bug and SmolVLM padding calculation bug affect different processors." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "GraniteSpeech stale CI test and auto_mappings referencing removed Sam3LiteText configs are separate CI breakages." + } + ] + }, + "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": 24, + "review_unit_index": 37, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:29870", + "issue:35141", + "issue:36246", + "issue:38617", + "issue:41628", + "issue:42175", + "issue:42796", + "issue:43295", + "issue:43526", + "issue:43540", + "issue:43643", + "issue:43650", + "issue:43854", + "issue:44112", + "issue:44186", + "issue:44220", + "issue:44246", + "issue:44279", + "issue:44462", + "issue:44496", + "issue:44521", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44779", + "issue:44938", + "issue:44964", + "issue:45003", + "issue:45245", + "issue:45276", + "issue:45290", + "issue:45381", + "issue:45542", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46082", + "issue:46121", + "issue:46234", + "issue:46369", + "issue:46489", + "issue:46697", + "issue:46752" + ], + "soft_pairs": [ + "issue:44779|issue:46489", + "issue:43526|issue:46697", + "issue:44617|issue:46369", + "issue:43540|issue:44734", + "issue:43295|issue:45276", + "issue:45290|issue:46752", + "issue:44737|issue:45803", + "issue:44462|issue:45907", + "issue:44521|issue:45381", + "issue:45276|issue:45584", + "issue:44938|issue:45542", + "issue:25251|issue:29870", + "issue:45632|issue:46234", + "issue:44246|issue:45003", + "issue:43854|issue:44964", + "issue:44964|issue:45600", + "issue:43643|issue:44496", + "issue:44112|issue:44964", + "issue:27957|issue:35141", + "issue:46082|issue:46121", + "issue:45245|issue:45584", + "issue:42175|issue:44279", + "issue:42796|issue:44521", + "issue:44186|issue:44734", + "issue:36246|issue:41628", + "issue:43650|issue:44220", + "issue:36246|issue:38617" + ], + "prepared_review_unit_hash": "d4f020aa79548737dccd1bdba78efa9230ab0a1bdda7023a411e13b92d449c27", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35350, + "estimated_input_tokens": 8838, + "estimated_eval_tokens": 17932 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is highly heterogeneous: the soft-similarity pairs mostly share broad subsystem themes or surface wording, but they do not look like the same underlying bug/change. No PRs are present.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest representative issue here: it is open, has the most inbound references, and describes a concrete, well-scoped multimodal apply_chat_template bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the most central and actionable issue in this set, with clear impact and active references; it is the best single issue to anchor the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44779", + "right": "issue:46489", + "accept": false, + "reason": "Same tokenizer family and both mention v5 regressions, but one is DeepSeek generally and the other is DeepSeek-Coder v1 with a separate PR-gap claim; not clearly the same concrete bug." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "Both are image-processor edge cases, but they affect different models and different behaviors (label reduction vs over-padding)." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3 memory-related, but they concern different model variants and likely different causes; not one duplicate bug." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Different code paths: Qwen3OmniMoe video input validation versus serve KV-cache continuation indexing." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Both involve regressions around model/processor behavior, but the concrete failures are unrelated (processor.tokenizer access vs resize_token_embeddings semantics)." + }, + { + "left": "issue:45290", + "right": "issue:46752", + "accept": false, + "reason": "Both are apply_chat_template crashes, but the triggering inputs differ materially: assistant tool-call messages with no content versus an empty messages list." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "Different subsystems and failure modes: CPU tensor allocation in XLNet versus a broad exception handler in FuyuBatchFeature." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "One is AutoTokenizer ignoring tokenizer.json, the other is hub file listing not forwarding revision; related repo-loading area but not the same bug." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Both are multimodal-related, but one is assistant mask generation and the other is vision_position_ids for Qwen2.5-VL video inputs." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Unrelated bugs: Gemma4 embedding resizing versus Whisper empty-transcription generation failure." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import failure is unrelated to missing TensorFlow backend wiring for the torch extra." + }, + { + "left": "issue:25251", + "right": "issue:29870", + "accept": false, + "reason": "Different generation/pipeline bugs with different symptoms and code paths (nested-list output vs decoder_attention_mask assignment)." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "One is a trust_remote_code cache-path collision, the other is a missing import_utils helper symbol." + }, + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "Import slowness and unsafe sys.modules access are not the same concrete defect." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Both are model-loading issues, but they involve different models and likely different root causes." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Different failures: model loading regression versus stale auto_mappings references to removed configs." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "Both are config-loading problems, but one is trust_remote_code returning incomplete configs while the other is an unrecognized model missing model_type." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "CI failure in GraniteSpeech device override is unrelated to a Phi-4-multimodal load regression." + }, + { + "left": "issue:27957", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention support for XLMRoberta and embedding reinitialization after resize_token_embeddings are unrelated issues." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "Different config-validation failures: Llama head_dim rejection versus JSON-loaded rope-ignore key handling." + }, + { + "left": "issue:45245", + "right": "issue:45584", + "accept": false, + "reason": "Category-cardinality overflow is unrelated to Whisper empty-transcription generation failure." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "One is a specific backend-install packaging complaint; the other is a vague dependency issue without a matching concrete defect." + }, + { + "left": "issue:42796", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer BPE-marker handling and multimodal assistant-mask generation are different bugs in different components." + }, + { + "left": "issue:44186", + "right": "issue:44734", + "accept": false, + "reason": "LayoutLMv2Tokenizer padding/NER crash and serve KV-cache continuation indexing are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are ImportErrors, but they concern different missing exports in different modules." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "The titles are too generic to infer a shared underlying bug; the described areas are unrelated." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different missing symbols in different modules; no evidence they stem from the same change or 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": 24, + "review_unit_index": 38, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:32090", + "issue:35141", + "issue:41950", + "issue:42907", + "issue:43295", + "issue:43299", + "issue:43381", + "issue:43404", + "issue:43441", + "issue:43479", + "issue:43526", + "issue:43531", + "issue:43673", + "issue:43824", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44220", + "issue:44279", + "issue:44295", + "issue:44327", + "issue:44336", + "issue:44496", + "issue:44521", + "issue:44683", + "issue:44792", + "issue:44933", + "issue:44987", + "issue:45042", + "issue:45083", + "issue:45216", + "issue:45335", + "issue:45375", + "issue:45446", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45600", + "issue:45759", + "issue:46123", + "issue:46369", + "issue:46424", + "issue:46506", + "issue:46515", + "issue:46693", + "issue:46752" + ], + "soft_pairs": [ + "issue:43404|issue:45216", + "issue:32090|issue:35141", + "issue:43479|issue:45375", + "issue:44295|issue:46506", + "issue:43673|issue:46424", + "issue:41950|issue:44016", + "issue:43531|issue:45600", + "issue:44016|issue:44220", + "issue:43976|issue:44987", + "issue:43295|issue:45335", + "issue:45083|issue:46506", + "issue:43381|issue:45593", + "issue:42907|issue:43441", + "issue:44683|issue:45446", + "issue:43673|issue:44336", + "issue:43526|issue:43901", + "issue:44496|issue:45042", + "issue:43901|issue:44496", + "issue:43976|issue:44279", + "issue:44792|issue:46424", + "issue:43441|issue:45588", + "issue:44496|issue:45759", + "issue:44016|issue:44062", + "issue:43824|issue:44933", + "issue:44327|issue:46752", + "issue:46123|issue:46369", + "issue:44521|issue:46693", + "issue:43299|issue:45216", + "issue:44279|issue:46515", + "issue:45335|issue:45584" + ], + "prepared_review_unit_hash": "d1f8dba1ba9608292d5783de308416d83d07b9edd833e5cae15403c23fff0191", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36076, + "estimated_input_tokens": 9019, + "estimated_eval_tokens": 18294 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are a grab bag of distinct Transformers bugs and regressions rather than a duplicate set. The soft-similarity pairs mostly share broad themes like configs, generation, or model loading, but they target different code paths and should stay separate.", + "confidence": 0.92, + "canonical_issue_reason": "issue:44521 is the strongest representative overall: it is open, has the most inbound references and discussion, and describes a concrete multimodal API regression with a clear repro.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the best global issue candidate because it is active, well-scoped, and appears most likely to benefit from continued tracking or follow-up.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both concern model weight tying/save behavior, but they hit different models and different failure modes; not the same underlying bug." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "One is a Trainer GPU broadcast TypeError; the other is post-init reinitialization after resizing embeddings. Related area, different bug." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Both are config-handling issues, but one is default init for Phi4Multimodal and the other is a missing field silently dropped for Qwen3_5MoeVisionConfig." + }, + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "Unrelated failures: buffer access for position_ids versus CPU/CUDA result divergence in PPDocLayoutV3." + }, + { + "left": "issue:43673", + "right": "issue:46424", + "accept": false, + "reason": "Both involve cache settings, but one is a missing GenerationMixin cache in chunked prefill while the other is max_cache_length being honored only for quantized cache." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "A video-classification processor lookup bug and a notebook syntax error are completely different artifacts." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and stale auto_mappings references are unrelated." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error versus an audio feature extraction issue; no shared underlying code path." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "One is a Python-version compatibility problem; the other is a specific model load failure for physical-intelligence/fast." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Both mention token/embedding handling, but one is a multimodal processor regression and the other is a t5gemma decoder embedding update bug." + }, + { + "left": "issue:45083", + "right": "issue:46506", + "accept": false, + "reason": "Different helper behavior in qwen3_omni_moe versus device-dependent output drift in PPDocLayoutV3." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction is unrelated to D-FINE auxiliary-loss behavior." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Both touch Ministral/Devstral areas, but one is save/export of dequantized models and the other is FlashAttention loading failure." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both mention flex_attention, but one is a torch>=2.9 compiled failure and the other is a version-check/import bug for AuxRequest." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Cache/memory-related naming aside, chunked-prefill cache handling and ANSI output formatting are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "ImageProcessor label reduction and text-classification docs mismatch are different subsystems and failures." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "An unrecognized-model config issue is not the same as PIL image processors incorrectly requiring torchvision." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Documentation mismatch for pipelines and model config import failure do not share an underlying bug." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "The former is a Python compatibility regression; the latter is a vague dependency problem." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "A failed janus image-generation test and a cache-length configuration issue are different failures." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Ministral FlashAttention loading and sink-less flash_attention aux handling are separate code paths." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Unrecognized model config versus from_config not unwrapping text_config for composite multimodal configs are related only at a very high level." + }, + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Notebook syntax error is unrelated to the AddedToken keyword-argument TypeError." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but they involve different missing symbols from different modules." + }, + { + "left": "issue:44327", + "right": "issue:46752", + "accept": false, + "reason": "QA decode_spans top-k bounds error and empty-list apply_chat_template crash are unrelated." + }, + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "Partial-batch mask generation dropping results is not the same as memory overflow in sam3-litetext." + }, + { + "left": "issue:44521", + "right": "issue:46693", + "accept": false, + "reason": "Assistant-mask generation for multimodal chat templates is unrelated to FA2 long-context performance regression." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3VL MoE loading breakage and Qwen3.5 checkpoint save corruption are different model/version regressions." + }, + { + "left": "issue:44279", + "right": "issue:46515", + "accept": false, + "reason": "A generic dependency issue is not the same as a specific torch 2.4.1 incompatibility." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "t5gemma embedding resize and Whisper empty-transcription generation failures 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": 24, + "review_unit_index": 39, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:33290", + "issue:39692", + "issue:41950", + "issue:42796", + "issue:43295", + "issue:43299", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43577", + "issue:43650", + "issue:43653", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44075", + "issue:44242", + "issue:44279", + "issue:44291", + "issue:44351", + "issue:44361", + "issue:44368", + "issue:44462", + "issue:44466", + "issue:44683", + "issue:44779", + "issue:45020", + "issue:45081", + "issue:45103", + "issue:45200", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45357", + "issue:45372", + "issue:45405", + "issue:45542", + "issue:45593", + "issue:45656", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:46082", + "issue:46209", + "issue:46399", + "issue:46512" + ], + "soft_pairs": [ + "issue:46082|issue:46399", + "issue:44279|issue:45405", + "issue:43295|issue:44779", + "issue:39692|issue:43901", + "issue:44075|issue:44368", + "issue:43976|issue:45405", + "issue:46082|issue:46512", + "issue:43577|issue:45216", + "issue:44351|issue:45278", + "issue:42796|issue:43650", + "issue:41950|issue:45020", + "issue:44779|issue:45812", + "issue:43299|issue:45357", + "issue:43479|issue:43525", + "issue:44242|issue:45200", + "issue:43299|issue:43854", + "issue:44466|issue:45276", + "issue:41950|issue:45823", + "issue:43653|issue:45593", + "issue:44683|issue:45542", + "issue:43650|issue:46209", + "issue:44462|issue:45081", + "issue:44016|issue:44291", + "issue:44361|issue:45593", + "issue:43441|issue:45834", + "issue:45103|issue:45372", + "issue:33290|issue:45656" + ], + "prepared_review_unit_hash": "fd88ccb349a0862e77543d4e69f05b72fbcf524b6bb16087b091b32e554cee3f", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35388, + "estimated_input_tokens": 8847, + "estimated_eval_tokens": 17950 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly a grab-bag of unrelated model/tokenizer/config regressions and docs/infra issues, not a single duplicate cluster. No soft edge looks like the same underlying bug/change.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the broadest umbrella report here because it covers recent-version breakage in model loading, but the overall set is too heterogeneous to treat as one duplicate family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "Different bugs: LlamaConfig head_dim validation vs Qwen3.5 rotary-kernel registration failure." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Unrelated: generic transformers dependency issue vs PEFT minimum-version bump." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Both tokenizer regressions, but different models and failure modes (processor/tokenizer images vs Deepseek token IDs)." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "Different docs issues in different pipelines; no shared code-path bug." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "SGD optimizer args vs Qwen3.5 tie_word_embeddings warning are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python-version compatibility break is not the same as a PEFT version requirement change." + }, + { + "left": "issue:46082", + "right": "issue:46512", + "accept": false, + "reason": "Config validation bug vs stale dtype metadata after from_config; different problem." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "BLIP2 dtype-loading issue vs Qwen3.5 save_pretrained key serialization bug." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "HybridCache import failure is only one possible import error; not the same as the broad import-breakage report." + }, + { + "left": "issue:42796", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer bug vs non-bug placeholder issue; no relation." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification processor lookup bug is separate from remote_code model-loading regressions." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Both tokenizer regressions, but different model families and likely different root causes." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Qwen3VL load failure vs Qwen3.5 save_pretrained serialization issue; different code paths." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Different config bugs: multimodal default initialization vs missing pad_token_id attribute." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss bug is unrelated to Gemma 4 mm_token_type_ids defaults." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Different model load failures with different stacks and symptoms." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "Both touch weights/embeddings, but one is serialization on device and the other is resize_token_embeddings propagation." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Pipeline processor-selection bug vs Gemma4 device-mapping mismatch; unrelated." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "Tokenizer special-token registration vs D-FINE auxiliary-loss omission are unrelated." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Torch 2.9 flex_attention failure is unrelated to the TensorFlow/tensorboard backend error." + }, + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "Placeholder issue vs FP8/B200 inference failure; no duplicate relationship." + }, + { + "left": "issue:44462", + "right": "issue:45081", + "accept": false, + "reason": "Both tokenizer-loading issues, but one is AutoTokenizer ignoring tokenizer.json and the other is a Mistral regex patch crash." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error vs init_empty_weights TypeError; different layers entirely." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "MLukeTokenizer task error is unrelated to D-FINE loss handling." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Different model-specific runtime bugs: FlashAttention on Ministral vs OCR index error on Kosmos2.5." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "auto_docstring annotations crash is unrelated to Gemma 4 processor-loading import failure." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Different DeepSpeed problems: Adafactor OOM vs optimizer.step being called twice." + } + ] + }, + "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": 24, + "review_unit_index": 40, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:41628", + "issue:42175", + "issue:42994", + "issue:43317", + "issue:43441", + "issue:43525", + "issue:43643", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43784", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44162", + "issue:44188", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44297", + "issue:44336", + "issue:44373", + "issue:44496", + "issue:44589", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:45072", + "issue:45081", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45600", + "issue:45712", + "issue:45800", + "issue:45823", + "issue:45907", + "issue:46291", + "issue:46399", + "issue:46655", + "issue:46726", + "issue:46829" + ], + "soft_pairs": [ + "issue:42994|issue:43317", + "issue:45081|issue:45907", + "issue:45600|issue:45712", + "issue:44623|issue:44749", + "issue:44261|issue:44263", + "issue:42175|issue:43976", + "issue:44623|issue:45397", + "issue:44671|issue:44779", + "issue:43901|issue:44246", + "issue:43723|issue:44162", + "issue:43673|issue:43761", + "issue:43441|issue:44683", + "issue:45542|issue:45800", + "issue:43824|issue:43976", + "issue:43525|issue:43881", + "issue:44016|issue:45823", + "issue:46291|issue:46399", + "issue:44188|issue:44336", + "issue:44589|issue:45362", + "issue:43525|issue:43643", + "issue:44336|issue:44373", + "issue:46655|issue:46726", + "issue:41628|issue:43976", + "issue:43784|issue:45542", + "issue:44246|issue:44496", + "issue:44297|issue:45397", + "issue:45561|issue:45563", + "issue:45072|issue:46829" + ], + "prepared_review_unit_hash": "57057f61a2e5a4f54aca6f87bca76ee33dd40661b6f24b9abfdeeff209343871", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36097, + "estimated_input_tokens": 9025, + "estimated_eval_tokens": 18306 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of unrelated Transformers v5 regressions, packaging/import issues, and model-specific bugs. I do not see any pair that is clearly the same underlying defect, so all soft edges should be rejected.", + "confidence": 0.82, + "canonical_issue_reason": "Issue 43723 is the broadest user-facing regression in the set (tokenizer loading in v5) and is the closest thing to a representative umbrella issue, but it is not a true duplicate hub for the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "43723 has the clearest general bug report and the best chance of representing the cluster\u2019s overall v5 loading-regression theme, even though the rest of the items are not duplicates of it.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both involve quantization/offload, but one is saving a quantized model and the other is loading a dequantized model with device_map=auto; different code paths and failure modes." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regex patching crashes versus list_repo_files revision forwarding; these are unrelated utilities and not the same bug." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both touch generated/mapping files, but one is stale auto_mappings config references and the other is leftover dummy classes leaking into exports; distinct cleanup bugs." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "processor.save_pretrained missing files is unrelated to the reported slowdown in data filtering after upgrade." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "These affect different model internals: MLA q_a_layernorm config precision versus GlmMoeDsaIndexer torch.split return handling." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "One is backend extras installation for TensorFlow, the other is Python version compatibility for Transformers 5.1.0; not the same defect." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Saving processor artifacts missing files and Gemma4 ZeRO3 from_pretrained are separate load/save paths with no concrete shared bug." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "CamemBERT MLM prediction regression and Deepseek tokenizer regression are different model-specific failures." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch for return_all_scores versus slow import time are unrelated issues." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer loading regression in v5 and ESM2 being broken are different model/loading problems." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Generation cache missing during chunked prefill and CLIPVisionModel hidden_states behavior are different code paths and outputs." + }, + { + "left": "issue:43441", + "right": "issue:44683", + "accept": false, + "reason": "Both are attention-kernel related, but one is Ministral flash-attention failure and the other is compiled flex_attention failure on newer torch; not the same bug." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "TensorFlow backend installation gap and torch/Transformers version incompatibility are different packaging/version problems." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Missing Qwen2_5_VLForConditionalGeneration export is unrelated to Python 3.9+/3.10+ compatibility." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Llama4Config missing pad_token_id and glm-4v-9b loading failure are model-specific and do not share a concrete code path." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error and Gemma4 device mismatch are unrelated issues." + }, + { + "left": "issue:46291", + "right": "issue:46399", + "accept": false, + "reason": "Kernels v0.15 revision/version requirement when importing pipelines is different from a Qwen3.5 kernel registration crash." + }, + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "torch.compile attention-kernel divergence and ANSI codes in loading_report are entirely different subsystems." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage lookup failure and Qwen3.5 chat crash are different failures with no shared root cause shown." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "A missing pad_token_id field on one config is not the same as trust_remote_code returning an object with missing fields." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "Terminal ANSI output bug and wrong docstring for position_ids are unrelated." + }, + { + "left": "issue:46655", + "right": "issue:46726", + "accept": false, + "reason": "Incorrect video padding calculation and missing-vision-backend image processor error are different processor bugs." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "Cannot import AutoImageProcessor and Transformers Python-version support are unrelated compatibility issues." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Import-time NameError from sentence-transformers integration and missing TensorFlow backend packaging are different dependency problems." + }, + { + "left": "issue:44246", + "right": "issue:44496", + "accept": false, + "reason": "Slow import timing and unrecognized model/config loading are not the same underlying defect." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma4 zero3 loading." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "pytest-xdist capture-file race and stale generation warning are separate issues in different parts of the test/runtime stack." + }, + { + "left": "issue:45072", + "right": "issue:46829", + "accept": false, + "reason": "dtype mismatches in bfloat16 inference and incorrect hidden_states assignment are different model output 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": 24, + "review_unit_index": 41, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:36246", + "issue:41950", + "issue:42491", + "issue:42757", + "issue:42886", + "issue:43257", + "issue:43299", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43673", + "issue:43901", + "issue:44162", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44336", + "issue:44351", + "issue:44451", + "issue:44479", + "issue:44485", + "issue:44496", + "issue:44561", + "issue:44734", + "issue:44779", + "issue:44871", + "issue:45003", + "issue:45103", + "issue:45137", + "issue:45200", + "issue:45292", + "issue:45362", + "issue:45375", + "issue:45397", + "issue:45561", + "issue:45593", + "issue:45759", + "issue:46132", + "issue:46399", + "issue:46424", + "issue:46506", + "issue:46614", + "issue:46655", + "issue:46665", + "issue:46752", + "issue:46821" + ], + "soft_pairs": [ + "issue:45103|issue:46399", + "issue:46614|issue:46665", + "issue:44242|issue:45593", + "issue:43526|issue:45003", + "issue:42491|issue:43299", + "issue:43540|issue:46752", + "issue:43531|issue:45362", + "issue:44295|issue:45397", + "issue:44246|issue:44871", + "issue:44336|issue:44485", + "issue:45003|issue:46821", + "issue:44451|issue:46752", + "issue:44336|issue:46424", + "issue:42886|issue:46132", + "issue:43550|issue:46506", + "issue:41950|issue:44220", + "issue:44779|issue:45292", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:43901|issue:45003", + "issue:44734|issue:46424", + "issue:43257|issue:45137", + "issue:45003|issue:45561", + "issue:43540|issue:46655", + "issue:44496|issue:45003", + "issue:43673|issue:44561", + "issue:44162|issue:45200", + "issue:44479|issue:46655", + "issue:45375|issue:45759" + ], + "prepared_review_unit_hash": "7d1cab9cc16afe5004f536cb904784ac16c76b85e94cdefe7b6ea95249aeb787", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36318, + "estimated_input_tokens": 9080, + "estimated_eval_tokens": 18416 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous issues with only loose topical similarity; I would not merge any soft edge pairs as duplicates. The closest theme is Qwen/multimodal/video-regression reports, but the concrete failures differ.", + "confidence": 0.74, + "canonical_issue_reason": "Issue 44479 is the broadest concrete regression report in the set, covering multiple Qwen multimodal/video-input models under a single v5.3.0 regression. It is the best anchor only in the sense of breadth, not because the whole cluster is one duplicate thread.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 44479 is the most representative issue to anchor the cluster because it spans several related models and a clear version regression, making it the strongest single candidate among otherwise unrelated reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "Different failure modes: auto-docstring parsing with future annotations vs a Qwen3.5 kernel-registration crash." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Unrelated subsystems: stale phonemizer language state vs from_pretrained behavior inside Python classes." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Both concern missing losses, but they affect different models and code paths (router loss vs D-FINE auxiliary losses)." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "Different areas entirely: Beit image processor label reduction vs unsafe sys.modules access in modeling_utils." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen MoE loading, but the concrete breakage and model families differ." + }, + { + "left": "issue:43540", + "right": "issue:46752", + "accept": false, + "reason": "Video-input ValueError in Qwen3OmniMoe is unrelated to empty-list chat template handling." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Sliding-window behavior in Qwen3-MoE is not the same bug as a chat crash in Qwen3.5-35B." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "Different bugs: position_ids buffer access vs gemma-4 ZeRO3 from_pretrained." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import slowness is unrelated to inconsistent eos_token_id configuration." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "ANSI codes in loading_report is unrelated to GLM-5 RoPE implementation." + }, + { + "left": "issue:45003", + "right": "issue:46821", + "accept": false, + "reason": "Unsafe sys.modules access is unrelated to a TimesFM window_size AttributeError." + }, + { + "left": "issue:44451", + "right": "issue:46752", + "accept": false, + "reason": "Model loading failure for ScandiBERT is unrelated to apply_chat_template on an empty list." + }, + { + "left": "issue:44336", + "right": "issue:46424", + "accept": false, + "reason": "Terminal color output bug is unrelated to cache max length behavior." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline cache loading is unrelated to AttentionInterface.register behavior changes." + }, + { + "left": "issue:43550", + "right": "issue:46506", + "accept": false, + "reason": "Torch.compile/SDPA failure in Bamba is unrelated to CPU vs CUDA divergence in PPDocLayoutV3." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video-classification image-processor lookup and fbank feature extraction are different pipelines." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer correctness regression is unrelated to resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but for different symbols and different libraries/modules." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Missing Qwen2_5_VLImageProcessor export is not the same as missing HybridCache export." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs mentioning return_all_scores is unrelated to sys.modules access." + }, + { + "left": "issue:44734", + "right": "issue:46424", + "accept": false, + "reason": "KV-cache continuation crash and cache max_length semantics are distinct cache bugs." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Qwen3 MoE weight conversion under accelerate+deepspeed is unrelated to ZeRO3 deque underflow." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Modeling_utils import safety and pytest-xdist capture-file races are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-related, but Qwen3OmniMoe input validation and SmolVLM padding calculation are different code paths." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config loading is unrelated to sys.modules access." + }, + { + "left": "issue:43673", + "right": "issue:44561", + "accept": false, + "reason": "Chunked-prefill cache handling and removal of is_torch_fx_available are separate regressions." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are unrelated." + }, + { + "left": "issue:44479", + "right": "issue:46655", + "accept": false, + "reason": "Both involve video models, but the reported regressions are model-specific and not the same underlying bug." + }, + { + "left": "issue:44485", + "right": "issue:46655", + "accept": false, + "reason": "GLM-5 RoPE implementation is unrelated to SmolVLM video padding." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Both are Qwen multimodal config issues, but one drops a config field while the other mishandles text_config unwrapping." + } + ] + }, + "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": 24, + "review_unit_index": 42, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:36246", + "issue:42886", + "issue:43262", + "issue:43295", + "issue:43352", + "issue:43475", + "issue:43540", + "issue:43596", + "issue:43653", + "issue:44038", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44368", + "issue:44466", + "issue:44479", + "issue:44561", + "issue:44623", + "issue:44749", + "issue:44857", + "issue:44877", + "issue:44977", + "issue:44987", + "issue:44995", + "issue:45137", + "issue:45200", + "issue:45229", + "issue:45250", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45412", + "issue:45459", + "issue:45584", + "issue:45600", + "issue:45859", + "issue:46123", + "issue:46170", + "issue:46369", + "issue:46399", + "issue:46506", + "issue:46726", + "issue:46832" + ], + "soft_pairs": [ + "issue:45859|issue:46399", + "issue:44242|issue:44857", + "issue:44295|issue:44749", + "issue:43262|issue:44246", + "issue:43475|issue:46369", + "issue:36246|issue:43596", + "issue:45137|issue:45600", + "issue:44877|issue:45375", + "issue:45292|issue:45584", + "issue:43295|issue:45292", + "issue:44561|issue:46399", + "issue:44623|issue:46726", + "issue:44466|issue:45335", + "issue:44038|issue:46506", + "issue:29870|issue:30333", + "issue:44987|issue:45459", + "issue:43352|issue:45250", + "issue:44479|issue:46170", + "issue:45412|issue:46123", + "issue:43653|issue:44162", + "issue:44060|issue:44077", + "issue:44368|issue:44977", + "issue:46170|issue:46832", + "issue:44995|issue:45200", + "issue:42886|issue:45229", + "issue:44877|issue:46170", + "issue:43475|issue:43540" + ], + "prepared_review_unit_hash": "bd61730bb4129f94c104e8e891a67bededb78bc99ba51a47d64f575ad566e882", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36001, + "estimated_input_tokens": 9001, + "estimated_eval_tokens": 18258 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of unrelated issues across tokenizers, multimodal processors, model configs, generation, memory, and infrastructure. The soft pair similarities are topical rather than duplicate-level, so none should be merged.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue is safe here: the items are heterogeneous and the candidate pairs do not share the same concrete bug path.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a strong global representative because none of the candidates are true duplicates; they span different models and failure modes.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45859", + "right": "issue:46399", + "accept": false, + "reason": "Both involve Qwen3.5/MoE rotary-attention code, but one is CPU-offload incompatibility and the other is a kernel registration crash; different failure paths." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "Mixture-of-experts load-balancing loss and RT-DETR AMP CUDA crashes are unrelated bugs." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "One is a position_ids buffer access error; the other is a slowdown regression after upgrade." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and slow imports are unrelated issues." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "Both are SAM3-related, but one is a missing attribute in the vision encoder output and the other is a memory overflow in sam3-litetext." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "Qwen2.5-VL image processor import failure and DeepSpeed ZeRO3/BertModel index error do not share a code path." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "Empty-deque ZeRO3 failure and stale auto_mappings references are different kinds of breakage." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Both are config-related, but they concern different models and different bugs: strict loading vs a missing field being dropped." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Resize-token-embeddings behavior and Whisper empty-transcription generation are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "A processor.tokenizer regression with image inputs is not the same bug as resize_token_embeddings not updating output embeddings." + }, + { + "left": "issue:44561", + "right": "issue:46399", + "accept": false, + "reason": "Removal of is_torch_fx_available affecting trust_remote_code is unrelated to Qwen3.5 rotary kernel registration." + }, + { + "left": "issue:44623", + "right": "issue:46726", + "accept": false, + "reason": "Saving processor files and missing vision backends causing AutoProcessor errors are different loader/saver problems." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Both mention tied/resize embeddings, but one is serialization inconsistency and the other is a model-specific resize propagation bug." + }, + { + "left": "issue:44038", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3-VL-Moe and PPDocLayoutV3 issues are entirely separate model families and symptoms." + }, + { + "left": "issue:29870", + "right": "issue:30333", + "accept": false, + "reason": "Bart generate UnboundLocalError and MLFlow job-status reporting are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45459", + "accept": false, + "reason": "Tokenizer load failure on transformers>=5.1.0 and protobuf error masking are both loading-related but not the same defect." + }, + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "A model not supporting Flash Attention 2.0 and a generic Flash Attention 2.0 issue are not the same concrete bug." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "A video regression in Qwen models and a Windows ROCm import/dependency issue are unrelated." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "Memory not released on RT-DETR deletion and MaskGenerationPipeline partial-batch handling are different subsystems." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and ESM2 breakage are not the same underlying problem." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "A tied-weights warning bug and patchtsmixer post_init policy are unrelated." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen3.5 and flash attention/tied weights, but one is a warning about config tying and the other is a generation failure." + }, + { + "left": "issue:46170", + "right": "issue:46832", + "accept": false, + "reason": "Windows ROCm torch.distributed dependency issues and RTDetrModel feature-level shape errors do not overlap." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "A stale cache crash in GlmMoeDsa and Gemma4 mm_token_type_ids defaults are unrelated." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma4 multi-GPU OOM are different problems." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "Strict config loading for granite_speech and Windows ROCm dependency/import errors are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "Both are multimodal/video issues, but they affect different models and fail for different reasons." + } + ] + }, + "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": 24, + "review_unit_index": 43, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:42994", + "issue:43065", + "issue:43232", + "issue:43408", + "issue:43441", + "issue:43475", + "issue:43482", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43701", + "issue:43723", + "issue:43761", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44162", + "issue:44188", + "issue:44242", + "issue:44297", + "issue:44361", + "issue:44368", + "issue:44483", + "issue:44623", + "issue:44749", + "issue:44871", + "issue:44898", + "issue:44995", + "issue:45375", + "issue:45397", + "issue:45478", + "issue:45563", + "issue:45588", + "issue:45656", + "issue:45685", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46132", + "issue:46202", + "issue:46421", + "issue:46431", + "issue:46710", + "issue:46726", + "issue:46752", + "issue:46821", + "issue:46829" + ], + "soft_pairs": [ + "issue:45397|issue:45656", + "issue:43065|issue:43408", + "issue:44368|issue:45563", + "issue:43931|issue:44368", + "issue:43723|issue:46710", + "issue:44995|issue:45588", + "issue:44188|issue:46202", + "issue:45987|issue:46082", + "issue:44898|issue:46829", + "issue:44483|issue:46202", + "issue:43906|issue:46202", + "issue:44188|issue:46421", + "issue:45685|issue:46431", + "issue:43065|issue:46123", + "issue:44297|issue:44623", + "issue:44361|issue:44749", + "issue:43994|issue:46726", + "issue:43761|issue:44242", + "issue:43646|issue:44162", + "issue:44749|issue:45656", + "issue:43701|issue:44749", + "issue:44871|issue:46132", + "issue:44623|issue:45656", + "issue:43232|issue:43673", + "issue:42994|issue:43482", + "issue:43441|issue:46752", + "issue:45375|issue:45478", + "issue:43475|issue:43643", + "issue:43475|issue:46821" + ], + "prepared_review_unit_hash": "b2bb76cb17a4e238c139115fdff818abc37a8fa8c0a35055952c13711e46b5ad", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35860, + "estimated_input_tokens": 8965, + "estimated_eval_tokens": 18186 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is very heterogeneous: mostly unrelated Transformers v5 regressions, model-specific loading bugs, and a few generation/cache issues. There is no strong duplicate hub, so all soft links should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "No single issue cleanly covers this cluster; if forced to pick one representative, issue 43646 is the broadest Transformers v5 breakage, but it still only matches a subset of the items.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43646 is the best umbrella candidate because it is broad and version-wide ('Transformers 5.0.0 breaks custom model initialization'), making it the closest representative of the cluster's general theme.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Different bugs: one is a Gemma-4 ZeRO/from_pretrained loading issue, the other is a deepspeed optimizer-step duplication problem." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both mention SAM3, but one is a dummy Conv2d implementation detail and the other is a model-class mismatch warning." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Both are warnings, but about unrelated behaviors: tie_word_embeddings vs a stale num_return_sequences warning." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Different subsystems and failures: Qwen3-VL weight-shape mismatch vs a Qwen3.5 LoRA warning." + }, + { + "left": "issue:43723", + "right": "issue:46710", + "accept": false, + "reason": "One is tokenizer loading in v5; the other is an accuracy regression in a specific model after upgrade." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Different runtime paths: stale indexer cache on second forward pass vs flash-attention crashing on s_aux=None." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "Torch.compile attention-kernel branching is unrelated to Gemma token roundtrip failure on \u2581." + }, + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "Cache return-type bug vs LlamaConfig head_dim validation; same broad area but not the same defect." + }, + { + "left": "issue:44898", + "right": "issue:46829", + "accept": false, + "reason": "Different vision-model bugs: Perceiver resolution handling vs VideoPrism hidden_states/output mapping." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "API request rejection in chat/completions is unrelated to tokenizer/pipeline roundtrip behavior." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "An isolated reproduction of another issue is not the same as Gemma pipeline token roundtrip failure." + }, + { + "left": "issue:44188", + "right": "issue:46421", + "accept": false, + "reason": "Both touch generation/chunked prefill, but one is kernel branching divergence and the other is cache lazy-init recompilation." + }, + { + "left": "issue:45685", + "right": "issue:46431", + "accept": false, + "reason": "MPS histogram/int kernel error is unrelated to missing chat-template support in a processor." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel decoder implementation detail is unrelated to MaskGenerationPipeline's final-partial-batch logic." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Both are save_pretrained bugs, but tokenizer metadata mismatch and processor file omission are different code paths." + }, + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "Tokenizer AttributeError and a multilingual performance regression are not the same underlying bug." + }, + { + "left": "issue:43994", + "right": "issue:46726", + "accept": false, + "reason": "One is wrong outputs from AutoModel/pipeline, the other is a missing-vision-backend processor error." + }, + { + "left": "issue:43761", + "right": "issue:44242", + "accept": false, + "reason": "CLIPVision hidden_states regression is unrelated to MoE load-balancing loss gating." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Both are broad regressions, but one is custom model initialization under v5 and the other is an ESM2-specific breakage." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Performance slowdown after upgrade is not the same as deepspeed stepping the optimizer twice." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "Checkpoint-key mismatch and a runtime throughput regression are different failure modes." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "EOS token config inconsistency is unrelated to AttentionInterface.register changing function behavior." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save-pretrained file omission is unrelated to duplicate optimizer stepping." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both involve generation/caching, but one is sync_gpus/update_model_kwargs ordering and the other is missing cache in chunked_prefill." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving and Qwen2.5-GGUF loading are opposite operations with different failure modes." + }, + { + "left": "issue:43441", + "right": "issue:46752", + "accept": false, + "reason": "FlashAttention crash in Ministral-3 is unrelated to apply_chat_template failing on an empty list." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 variants, but one is a strict-field drop in config and the other is a from_pretrained loading error." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "Missing fpn_position_embeddings in SAM3 video output is unrelated to trust_remote_code AutoConfig field loss." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "SAM3 video encoder attribute loss and TimesFM window_size AttributeError are different model-specific bugs." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "Same as above: no shared underlying code path or fix." + } + ] + }, + "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": 24, + "review_unit_index": 44, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:36010", + "issue:39692", + "issue:41628", + "issue:42371", + "issue:42673", + "issue:42831", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43504", + "issue:43525", + "issue:43704", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43873", + "issue:43986", + "issue:44038", + "issue:44188", + "issue:44222", + "issue:44242", + "issue:44351", + "issue:44623", + "issue:44792", + "issue:44869", + "issue:44933", + "issue:44945", + "issue:44964", + "issue:45081", + "issue:45092", + "issue:45103", + "issue:45137", + "issue:45230", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45397", + "issue:45406", + "issue:45561", + "issue:45753", + "issue:46202", + "issue:46326", + "issue:46399", + "issue:46701", + "issue:46726", + "issue:46832" + ], + "soft_pairs": [ + "issue:28282|issue:30990", + "issue:43986|issue:46726", + "issue:45310|issue:45375", + "issue:41628|issue:44933", + "issue:44964|issue:45137", + "issue:44869|issue:45356", + "issue:45230|issue:45397", + "issue:44242|issue:46832", + "issue:39692|issue:46726", + "issue:44038|issue:44623", + "issue:42994|issue:43525", + "issue:43723|issue:46326", + "issue:42371|issue:45092", + "issue:43232|issue:44188", + "issue:43784|issue:46399", + "issue:45406|issue:46202", + "issue:43873|issue:44945", + "issue:44945|issue:45753", + "issue:41628|issue:44351", + "issue:42673|issue:43704", + "issue:42994|issue:43824", + "issue:45081|issue:45103", + "issue:44038|issue:45310", + "issue:46202|issue:46701", + "issue:44222|issue:45310", + "issue:43278|issue:45137", + "issue:36010|issue:41628", + "issue:42831|issue:44222", + "issue:43232|issue:44792", + "issue:43504|issue:45561" + ], + "prepared_review_unit_hash": "9081bb5291f795e841863d37cb265ade26c25a1c2b6e5483c60b427394e7c994", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36475, + "estimated_input_tokens": 9119, + "estimated_eval_tokens": 18494 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly heterogeneous issue reports around unrelated transformers/model bugs, so nearly all soft matches should be rejected. The only clear duplicate-looking pair is the repeated Qwen3ForCausalLM VRAM leak report (42673/43704), which appears to be the same underlying bug.", + "confidence": 0.92, + "canonical_issue_reason": "issue:42673 is the stronger canonical issue because it is the earlier, more discussed report of the exact same Qwen3ForCausalLM VRAM leak later repeated as issue:43704.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42673 is the best representative issue in this set because it has the clearest duplicate twin, the most discussion, and a concrete reproducible bug description.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "Missing PyTorch dependency error is unrelated to a sentence-transformers loading hang." + }, + { + "left": "issue:43986", + "right": "issue:46726", + "accept": false, + "reason": "Both involve processor/backend availability, but one is a video-model torchvision crash and the other is a generic unrecognized image processor error." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": false, + "reason": "Qwen3.5 MoE from_pretrained failure is not the same bug as a missing deepstack_visual_indexes config field." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Different missing-import failures from different modules; not the same underlying symbol/export bug." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Model loading regression and DeepSpeed ZeRO3 deque underflow are unrelated code paths." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding crash is unrelated to the Kimi tokenizer codec regression." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A placeholder bug report does not match a specific gemma-4 zero3 from_pretrained failure." + }, + { + "left": "issue:44242", + "right": "issue:46832", + "accept": false, + "reason": "Load-balancing loss logic and RT-DETR backbone feature-level mismatch are different bugs." + }, + { + "left": "issue:39692", + "right": "issue:46726", + "accept": false, + "reason": "Documentation example errors are not the same as the AutoProcessor backend-missing crash." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe import/runtime issue is unrelated to processor.save_pretrained missing files." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized model saving failure and missing pad_token_id on Llama4Config are different issues." + }, + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer loading in v5 and continue_final_message behavior for mistral_common are different failure modes." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance is unrelated to the InternVL2 meta-initialization incompatibility." + }, + { + "left": "issue:43232", + "right": "issue:44188", + "accept": false, + "reason": "Generation kwargs update after sync_gpus is not the same as attention-kernel divergence under torch.compile." + }, + { + "left": "issue:43784", + "right": "issue:46399", + "accept": false, + "reason": "sentence-transformers import NameError is unrelated to a missing decorator on apply_rotary_pos_emb registration." + }, + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on \u2581 are different bugs." + }, + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Quantization offloading behavior is unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "Pipeline-parallel output corruption is not the same as an infinite loop on a specific image." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they concern different missing transformers symbols and different fixes." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "These have the same title and describe the same Qwen3ForCausalLM VRAM leak in multiple dataloader threads; 43704 is a duplicate report." + }, + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Quantized saving failure and missing Qwen2_5_VLForConditionalGeneration import are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Mistral regex patch crash and auto-docstring future-annotations crash are different AttributeError sites." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen MoE, but one is a Qwen3-VL-Moe crash and the other is a Qwen3.5-Moe from_pretrained error." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Gemma tokenizer roundtrip failure and DiffusionGemma image-text pipeline failure are different concrete problems." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "FP8 save_pretrained on moe models is not the same as the Qwen3.5-Moe from_pretrained error." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Embedding dtype drift between train/eval is unrelated to the DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "GenerationMixin import failure and AutoImageProcessor import failure are separate missing-symbol regressions." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression is not the same as FP8 save_pretrained behavior." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation state update bug and janus image generation test failure are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "BEiT legacy-field loading failure is unrelated to the pytest-xdist captured_info.txt race." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is well grounded in the packet and stays conservative on the soft matches. The only accepted duplicate pair, issue:42673 and issue:43704, is supported by identical titles, earlier date, and higher discussion on 42673. No PR-mergeability concerns apply here since these are issues, not PRs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 45, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:27957", + "issue:30333", + "issue:30990", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:42175", + "issue:42222", + "issue:42673", + "issue:42898", + "issue:43116", + "issue:43295", + "issue:43479", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:43976", + "issue:44016", + "issue:44060", + "issue:44079", + "issue:44190", + "issue:44295", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44933", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45200", + "issue:45310", + "issue:45356", + "issue:45405", + "issue:45561", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45812", + "issue:46077", + "issue:46123", + "issue:46231", + "issue:46693", + "issue:46829" + ], + "soft_pairs": [ + "issue:45561|issue:45600", + "issue:30333|issue:33290", + "issue:42175|issue:44933", + "issue:44496|issue:45561", + "issue:43976|issue:44933", + "issue:30990|issue:35141", + "issue:42175|issue:45712", + "issue:42898|issue:44568", + "issue:43479|issue:46829", + "issue:46077|issue:46123", + "issue:44623|issue:45310", + "issue:43479|issue:44869", + "issue:42673|issue:46693", + "issue:42898|issue:45812", + "issue:43723|issue:44568", + "issue:44016|issue:46231", + "issue:42898|issue:44779", + "issue:43479|issue:44079", + "issue:42222|issue:45200", + "issue:44734|issue:45103", + "issue:43784|issue:43824", + "issue:43874|issue:45092", + "issue:27957|issue:30333", + "issue:30333|issue:34689", + "issue:42898|issue:45356", + "issue:42898|issue:43295", + "issue:44295|issue:45656", + "issue:43116|issue:44190", + "issue:44060|issue:45125", + "issue:45405|issue:45712", + "issue:30333|issue:35141" + ], + "prepared_review_unit_hash": "d4f39c3f2eafdfadb4689e209bdb7706044e6817d5849cf5b8bf9e9f00bda5d5", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36727, + "estimated_input_tokens": 9182, + "estimated_eval_tokens": 18620 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is highly heterogeneous: tokenizer regressions, multimodal model bugs, training/runtime issues, CI/docs problems, and example-script failures. No single canonical duplicate stands out, and every soft-edge pair is a distinct issue rather than the same underlying bug/change.", + "confidence": 0.97, + "canonical_issue_reason": "No coherent shared bug/change spans the cluster; the issues cover many unrelated subsystems and failure modes.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue is a good global canonical for this set; at best, some tokenizer-related reports are broadly similar, but they do not represent the same defect.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45561", + "right": "issue:45600", + "accept": false, + "reason": "Different areas: pytest-xdist temp-file race vs a broken auto_mappings config reference." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLflow reporting after failed training is unrelated to deepspeed Adafactor OOM." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging/backend dependency issue vs a bad import in image_utils." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading failure is unrelated to a testing-utils race condition." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python version compatibility bug vs nonexistent image_utils import." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Sentence-transformers loading hang is unrelated to embedding resize/post_init reinit." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Missing tensorflow backend in extras is unrelated to leftover dummy PT classes." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer-related regressions, but they describe different concrete behaviors and code paths." + }, + { + "left": "issue:43479", + "right": "issue:46829", + "accept": false, + "reason": "Different multimodal output/config initialization bug vs classifier output field mix-up." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Incorrect attention mask construction is unrelated to final-batch pipeline bookkeeping." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Processor save_pretrained file omission is unrelated to Qwen3.5 MoE from_pretrained failure." + }, + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Multimodal config defaults vs Whisper timestamp decode crash are unrelated." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "VRAM leak in multi-thread dataloading is not the same as FA2 long-context performance regression." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both involve tokenizers in v5, but the reported defects and affected models differ." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "AutoTokenizer loading issue is distinct from missing BOS/EOS insertion for a specific tokenizer." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Notebook syntax error is unrelated to a documentation parameter-name mismatch." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Generic clean_up_tokenization_spaces behavior change is not the same defect as DeepSeek tokenizer regression." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Phi4MultimodalConfig initialization bug is unrelated to ModelOutput key assignment logic." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "Broken vitpose models and Gemma 4 mm_token_type_ids defaulting are different model-specific issues." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "KV-cache tensor indexing crash is unrelated to auto-docstring annotation handling." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Sentence-transformers import NameError and Qwen2.5-VL import error are distinct failures." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Missing image-patch helper in Glm46VImageProcessorFast is not the same as remote-code meta-init incompatibility." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 support for XLM-R is unrelated to MLflow reporting on failed jobs." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Training-job MLflow reporting and Llama 3.2 Vision model loading are unrelated." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they concern different models and different underlying codec/regex behavior." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Processor.tokenizer access breakage is a different compatibility issue than clean_up_tokenization_spaces behavior." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Position_ids buffer access error is unrelated to deepspeed optimizer double-step behavior." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Example script returns empty multi-label results vs local dataset loading failure are different bugs." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Incorrect tied-weights warning in Qwen3-Next is unrelated to missing _tp_plan in Qwen3.5 MoE." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Dependency version bump policy issue is unrelated to dummy class cleanup." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "MLflow failure reporting and embedding reinitialization after resize 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": 24, + "review_unit_index": 46, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:22355", + "issue:28282", + "issue:41628", + "issue:42757", + "issue:42796", + "issue:42831", + "issue:43122", + "issue:43232", + "issue:43475", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:44075", + "issue:44190", + "issue:44206", + "issue:44222", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44448", + "issue:44484", + "issue:44485", + "issue:44623", + "issue:44734", + "issue:44933", + "issue:44977", + "issue:45071", + "issue:45092", + "issue:45103", + "issue:45245", + "issue:45310", + "issue:45656", + "issue:45712", + "issue:45834", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46231", + "issue:46305", + "issue:46421", + "issue:46424", + "issue:46431", + "issue:46655", + "issue:46710", + "issue:46762" + ], + "soft_pairs": [ + "issue:44222|issue:44297", + "issue:22355|issue:28282", + "issue:44075|issue:46231", + "issue:44933|issue:45712", + "issue:44360|issue:46762", + "issue:42831|issue:44623", + "issue:44448|issue:46132", + "issue:44297|issue:45656", + "issue:42796|issue:43723", + "issue:44222|issue:44623", + "issue:43479|issue:43644", + "issue:45245|issue:46132", + "issue:41628|issue:42757", + "issue:44373|issue:44485", + "issue:45071|issue:46082", + "issue:44297|issue:45310", + "issue:46132|issue:46710", + "issue:44222|issue:44484", + "issue:43723|issue:44977", + "issue:44190|issue:46082", + "issue:43232|issue:46424", + "issue:43122|issue:46209", + "issue:44206|issue:45103", + "issue:43232|issue:45834", + "issue:44263|issue:45092", + "issue:44734|issue:46421", + "issue:46209|issue:46431", + "issue:43475|issue:46305", + "issue:46305|issue:46655", + "issue:44295|issue:44623", + "issue:44263|issue:44360", + "issue:44263|issue:44373" + ], + "prepared_review_unit_hash": "b48cfbdd4f307026e5571f7b753f901364b8fbd60e09b7cd85da812645552ef3", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 32, + "serialized_chars": 36373, + "estimated_input_tokens": 9094, + "estimated_eval_tokens": 18444 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a heterogeneous set of unrelated Transformers bug reports. The only mild local cohesion is around the DSA indexer and a few save_pretrained/tokenizer issues, but none of the soft pairs look like true duplicates. No PRs are present.", + "confidence": 0.34, + "canonical_issue_reason": "issue:44263 is the strongest concrete bug report in the set and sits near the small DSA-indexer mini-cluster, so it is the best standalone representative among otherwise mixed topics.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44263 is the most substantive, code-path-specific issue here; the neighboring DSA-indexer reports are related enough to make it the best issue candidate, even though they are not duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Both involve save_pretrained, but one is an FP8 MoE save issue and the other is tokenizer_class metadata mismatch; different bugs and failure modes." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Both are import errors, but for unrelated symbols and causes: transformers.onnx vs missing PyTorch in the environment." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "Optimizer-related only at a broad level; one is ignored SGD args, the other is a docs/parameter-name mismatch in set_training." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both touch imports/dummy objects, but one is a missing image_utils import and the other is leftover dummy_pt_objects cleanup." + }, + { + "left": "issue:44360", + "right": "issue:46762", + "accept": false, + "reason": "Both are numerical/implementation complaints, but they target different attention/indexing code paths and are not the same bug." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "FP8 accuracy regression and processor.save_pretrained missing files are unrelated defects." + }, + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "One reports version-to-version output differences; the other is a behavior change in AttentionInterface.register." + }, + { + "left": "issue:44297", + "right": "issue:45656", + "accept": false, + "reason": "Tokenizer metadata save bug vs optimizer step being called twice under deepspeed; no shared code-path bug." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Both are tokenizer/import-adjacent, but one is BPE marker handling and the other is AutoTokenizer loading failure in v5." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both mention save_pretrained, but one is FP8 MoE model saving and the other is processor file emission; different artifacts." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Config default initialization with None and non-persistent buffer junk are different model-initialization problems." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Category-count overflow and AttentionInterface.register behavior are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they concern different packages/symbols and different root causes." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "Docstring mismatch for position_ids vs GLM-5 RoPE implementation; one is documentation, the other is model logic." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "Type-checking regression in v5.4.0 vs LlamaConfig head_dim validation; not the same underlying bug." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer save metadata bug and Qwen3.5 MoE from_pretrained error are separate issues." + }, + { + "left": "issue:46132", + "right": "issue:46710", + "accept": false, + "reason": "Attention registration semantics and DeepSeek accuracy regression do not share the same concrete bug." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 model save bug vs a question about max_shard_size defaults; no duplicate relationship." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading failure and flash-attention generation issue are unrelated." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Dataset loading script issue vs config validation on head_dim; different subsystem and bug." + }, + { + "left": "issue:43232", + "right": "issue:46424", + "accept": false, + "reason": "Generation kwargs after sync_gpus vs cache max_length semantics are separate cache/generation behaviors." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenization regression across versions and FP8 inference on B200 are unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45103", + "accept": false, + "reason": "Feature-extractor center-arg crash vs auto_docstring annotations crash; different code paths." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwarg handling vs Kosmos2.5 OCR index error are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "DSA indexer torch.split return values and remote-code checkpoint meta-init incompatibility are different bugs." + }, + { + "left": "issue:44734", + "right": "issue:46421", + "accept": false, + "reason": "KV-cache continuation indexing crash vs lazy cache initialization causing recompilation are not the same defect." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "FP8 inference on B200 and missing chat template on a processor are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "SAM3 vision encoder attribute error vs MoonshineStreaming sliding-window off-by-one are different model bugs." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Both are sequence/padding style bugs, but in different processors with different off-by-one mechanics." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Buffer access error for position_ids vs missing processor save_pretrained files are unrelated." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Both mention the DSA indexer, but one is a torch.split return-value issue and the other is a missing ReLU; related area, not the same bug." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "DSA indexer bug vs a docstring error for position_ids; clearly not duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 47, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28218", + "issue:30990", + "issue:33357", + "issue:38175", + "issue:39401", + "issue:41628", + "issue:42371", + "issue:42757", + "issue:42796", + "issue:42831", + "issue:42886", + "issue:42907", + "issue:42994", + "issue:43066", + "issue:43122", + "issue:43232", + "issue:43408", + "issue:43576", + "issue:43644", + "issue:43646", + "issue:43784", + "issue:43824", + "issue:44038", + "issue:44188", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44373", + "issue:44483", + "issue:44568", + "issue:44829", + "issue:44945", + "issue:45092", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45381", + "issue:45406", + "issue:45593", + "issue:45753", + "issue:46002", + "issue:46050", + "issue:46305", + "issue:46328", + "issue:46369", + "issue:46402" + ], + "soft_pairs": [ + "issue:30990|issue:38175", + "issue:42831|issue:45310", + "issue:44279|issue:46328", + "issue:45381|issue:46305", + "issue:42907|issue:46050", + "issue:45092|issue:45357", + "issue:43644|issue:46305", + "issue:44829|issue:46002", + "issue:22355|issue:27957", + "issue:44261|issue:45092", + "issue:44483|issue:45406", + "issue:42831|issue:44038", + "issue:41628|issue:43824", + "issue:43408|issue:46402", + "issue:42831|issue:45753", + "issue:22355|issue:28218", + "issue:39401|issue:44568", + "issue:27957|issue:30990", + "issue:42886|issue:43066", + "issue:42757|issue:45278", + "issue:44188|issue:44373", + "issue:42886|issue:42994", + "issue:43408|issue:46369", + "issue:43232|issue:45593", + "issue:44261|issue:44373", + "issue:42796|issue:43122", + "issue:43646|issue:46002", + "issue:42907|issue:44945", + "issue:30990|issue:33357", + "issue:43576|issue:43784", + "issue:42371|issue:44263" + ], + "prepared_review_unit_hash": "ddd044e59eba6fae79851d2b0b2fa7e4d3639d460107d0b435629d2194c2d757", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36488, + "estimated_input_tokens": 9122, + "estimated_eval_tokens": 18500 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is very heterogeneous: most pairs are different bug reports with only superficial topical overlap. The clearest umbrella issue is the broad import-regression report, but there is no PR in the set.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45278 is the broadest umbrella here: it captures multiple post-upgrade import failures and can represent the import-error subgroup better than the narrower, model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick one representative issue, issue:45278 is the best single anchor for the cluster because it is broad and cross-cutting, though the overall set is still not a true duplicate cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Different symptoms and different code paths: loading hangs in Sentence-Transformers vs zero probabilities in a SigLIP2 model." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "Both mention model/precision problems, but one is a FineGrainedFP8 accuracy regression and the other is a Qwen3.5-MoE from_pretrained error." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "A vague dependency issue is not the same bug as a SequenceFeatureExtractor.pad performance regression." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Both are offset/masking style bugs, but they affect different models and different code paths." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both involve quantized model persistence, but one is a dequantized Ministral/Devstral save failure and the other is a sinq save/load regression." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Different bugs: remote-code/meta-init incompatibility for old InternVL2 vs incorrect visual encoder keys on Qwen3.5 save_pretrained." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Unrelated regressions: non-persistent buffer initialization junk vs a MoonshineStreaming sliding-window off-by-one." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Different failure modes: degenerate training with flash_attention_3 vs NaNs at checkout/init in classification heads." + }, + { + "left": "issue:22355", + "right": "issue:27957", + "accept": false, + "reason": "Import failure for transformers.onnx is unrelated to an XLMRoberta Flash Attention 2 issue." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "Separate problems: missing rms_norm_eps in MLA layernorm vs remote-code meta-initialization incompatibility." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "One is an API request rejection in chat/completions; the other is a processor attribute crash in transformers serve." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "Both mention Qwen/MoE, but the reported failures are different and do not point to the same code path." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but for different symbols and likely different missing-export regressions." + }, + { + "left": "issue:43408", + "right": "issue:46402", + "accept": false, + "reason": "Both involve sam3, but a model-type warning is not the same as missing text-encoder weights loading." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "Different issues: FP8 accuracy regression vs infinite loop on a specific image in Qwen3_5." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "An ONNX import error is unrelated to tokenizer adding an extra space after an added token." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer bugs, but one is offset_mapping on Qwen3 and the other is BOS/EOS handling on mdeberta-v3-base." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Different subsystems and symptoms: Flash Attention 2 on XLMRoberta vs Sentence-Transformers loading hang." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache loading and wrong decoder type are different tokenizer/version regressions." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "A specific huggingface_hub import error is not the same underlying bug as a broad set of transformers import failures." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to a wrong docstring for position_ids." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Tokenizer offline cache loading and quantized model saving are different problems." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "Same model family, but a model-type warning is not the same as a memory overflow in sam3-litetext." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation sync_gpus state handling is unrelated to D-FINE auxiliary-loss behavior." + }, + { + "left": "issue:44261", + "right": "issue:44373", + "accept": false, + "reason": "A precision/config bug in MLA q_a_layernorm is not the same as a documentation mismatch." + }, + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Tokenizer BPE-marker handling for a specific model is not the same as a version-wide tokenization change report." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Custom model initialization breakage and NaNs in classification heads are different failure modes." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Quantized save failures are unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Loading stuck in one model stack is not the same as a MacOS bus error in another." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "A broken CLI env command is not the same bug as a NameError when importing sentence-transformers." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration guidance and a torch.split return-value 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": 24, + "review_unit_index": 48, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:29159", + "issue:36296", + "issue:39401", + "issue:41628", + "issue:42175", + "issue:42757", + "issue:42907", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43756", + "issue:43784", + "issue:43824", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44060", + "issue:44117", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44360", + "issue:44485", + "issue:44488", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44938", + "issue:44945", + "issue:44964", + "issue:45200", + "issue:45216", + "issue:45278", + "issue:45290", + "issue:45478", + "issue:45600", + "issue:46172", + "issue:46326", + "issue:46614", + "issue:46762" + ], + "soft_pairs": [ + "issue:44295|issue:44297", + "issue:43756|issue:44060", + "issue:39401|issue:44488", + "issue:43976|issue:44945", + "issue:46172|issue:46762", + "issue:42907|issue:44964", + "issue:42907|issue:43937", + "issue:44117|issue:44704", + "issue:43784|issue:45600", + "issue:41628|issue:45278", + "issue:44704|issue:45290", + "issue:44625|issue:44829", + "issue:39401|issue:44521", + "issue:44360|issue:44485", + "issue:43408|issue:43475", + "issue:43122|issue:44521", + "issue:44038|issue:44297", + "issue:43122|issue:46326", + "issue:42907|issue:43452", + "issue:39401|issue:46614", + "issue:22355|issue:29159", + "issue:44038|issue:44295", + "issue:45216|issue:45478", + "issue:43116|issue:45200", + "issue:44038|issue:44222", + "issue:43824|issue:44351", + "issue:43122|issue:44779", + "issue:42757|issue:44351", + "issue:36296|issue:43116", + "issue:43278|issue:44945", + "issue:22355|issue:29127", + "issue:42175|issue:44938" + ], + "prepared_review_unit_hash": "e3a1cd42e9a7200d5fc949132cac12d3fb49c5cff2c23d700438fdde92b97732", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "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 very noisy and mostly consists of unrelated model-, tokenizer-, import-, and training-regression reports. I do not see any soft edge that clearly represents the same underlying bug/change.", + "confidence": 0.24, + "canonical_issue_reason": "Issue 45278 is the broadest umbrella in the set: it captures a general import-regression report after a version upgrade, which is the closest central theme among the otherwise heterogeneous issues.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best available anchor for this cluster because it is the most general and could subsume some of the narrower import-error reports, though the overall cluster is not a single coherent duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Different bugs: buffer registration/position_ids access versus tokenizer_class metadata saved in tokenizer_config.json." + }, + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Both are model-specific, but one is RoPE-layer dropping and the other is a tied-weights warning in Qwen3-Next; no shared code-path bug." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Both involve tokenizers, but one is wrong offset_mapping for Qwen3 and the other is a model loading failure for cjvt/sleng-bert." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Unrelated failures: Python-version compatibility versus incorrect output under pipeline parallelism." + }, + { + "left": "issue:46172", + "right": "issue:46762", + "accept": false, + "reason": "Different model implementations and symptoms; no evidence they are the same numerical bug." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Different model families and different failure modes: save-dequantized Minstrals/Devstrals versus loading Phi-4 multimodal." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Different regressions: dequantized save failure versus GLM-5 GenerationConfig validation." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "Tokenizer mapping fallback bug versus AutoProcessor not forwarding kwargs to cached_file; separate code paths." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Both are import/config issues, but one is a NameError in sentence-transformers integration and the other is stale auto_mappings references for Sam3LiteText." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Both are import-related, but 45278 is a broad report of many import errors and 41628 is one specific missing export; not enough to call them the same bug." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "AutoProcessor kwargs forwarding is unrelated to apply_chat_template crashing on assistant tool-call messages." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "Different issues: num_labels propagation in Qwen3.5 versus degenerate training with flash_attention_3." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer offset_mapping bug versus multimodal assistant_mask generation; different behavior and code paths." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "Completely different topics: DSA indexer ReLU discussion versus GLM-5 RoPE implementation." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Both are SAM 3 Video-related, but one is a model-type warning and the other is a missing attribute in vision encoder output." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Tokenization regression versus multimodal chat-template assistant masks; not the same underlying defect." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe loading/training issue versus tokenizer save_pretrained metadata mismatch." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "Different tokenizer bugs: version-to-version tokenization differences versus continue_final_message on mistral_common tokenizers." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Saving dequantized Minstrals/Devstrals is unrelated to gguf_file breaking AutoTokenizer/AutoModel loading." + }, + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Both mention tokenizers, but one is Qwen3 offset_mapping and the other is stale phonemizer language state in Wav2Vec2PhonemeCTCTokenizer." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "No module named transformers.onnx is unrelated to slow-vs-fast tokenizer inconsistencies." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Different problems: Qwen3-VL-Moe bug versus position_ids buffer access error." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5 and newer versions, but one is save_pretrained checkpoint corruption and the other is from_pretrained loading failure; not clearly the same bug." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Example-script multi-label output bug versus Gemma 4 mm_token_type_ids defaulting; unrelated." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "Qwen3-VL-Moe loading issue versus FP8 save_pretrained moe issue; related only by broad model family." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Different missing symbols from different APIs; both are import errors but not the same concrete bug." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but one is generic tokenization changes in v5 and the other is Deepseek-specific incorrect results." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Different packages and failure modes: huggingface_hub missing is_offline_mode versus transformers missing HybridCache." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training bug is unrelated to the example script returning empty multi-label results." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Dtype mismatch during evaluate is unrelated to incorrect output under pipeline parallelism." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "No module import failure versus a LayoutLMv3 error-message clarity issue; not duplicates." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Backend packaging/install issue versus Python 3.14 load failure; different root causes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 49, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28348", + "issue:38617", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42898", + "issue:43066", + "issue:43116", + "issue:43441", + "issue:43493", + "issue:43611", + "issue:43618", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44188", + "issue:44263", + "issue:44355", + "issue:44485", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44936", + "issue:44995", + "issue:45072", + "issue:45083", + "issue:45092", + "issue:45278", + "issue:45357", + "issue:45446", + "issue:45478", + "issue:45753", + "issue:46002", + "issue:46231", + "issue:46328", + "issue:46614", + "issue:46640", + "issue:46697", + "issue:46710", + "issue:46752", + "issue:46762", + "issue:46829" + ], + "soft_pairs": [ + "issue:43493|issue:45753", + "issue:44188|issue:46328", + "issue:42898|issue:46614", + "issue:43937|issue:44485", + "issue:38617|issue:45278", + "issue:43066|issue:44779", + "issue:43611|issue:46002", + "issue:44355|issue:46231", + "issue:43116|issue:43493", + "issue:44038|issue:45478", + "issue:43976|issue:45446", + "issue:42673|issue:45072", + "issue:45357|issue:45478", + "issue:44625|issue:46002", + "issue:45083|issue:45446", + "issue:42371|issue:43874", + "issue:43618|issue:46829", + "issue:43116|issue:43825", + "issue:44075|issue:44936", + "issue:43992|issue:44704", + "issue:42222|issue:43441", + "issue:44995|issue:46697", + "issue:43066|issue:46710", + "issue:28348|issue:46640", + "issue:44263|issue:46762", + "issue:46614|issue:46752", + "issue:45092|issue:46762", + "issue:43827|issue:45446" + ], + "prepared_review_unit_hash": "8d73576e22545e149b0fedf0670b1e5be32b4f8cf2405e0dac39de7044780ab7", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35186, + "estimated_input_tokens": 8797, + "estimated_eval_tokens": 17850 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose theme match around Transformers v5 regressions, but the individual issues point to different concrete bugs across models, tokenizers, processors, import paths, and docs. I don\u2019t see any soft-edge pair that is clearly the same underlying defect.", + "confidence": 0.88, + "canonical_issue_reason": "No single issue cleanly represents the whole cluster; if one must be chosen as the broadest umbrella, issue 45278 is the closest because it covers general import regressions after upgrading Transformers, but it is still not a true duplicate of the rest.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the best global representative only in a very loose sense because it is the broadest regression report in the set; however, the cluster is heterogeneous and it should not be treated as a duplicate anchor for the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "Different models and symptoms: SigLIP2 implementation discrepancy vs a Qwen3_5 infinite-loop bug on one image." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "Unrelated code paths: attention-kernel branching under torch.compile vs a feature-extractor padding performance issue." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "Different subsystems: tokenizer-space cleanup behavior in v5 vs stale espeak backend state in phonemization." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "Both are GLM-5-related, but one is GenerationConfig validation and the other is RoPE implementation; not the same bug." + }, + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "45278 is a broad import-error report, but 38617 is one specific missing symbol import; not enough evidence they are the same underlying defect." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both involve tokenizers, but one is a decoder-type regression and the other is incorrect DeepSeek tokenization output." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "Model loading via base_model_prefix is unrelated to NaNs in classification heads during checkout/init." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled-Python runtime errors are unrelated to a documentation parameter-name mismatch." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script multi-label output bug is unrelated to a SigLIP2 implementation discrepancy." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Related Qwen/MoE area, but different models and different failure modes: one bug in Qwen3-VL-Moe, the other a from_pretrained error in Qwen3.5 Moe." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version incompatibility is a different problem from a flex_attention import/version check bug." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM is unrelated to bfloat16 dtype mismatches in different models." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Both touch Qwen3.5 model handling, but save_pretrained key mapping and from_pretrained loading failure are distinct bugs." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "num_labels propagation is a config-wiring bug, not the same as NaNs in classification-head initialization." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "A helper length-calculation bug in qwen3_omni_moe is unrelated to a PyTorch version-check import issue." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API guidance and missing image-patch-count method are unrelated." + }, + { + "left": "issue:43618", + "right": "issue:46829", + "accept": false, + "reason": "Both are output-field regressions, but they affect different model output classes and are not clearly the same code-path bug." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "A bad example-script result is not the same as an incorrect pipeline error message about translation support." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "Optimizer argument handling and trainer evaluate-after-train failure are different issues." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "A missing embed_tokens weight in UMT5Encoder loading is unrelated to AutoProcessor kwargs forwarding." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "VitPose breakage and Ministral FlashAttention failure are different model-specific regressions." + }, + { + "left": "issue:44995", + "right": "issue:46697", + "accept": false, + "reason": "A stale GlmMoeDsa indexer cache bug is unrelated to Swin2SR image over-padding." + }, + { + "left": "issue:43066", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer decoder-type regression is not the same as a downstream accuracy regression in DeepSeek-R1-Distill-Llama-8B." + }, + { + "left": "issue:28348", + "right": "issue:46640", + "accept": false, + "reason": "One is a feature request for GPT2 flash-attention support; the other is an unspecified Flash Attention 2 issue." + }, + { + "left": "issue:44263", + "right": "issue:46762", + "accept": false, + "reason": "torch.split return handling in GlmMoeDsaIndexer is unrelated to Minimax MSA numerical-difference regressions." + }, + { + "left": "issue:46614", + "right": "issue:46752", + "accept": false, + "reason": "Stale phonemizer language state and empty-list chat-template IndexError are unrelated bugs." + }, + { + "left": "issue:45092", + "right": "issue:46762", + "accept": false, + "reason": "Remote-code checkpoint/meta-init incompatibility is unrelated to attention numerical differences." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Docs still referencing pipeline() after removals is unrelated to the flex_attention import/version-check 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": 24, + "review_unit_index": 50, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:33453", + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42617", + "issue:42673", + "issue:42796", + "issue:43064", + "issue:43122", + "issue:43408", + "issue:43450", + "issue:43673", + "issue:43742", + "issue:43937", + "issue:43986", + "issue:44117", + "issue:44291", + "issue:44380", + "issue:44410", + "issue:44466", + "issue:44484", + "issue:44779", + "issue:44829", + "issue:44877", + "issue:44977", + "issue:45071", + "issue:45081", + "issue:45216", + "issue:45310", + "issue:45341", + "issue:45356", + "issue:45362", + "issue:45405", + "issue:45446", + "issue:45561", + "issue:45563", + "issue:45588", + "issue:46002", + "issue:46077", + "issue:46178", + "issue:46231", + "issue:46298", + "issue:46328", + "issue:46512", + "issue:46614", + "issue:46655", + "issue:46752" + ], + "soft_pairs": [ + "issue:43064|issue:43673", + "issue:42673|issue:42796", + "issue:43937|issue:45563", + "issue:33453|issue:39401", + "issue:45356|issue:46614", + "issue:43673|issue:46328", + "issue:44380|issue:46077", + "issue:42673|issue:46655", + "issue:42222|issue:46752", + "issue:44291|issue:45071", + "issue:45405|issue:45446", + "issue:44466|issue:46002", + "issue:44117|issue:45588", + "issue:39401|issue:44779", + "issue:44410|issue:45310", + "issue:43937|issue:44977", + "issue:45071|issue:46512", + "issue:45341|issue:46328", + "issue:44829|issue:45216", + "issue:45341|issue:45561", + "issue:43450|issue:43986", + "issue:44117|issue:45081", + "issue:42617|issue:45341", + "issue:38617|issue:41628", + "issue:43122|issue:46298", + "issue:43408|issue:43742", + "issue:45362|issue:46328", + "issue:44877|issue:46231", + "issue:42222|issue:46178", + "issue:44484|issue:45310" + ], + "prepared_review_unit_hash": "99973ae48a4e236842c23c33e178d5946e5cf8a981849a4894d17f6686e7ad24", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36692, + "estimated_input_tokens": 9173, + "estimated_eval_tokens": 18602 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of distinct Transformers issues around tokenizers, model loading, generation, and testing utilities. I did not find any soft pair that looks like the same underlying bug/change.", + "confidence": 0.36, + "canonical_issue_reason": "issue:43122 is the most representative broad regression here: a generic tokenizer behavior change across versions, with some nearby tokenizer issues but no exact duplicate in the provided set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43122 is the best global issue anchor because it is broad, version-related, and close to the main tokenizer-regression theme that many other items in the cluster orbit around.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Different areas and failure modes: FSDP/PEFT optimizer state handling vs generation cache missing during chunked prefill." + }, + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "Unrelated bugs: VRAM leak in Qwen3ForCausalLM multithreaded dataloaders vs tokenizer BPE-marker handling." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Both mention generation, but one is invalid GenerationConfig for GLM-5 and the other is a stale warning about num_return_sequences." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Both are tokenizer issues, but one is a generic loading regression and the other is Qwen3 offset_mapping correctness; different bugs." + }, + { + "left": "issue:45356", + "right": "issue:46614", + "accept": false, + "reason": "Different tokenizers and code paths: Kimi-K2.5 codec/fix_mistral_regex regression vs Wav2Vec2 phonemizer backend state bug." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Generation cache bug vs SequenceFeatureExtractor.pad performance issue; not the same underlying defect." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Both are attention-related configs, but one ignores GPT2 attention scaling under SDPA/FlashAttention while the other builds the wrong mask for Gemma4 bidirectional attention." + }, + { + "left": "issue:42673", + "right": "issue:46655", + "accept": false, + "reason": "Different subsystems: VRAM leak in model threading vs video processor padding calculation." + }, + { + "left": "issue:42222", + "right": "issue:46752", + "accept": false, + "reason": "Vitpose model breakage is unrelated to apply_chat_template crashing on an empty list." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Both are v5 compatibility regressions, but one is unexpected _is_hf_initialized in init_empty_weights and the other is PretrainedConfig type-checking behavior." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "Version-check issues in different dependencies/code paths: PEFT minimum version bump vs PyTorch version check for flex_attention AuxRequest." + }, + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Serialization of tied weights depending on device vs NaNs in classification heads after checkout/init are different bugs." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES None handling is unrelated to a flash_attention.py crash on s_aux=None." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but they involve different model families and distinct symptoms (offset mapping vs incorrect tokenization results)." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Missing qwen3next projections and Qwen3.5 MoE from_pretrained error are separate loading/model-definition issues." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "Different generation bugs: invalid GenerationConfig for GLM-5 vs Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:45071", + "right": "issue:46512", + "accept": false, + "reason": "PretrainedConfig type-checking breakage is not the same as stale dtype propagation from from_config(dtype=...)." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "One is a vague testing_utils bug report; the other is a specific SequenceFeatureExtractor.pad inefficiency in array conversion." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "Different failure modes: flash_attention_3 causes degenerate training vs save_pretrained saving an incorrect Qwen3.5 checkpoint." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "The second is a specific xdist race on captured_info.txt; the first is too generic and does not clearly describe the same bug." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Video processor batch shape bug vs AutoProcessor crash without torchvision are related broadly to video, but not the same underlying defect." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer-loading failures: missing mapping entry handling vs Mistral regex patch crashing on backend_tokenizer absence." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "3d_parallel.py runtime failure is unrelated to a testing_utils bug." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and code paths; no sign of one shared regression." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Tokenizer behavior regression vs noisy logging message; clearly unrelated." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "SAM3 tracker/video warning mismatch is unrelated to MobileLLM-125M key-loading failure." + }, + { + "left": "issue:45362", + "right": "issue:46328", + "accept": false, + "reason": "Qwen3.5 chat crash and SequenceFeatureExtractor.pad inefficiency are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "Strict config loading for granite_speech is unrelated to a documentation parameter-name typo." + }, + { + "left": "issue:42222", + "right": "issue:46178", + "accept": false, + "reason": "Vitpose model breakage and BACKENDS_MAPPING tensorflow_text error are different issues." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "save_pretrained shard-size question is unrelated to Qwen3.5 MoE from_pretrained loading errors." + } + ] + }, + "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": 24, + "review_unit_index": 51, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:27957", + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43352", + "issue:43441", + "issue:43452", + "issue:43577", + "issue:43673", + "issue:43874", + "issue:43931", + "issue:43937", + "issue:43986", + "issue:44038", + "issue:44075", + "issue:44220", + "issue:44261", + "issue:44263", + "issue:44360", + "issue:44410", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44493", + "issue:44496", + "issue:44829", + "issue:44933", + "issue:45003", + "issue:45071", + "issue:45278", + "issue:45341", + "issue:45464", + "issue:45478", + "issue:45542", + "issue:45656", + "issue:45753", + "issue:46172", + "issue:46209", + "issue:46231", + "issue:46369", + "issue:46617", + "issue:46762" + ], + "soft_pairs": [ + "issue:42371|issue:44261", + "issue:39401|issue:46209", + "issue:43937|issue:44496", + "issue:44410|issue:45478", + "issue:42371|issue:46617", + "issue:44261|issue:46762", + "issue:38617|issue:44220", + "issue:44075|issue:45656", + "issue:44933|issue:45278", + "issue:44263|issue:44485", + "issue:43066|issue:43122", + "issue:43097|issue:44493", + "issue:43673|issue:45341", + "issue:41628|issue:45071", + "issue:43931|issue:44410", + "issue:42222|issue:45542", + "issue:45071|issue:45542", + "issue:44484|issue:45478", + "issue:44483|issue:45464", + "issue:44263|issue:46172", + "issue:44360|issue:46172", + "issue:43874|issue:44261", + "issue:43012|issue:43577", + "issue:27957|issue:43352", + "issue:43931|issue:44038", + "issue:43441|issue:44829", + "issue:43097|issue:43452", + "issue:45003|issue:45341", + "issue:43986|issue:45341", + "issue:44933|issue:46231", + "issue:44261|issue:44485", + "issue:45753|issue:46369" + ], + "prepared_review_unit_hash": "c54c6bf4aa66979e2ce519b3ed53d8dc848ca7e160505ee92a67d88d3808bfb5", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 36741, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab bag of Transformers regressions across v5: import/API breakages, tokenizer changes, model-specific loading bugs, and a few unrelated numerical/runtime issues. I don\u2019t see any soft pair that clearly describes the same underlying bug or a mergeable duplicate PR-sized change.", + "confidence": 0.93, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here: it describes multiple import failures after upgrading to v5.5.0, which is the closest fit for the cluster\u2019s import/API regression subset.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best single representative because it is the most general and central regression report among these issues; the rest are narrower symptom-specific bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 API warning/deprecation vs MLA layernorm epsilon precision bug; different subsystem and failure mode." + }, + { + "left": "issue:39401", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer offset mapping bug vs FP8 inference failure on B200; unrelated behavior and code path." + }, + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "GLM-5 generation config validation error vs unrecognized model/config loading error; different root causes." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Missing qkv/gate projections in qwen3next layer 0 vs Qwen3.5 Moe from_pretrained error; distinct model families and symptoms." + }, + { + "left": "issue:42371", + "right": "issue:46617", + "accept": false, + "reason": "TF32 settings warning vs cache FileNotFoundError for trust_remote_code models; not the same bug." + }, + { + "left": "issue:44261", + "right": "issue:46762", + "accept": false, + "reason": "Layernorm epsilon precision issue vs numerical differences in a different attention implementation; same broad area, but not the same concrete bug." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "Missing import from configuration_utils vs audio fbank feature extraction issue; unrelated code paths." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "Optimizer SGD args ignored vs optimizer step called twice with deepspeed; different optimizer behavior." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "Nonexistent image_utils import and broad import-error report overlap thematically, but the specific missing symbol and upgrade scope are not the same issue." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "torch.split return values bug in GlmMoeDsaIndexer vs GLM-5 RoPE implementation; both GLM-related but different concrete failures." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Wrong tokenizer decoder type vs different tokenization output across versions; related area, but one is type selection and the other is behavioral regression." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Removed tie_embeddings_and_encoder_decoder API vs unexpected position-id keys in model outputs; unrelated breakages." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache missing during chunked prefill vs a testing_utils bug; no shared underlying defect." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "AutoImageProcessor import failure vs PreTrainedConfig type checking breakage; both API regressions, but not the same concrete one." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Qwen3-VL-30B-A3B-Instruct weight-shape mismatch vs qwen3next missing projections; different model load failures." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose model breakage vs TensorBoard/TensorFlow backend detection bug; unrelated subsystems." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "Config type-checking regression vs undefined tf backend when only tensorboard is installed; different API paths." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained sharding-size question vs Qwen3.5 Moe from_pretrained error; not the same bug." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "chat/completions endpoint rejection vs streaming inference failure on Qwen3.5-0.8B; both API-level, but distinct failures." + }, + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "GlmMoeDsaIndexer split behavior vs deepseek-v4 dtype/transpose bug; different model internals." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "DSA indexer missing ReLU vs deepseek-v4 comb transpose bug; unrelated implementation issues." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing get_number_of_image_patches in image processor vs MLA layernorm eps precision bug; different components." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "PyTorch warning when compiling bf16 models vs BLIP2 dtype remaining float32; related only by dtype, not the same defect." + }, + { + "left": "issue:27957", + "right": "issue:43352", + "accept": false, + "reason": "XLMRoberta Flash Attention 2 support vs NemotronHForCausalLM Flash Attention 2.0 support; same feature area, but separate model-specific support requests." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Qwen3-VL weight mismatch vs Qwen3-VL-Moe bug; both VL models, but the reported load failures are different." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Ministral-3 FlashAttention failure vs flash_attention_3 training degeneration; both attention-related, but different code paths and symptoms." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "Removed config API vs gguf_file loading breakage; unrelated regression types." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafe sys.modules access in modeling_utils vs testing_utils bug; different modules and bug classes." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "AutoProcessor crash without torchvision vs testing_utils bug; no shared root cause." + }, + { + "left": "issue:44933", + "right": "issue:46231", + "accept": false, + "reason": "Missing image_utils import vs docs mismatch for set_training(); one is a runtime import error, the other documentation only." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "MLA layernorm epsilon precision issue vs GLM-5 RoPE implementation; both numeric, but not the same bug." + }, + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "Qwen3_5 infinite loop on a specific image vs sam3-litetext memory overflow; entirely different failure modes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 52, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:38617", + "issue:42757", + "issue:42890", + "issue:42907", + "issue:43012", + "issue:43064", + "issue:43097", + "issue:43502", + "issue:43531", + "issue:43673", + "issue:43873", + "issue:43874", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44062", + "issue:44117", + "issue:44263", + "issue:44315", + "issue:44351", + "issue:44380", + "issue:44410", + "issue:44484", + "issue:44655", + "issue:44704", + "issue:44936", + "issue:44938", + "issue:44987", + "issue:45103", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45440", + "issue:45588", + "issue:45753", + "issue:46050", + "issue:46123", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365", + "issue:46421", + "issue:46655", + "issue:46664", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821" + ], + "soft_pairs": [ + "issue:44117|issue:44987", + "issue:43531|issue:45341", + "issue:43874|issue:44263", + "issue:45440|issue:46172", + "issue:43673|issue:46123", + "issue:43097|issue:45588", + "issue:44410|issue:45375", + "issue:44038|issue:44410", + "issue:38617|issue:42757", + "issue:43064|issue:44315", + "issue:43986|issue:45372", + "issue:43097|issue:46050", + "issue:43986|issue:44484", + "issue:46655|issue:46664", + "issue:42890|issue:46298", + "issue:46298|issue:46752", + "issue:46172|issue:46710", + "issue:38617|issue:44351", + "issue:43097|issue:43873", + "issue:43012|issue:46821", + "issue:44062|issue:44117", + "issue:44117|issue:46701", + "issue:43502|issue:43992", + "issue:44936|issue:46365", + "issue:44704|issue:45103", + "issue:44380|issue:46421", + "issue:44938|issue:46178", + "issue:43986|issue:44117", + "issue:42907|issue:44655", + "issue:42907|issue:45753" + ], + "prepared_review_unit_hash": "60cbce8c38a3bce797a88feb3491014384f683070ab0e785364a4185176f477f", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36464, + "estimated_input_tokens": 9116, + "estimated_eval_tokens": 18488 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: it mixes unrelated import errors, model-specific regressions, trainer/distributed bugs, processor issues, and deprecation complaints. None of the soft-edge pairs look like the same underlying defect, so no pair should be merged as duplicates. If one issue must anchor the cluster, the broadest and most user-visible is the v5 breaking-change report around `tie_embeddings_and_encoder_decoder` removal.", + "confidence": 0.93, + "canonical_issue_reason": "issue:43097 is the strongest representative: it describes a concrete v5 breaking change with broader user impact and inbound references, unlike the other narrowly scoped or one-off model/runtime bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43097 is the best global issue candidate because it is a clear, high-impact regression/deprecation complaint that is more representative of the cluster than the many unrelated niche failures.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Both involve tokenizer/model loading, but one is a mapping-lookup bug and the other is a model load failure; not the same concrete failure path." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Unrelated topics: Qwen3-MoE sliding-window behavior vs a testing_utils bug." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "Different GLM-related code paths; image processor multimodal token counting vs a torch.split return-value issue." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Both mention DeepSeek, but they describe different model implementations and different tensor/code-path bugs." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Cache initialization during chunked prefill is not the same as a pipeline batch-finalization bug." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "A deprecation/removal complaint is unrelated to a flash-attention AttributeError on sink-less models." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Different Qwen3 variants and different missing fields/projections; not one shared bug." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Both are Qwen3-family issues, but one is a transformers version bug and the other is a missing projection in layer 0." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are ImportErrors, but for different missing symbols from different packages and likely different fixes." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "Distributed optimizer-state corruption and Liger-kernel application are separate training pipeline problems." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Both are loading crashes, but one is torchvision-related video processor loading and the other is mistral_common/Gemma processor loading." + }, + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "Saving a dequantized model is a different defect from sinq quantization breakage across save/load." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Processor loading without torchvision is unrelated to the `max_shard_size` default question." + }, + { + "left": "issue:46655", + "right": "issue:46664", + "accept": false, + "reason": "Same SmolVLM video processor area, but padding math and `num_frames` semantics are distinct bugs." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "A flaky test setup issue and a noisy logging message are unrelated." + }, + { + "left": "issue:46298", + "right": "issue:46752", + "accept": false, + "reason": "Noisy logging and an `apply_chat_template` empty-list crash have no shared failure path." + }, + { + "left": "issue:46172", + "right": "issue:46710", + "accept": false, + "reason": "DeepSeek tensor-shape code is unrelated to the reported accuracy regression in a different model family." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Different missing imports (`layer_type_validation` vs `HybridCache`) and likely different version-skew fixes." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "A v5 removal/deprecation issue is not the same as offloading failing under quantization." + }, + { + "left": "issue:43012", + "right": "issue:46821", + "accept": false, + "reason": "PyTorch warning behavior for bfloat16 compilation is unrelated to a TimesFM `window_size` AttributeError." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer construction error vs tokenizer-mapping `None` handling; different code paths." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "A tokenizer mapping assumption bug is not the same as a DiffusionGemma image-text pipeline failure." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "Both are from_pretrained-related, but one is offline/download policy leakage and the other is a missing weight in UMT5 loading." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "Trainer evaluate-after-train behavior is unrelated to an import failure caused by datasets." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "Passing kwargs to cached_file and auto_docstring annotation handling are entirely separate bugs." + }, + { + "left": "issue:44380", + "right": "issue:46421", + "accept": false, + "reason": "Attention-scaling config ignored in GPT2 backends is not the same as chunked-prefill cache recompilation." + }, + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Python 3.14 import/load failure is unrelated to the BACKENDS_MAPPING tensorflow_text error." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Optional-dependency processor loading and tokenizer-mapping `None` handling are different failures." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Saving a dequantized model and saving Pipeline objects involve different serialization paths and bugs." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Model save/load breakage and a Qwen3_5 image infinite loop are unrelated defects." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 24, + "review_unit_index": 53, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:36683", + "issue:38617", + "issue:42175", + "issue:42490", + "issue:42757", + "issue:42994", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "issue:43618", + "issue:43756", + "issue:43761", + "issue:43824", + "issue:43825", + "issue:43867", + "issue:43906", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44077", + "issue:44162", + "issue:44222", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44625", + "issue:44655", + "issue:45478", + "issue:45479", + "issue:45584", + "issue:45712", + "issue:46132", + "issue:46172", + "issue:46298", + "issue:46365" + ], + "soft_pairs": [ + "issue:43825|issue:44655", + "issue:43208|issue:43550", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:43012|issue:44368", + "issue:43992|issue:44484", + "issue:43066|issue:43906", + "issue:43352|issue:43976", + "issue:43352|issue:44380", + "issue:43097|issue:46298", + "issue:44162|issue:46172", + "issue:44380|issue:46132", + "issue:45584|issue:46365", + "issue:44077|issue:46365", + "issue:44222|issue:45478", + "issue:36683|issue:43582", + "issue:44380|issue:45479", + "issue:42175|issue:46365", + "issue:43064|issue:44380", + "issue:42757|issue:43824", + "issue:42490|issue:43012", + "issue:43208|issue:43526", + "issue:44483|issue:44514", + "issue:43618|issue:44380", + "issue:38617|issue:43502", + "issue:44038|issue:44230", + "issue:43618|issue:43761", + "issue:43756|issue:44485", + "issue:42994|issue:44655", + "issue:43299|issue:43867", + "issue:44625|issue:46365" + ], + "prepared_review_unit_hash": "562963a2bd591cfbca562c05b6bbdd9c081a2bae40be221f79520531021adfd4", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 31, + "serialized_chars": 36149, + "estimated_input_tokens": 9038, + "estimated_eval_tokens": 18332 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues. The soft pairs are similar at a surface level (imports, attention, save/load, Qwen/CLIP, etc.) but do not look like the same underlying bug or change, so none should be merged as duplicates.", + "confidence": 0.74, + "canonical_issue_reason": "issue:46365 is the best representative issue here: it is currently open, user-facing, and describes a concrete import conflict affecting Trainer. The other issues are either narrower, model-specific, already closed, or about different code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:46365 is the strongest overall issue to anchor triage because it is broad, actionable, and high-impact. It is a clearer canonical candidate than the other reports in this set, which are mostly isolated regressions or feature-specific bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Different failures: one is a pipeline task-support error message, the other is save_pretrained failing for Pipeline objects. Same area, but not the same bug." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "Different models and code paths: xLSTM training blockers vs Bamba torch.compile/SDPA failure." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A PyTorch precision warning on bfloat16 compilation is not the same as a model-specific torch.compile SDPA crash." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "One is about leftover dummy classes leaking into the API surface; the other is just a noisy logging message. No shared bug." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Both mention warnings, but they come from different causes: compiler precision behavior vs tie_word_embeddings config warnings in Qwen3.5 LoRA." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "UMT5 missing embed_tokens.weight on load is unrelated to the max_shard_size default in save_pretrained." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Wrong tokenizer decoder type and an isolated reproduction of another issue are not the same underlying defect." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Nemotron Flash Attention gating and Python 3.9/3.10 compatibility are separate issues." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Both touch attention backends, but one is unsupported Flash Attention for a model and the other is GPT2 attention-scaling config being ignored. Different bugs." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "Removal of tie_embeddings_and_encoder_decoder is unrelated to a noisy logging message." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 breakage and a DeepSeek-V4 tensor transpose issue are different model-specific failures." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "GPT2 backend config loss is not the same as AttentionInterface.register changing the registered function behavior." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper empty-transcription failure after token alignment is unrelated to the Trainer import conflict." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "patchtsmixer post_init validation and importing Trainer with datasets installed are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "Both are MoE/save/load themed, but one is FP8 save_pretrained and the other is a Qwen3.5 Moe from_pretrained regression on newer transformers. Not the same bug." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Gemma3Config missing vocab_size and an Apple Silicon TypeError in caching_allocator_warmup are unrelated." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "GPT2 attention scaling under SDPA/FlashAttention is unrelated to the degenerate zero-loss classification bug." + }, + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "TensorFlow backend packaging and a Trainer import conflict involving datasets are different dependency issues." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer state corruption is a distributed training bug, not an attention-scaling regression." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are ImportErrors, but they involve different symbols, packages, and likely causes." + }, + { + "left": "issue:42490", + "right": "issue:43012", + "accept": false, + "reason": "Local path trailing-slash warning is unrelated to a bfloat16 precision warning during compile." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training blockers and BeitImageProcessorFast reduce_labels returning one label are unrelated." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "A chat-completions API rejection is not the same as Qwen2_5_VLProcessor crashing on batched apply_chat_template input." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIPOutput attentions assignment and GPT2 attention scaling under backends are different regressions." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "Missing layer_type_validation import and local_files_only still triggering network calls are unrelated import/network bugs." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "One is a general Qwen3-VL-Moe breakage on transformers 5.0, the other is a request for FP8 inference support. Not the same change." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both involve CLIP, but one loses attentions and the other loses hidden_states. Different outputs, different failure modes." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "Smollm3 RoPE layer dropping and GLM-5 RoPE implementation are model-specific and not the same bug." + }, + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Quantized model saving failure and saving Pipeline objects are both save-related, but they are different object types and failure paths." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Qwen3VL-Moe loading regression and a state_dict sorting load error are not the same underlying issue." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "num_labels propagation in Qwen3.5 is unrelated to the Trainer import conflict with datasets." + } + ] + }, + "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": 24, + "review_unit_index": 54, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:29870", + "issue:33453", + "issue:36010", + "issue:36296", + "issue:36683", + "issue:41669", + "issue:42757", + "issue:42890", + "issue:43010", + "issue:43064", + "issue:43208", + "issue:43352", + "issue:43475", + "issue:43525", + "issue:43575", + "issue:43611", + "issue:43618", + "issue:43867", + "issue:43901", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44075", + "issue:44077", + "issue:44246", + "issue:44355", + "issue:44483", + "issue:44492", + "issue:44625", + "issue:44655", + "issue:44855", + "issue:44938", + "issue:45071", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45310", + "issue:45357", + "issue:45375", + "issue:45478", + "issue:45561", + "issue:45850", + "issue:46077", + "issue:46082", + "issue:46202", + "issue:46365", + "issue:46369", + "issue:46829" + ], + "soft_pairs": [ + "issue:45276|issue:46077", + "issue:36683|issue:43475", + "issue:36010|issue:42757", + "issue:43352|issue:43937", + "issue:44483|issue:46369", + "issue:43611|issue:44655", + "issue:43901|issue:44655", + "issue:43867|issue:45216", + "issue:36683|issue:44077", + "issue:43976|issue:44938", + "issue:44625|issue:46077", + "issue:46077|issue:46082", + "issue:43208|issue:46829", + "issue:43618|issue:45216", + "issue:36683|issue:43525", + "issue:36683|issue:44625", + "issue:29870|issue:36296", + "issue:43064|issue:44492", + "issue:43992|issue:45310", + "issue:43010|issue:45278", + "issue:44075|issue:46365", + "issue:29870|issue:33453", + "issue:44355|issue:44855", + "issue:44938|issue:45071", + "issue:43867|issue:46202", + "issue:36683|issue:43064", + "issue:43867|issue:45375", + "issue:42890|issue:45561", + "issue:41669|issue:44246", + "issue:43010|issue:45850", + "issue:36296|issue:43575", + "issue:43010|issue:43976", + "issue:43618|issue:44077", + "issue:43992|issue:45478", + "issue:43867|issue:45357" + ], + "prepared_review_unit_hash": "3e159928e082047efcf8955938674f73f2b73d94936a85b82c69d9ba37a79027", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37819, + "estimated_input_tokens": 9455, + "estimated_eval_tokens": 19166 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose collection of unrelated Transformers regressions; the soft pairs share broad themes like loading, importing, or config handling, but not the same concrete bug. No soft edge looks safe to merge as a duplicate/combined PR candidate.", + "confidence": 0.42, + "canonical_issue_reason": "issue:45278 is the broadest representative of the set: it captures the general post-upgrade import-regression theme that recurs across several items, though the cluster is still too heterogeneous for a tight duplicate grouping.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best single issue to anchor the cluster because it is the most umbrella-like and closest to the repeated Transformers upgrade breakage reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Different Gemma4 bugs: token embedding resizing vs causal attention mask construction." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Different models and code paths: Gemma3 config vocab_size vs SAM3 vision output missing field." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but from different packages and causes: GenerationMixin vs huggingface_hub is_offline_mode." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Unrelated failures: Flash Attention support in Nemotron vs invalid GenerationConfig in GLM-5." + }, + { + "left": "issue:44483", + "right": "issue:46369", + "accept": false, + "reason": "Different problems: chat/completions request handling vs Trainer import being blocked by datasets." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Different code paths: model loading with base_model_prefix vs saving Pipeline objects." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for return_all_scores vs pipeline serialization; not the same bug." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Both involve model I/O, but one is a sorted-state_dict load error and the other is a Qwen3.5 save_pretrained regression." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Different model/config bugs: Gemma3 vocab_size vs patchtsmixer post_init policy." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility reports, but for different versions and failure modes." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Qwen3.5 num_labels propagation is unrelated to Gemma4 attention masking." + }, + { + "left": "issue:46077", + "right": "issue:46082", + "accept": false, + "reason": "Different config validations: Gemma4 bidirectional attention vs LlamaConfig head_dim constraints." + }, + { + "left": "issue:43208", + "right": "issue:46829", + "accept": false, + "reason": "Unrelated model families and symptoms: xLSTM training blockers vs VideoPrism output field mismatch." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to Qwen3.5 checkpoint saving." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Different missing-config-attribute bugs in different models: vocab_size vs pad_token_id." + }, + { + "left": "issue:36683", + "right": "issue:44625", + "accept": false, + "reason": "Both are config propagation issues, but for different fields and models." + }, + { + "left": "issue:29870", + "right": "issue:36296", + "accept": false, + "reason": "Bart generate UnboundLocalError is unrelated to the tensor-parallel training bug." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Wrong optimizer states under FSDP2/PEFT is unrelated to a typo in cache strategies." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both are from_pretrained-related, but they affect different models and different missing-key behavior." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Cache update no_grad behavior is unrelated to general import errors after upgrade." + }, + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "Optimizer SGD args not used is unrelated to Trainer import resolution with datasets installed." + }, + { + "left": "issue:29870", + "right": "issue:33453", + "accept": false, + "reason": "Different regressions: Bart generate mask error vs tokenizer loading regression." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Both are import-time failures, but one is compiled-file handling and the other is a Python 3.13 torch.jit parse issue." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Different compatibility breakages: Python 3.14 load failure vs PretrainedConfig type checking." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "Different model I/O bugs: sorted state_dict load error vs Gemma pipeline roundtrip failure." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3 config attribute bug is unrelated to FSDP2 optimizer-state corruption." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Both touch config/model loading, but the actual missing-field and save/load issues are different." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "Both are test flakiness reports, but with different causes: missing set_seed vs xdist file race." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Similar broad import-performance theme, but not enough to show the same concrete underlying cause." + }, + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "A generic complaint that 5.0 breaks many models is too broad and not the same bug as cache update behavior." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Both mention tensor parallelism, but one is a training bug and the other is OOM during model loading." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Cache no_grad behavior is unrelated to Python-version compatibility." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different model issues: CLIP attentions regression vs patchtsmixer post_init validation." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Similar title shape, but these are different models and likely different loading failures." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Different checkpoint/load issues: sorted state_dict load error vs wrong visual encoder keys on save_pretrained." + } + ] + }, + "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": 24, + "review_unit_index": 55, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28282", + "issue:29870", + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:42548", + "issue:42757", + "issue:42890", + "issue:42898", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43299", + "issue:43502", + "issue:43576", + "issue:43618", + "issue:43723", + "issue:43824", + "issue:43827", + "issue:43874", + "issue:44062", + "issue:44112", + "issue:44222", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44655", + "issue:44734", + "issue:44749", + "issue:44961", + "issue:44964", + "issue:45071", + "issue:45125", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45600", + "issue:46132", + "issue:46285", + "issue:46617", + "issue:46664", + "issue:46726" + ], + "soft_pairs": [ + "issue:28282|issue:33453", + "issue:43618|issue:46285", + "issue:44360|issue:45440", + "issue:42890|issue:44964", + "issue:42890|issue:44112", + "issue:42890|issue:45600", + "issue:42898|issue:46664", + "issue:43502|issue:44062", + "issue:44734|issue:46617", + "issue:43723|issue:44556", + "issue:44387|issue:44556", + "issue:44222|issue:45125", + "issue:43824|issue:46617", + "issue:43502|issue:44485", + "issue:43874|issue:46617", + "issue:43010|issue:43065", + "issue:43576|issue:46132", + "issue:45125|issue:45478", + "issue:30333|issue:33453", + "issue:44355|issue:44749", + "issue:42548|issue:44351", + "issue:45125|issue:45375", + "issue:43010|issue:43232", + "issue:46664|issue:46726", + "issue:30990|issue:31795", + "issue:43502|issue:44222", + "issue:43502|issue:44961", + "issue:43827|issue:44961", + "issue:30333|issue:36296", + "issue:43299|issue:44556", + "issue:29870|issue:31795", + "issue:44355|issue:45071", + "issue:33453|issue:34689", + "issue:44355|issue:44655", + "issue:42757|issue:43576", + "issue:43576|issue:45071" + ], + "prepared_review_unit_hash": "08049a09d4ef2fe197d3fc119c3886ee2e71defb1e9c20bab610a59cadb0f277", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 36, + "serialized_chars": 38189, + "estimated_input_tokens": 9548, + "estimated_eval_tokens": 19352 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is noisy and mostly contains unrelated loading/import/regression issues; none of the soft pairs look like confident duplicates.", + "confidence": 0.31, + "canonical_issue_reason": "Issue 33453 is the broadest and most central loading-regression report in the set, and it best represents the tokenizer/model-loading subset of the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "33453 is the most generally framed issue and the best candidate to anchor triage for the loading-regression subgroup, even though the cluster overall is heterogeneous.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "PyTorch-missing import error vs tokenizer loading regression; different failure modes and causes." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Both touch CLIP internals, but one is an attentions assignment bug and the other is a model-structure inconsistency." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "DSA indexer/ReLU discussion is unrelated to DeepSeekV3 implementation divergence." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "CI flake from missing set_seed is unrelated to Phi-4 multimodal model loading." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "A flaky SAM HQ integration test is not the same bug as a GraniteSpeech stale device override failure." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Test seeding issue vs stale auto_mappings references; different code paths." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenization cleanup behavior change is unrelated to video processor num_frames semantics." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "local_files_only network requests and AddedToken keyword collision are unrelated." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation crash and trust_remote_code cache FileNotFound are different loading bugs." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Both are loading-related, but one is tokenizer loading and the other is checkpoint reload after version upgrade." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Int4 quantization OOM is a different problem from reload failure after upgrade." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained moe and missing _tp_plan for tensor parallelism are distinct issues." + }, + { + "left": "issue:43824", + "right": "issue:46617", + "accept": false, + "reason": "Cannot import a specific Qwen2.5 VL class is unrelated to cache file lookup failure." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "Network requests despite local_files_only and GLM-5 RoPE implementation are unrelated." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "Missing image-processor method and cache FileNotFound are different bugs." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache update no_grad and dummy Conv2d in Sam3PixelDecoder do not share the same code-path problem." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "Broken env command and changed AttentionInterface.register behavior are separate API issues." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": false, + "reason": "Same model family, but one is missing tensor-parallel metadata and the other is a generic from_pretrained error; not enough to merge." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "Failed MLFlow reporting and tokenizer loading regression are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors and inference slowdown after upgrade are different problems." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Different missing symbols with different root causes; both are import errors but not the same bug." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Missing _tp_plan and missing deepstack_visual_indexes are separate Qwen3.5 MoE config issues." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Cache update decoration and generation kwargs sync_gpus handling are distinct generation internals." + }, + { + "left": "issue:46664", + "right": "issue:46726", + "accept": false, + "reason": "Video num_frames semantics and missing vision backend handling are unrelated processor issues." + }, + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Sentence Transformers loading stall and forward() documentation confusion are not the same bug." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "local_files_only networking bug and FP8 save_pretrained moe are unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "No meaningful overlap beyond being issues; racoon is unrelated." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs still referencing pipeline() and racoon are unrelated." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLFlow reporting and tensor parallel training bug are different subsystems." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Both mention version upgrades and loading, but they are different models and different failure modes." + }, + { + "left": "issue:29870", + "right": "issue:31795", + "accept": false, + "reason": "Generate() UnboundLocalError in BART evaluation and documentation ambiguity are unrelated." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled-file runtime errors and PretrainedConfig type checking are different issues." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading-related, but one is a tokenizer regression and the other is a specific multimodal model-loading breakage." + }, + { + "left": "issue:44355", + "right": "issue:44655", + "accept": false, + "reason": "Compiled Python file errors and inability to save Pipeline objects are unrelated." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Hub import failure and broken env command are different v5 regressions." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "Broken env command and type-checking regression 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": 24, + "review_unit_index": 56, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:31795", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:41669", + "issue:42371", + "issue:42548", + "issue:43408", + "issue:43475", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43824", + "issue:43901", + "issue:44038", + "issue:44222", + "issue:44230", + "issue:44273", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44908", + "issue:44938", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45092", + "issue:45125", + "issue:45230", + "issue:45278", + "issue:45362", + "issue:45468", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46095", + "issue:46143", + "issue:46144", + "issue:46209", + "issue:46218", + "issue:46617", + "issue:46664" + ], + "soft_pairs": [ + "issue:44315|issue:44360", + "issue:31795|issue:38617", + "issue:43576|issue:43824", + "issue:44961|issue:45468", + "issue:44351|issue:46617", + "issue:44485|issue:44961", + "issue:44908|issue:44961", + "issue:43576|issue:44938", + "issue:36296|issue:43901", + "issue:46143|issue:46144", + "issue:43475|issue:46617", + "issue:44222|issue:44230", + "issue:41628|issue:42548", + "issue:44315|issue:46209", + "issue:44485|issue:46664", + "issue:43673|issue:44492", + "issue:45362|issue:45850", + "issue:44273|issue:45542", + "issue:44623|issue:45230", + "issue:44360|issue:46664", + "issue:44038|issue:45125", + "issue:44273|issue:45003", + "issue:44393|issue:45753", + "issue:44393|issue:45071", + "issue:43576|issue:44556", + "issue:41669|issue:43673", + "issue:43525|issue:45850", + "issue:44556|issue:45092", + "issue:43572|issue:45850", + "issue:44492|issue:45850", + "issue:45278|issue:45850", + "issue:42371|issue:44492", + "issue:44230|issue:46095", + "issue:41669|issue:46218", + "issue:43408|issue:44315" + ], + "prepared_review_unit_hash": "0df14c4cf991f3ace882a2ef49280f9568e0a681a1e8225357d676eed3ed5200", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 37294, + "estimated_input_tokens": 9324, + "estimated_eval_tokens": 18904 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is noisy and mostly spans unrelated bugs. A few issues share a generic v5 import/export-breakage theme, but none of the soft pairs look like the same concrete fix. Issue 45278 is the best umbrella representative.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45278 is the clearest broad-but-concrete umbrella: it describes many import errors after upgrading to 5.5.0, which matches the largest recurring theme in the set better than the narrower or more speculative items.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the most suitable global representative because it captures the shared post-v5 import/regression pattern without being as vague as the catch-all complaints.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Different subsystems and failures: Liger Kernel/model_init vs a DSA indexer ReLU discussion." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Documentation confusion about forward args is unrelated to an import error for layer_type_validation." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Broken `transformers env` command and missing Qwen2_5_VLForConditionalGeneration are different problems." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "No shared bug: placeholder issue `racoon` versus a Gemma-4 audio positional encoding bug." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "Both are import/load failures, but one is a missing symbol export and the other is a cache FileNotFoundError." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion is unrelated to the `racoon` issue." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwarg handling has nothing to do with the `racoon` report." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "`transformers env` breakage and Python 3.14 load failure are distinct compatibility issues." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor parallel training bug is unrelated to a TextClassificationPipeline docs mismatch." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Both involve RoFormer, but one is kwargs passthrough and the other is attention-interface usage; not the same concrete bug." + }, + { + "left": "issue:43475", + "right": "issue:46617", + "accept": false, + "reason": "SAM 3 video attribute error is unrelated to a trust_remote_code cache FileNotFoundError." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Both mention FP8/moe, but one is save_pretrained behavior and the other is inference support." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Two different missing top-level exports; no evidence they stem from one identical regression." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "Liger Kernel/model_init and FP8 inference on B200 are unrelated code paths." + }, + { + "left": "issue:44485", + "right": "issue:46664", + "accept": false, + "reason": "GLM-5 RoPE implementation is unrelated to SmolVLM video-frame semantics." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "A GenerationMixin cache bug and a cache-strategy typo are not the same underlying issue." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Specific Qwen3.5 chat crash versus a broad claim that v5 breaks many models; too different to merge." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading and TensorFlow backend detection are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "Missing processor.save_pretrained files is not the same as a generic bug report." + }, + { + "left": "issue:44360", + "right": "issue:46664", + "accept": false, + "reason": "DSA indexer ReLU discussion is unrelated to SmolVLM num_frames behavior." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Both mention Qwen/MoE, but one is a vague v5 bug report and the other is a missing tensor-parallel plan." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading complaints and unsafe sys.modules access may be adjacent, but they are not clearly the same bug." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Both are Qwen3-VL/Qwen3.5 vision issues, but one is 2D box hallucination and the other is an infinite loop on a specific image." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "A model-specific vision output bug is unrelated to PretrainedConfig type checking." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "An env command regression and checkpoint reload incompatibility after upgrade are different symptoms." + }, + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "Import-star removal concerns are unrelated to a missing GenerationMixin cache in chunked prefill." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Missing `pad_token_id` on Llama4Config is one model/config bug, not the broad v5 breakage complaint." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both are upgrade regressions, but one is checkpoint reload failure and the other is remote-code meta-init incompatibility." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "StableLmConfig missing `pad_token_idx` is too specific to merge with a generic v5 breakage claim." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A cache-strategy typo is not the same as a broad version-breakage report." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "Both mention v5 regressions, but 45278 is a concrete import-error umbrella while 45850 is too vague to be the same underlying bug." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API settings and cache-strategy typo are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 inference support for Qwen models is unrelated to a DeepSeek weight conversion question." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "Removing import-star usage is not the same as a missing Qwen3_5ForCausalLM import." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "SAM3 video/tracker class mismatch is unrelated to Liger Kernel application in model_init." + } + ] + }, + "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": 24, + "review_unit_index": 57, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:34689", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43298", + "issue:43352", + "issue:43502", + "issue:43519", + "issue:43576", + "issue:43606", + "issue:43704", + "issue:43827", + "issue:43976", + "issue:44077", + "issue:44246", + "issue:44261", + "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:45310", + "issue:45341", + "issue:45468", + "issue:45478", + "issue:45850", + "issue:46018", + "issue:46095", + "issue:46172", + "issue:46218", + "issue:46328", + "issue:46655" + ], + "soft_pairs": [ + "issue:42890|issue:44908", + "issue:44246|issue:44273", + "issue:43576|issue:44908", + "issue:43352|issue:44393", + "issue:43976|issue:44393", + "issue:44908|issue:45071", + "issue:44704|issue:44908", + "issue:44295|issue:45230", + "issue:44297|issue:45230", + "issue:41669|issue:44351", + "issue:44492|issue:46328", + "issue:41669|issue:45341", + "issue:44393|issue:45478", + "issue:43976|issue:45850", + "issue:43976|issue:46218", + "issue:44077|issue:44861", + "issue:43827|issue:44908", + "issue:31795|issue:33453", + "issue:43576|issue:44273", + "issue:44315|issue:44829", + "issue:44360|issue:45468", + "issue:43519|issue:46655", + "issue:44273|issue:44908", + "issue:44655|issue:44861", + "issue:43576|issue:46218", + "issue:45230|issue:45310", + "issue:43519|issue:44485", + "issue:43519|issue:44908", + "issue:41669|issue:43704", + "issue:31795|issue:34689", + "issue:46018|issue:46172", + "issue:43704|issue:44485", + "issue:44485|issue:45468", + "issue:43502|issue:43519", + "issue:42371|issue:43704", + "issue:45850|issue:46218", + "issue:43519|issue:44623", + "issue:44263|issue:45468", + "issue:43606|issue:45468", + "issue:44261|issue:45468", + "issue:43298|issue:46095", + "issue:43519|issue:43704", + "issue:44351|issue:46218", + "issue:43352|issue:46218" + ], + "prepared_review_unit_hash": "24b19f5a7499098d5aae8f1deec9c7b8d04ad1c5bbb18f749e06e21237d67f33", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 44, + "serialized_chars": 39635, + "estimated_input_tokens": 9909, + "estimated_eval_tokens": 20074 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly noisy: the soft edges are mostly title-level similarity across unrelated bugs (loading, processors, docs, model-specific regressions, and utilities). I did not find any convincing duplicate pairs, and there are no PRs in the cluster.", + "confidence": 0.27, + "canonical_issue_reason": "issue:44908 is the most connected node in the similarity graph and the closest thing to a hub, but the cluster is too heterogeneous for it to be a strong duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44908 is the least-bad anchor only because it has the broadest linkage; it still does not cleanly represent a single underlying bug family.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test flakiness from missing seeds vs scheduler kwargs being ignored." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Import-time slowness vs lazy-loading malfunction; related theme, different bug." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Broken env command vs inverse_sqrt scheduler kwargs handling." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash-Attention support gating vs Qwen3-VL bbox output error." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python-version compatibility regression vs vision-model output bug." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwarg propagation vs PretrainedConfig type-checking regression." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor cached_file kwargs bug vs scheduler kwargs bug." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "position_ids buffer access issue vs an unspecified generic bug report." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "tokenizer save metadata mismatch vs an unspecified generic bug report." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star cleanup/perf issue vs missing HybridCache import." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Cache-strategy typo vs SequenceFeatureExtractor.pad performance bug." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Import-star cleanup/perf issue vs testing_utils bug." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Different Qwen3.5/Qwen3-VL model failures, not the same code-path." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "Python support regression vs vague complaint about many broken models." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python-version regression vs missing Qwen3_5 import." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "PatchTSMixer post_init policy vs tied-weight key AttributeError." + }, + { + "left": "issue:43827", + "right": "issue:44908", + "accept": false, + "reason": "Docs still mention pipeline() vs scheduler kwargs bug." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Forward-doc confusion vs tokenizer loading regression." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Env command breakage vs lazy-loading behavior." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel application with model_init vs flash_attention_3 training degeneration." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer activation issue vs Gemma-4 audio positional encoding bug." + }, + { + "left": "issue:43519", + "right": "issue:46655", + "accept": false, + "reason": "Qwen3VL timestamp math vs SmolVLM video padding calculation." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy-loading bug vs scheduler kwargs bug." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Pipeline save_pretrained issue vs tied-weight key crash." + }, + { + "left": "issue:43576", + "right": "issue:46218", + "accept": false, + "reason": "v5 env command regression vs missing Qwen3_5 import." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report vs specific Qwen3.5 MoE from_pretrained regression." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Timestamp calculation bug vs GLM-5 RoPE implementation." + }, + { + "left": "issue:43519", + "right": "issue:44908", + "accept": false, + "reason": "Processor timestamp math vs scheduler kwargs bug." + }, + { + "left": "issue:41669", + "right": "issue:43704", + "accept": false, + "reason": "Import-star cleanup/perf issue vs VRAM leak in multi-thread use." + }, + { + "left": "issue:31795", + "right": "issue:34689", + "accept": false, + "reason": "Documentation confusion vs model-loading failure." + }, + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "Shared-expert gating issue vs tensor transpose/weight-conversion bug." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "VRAM leak vs RoPE implementation." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "GLM-5 RoPE implementation vs Gemma-4 audio positional encoding." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "local_files_only network leak vs timestamp calculation." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 settings API change vs VRAM leak." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "Broad complaint about broken models vs missing Qwen3_5 import." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Timestamp calculation bug vs processor save_pretrained missing files." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "torch.split return-value issue vs audio positional encoding bug." + }, + { + "left": "issue:43606", + "right": "issue:45468", + "accept": false, + "reason": "CPU offload device mismatch vs audio positional encoding." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "RMS norm epsilon precision issue vs audio positional encoding." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Continuous batching support gap vs DeepSeek-V4 weight-conversion question." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Timestamp calculation bug vs VRAM leak." + }, + { + "left": "issue:44351", + "right": "issue:46218", + "accept": false, + "reason": "Missing HybridCache import vs missing Qwen3_5 import." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash-Attention 2.0 support gating vs missing Qwen3_5 import." + } + ] + }, + "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": 24, + "review_unit_index": 58, + "review_unit_count": 58, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-487", + "nodes": [ + "issue:28348", + "issue:29870", + "issue:30333", + "issue:31515", + "issue:31795", + "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:44492", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44938", + "issue:44998", + "issue:45250", + "issue:45290", + "issue:45310", + "issue:45850", + "issue:46018", + "issue:46144" + ], + "soft_pairs": [ + "issue:42913|issue:43066", + "issue:28348|issue:45250", + "issue:44908|issue:44998", + "issue:44492|issue:44998", + "issue:43618|issue:44998", + "issue:44998|issue:45850", + "issue:44938|issue:44998", + "issue:44861|issue:45290", + "issue:31515|issue:38617", + "issue:43572|issue:44861", + "issue:43525|issue:44861", + "issue:44230|issue:44623", + "issue:28348|issue:43352", + "issue:29870|issue:31515", + "issue:30333|issue:31515", + "issue:31515|issue:36296", + "issue:31515|issue:31795", + "issue:44230|issue:45310", + "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": "6f7dd66e9e9849d541d061ecfe67e15f1d8f2cfa58a094a2b56837ce15260787", + "component_budget": { + "node_count": 487, + "item_count": 487, + "soft_pair_count": 1796, + "serialized_chars": 930665, + "estimated_input_tokens": 232667, + "estimated_eval_tokens": 465590 + }, + "budget": { + "node_count": 34, + "item_count": 34, + "soft_pair_count": 31, + "serialized_chars": 28240, + "estimated_input_tokens": 7060, + "estimated_eval_tokens": 14376 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "No genuine duplicates detected; this cluster is a grab bag of unrelated tokenizer, model, training, and docs/config issues, with only loose topical overlap in a few version-regression titles. All soft edges should be rejected.", + "confidence": 0.77, + "canonical_issue_reason": "issue:42913 is the broadest version-regression/tokenizer anchor in the set, but the cluster is too heterogeneous for a true duplicate canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42913 is the most representative high-level regression report among the issues shown; it is the closest thing to a cluster anchor even though the set is not a real duplicate group.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Related tokenizer/v5 symptoms, but one is generic behavior change and the other is a decoder-type bug; not the same fix." + }, + { + "left": "issue:28348", + "right": "issue:45250", + "accept": false, + "reason": "Both mention Flash Attention 2.0, but they target different model families and likely different implementation gaps." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "Inverse-sqrt scheduler kwargs bug is unrelated to the spurious 'Unemployment' issue." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Cache-strategy typo and the unrelated 'Unemployment' issue are not duplicates." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the spurious 'Unemployment' issue." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "'Unemployment' is unrelated to the broad '5.0 breaks too many models' complaint." + }, + { + "left": "issue:44938", + "right": "issue:44998", + "accept": false, + "reason": "Python 3.14 import/load failure is unrelated to the spurious 'Unemployment' issue." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Tied-weight key handling and chat-template tool-call crashing are different code paths and bugs." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Slow from_pretrained loading is unrelated to the ImportError for layer_type_validation." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig is unrelated to tied-weight key attribute handling." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Llama4 pad_token_id missing is a config regression, not the tied-weight crash." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support and missing processor.save_pretrained files are unrelated issues." + }, + { + "left": "issue:28348", + "right": "issue:43352", + "accept": false, + "reason": "Same Flash Attention 2.0 wording, but GPT2 support and Nemotron support are separate model-specific changes." + }, + { + "left": "issue:29870", + "right": "issue:31515", + "accept": false, + "reason": "BART generate() decoder_attention_mask bug is unrelated to from_pretrained slowness." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow run-finalization on failure is unrelated to from_pretrained slowness." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Loading slowdown and tensor-parallel training bug are different problems." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Loading slowdown and forward() documentation confusion are unrelated." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "FP8 support request and Qwen3.5 from_pretrained error are different model/version issues." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Continuous batching for audio models and FP8 inference support are different feature gaps." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba init_weights behavior and RoFormer attention interface usage are unrelated model implementation issues." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2Former ignore_value handling and DeepSeek-V4 shared expert gating are unrelated." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "A config type-hint bug is unrelated to the tokenizer regex warning regression." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Mask2Former ignore_value usage and loss-kwargs propagation are different code paths." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Loss-function kwargs propagation and RoFormer attention interface are unrelated bugs." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2Former ignore_value handling and tokenizer regex warnings do not share a concrete bug." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "Mask2Former ignore_value handling and RoFormer attention interface usage are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "Mask2Former ignore_value handling and the quantization_operation typo are unrelated." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Loss kwargs propagation and ModernBERT type-hint correction are different bug classes." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "ModernBERT norm_eps typing and a quantization-operation typo are unrelated." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Llama4 norm behavior question and ModernBERT type-hint issue are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Mask2Former ignore_value handling and ModernBERT type-hint correction are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "7723390dbce22cccff85211912e5a1819596d2dda385322f93151d2b0fab26b4", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2877, + "estimated_input_tokens": 720, + "estimated_eval_tokens": 1696 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2877, + "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 unrelated and should not be clustered as duplicates. They cover different topics: tokenizer truncation behavior, learning-rate scheduler functionality, and Wav2Vec2 word boundary offsets.", + "confidence": 0.99, + "canonical_issue_reason": "Issue 10519 is the most concrete and self-contained request, but it is not actually equivalent to the others, so there is no true canonical duplicate in this set.", + "canonical_pr_reason": null, + "best_issue_reason": "10519 is the clearest issue to use as a representative only because it states a specific feature/behavior request; the other two are unrelated concerns.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Tokenizer truncation direction and Wav2Vec2 word-offset extraction are different features in different code paths." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "Truncation behavior for tokenizers is unrelated to ReduceLROnPlateau-like scheduler functionality." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "A learning-rate scheduler feature request does not match a Wav2Vec2 alignment/time-offset issue." + } + ] + }, + "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": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-13244-3", + "nodes": [ + "issue:13244", + "issue:15980", + "issue:16618" + ], + "soft_pairs": [ + "issue:13244|issue:16618", + "issue:15980|issue:16618" + ], + "prepared_review_unit_hash": "36e9f630ab5dba8a921a46038c7cdcfcbb49c6afcf8220db72fe69c22d29a18e", + "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 issues are not duplicates: one is about tokenizer output mismatch, one about authentication/error handling when downloading private models, and one about a model-loading failure for bert-base-uncased. The only loose overlap is around model loading, but the concrete failure modes differ.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 16618 is the broadest of the three and the only one directly about model loading, but it is not a strong canonical match for the other two since their root problems are different.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 16618 is the best representative of the cluster only by being the most general/loading-related title; however, it should not be treated as a duplicate target for the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Tokenizer discrepancies for TAPAS and model-loading failures for bert-base-uncased are different bugs in different code paths." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "A private-model authentication/download error is not the same underlying issue as a generic inability to load bert-base-uncased." + } + ] + }, + "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": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-26350-2", + "nodes": [ + "issue:26350", + "issue:33260" + ], + "soft_pairs": [ + "issue:26350|issue:33260" + ], + "prepared_review_unit_hash": "372ee279eae1739978e3a99287d76ce7c84f558c7cf1d669dcbcde910f5d77b6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1589, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicates: one requests Flash Attention 2 support for more architectures, the other requests GGUF support for more architectures. They share only a broad \u201csupport more architectures\u201d theme, not the same underlying change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Different features in different code paths: Flash Attention 2 support vs GGUF support. Similar high-level goal, but not the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 5, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "9ae707f4bf124ef02dc2759b13264d3b7905c769679c58580cce26046cc46daf", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13526, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13526, + "estimated_input_tokens": 3382, + "estimated_eval_tokens": 7020 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "46222/46226 is a clear exact duplicate pair. The generic i18n template issues 43834/44365/44370 also look like duplicates of the same underlying request based on identical placeholder titles and near-identical similarity scores. The remaining language-coded translation issues do not have enough evidence here to safely collapse into those groups.", + "confidence": 0.69, + "canonical_issue_reason": "issue:46222 is the strongest canonical issue because it has the clearest exact-duplicate match (46226) plus inbound references, making it the most defensible representative in this cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:46222 is the best global issue representative because it anchors the only explicit exact-duplicate pair and has some outside references/activity; the generic template issues are highly similar but less clearly established from the packet alone.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Both are docs-translation issues, but the packet does not show they are the same underlying request." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "High template similarity alone is not enough to prove a duplicate here." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "The titles differ materially and the evidence does not establish the same request." + }, + { + "left": "issue:27088", + "right": "issue:38435", + "accept": false, + "reason": "These are different language-coded translation requests; the packet does not support merging them." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Turkish and Spanish translation requests are distinct and not shown to be the same issue." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Similarity is too indirect to conclude these are duplicates." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "The packet does not show the same underlying translation request." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Spanish and Romanian translation requests are separate language-specific tasks." + }, + { + "left": "issue:28936", + "right": "issue:44370", + "accept": false, + "reason": "Shared docs/i18n wording is not enough to merge them as duplicates." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "The Romanian issue and the generic template issue are not clearly the same request from the available evidence." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "The packet does not show that these are the same translation request." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "The evidence is suggestive but still insufficient to treat them as the same issue." + }, + { + "left": "issue:38435", + "right": "issue:46222", + "accept": false, + "reason": "Different titles and no direct evidence of a shared underlying request." + }, + { + "left": "issue:38435", + "right": "issue:46226", + "accept": false, + "reason": "Different titles and no direct evidence of a shared underlying request." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": true, + "reason": "Identical placeholder-style title and very high similarity strongly suggest the same underlying template issue." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": true, + "reason": "The titles are identical placeholders and the similarity score is near-perfect, consistent with a duplicate." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "Despite high similarity, the available evidence is not enough to treat the generic template issue and the Thai issue as the same request." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "Despite high similarity, the available evidence is not enough to treat the generic template issue and the Thai issue as the same request." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": true, + "reason": "Near-identical generic i18n template titles and extremely high similarity make this look like a duplicate." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "No direct evidence that the generic template issue and the Thai issue are the same underlying request." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "No direct evidence that the generic template issue and the Thai issue are the same underlying request." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "The packet does not establish that the generic template issue and the Thai issue are duplicates." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "The packet does not establish that the generic template issue and the Thai issue are duplicates." + }, + { + "left": "issue:46222", + "right": "issue:46226", + "accept": true, + "reason": "Same exact Thai title and identical body length make this a clear duplicate pair." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The 46222/46226 duplicate call is grounded, but the rest of the summary overreaches. The packet only shows identical generic i18n template titles for 43834/44365/44370 and very high similarity scores; it does not show the claimed language-specific differences. Treating all of those as clearly separate requests is not well supported, so several soft-edge false verdicts are too strong for the evidence provided." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-36817-18", + "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", + "issue:46645", + "issue:46653" + ], + "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:42738|issue:46653", + "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:43909|issue:46645", + "issue:44121|issue:45322", + "issue:43837|issue:45027", + "issue:37015|issue:45027", + "issue:43837|issue:45177", + "issue:45322|issue:46133", + "issue:42503|issue:46653", + "issue:45027|issue:46653" + ], + "prepared_review_unit_hash": "f6bb6e4f55e7b9ec19f79a19ce51c3366db7567871237c1ddee4071ccf3f4993", + "component_budget": { + "node_count": 18, + "item_count": 18, + "soft_pair_count": 40, + "serialized_chars": 24143, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 + }, + "budget": { + "node_count": 18, + "item_count": 18, + "soft_pair_count": 40, + "serialized_chars": 24143, + "estimated_input_tokens": 6036, + "estimated_eval_tokens": 12328 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a mixed bag of unrelated model-addition requests (different model families, modalities, and features). They are superficially similar because many titles say \"Add model/support\", but they do not describe the same underlying bug or change, so they should not be deduplicated together.", + "confidence": 0.97, + "canonical_issue_reason": "No single issue is a clear duplicate canonical for the cluster because the items request different models/features (e.g. EuroBert, RF-DETR, NeoBERT, MiniCPM3, Qwen3-TTS/ASR, etc.).", + "canonical_pr_reason": null, + "best_issue_reason": "If forced to pick a representative, issue 45322 is one of the clearer, well-scoped model request reports with a concrete target and useful context; however, it is not a true canonical for the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "EuroBert config support and DEIMv2 are unrelated model requests." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert and Qwen3-TTS are different models and different modalities." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "EuroBert and RF-DETR are unrelated architecture additions." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are different detection model requests, not the same change." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "RF-DETR is a vision detection model; LFM2.5 Audio is an audio model request." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR and OpenAI weight-sparse transformer are distinct model families and code paths." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "RF-DETR and Qwen3-ASR are different tasks/models." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR and Qwen3-TTS are unrelated." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are different model support requests." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "NeoBERT and circuit-sparsity transformer are not the same underlying model change." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT is a BERT-style LM; LFM2.5 Audio is an audio model." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR and NeoBERT are unrelated model additions." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are different BERT-like models, not duplicates." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 architecture support and Qwen3-ASR are different model requests." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 support and weight-sparse transformer support are unrelated." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "MiniCPM3 and ModernVBERT are different model families." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "MiniCPM3 and Voxtral TTS support are different modalities/models." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are unrelated 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 not the same change." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT and BERT-like models with RoPE are related only loosely; they are not the same concrete request." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "RoPE-enabled BERT-like models and Voxtral TTS are unrelated." + }, + { + "left": "issue:42738", + "right": "issue:46653", + "accept": false, + "reason": "BERT-like RoPE support and a profile/name issue have no overlap." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Qwen3-TTS and Qwen3-ASR share a prefix but target different speech tasks and likely different implementations." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS and Voxtral-4B-TTS are different TTS model support requests." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS and DeepSeek-OCR2 are unrelated model requests." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-TTS and weight-sparse transformer are different model changes." + }, + { + "left": "issue:43671", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-TTS and LFM2.5 Audio are different audio model requests." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Qwen3-ASR and LFM2.5 Audio are different speech/audio model requests." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR and EUPE are different model requests with no shared code path shown." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "Qwen3-ASR and circuit-sparsity transformer are unrelated." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral TTS are different speech tasks/models." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are different model domains." + }, + { + "left": "issue:43909", + "right": "issue:46645", + "accept": false, + "reason": "LFM2.5 Audio support and Granite SWA+sink are different changes; one is model support, the other is an architectural option." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "OpenAI weight-sparse transformer and EUPE are unrelated model requests." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-ASR and Voxtral-4B-TTS are different speech model requests." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral-4B-TTS are unrelated model additions." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model domains." + }, + { + "left": "issue:45322", + "right": "issue:46133", + "accept": false, + "reason": "EUPE and TIPSv2 are different Meta/Google model requests with different purposes." + }, + { + "left": "issue:42503", + "right": "issue:46653", + "accept": false, + "reason": "ModernVBERT models and a closed personal-name issue are unrelated." + }, + { + "left": "issue:45027", + "right": "issue:46653", + "accept": false, + "reason": "Voxtral TTS support and a personal-name 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": 7, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-38740-2", + "nodes": [ + "issue:38740", + "issue:46443" + ], + "soft_pairs": [ + "issue:38740|issue:46443" + ], + "prepared_review_unit_hash": "2f975088240b738faec55763a42075a0601ee2668ab1de67f7b7fe7200faed28", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1505, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1505, + "estimated_input_tokens": 377, + "estimated_eval_tokens": 1010 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are unrelated: one is a documentation request about adding `pruna` as an optimization framework, and the other is a Vision RoPE refactor. No duplicate relationship is supported.", + "confidence": 0.99, + "canonical_issue_reason": "If a single issue must represent the cluster, issue 46443 is the better representative because it is a concrete code refactor rather than a docs request; however, the pair is not a duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46443 is the most suitable standalone issue because it describes an actual implementation refactor, whereas issue 38740 is only a documentation addition request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38740", + "right": "issue:46443", + "accept": false, + "reason": "Different topics and artifact types: a docs request for adding `pruna` versus a Vision RoPE refactor. They do not describe the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 8, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41084-2", + "nodes": [ + "issue:41084", + "issue:43037" + ], + "soft_pairs": [ + "issue:41084|issue:43037" + ], + "prepared_review_unit_hash": "e5075a69797bcde4ee5af32aaae787ffbd67a4c9b6c9b3eb53ebc515496789ff", + "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 appear unrelated: one is about set block decoding mechanics, the other about adding DeepSeek v3.2 support. There is no evidence they describe the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "If a representative issue is needed, \"DeepSeek v3.2 support\" is the more concrete feature request, but it is not a duplicate of the block decoding issue.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different problem domains with no shared files, targets, or explicit linkage; one concerns decoding behavior, the other model 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": 9, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41306-2", + "nodes": [ + "issue:41306", + "issue:41826" + ], + "soft_pairs": [ + "issue:41306|issue:41826" + ], + "prepared_review_unit_hash": "b1d12d6c466a4d62e69219ff8ab7dc5f941c8f88e3eba4f1f70e406df69a8fb8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are both about tiled/efficient model components, but they target different features and code paths: FLCE/LigerFusedLinearCrossEntropy flag handling in `from_pretrained()` versus integrating TiledMLP for reduced memory. They are not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": "Issue 41306 is the more concrete and actionable report, focused on a specific loading/configuration gap for TiledFusedLogitsLoss in `from_pretrained()`.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 41306 is the better representative because it names a precise component, API surface, and expected behavior, making it easier to triage and resolve than the broader TiledMLP integration request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41306", + "right": "issue:41826", + "accept": false, + "reason": "They concern different mechanisms and fixes: one is about enabling a loss-function flag during model loading, the other about adding TiledMLP for memory reduction. Same general theme, but not the same underlying 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": 10, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43598-2", + "nodes": [ + "issue:43598", + "issue:43600" + ], + "soft_pairs": [ + "issue:43598|issue:43600" + ], + "prepared_review_unit_hash": "69d579f1bde9b2e53a2df7d800eca2806aebb6536ecec240321185786a5d5fc7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1584, + "estimated_input_tokens": 396, + "estimated_eval_tokens": 1048 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues share only a loose theme around custom trainers, but they describe different Trainer behaviors: one is about `remove_unused_column` customizability, the other about FLOPS estimation. They should not be deduplicated.", + "confidence": 0.91, + "canonical_issue_reason": "issue #43598 is the broader, more central Trainer customizability request; issue #43600 is a separate FLOPS-estimation concern.", + "canonical_pr_reason": null, + "best_issue_reason": "#43598 is the better representative issue because it focuses on a core Trainer customization behavior rather than a narrower metrics/estimation edge case.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Both mention custom trainers, but they target different code paths and user problems: unused-column removal vs FLOPS counting. They are not the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43684-2", + "nodes": [ + "issue:43684", + "issue:44405" + ], + "soft_pairs": [ + "issue:43684|issue:44405" + ], + "prepared_review_unit_hash": "efda7dcd3abfb8c99ad78b19bbac78b0a60ac9591dcefe8141a9d31d4f13cf17", + "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 at a very high level (both ask for new Qwen model support in auto classes), but they target different model families and different APIs, so they should not be treated as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": "Issue #43684 is the broader registration request and is the better anchor if one issue must represent the cluster, but it is still distinct from #44405 rather than a true duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "#43684 is the more general and reusable request: it asks for AutoModel and AutoModelForConditionalGeneration registration for Qwen3-Omni, which is closer to a canonical tracking issue than the narrower Qwen3.5 sequence-classification request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying changes: one is registration for Qwen3-Omni in AutoModel/AutoModelForConditionalGeneration, the other is adding AutoModelForSequenceClassification support for Qwen3.5. Same broad subsystem, but not the same bug or mergeable fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 12, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "4553bcd22152f2b27407144da2bb8ae8b173fd1686c5074d173f236e4ad7b94c", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4265, + "estimated_input_tokens": 1067, + "estimated_eval_tokens": 2390 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Three issues are effectively indistinguishable low-content reports titled \"huggingface \"; the \"Fett\" issue appears separate. Among the huggingface trio, the earliest report is the best canonical candidate.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 43845 is the earliest of the near-identical \"huggingface \" reports, so it is the best anchor for the duplicate cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43845 is the strongest global issue candidate because it is the first of the repeated identical-title reports and there is no richer signal to prefer another.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different titles and no content evidence tying \"Fett\" to the \"huggingface \" reports." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": true, + "reason": "Same title, near-identical low-content reports, and created one minute apart; likely the same underlying issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": true, + "reason": "Same title and similarly sparse reporting; likely duplicate instances of the same issue." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": true, + "reason": "Same title, close timing, and no distinguishing details; likely the same underlying issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: 43835 is the outlier by title, while 43845/43846/43859 are the only identical-title reports with similarly sparse bodies. The canonical choice of 43845 as the earliest of that trio is conservative and supported. The soft-edge accepts are also reasonable given the title match and low-content signals, with no unsupported leap beyond the available metadata." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "32482ae257a5a9107203e013c27e2634044454120aa2bcb0f171b26d6176e601", + "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 low-signal, likely unrelated placeholder/spam issues with no substantive evidence of a shared bug or feature request. No soft duplicates should be merged.", + "confidence": 0.93, + "canonical_issue_reason": "issue:44371 is the earliest and most generic placeholder title, but it is not a strong canonical target because the cluster lacks a real shared problem statement.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44371 is the least specific and earliest of the spam-like issues, so it is only the best placeholder representative, not a true duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are generic '' placeholders, but there is no underlying bug/change to unify." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Same placeholder pattern, but these are not substantive duplicate reports." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "Titles are different and there is no shared problem description beyond weak similarity." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "No shared technical issue; 'Google LLC' is unrelated to the spam placeholder." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Documentation suggestion vs. spam placeholder: not the same underlying issue." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Documentation suggestion vs. spam placeholder: not a duplicate pair." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Identical spam titles, but they do not describe a concrete mergeable bug or change." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "No common issue content; one is spam, the other is an unrelated named topic." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "No common issue content; one is spam, the other is an unrelated named topic." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 14, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45145-2", + "nodes": [ + "issue:45145", + "issue:45151" + ], + "soft_pairs": [ + "issue:45145|issue:45151" + ], + "prepared_review_unit_hash": "f4da3f585eb1084379aaac9e6fe218cb5efac4ab97a2fbdcb84c44ea125472ef", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1645, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues look like a likely duplicate pair: they have the same title, similar body sizes, and moderate soft-similarity, but the evidence is not strong enough to call them near-identical with high confidence. No PRs are involved.", + "confidence": 0.79, + "canonical_issue_reason": "Issue #45145 is the better canonical because it was created first and already has an inbound reference, so it is the more established thread.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45145 is the strongest representative of the cluster for the same reason: earlier creation and existing reference activity.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "Same title and moderate soft-similarity suggest a likely duplicate, but the evidence is not strong enough to confirm the same underlying issue. With only Jaccard 0.63, similar body length, and no explicit links or shared targets, this should stay conservative." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The canonical-selection rationale is grounded (issue #45145 was created earlier and has an inbound reference), but the cluster summary and soft-edge verdict are a bit too strong for the evidence. We only have identical titles plus soft-similarity/Jaccard 0.63 and similar body lengths, which supports a likely duplicate but not confidently a \"near-identical\" or definitively same underlying issue. Please soften the wording and keep the verdict conservative." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": 16, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45672-2", + "nodes": [ + "issue:45672", + "issue:45674" + ], + "soft_pairs": [ + "issue:45672|issue:45674" + ], + "prepared_review_unit_hash": "ae5adbbf6ccb2b0d596567cf565d58b4aa0f02bfea6a29c0e6f288a609f988ee", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1673, + "estimated_input_tokens": 419, + "estimated_eval_tokens": 1094 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related to 4-bit inference regressions, but they describe different failure modes and likely different fixes: one is a torch.finfo()/uint8 path in Gemma4 audio modules, the other is a BitsAndBytesConfig skip-module interaction that drops the default lm_head exclusion and triggers an AssertionError.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45674 is the more general and clearly described 4-bit inference regression, but it is not a duplicate of 45672 because the root cause and failing code path differ.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Different bugs: 45672 is a torch.finfo TypeError on uint8 weights in Gemma4 audio modules; 45674 is a BitsAndBytesConfig skip-module bug that removes the default lm_head exclusion and causes an AssertionError. Same broad area, not the same underlying defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45762-2", + "nodes": [ + "issue:45762", + "issue:45763" + ], + "soft_pairs": [ + "issue:45762|issue:45763" + ], + "prepared_review_unit_hash": "e6c47a131ec34fc98fa7d01746135dd745bb49b4d6e9b001d9dc0db8b171c3a9", + "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 cluster contains two unrelated closed issues with different person names and different dates; there is no evidence they describe the same underlying problem or request.", + "confidence": 0.97, + "canonical_issue_reason": "Neither issue has any body, references, or shared targets/filenames; the titles refer to different names and dates, so there is no clear canonical duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "Both issues are equally sparse, but issue #45762 is marginally a better canonical placeholder only because it appears first; there is no substantive basis to prefer it.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "The titles identify different people and different dates, and there is no supporting discussion or metadata tying them to the same bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 18, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45764-2", + "nodes": [ + "issue:45764", + "issue:45765" + ], + "soft_pairs": [ + "issue:45764|issue:45765" + ], + "prepared_review_unit_hash": "50d5e79d5ce959435080f868ebbc0b3a4c15a5c235cbcac1ede8082c0495605e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues do not look like duplicates or part of the same technical problem; they appear to be unrelated name/date entries.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue candidate: the issues are unrelated and do not describe the same bug or request.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a good representative for the cluster because there is no coherent shared technical topic.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "The titles are different personal-name/date strings with no shared technical context, so they do not appear to be 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": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45810-2", + "nodes": [ + "issue:45810", + "issue:46295" + ], + "soft_pairs": [ + "issue:45810|issue:46295" + ], + "prepared_review_unit_hash": "652ec25c436609ca5749939274f16b52ebc135640e20d57f9933be63eb929fd7", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1540, + "estimated_input_tokens": 385, + "estimated_eval_tokens": 1026 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: one requests a Qwen3_5 token-classification model for value-model use, while the other asks for FP8 training support. They concern different features and code paths.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue #45810 is the more concrete and actionable request; #46295 is broader and more general.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "Different underlying requests: model-class support for token classification vs FP8 training support. No shared bug/change target." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 20, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-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": "bd7ee2424c837fa074de051476e5f92cfba4b1b748ba0528bb977c7aec02e9ed", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two exact duplicate issue pairs stand out: 45870/45871 and 45872/45873. Issue 45869 is a separate dataset-promo post and is not the same underlying request. The example-request pair and the scraping-service pair are related only by theme ('financial time series') but describe different artifacts and should not be merged.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue for the full set; the data splits into two separate exact-duplicate pairs (45870/45871 and 45872/45873) plus an unrelated dataset announcement (45869).", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must be chosen as the best representative, 45870 is a reasonable pick because it is the earlier half of an exact duplicate pair and has the clearest concrete example-request title.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Dataset/sample promo vs example/tutorial request; same broad finance theme, but not the same underlying issue." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Dataset/sample promo vs the duplicate example request; unrelated artifact types and goals." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Exact same title and substantially the same body; clear duplicate pair." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Both mention financial time series, but one is an example request and the other is a scraping-service request; different underlying change." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Different concrete request: example/tutorial vs scraping service integration; not mergeable as one issue." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Same as 45870/45872: shared theme only, but distinct requested outcomes." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Same as 45870/45873: not the same bug or change, just a related topic." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Exact same title and substantially the same body; clear duplicate pair." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative on the soft edges. The two high-similarity pairs are treated as duplicates only where the evidence is strongest (45870/45871 and 45872/45873), and the cross-pair links are correctly rejected as related-only. 45869 is also reasonably separated as a different dataset-style post rather than the same request." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 21, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46439-2", + "nodes": [ + "issue:46439", + "issue:46441" + ], + "soft_pairs": [ + "issue:46439|issue:46441" + ], + "prepared_review_unit_hash": "9d41aa86d52b0698d6f13a4744065c9f1fa0fcec45ac40579313dc41a23afb40", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are related at the theme level (`cache_implementation=\"static\"` / `linear_attention`), but the packet does not establish a shared root cause. One reports wrong-shape corruption during `early_initialization`, the other a model-specific `KeyError` on Qwen3.5, so they should be clustered cautiously rather than treated as the same defect.", + "confidence": 0.43, + "canonical_issue_reason": "Issue #46439 is the better canonical issue because it is broader and more root-cause-oriented in its wording, covering the static-cache `early_initialization` path and the linear-attention breakage without being tied to one model.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46439 is the strongest representative for the cluster since it describes the more general failure mode, while #46441 reads as a narrower manifestation on a specific model and cache setting.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46439", + "right": "issue:46441", + "accept": false, + "reason": "Reject as a same-bug match: the overlap is only the static-cache/linear_attention context, and the packet does not show a common root cause or a clearly shared fix path." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The similarity is grounded only in the shared static-cache/linear_attention theme, but the packet does not establish a common root cause or a registry/early_initialization bug. The summary and canonicalization language go a bit too far beyond the titles, so the soft-edge accept should be more tentative rather than asserting the same underlying bug class." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 22, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46551-2", + "nodes": [ + "issue:46551", + "issue:46552" + ], + "soft_pairs": [ + "issue:46551|issue:46552" + ], + "prepared_review_unit_hash": "20cab18958af5d5eab95ce91f3f947dc4ee4e6593fc8188b772d0553e3ba0be2", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1707, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are duplicate issues with identical titles and matching descriptions about Wav2Vec2CTCTokenizer v5 incorrectly treating the word delimiter as a special token. The open issue is the better canonical tracker.", + "confidence": 0.99, + "canonical_issue_reason": "Prefer issue #46552 because it is open and already has inbound references, making it the more active tracker for the same bug; #46551 appears to be the older closed duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46552 is the best issue to keep because it is open, currently referenced, and describes the exact same problem as #46551.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46551", + "right": "issue:46552", + "accept": true, + "reason": "Same title and same underlying tokenizer bug; the descriptions match and there is no meaningful distinction beyond open vs closed state." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: both issues have identical titles and matching body length, and issue #46552 is open with inbound references while #46551 is closed. The soft-edge verdict is conservative and supported by the evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 23, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46657-2", + "nodes": [ + "issue:46657", + "issue:46692" + ], + "soft_pairs": [ + "issue:46657|issue:46692" + ], + "prepared_review_unit_hash": "9f10787b81e226d70eccb0f94704ee017538d1b642809e16f5f5079fe0231c37", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1569, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1569, + "estimated_input_tokens": 393, + "estimated_eval_tokens": 1042 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are not duplicates: one reports a tensor shape mismatch during object-detection training, while the other reports an out-of-memory failure after several epochs in RT-DETR v2. They describe different failure modes and likely different root causes.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue fits because the reports are about different symptoms and different code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46657 is the more general training-bug report, but it still does not subsume 46692; the latter is a distinct RT-DETR v2 OOM problem.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46657", + "right": "issue:46692", + "accept": false, + "reason": "Different bugs: tensor size mismatch vs memory exhaustion after epochs. 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": 24, + "component_count": 24, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46822-2", + "nodes": [ + "issue:46822", + "issue:46823" + ], + "soft_pairs": [ + "issue:46822|issue:46823" + ], + "prepared_review_unit_hash": "d12d19e1f71b18f102de502235631b8061e3191a5e1b1aef1924731b88e46266", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1672, + "estimated_input_tokens": 418, + "estimated_eval_tokens": 1092 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1672, + "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 two superficially similar but distinct bug reports about model loading/dispatch, not duplicates: one is a ZeRO-3 zero.Init partitioning/OOM problem for a composite multimodal submodule, the other is a 4-bit device_map/max_memory dispatch-to-CPU ValueError.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No single canonical issue stands out because the reports describe different code paths and failure modes; if one must be picked as the cluster representative, 46822 is slightly more concrete about the affected submodule and OOM symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46822", + "right": "issue:46823", + "accept": false, + "reason": "Both concern loading/dispatch, but the underlying bugs differ: ZeRO-3 zero.Init fails to partition a composite submodule causing OOM, while 4-bit device_map='auto' with max_memory incorrectly sends modules to CPU and raises ValueError. They do not look like the same fixable defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 1, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-40546-2", + "nodes": [ + "pull_request:40546", + "pull_request:42781" + ], + "soft_pairs": [ + "pull_request:40546|pull_request:42781" + ], + "prepared_review_unit_hash": "6d7039d86631c2ce089ba668a87c58747cf5e74dcaf552d61cf68118924cb2c1", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7223, + "estimated_input_tokens": 1806, + "estimated_eval_tokens": 3868 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7223, + "estimated_input_tokens": 1806, + "estimated_eval_tokens": 3868 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related VibeVoice implementation PRs, but they target different scopes: the first adds the base VibeVoice model/docs, while the second adds VibeVoice Realtime plus an acoustic tokenizer and extra generation/auto-processing plumbing. They overlap in shared model files, but are not the same concrete change set.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #42781 is the broader, newer VibeVoice deliverable and includes the base VibeVoice code plus realtime/tokenizer support, so it is the better representative PR for this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR #42781 is the strongest representative of the VibeVoice work because it bundles the main model integration with the additional realtime path and supporting components; #40546 is a narrower precursor implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Shared VibeVoice files alone are not enough: #40546 implements the base model, while #42781 adds realtime-specific functionality and an acoustic tokenizer with extra auto/generation changes. They are related but 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": 2, + "component_count": 36, + "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": "4cbfc830f0de97c5b90d2ca4443a3a0053513ffd8c65a789e96f06cfe8ac8f15", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4675, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are unrelated: one adds TimesFM forecasting with covariates, the other adds DINOv3 image-classification support. They do not share the same underlying bug or code change, so this is not a duplicate cluster.", + "confidence": 0.99, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR #41224 is the more self-contained feature addition, but it is still unrelated to PR #40755 and should not be treated as the canonical representative of a duplicate set.", + "soft_edge_verdicts": [ + { + "left": "[TimesFM] Add support for forecasting with covariates", + "right": "Add DINOv3ViTForImageClassification support", + "accept": false, + "reason": "Different models, different code paths, and different feature areas; no evidence they fix the same underlying issue or could be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-41356-2", + "nodes": [ + "pull_request:41356", + "pull_request:44339" + ], + "soft_pairs": [ + "pull_request:41356|pull_request:44339" + ], + "prepared_review_unit_hash": "aead1308c167b0bae6ff11bf9358340baa5e76f816618383418b505a10a9cd01", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6050, + "estimated_input_tokens": 1513, + "estimated_eval_tokens": 3282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two DEIMv2 pull requests target the same model family, but they are not the same change: PR 41356 is an early/basic model + image processor add, while PR 44339 is a much broader follow-up that adds loss code, auto-mappings, conversion tooling, modular files, and more integration work.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44339 is the more complete and representative DEIMv2 implementation; PR 41356 looks like an earlier partial version of the same model bring-up.", + "best_issue_reason": null, + "best_pr_reason": "PR 44339 is the strongest single artifact for the cluster because it subsumes the basic model work and includes the broader integration needed for a full Transformers addition.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41356", + "right": "pull_request:44339", + "accept": false, + "reason": "Same model area, but not the same concrete change: 44339 expands the implementation substantially and adds new code paths beyond the basic PR 41356. They look like related iterations, not mergeable duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 4, + "component_count": 36, + "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-41855-15", + "nodes": [ + "pull_request:41855", + "pull_request:42668", + "pull_request:43251", + "pull_request:43254", + "pull_request:43424", + "pull_request:43823", + "pull_request:43989", + "pull_request:43995", + "pull_request:44827", + "pull_request:45170", + "pull_request:45213", + "pull_request:45655", + "pull_request:46048", + "pull_request:46267", + "pull_request:46819" + ], + "soft_pairs": [ + "pull_request:41855|pull_request:43823", + "pull_request:41855|pull_request:43424", + "pull_request:42668|pull_request:43823", + "pull_request:43251|pull_request:46819", + "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:46819", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43989|pull_request:46819", + "pull_request:43995|pull_request:46819", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:46267", + "pull_request:44827|pull_request:46048", + "pull_request:44827|pull_request:46819", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:45655|pull_request:46819", + "pull_request:44827|pull_request:45655", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "ec35b27e7bdf8b56ce31f353a4206af9d28d2ba6fc9b69c29700eaef8cf16683", + "component_budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 24, + "serialized_chars": 40996, + "estimated_input_tokens": 10249, + "estimated_eval_tokens": 20754 + }, + "budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 24, + "serialized_chars": 40996, + "estimated_input_tokens": 10249, + "estimated_eval_tokens": 20754 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly unrelated PRs, with one likely duplicate pair around `fixed_cross_entropy` kwargs handling. I would keep `pull_request:43251` as the canonical representative for that duplicate fix and reject the rest as separate changes.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "`pull_request:43251` is the clearest representative of the only real duplicate pair in the set: it directly targets issue 43240 and fixes `fixed_cross_entropy` by forwarding the missing `cross_entropy` kwargs in `loss_utils.py`.", + "best_issue_reason": null, + "best_pr_reason": "`pull_request:43251` is the best PR anchor for the cluster because it is the most issue-linked and concrete fix among the near-duplicates; `43254` is essentially the same change but less directly anchored, while the remaining PRs are unrelated.", + "soft_edge_verdicts": [ + { + "left": "pull_request:41855", + "right": "pull_request:43823", + "accept": false, + "reason": "Mistral tokenizer method additions vs adding MobileLLM; different components and no shared bug/change." + }, + { + "left": "pull_request:41855", + "right": "pull_request:43424", + "accept": false, + "reason": "Tokenizer API gap fix vs ExecuTorch dynamic-shape test; unrelated changes." + }, + { + "left": "pull_request:42668", + "right": "pull_request:43823", + "accept": false, + "reason": "Processor auto-loading robustness vs a new model addition; not the same underlying change." + }, + { + "left": "pull_request:43251", + "right": "pull_request:46819", + "accept": false, + "reason": "`fixed_cross_entropy` kwargs forwarding vs beam-search cache reordering; different code paths and bugs." + }, + { + "left": "pull_request:43251", + "right": "pull_request:43823", + "accept": false, + "reason": "Loss utility fix and MobileLLM model add are unrelated." + }, + { + "left": "pull_request:43251", + "right": "pull_request:44827", + "accept": false, + "reason": "Cross-entropy kwargs fix vs Mistral4 test/model work; unrelated." + }, + { + "left": "pull_request:43424", + "right": "pull_request:44827", + "accept": false, + "reason": "ExecuTorch export test vs Mistral4 fixes; not the same underlying bug/change." + }, + { + "left": "pull_request:43424", + "right": "pull_request:43823", + "accept": false, + "reason": "Dynamic-shape export test vs model addition; unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:46819", + "accept": false, + "reason": "MobileLLM model addition vs beam-search cache fix; no common code-path problem." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45655", + "accept": false, + "reason": "New model support vs `cls.config` resolution order fix; unrelated." + }, + { + "left": "pull_request:43823", + "right": "pull_request:43995", + "accept": false, + "reason": "New model addition vs Falcon output-collection refactor; different feature areas." + }, + { + "left": "pull_request:43989", + "right": "pull_request:46819", + "accept": false, + "reason": "Auto video processor lookup fix vs beam-search cache handling; unrelated bugs." + }, + { + "left": "pull_request:43995", + "right": "pull_request:46819", + "accept": false, + "reason": "Falcon refactor vs beam-search cache reordering; not mergeable as one fix." + }, + { + "left": "pull_request:43995", + "right": "pull_request:45213", + "accept": false, + "reason": "Model refactor vs model-creation skill content; unrelated." + }, + { + "left": "pull_request:43995", + "right": "pull_request:44827", + "accept": false, + "reason": "Falcon output-interface refactor vs Mistral4 test fixes; different subsystems." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45170", + "accept": false, + "reason": "Mistral4 test/model fixes vs `layrnorm` mapping rename; separate model families and fixes." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46267", + "accept": false, + "reason": "Mistral4 test fixes vs chat response parsing update; unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46048", + "accept": false, + "reason": "Mistral4 test fixes vs TDT loss kernel addition; different features." + }, + { + "left": "pull_request:44827", + "right": "pull_request:46819", + "accept": false, + "reason": "Mistral4 test fixes vs beam-search cache fix; unrelated." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45213", + "accept": false, + "reason": "Mistral4 fixes vs model-creation skill; completely different purpose." + }, + { + "left": "pull_request:43823", + "right": "pull_request:45213", + "accept": false, + "reason": "MobileLLM model addition vs AI skill/doc scaffolding; unrelated." + }, + { + "left": "pull_request:45655", + "right": "pull_request:46819", + "accept": false, + "reason": "`cls.config` resolution order fix vs beam-search cache reordering; different code paths." + }, + { + "left": "pull_request:44827", + "right": "pull_request:45655", + "accept": false, + "reason": "Mistral4 test/model work vs `cls.config` resolution fix; no underlying bug overlap." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 36, + "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": "a031358726f6d92155050c1a671c98f8230885c97ead982f50abbadc9f3285dd", + "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 bump/maintenance changes, but they target different action families and different motivations, so they are not duplicates of the same underlying change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the broader and more clearly scoped workflow maintenance change, upgrading multiple GitHub Actions across many workflows for Node 24 compatibility.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the best representative if one PR must anchor the cluster because it covers the widest set of workflow-action updates and has a specific compatibility goal.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "They are both workflow dependency upgrades, but 42975 is about GitHub Actions version bumps for Node 24 compatibility, while 42976 upgrades Docker build actions to latest versions; different action types, different workflows, and not the same concrete bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 6, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-42978-7", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:45490", + "pull_request:46562", + "pull_request:46658", + "pull_request:46705" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:42978|pull_request:46705", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43448|pull_request:46705", + "pull_request:43451|pull_request:45490", + "pull_request:43451|pull_request:46705", + "pull_request:45490|pull_request:46562", + "pull_request:45490|pull_request:46705", + "pull_request:46562|pull_request:46658" + ], + "prepared_review_unit_hash": "4c6a4c05fca73ef77ff365870dcf138ceab3f85e40e56d6b745f1f4a170502a7", + "component_budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24699, + "estimated_input_tokens": 6175, + "estimated_eval_tokens": 12606 + }, + "budget": { + "node_count": 7, + "item_count": 7, + "soft_pair_count": 12, + "serialized_chars": 24699, + "estimated_input_tokens": 6175, + "estimated_eval_tokens": 12606 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: four separate model-addition PRs plus one distributed-config PR. The only mildly related pair is the Step3p7/Step_3_7_flash work, but they target different code paths and should not be merged as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No clear canonical PR: the items span unrelated model integrations and infra changes, so no single PR safely subsumes the others.", + "best_issue_reason": null, + "best_pr_reason": "No PR is globally suitable as a duplicate representative; at best, one of the larger model-addition PRs could serve as a topic anchor, but none is a true duplicate match for the rest.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model families and code paths: ViT NEPA vs Molmo." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "Different model families: ViT NEPA vs Molmo2." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "Different domains and implementations: vision model add vs time-series CTSM." + }, + { + "left": "pull_request:42978", + "right": "pull_request:46705", + "accept": false, + "reason": "Model addition PR vs distributed configuration infrastructure change." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Related naming, but distinct models and separate implementations: Molmo vs Molmo2." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Different architectures and tasks: Molmo vs CTSM." + }, + { + "left": "pull_request:43448", + "right": "pull_request:46705", + "accept": false, + "reason": "Different scope: model integration vs distributed config plumbing." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Different model families and tasks: Molmo2 vs CTSM." + }, + { + "left": "pull_request:43451", + "right": "pull_request:46705", + "accept": false, + "reason": "Different scope: model integration vs distributed config plumbing." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46562", + "accept": false, + "reason": "Both are model additions, but they target different products/code paths: CTSM vs Step3p7 Flash." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46705", + "accept": false, + "reason": "Different scope: time-series model addition vs distributed config infrastructure." + }, + { + "left": "pull_request:46562", + "right": "pull_request:46658", + "accept": false, + "reason": "Same broad family, but not the same concrete change: one adds the Step3p7 model, the other is a separate WIP Step 3 7 Flash script/mini-model effort." + } + ] + }, + "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": 36, + "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-43291-8", + "nodes": [ + "pull_request:43291", + "pull_request:43498", + "pull_request:43665", + "pull_request:44815", + "pull_request:45114", + "pull_request:46380", + "pull_request:46539", + "pull_request:46579" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43291|pull_request:46380", + "pull_request:43498|pull_request:46579", + "pull_request:43665|pull_request:46539", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:46380", + "pull_request:43665|pull_request:46579", + "pull_request:44815|pull_request:46539", + "pull_request:44815|pull_request:46380", + "pull_request:44815|pull_request:46579", + "pull_request:45114|pull_request:46579", + "pull_request:46380|pull_request:46539", + "pull_request:46380|pull_request:46579" + ], + "prepared_review_unit_hash": "01fb16fea64b647191074a0b5f38915ab1ab7ec01478748ca58814ca634308f5", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 14, + "serialized_chars": 23162, + "estimated_input_tokens": 5791, + "estimated_eval_tokens": 11838 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 14, + "serialized_chars": 23162, + "estimated_input_tokens": 5791, + "estimated_eval_tokens": 11838 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a grab bag of unrelated PRs: Whisper tokenizer/tests, tie_weights compatibility, dummy vision test assertions, FP8 dequant work, doctest/typo cleanup, CI infra, and generation-config fixes. The overlaps are only superficial, so none of the soft edges look like true duplicates.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "No clear canonical PR for duplicate-triage purposes; the cluster is heterogeneous. If one must be chosen as the best anchor, PR 46380 is the most substantive standalone runtime fix, but it is not a duplicate of the others.", + "best_issue_reason": null, + "best_pr_reason": "PR 46380 is the strongest standalone bug-fix candidate: it changes generation config validation/runtime behavior in a concrete code path, unlike the mostly test/doc/typo-only PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Whisper tokenizer/test fix vs. dummy assertion changes in CLIP/ViT tests; different bugs and code paths." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper decode logic vs. FP8 dequant/core loading changes; unrelated subsystems." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Whisper tokenizer fix vs. generation-config/Whisper validation fix; too broad and not the same underlying bug." + }, + { + "left": "pull_request:43498", + "right": "pull_request:46579", + "accept": false, + "reason": "Backward-compatibility alias for tie_weights vs. broad typo cleanup; different change types and no shared bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Intentional failing test edits vs. Phimoe CI/integration-test infrastructure; not the same issue." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Dummy test assertions vs. FP8 dequant fix; unrelated artifacts." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "Dummy test assertions vs. generation-validation/runtime fix; no common concrete bug." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46579", + "accept": false, + "reason": "Dummy test assertions vs. typo/doc cleanup across many code files; not duplicate work." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "FP8 dequant/model-loading fix vs. Phimoe CI memory/device-map setup; different problems." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "Both touch some shared utilities, but one is FP8 dequantization and the other is generation-config validation; not the same bug." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46579", + "accept": false, + "reason": "Concrete dequant fix vs. typo cleanup PR; no overlap in underlying change." + }, + { + "left": "pull_request:45114", + "right": "pull_request:46579", + "accept": false, + "reason": "Doctest formatting updates vs. general typo cleanup in code; similar surface area only, not same change." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46539", + "accept": false, + "reason": "Generation config/Whisper runtime fix vs. Phimoe test-job infrastructure; unrelated." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46579", + "accept": false, + "reason": "Generation validation/runtime fix vs. broad typo cleanup; not a duplicate or same bug family." + } + ] + }, + "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": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43488-3", + "nodes": [ + "pull_request:43488", + "pull_request:43532", + "pull_request:46747" + ], + "soft_pairs": [ + "pull_request:43488|pull_request:46747", + "pull_request:43488|pull_request:43532", + "pull_request:43532|pull_request:46747" + ], + "prepared_review_unit_hash": "9b33a20f3e739f9530dfbd34118031fa2e0aea86169ce0518bd86ce63f0dd415", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6349, + "estimated_input_tokens": 1588, + "estimated_eval_tokens": 3432 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6349, + "estimated_input_tokens": 1588, + "estimated_eval_tokens": 3432 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three pull requests are superficially similar as repo-maintenance/don't-merge changes, but they address different concrete areas: packaging/setup, CI comment automation, and PR security-gate workflow behavior. None look like duplicate fixes that could plausibly be merged into one PR.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical duplicate PR: the three PRs target unrelated changes and do not share the same underlying bug or code-path fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 46747 is the cleanest standalone representative because it is a single-file workflow change with one concrete behavior change, unlike the broader setup/workflow/test edits in the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:46747", + "accept": false, + "reason": "Different concerns: setup/version/dependency and a Vit test tweak versus changing the PR security-gate workflow trigger. Not the same underlying change." + }, + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "One changes packaging/dependencies and a test fixture; the other rewrites multiple GitHub workflow files and testing utilities. Different code paths and purposes." + }, + { + "left": "pull_request:43532", + "right": "pull_request:46747", + "accept": false, + "reason": "Both are workflow-related, but one handles self-comment CI/notification plumbing while the other changes the PR security-gate trigger model. Not a single 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": 9, + "component_count": 36, + "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": "8d9f542cce616734d82cdff1a17380025eaf0f02f7353cba572a02f0bbd4864f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6082, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are related but not duplicates: one adds a Promptable Concept Segmentation pipeline, the other adds a Promptable Visual Segmentation pipeline. They share pipeline/docs scaffolding, but target different features and model/test surfaces.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different underlying changes: concept segmentation vs visual segmentation are separate pipelines with different filenames, model integrations, and tests. The overlap is mostly shared infrastructure/docs, 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": 10, + "component_count": 36, + "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": "3366c015551af7bcd4431574a82a521d67a4ade4d3904d199e2516808e8669c9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4293, + "estimated_input_tokens": 1074, + "estimated_eval_tokens": 2404 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs both touch `trainer.py` but address different trainer behaviors: gradient-accumulation loss scaling vs. signature-column filtering for dataset column pruning. They are not the same underlying change and should not be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "No issue exists in this cluster; if one PR must be treated as the representative, `pull_request:43651` is the more self-contained code-path change, while `pull_request:43663` is an unrelated refactor of input-column filtering.", + "best_issue_reason": null, + "best_pr_reason": "`pull_request:43651` is the better representative because it fixes a specific training-step loss-scaling behavior with a focused property addition. `pull_request:43663` targets a different mechanism entirely (`_get_signature_columns`) and is not a duplicate of it.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Different trainer code paths and different intended behavior: one changes gradient accumulation loss normalization, the other changes model signature inspection / column filtering. Shared filename alone is not enough to treat them as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 11, + "component_count": 36, + "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": "5df3ba9feb0af633f67a7aed233fe1aaa9aee9ecaadf1a30bee7e334a082728e", + "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": "The two PRs are related by file and theme, but they fix different TypeAdapter-related failures in the CLI, so they should not be treated as duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 43842 is the narrower, more explicit fix for the runtime pydantic/TypeAdapter import problem and has a direct issue link; PR 43656 appears to address a separate annotation-evaluation NameError.", + "best_issue_reason": null, + "best_pr_reason": "If one PR must represent the cluster, 43842 is the better choice because it targets a concrete optional-dependency runtime failure with a specific fix, while 43656 is a different bug in the same file.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "Same file and similar wording, but different underlying problems: 43656 changes annotation handling to avoid a NameError from evaluated type hints, while 43842 makes TypeAdapter a TYPE_CHECKING-only import to avoid requiring pydantic at runtime. They are not the same fix and should not be merged as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 36, + "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": 13, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44024", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44018|pull_request:44068", + "pull_request:44066|pull_request:44071", + "pull_request: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:44722", + "pull_request:44068|pull_request:44116", + "pull_request:44018|pull_request:44116", + "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:44018|pull_request:44071", + "pull_request:44059|pull_request:44722", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request:43997|pull_request:44073", + "pull_request:44071|pull_request:44086", + "pull_request:44017|pull_request:44071", + "pull_request:44073|pull_request:44101", + "pull_request:44073|pull_request:44161", + "pull_request:44059|pull_request:44071", + "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:43997|pull_request:44018", + "pull_request:44056|pull_request:44161", + "pull_request:43997|pull_request:44161", + "pull_request:44074|pull_request:44161", + "pull_request:44018|pull_request:44161", + "pull_request:44010|pull_request:44071", + "pull_request:44010|pull_request:44073", + "pull_request:44018|pull_request:44129", + "pull_request:43997|pull_request:44074", + "pull_request:44010|pull_request:44018", + "pull_request:43997|pull_request:44129", + "pull_request:44071|pull_request:44129", + "pull_request:43996|pull_request:44161", + "pull_request:43996|pull_request:44017", + "pull_request:44010|pull_request:44044", + "pull_request:44044|pull_request:44074", + "pull_request:44066|pull_request:44722", + "pull_request:44085|pull_request:44722", + "pull_request:44085|pull_request:44129", + "pull_request:43996|pull_request:44007", + "pull_request:43996|pull_request:44129", + "pull_request:44044|pull_request:44129", + "pull_request:44044|pull_request:44154", + "pull_request:44024|pull_request:44076" + ], + "prepared_review_unit_hash": "37ce14a4b46d1b66d03098235d7792529a93b26e6095582259477e1fe82e816b", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84503, + "estimated_input_tokens": 21126, + "estimated_eval_tokens": 42508 + }, + "budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 72, + "serialized_chars": 76936, + "estimated_input_tokens": 19234, + "estimated_eval_tokens": 38724 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Cluster is mostly a pile of model-specific output-tracing refactors; only the GPT-Neo and GPT-J/CodeGen subgroups look like true duplicates.", + "confidence": 0.73, + "canonical_issue_reason": null, + "canonical_pr_reason": "44722 is the latest, most complete GPT-J/CodeGen output-tracing refactor and effectively supersedes the earlier GPT-J-focused variants.", + "best_issue_reason": null, + "best_pr_reason": "44722 best represents the duplicate core because it covers the same GPT-J/CodeGen code path as the earlier matching PRs and is the broadest revision.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Shared tracing theme only; different model code paths (CVT/FNet vs GPT-J)." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Different model families and implementations; only the same umbrella tracing refactor theme." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both are GPT-J output-tracing refactors on the same code path; one is a narrower subset of the other." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "ResNet/RT-DETR vs EfficientNet are different code paths, despite the shared tracing migration theme." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families; same standardized-output pattern is too broad to treat as a duplicate." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen changes are unrelated to EfficientNet's output tracing path." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTaV2 are distinct implementations; only the refactor pattern overlaps." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "Different underlying models/code paths (GPT-J/CodeGen vs MGP-STR)." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Both are GPT-Neo output-tracing refactors on the same model/code path." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen vs MPT; shared output-tracing infrastructure only." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model paths; not the same concrete bug/change." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen and GPT-Neo are different models; same tooling, not same fix." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet-related refactor does not fix the GPT-J/CodeGen code path." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen are unrelated code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are different model families; only the broad tracing initiative matches." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are unrelated code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are different implementations; not the same underlying fix." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo and GPT-J/CodeGen are different model code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert/XLM are unrelated fixes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert/XLM do not share the same concrete bug." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen are different model paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo vs GPT-J; same refactor style, different underlying model." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are different model implementations." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT2 and MGP-STR are unrelated concrete fixes." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert/XLM are different model paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are different model code paths." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT2 and GPT-J/CodeGen are different model paths; too broad to merge." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are unrelated implementations." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44722", + "accept": false, + "reason": "SegFormer and GPT-J/CodeGen are different model families." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44073", + "accept": false, + "reason": "SegFormer and VisualBert are different concrete code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-Neo and MPT are unrelated model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44073", + "accept": false, + "reason": "RegNet and VisualBert are different code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44086", + "accept": false, + "reason": "MPT and MGP-STR are unrelated model fixes." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44071", + "accept": false, + "reason": "SegFormer and MPT are different model families." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44101", + "accept": false, + "reason": "VisualBert and Flaubert/XLM are different code paths." + }, + { + "left": "pull_request:44073", + "right": "pull_request:44161", + "accept": false, + "reason": "VisualBert and LongT5 are unrelated fixes." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT2 and MPT are different underlying model paths." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet and SegFormer are different model implementations." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are unrelated concrete fixes." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are different model paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44018", + "accept": false, + "reason": "RegNet and GPT-Neo are different model families." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44161", + "accept": false, + "reason": "MPNet and LongT5 are unrelated fixes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 are different model paths." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet and LongT5 are unrelated implementations." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44161", + "accept": false, + "reason": "GPT-Neo and LongT5 are different concrete code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are unrelated model fixes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo and SpeechT5 are different model families." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated concrete changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are different model paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different implementations." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT2 and GPT-Neo are different model code paths." + }, + { + "left": "pull_request:44101", + "right": "pull_request:44161", + "accept": false, + "reason": "Flaubert/XLM and LongT5 are unrelated fixes." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are different model families." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "CVT/FNet and SegFormer are different code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTaV2 are unrelated model fixes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTaV2 and TextNet are different underlying model paths." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor; 44722 is the broader later revision." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both target the same GPT-J output-tracing change; 44722 subsumes the narrower GPT-J-only variant." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J and SpeechT5 are different model code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "CVT/FNet and ResNet are different model implementations." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated fixes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTaV2 and SpeechT5 are different concrete code paths." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTaV2 and VITS are unrelated model fixes." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are different model paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is conservative and consistent with the packet: it treats the same-model GPT-Neo and GPT-J/CodeGen PRs as duplicates, and rejects cross-model pairs rather than overclaiming mergeability. The canonical/best PR choices are grounded in the later, broader GPT-J/CodeGen revisions shown in the diffs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 36, + "review_unit_index": 2, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43997", + "pull_request:44010", + "pull_request:44056", + "pull_request:44073", + "pull_request:44074", + "pull_request:44129" + ], + "soft_pairs": [ + "pull_request:44073|pull_request:44074", + "pull_request:44056|pull_request:44073", + "pull_request:44056|pull_request:44074", + "pull_request:43997|pull_request:44010", + "pull_request:44010|pull_request:44129" + ], + "prepared_review_unit_hash": "6d82f7eb02cb3a86fe8ffd329e80b7eae494296137f2838be26a613758818663", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84503, + "estimated_input_tokens": 21126, + "estimated_eval_tokens": 42508 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 5, + "serialized_chars": 13067, + "estimated_input_tokens": 3267, + "estimated_eval_tokens": 6790 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "All items are model-specific PRs that standardize output tracing/capture across different architectures. They share a refactoring theme, but they do not appear to fix the same concrete code path or to be mergeable into one PR as duplicates.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44129 is the strongest cluster representative: it has the only explicit issue target, a focused model-specific refactor, and the clearest connection to the shared output-tracing migration theme.", + "best_issue_reason": null, + "best_pr_reason": "PR 44129 is the best standalone PR in the set because it is issue-linked and cleanly exemplifies the standardized output-capture refactor pattern.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch different models and code paths (VisualBert vs TextNet); same pattern, not the same change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "MPNet and VisualBert are separate model implementations; this is a shared refactor style, not a duplicate bug or one mergeable PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Different model targets and different implementation details; they are related by framework migration only, not by the same underlying fix." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert each migrate output tracing independently; no evidence they address the same concrete bug or code path." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "These are separate architecture migrations with different affected layers and outputs; too broad to treat as one duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 14, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43998-13", + "nodes": [ + "pull_request:43998", + "pull_request:43999", + "pull_request:44000", + "pull_request:44001", + "pull_request:44002", + "pull_request:44003", + "pull_request:44004", + "pull_request:44025", + "pull_request:44026", + "pull_request:44027", + "pull_request:44028", + "pull_request:44029", + "pull_request:44030" + ], + "soft_pairs": [ + "pull_request:43998|pull_request:44002", + "pull_request:43998|pull_request:44000", + "pull_request:43998|pull_request:44001", + "pull_request:43998|pull_request:44003", + "pull_request:43998|pull_request:44029", + "pull_request:43999|pull_request:44001", + "pull_request:43999|pull_request:44002", + "pull_request:43999|pull_request:44025", + "pull_request:43999|pull_request:44026", + "pull_request:43999|pull_request:44027", + "pull_request:44000|pull_request:44001", + "pull_request:44000|pull_request:44002", + "pull_request:44000|pull_request:44003", + "pull_request:44000|pull_request:44025", + "pull_request:44001|pull_request:44002", + "pull_request:44001|pull_request:44003", + "pull_request:44001|pull_request:44025", + "pull_request:44002|pull_request:44029", + "pull_request:44002|pull_request:44003", + "pull_request:44003|pull_request:44028", + "pull_request:44003|pull_request:44004", + "pull_request:43998|pull_request:44004", + "pull_request:44001|pull_request:44004", + "pull_request:44002|pull_request:44004", + "pull_request:44000|pull_request:44004", + "pull_request:44025|pull_request:44026", + "pull_request:44025|pull_request:44027", + "pull_request:44025|pull_request:44028", + "pull_request:44025|pull_request:44029", + "pull_request:44025|pull_request:44030", + "pull_request:44026|pull_request:44029", + "pull_request:44026|pull_request:44027", + "pull_request:44026|pull_request:44028", + "pull_request:44026|pull_request:44030", + "pull_request:44027|pull_request:44028", + "pull_request:44027|pull_request:44029", + "pull_request:44027|pull_request:44030", + "pull_request:44028|pull_request:44029", + "pull_request:44028|pull_request:44030", + "pull_request:44029|pull_request:44030" + ], + "prepared_review_unit_hash": "555d84ea5f1f3e8611428eb74fe2a7ad8d2cb6a1b335bbe20a62ed17580810d6", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40173, + "estimated_input_tokens": 10044, + "estimated_eval_tokens": 20344 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are thematically similar refactors around output tracing / `can_return_tuple`, but they target different model families and code paths. I do not see true duplicates or a single underlying bug/change shared across the set.", + "confidence": 0.87, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44003 is the broadest representative of the cluster: it applies the output-tracing refactor across two related Mamba-family files and introduces the shared capture plumbing, making it the most central example of this theme.", + "best_issue_reason": null, + "best_pr_reason": "PR 44003 is the best single representative in this set because it has the widest and clearest shared implementation change among closely related models, while the others are more isolated per-model refactors.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Same refactor pattern, but timm_backbone and upernet are different model paths with separate forward behavior; not the same underlying change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Both are output-tracing refactors, but they touch unrelated model families and code paths." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Different architectures and separate forward implementations; shared theme is too broad to treat as one change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "timm_backbone refactor is unrelated to the Mamba/Falcon-Mamba changes in 44003." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model families; no shared concrete bug or code path." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated implementations; same refactor style does not make them duplicates." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Separate models and separate forward signatures; only the general tracing refactor matches." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "MobileNetV1 and Depth Anything are different code paths; not one mergeable fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "Different subsystems with no shared concrete behavior change beyond output handling cleanup." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "MobileNetV1 and Speech-Encoder-Decoder are unrelated; this is only a broad refactor similarity." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "Vision-text dual encoder and UnivNet are unrelated models; no common underlying bug/change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "These are separate model refactors in different code paths, not the same fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "Shared tracing pattern, but CLIP-like vision-text and Mamba-family code are unrelated." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Different model families and forward logic; do not look mergeable into one PR." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are unrelated; same output-cleanup theme is too broad." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "Different model families and distinct code paths; not a duplicate change." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "UnivNet and Depth Anything are separate implementations with no shared concrete bug." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are unrelated models; only the refactor pattern overlaps." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "UperNet and Mamba-family code are different subsystems; not the same underlying change." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Mamba-family refactor and SuperPoint refactor are separate model-specific changes." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Both are output-signature refactors, but Mamba and CodeGen target different implementation details and are not one fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "timm_backbone and CodeGen are unrelated; shared refactor language is insufficient." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different models with distinct code paths." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen are unrelated beyond broad tracing cleanup." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "Vision-text dual encoder and CodeGen are not the same underlying change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "Depth Anything and Vision-Encoder-Decoder are separate model families; no shared concrete bug." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "Depth Anything and Speech-Encoder-Decoder are unrelated code paths." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "These are different model-specific refactors; no single fix to merge." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "Depth Anything and RWKV are unrelated models and implementation paths." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "Depth Anything and DPR are separate code paths; not a duplicate." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "Vision-Encoder-Decoder and RWKV are unrelated; only the output-tracing theme overlaps." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "Vision-Encoder-Decoder and Speech-Encoder-Decoder are different implementations and not the same change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "Vision-Encoder-Decoder and SuperPoint are unrelated model families." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "Different models and different forward paths; no duplicate fix." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "Speech-Encoder-Decoder and SuperPoint are unrelated code paths." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "Different model families; same refactor pattern is not enough." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "Speech-Encoder-Decoder and DPR are separate implementations with no shared bug." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are unrelated; not mergeable into one PR." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint and DPR are different model paths and not the same underlying fix." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are distinct models; only the generic tracing refactor overlaps." + } + ] + }, + "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": 36, + "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": "9000232f9688616c747369ab36366cb5c5644264577d0ce6ed53860dc3d148c4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6551, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two unrelated model-implementation PRs: one adds PenguinVL, the other adds Voxtral TTS. They share boilerplate auto-registry and docs paths, but target different models, codepaths, and test suites.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45401 is the better canonical PR for this cluster because it has an explicit issue target (45027) and is a focused, self-contained model addition.", + "best_issue_reason": null, + "best_pr_reason": "PR #45401 is the best PR to represent the cluster only in the sense that it is the most clearly scoped and traceable to a specific task; however, it is not a duplicate of #44662.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Reject: these are different model additions (PenguinVL vs Voxtral TTS) with different implementations, configs, processors, and tests. Shared auto-registration/docs edits are broad framework boilerplate, not evidence of the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 16, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-44979-2", + "nodes": [ + "pull_request:44979", + "pull_request:45363" + ], + "soft_pairs": [ + "pull_request:44979|pull_request:45363" + ], + "prepared_review_unit_hash": "af8c378707a694dc81f26f9bdda697272192e6c698201bbe7fb60a3c7a3ac348", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4406, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4406, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two pull requests are related by the broader idea of module/kernel fusion, but they implement different features and code paths: one adds a generic Module Fusion API, while the other adds n-to-1 kernel fusion through KernelConfig and model-loading integration. They are not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45363 is the better canonical representative because it is the more user-facing end-to-end feature and touches the actual kernel-fusion integration path, whereas PR 44979 is a lower-level API foundation.", + "best_issue_reason": null, + "best_pr_reason": "PR 45363 is the best representative of the cluster because it covers the concrete fusion workflow with config, integration, and example usage; PR 44979 is more of an enabling API layer.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Related theme, but not the same change: 44979 introduces a generic Module Fusion API, while 45363 implements KernelConfig-driven n-to-1 kernel fusion and model-loading integration. They do not appear mergeable into one duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 17, + "component_count": 36, + "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": "f9bc5de197ebddec69077274427d25e21db06edef5ef2d0065f838185b46be3b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4414, + "estimated_input_tokens": 1104, + "estimated_eval_tokens": 2464 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs implement the same Mistral tokenizer fix for the same AttributeError in `_patch_mistral_regex`, changing the same lines in the same file; 45317 adds regression tests on top of the identical code fix.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "pull_request:45317 is the most complete version of the shared fix: it applies the same code-path correction as 45086 and adds test coverage.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:45317 is the best representative because it contains the same bug fix plus regression tests, making it the stronger standalone change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying AttributeError fix in `_patch_mistral_regex`, same target file and code change; 45317 just extends it with tests." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 36, + "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-45115-3", + "nodes": [ + "pull_request:45115", + "pull_request:45569", + "pull_request:46565" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569", + "pull_request:45569|pull_request:46565" + ], + "prepared_review_unit_hash": "c16742028f2a3fbc00a007d53da26f40d4c39bf3b4d5a71b164f69347fe2f575", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 9122, + "estimated_input_tokens": 2281, + "estimated_eval_tokens": 4818 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 9122, + "estimated_input_tokens": 2281, + "estimated_eval_tokens": 4818 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three PRs are in the Nemotron family, but they target different changes: a NemotronH conversion/refactor, a NemotronH dense/sparse split and docs, and a separate Nemotron3.5 ASR addition. They are related contextually, but not duplicate fixes or mergeable as one PR.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45569 is the best representative of the cluster because it is the broadest NemotronH code change and the most central to the family split/refactor, touching the core model, auto-mapping, docs, and tests.", + "best_issue_reason": null, + "best_pr_reason": "PR #45569 is the strongest standalone PR for this cluster: it spans the main NemotronH implementation work rather than only a conversion mapping tweak (#45115) or an unrelated ASR model addition (#46565).", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Both are NemotronH-related, but #45115 is a conversion-mapping/inheritance refactor while #45569 is the larger architecture split into NemotronH dense/sparse variants. They do not look like the same concrete bug or change." + }, + { + "left": "pull_request:45569", + "right": "pull_request:46565", + "accept": false, + "reason": "These address different models and code paths: NemotronH dense/sparse refactoring versus Nemotron3.5 ASR streaming support. Shared family naming is not enough to treat them as duplicates." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 19, + "component_count": 36, + "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": "d72820a195f8f27352df4dd6017f465aded3b618cbc6e9ec06f058265f5d36a9", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5954, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5954, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are closely related parts of the same heterogeneous-model feature work. PR #45332 is the broader, more complete change set, while PR #45333 is a narrower config-only slice that overlaps heavily with it. The overlap is enough to treat them as the same change family for triage.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45332 is the better canonical PR because it covers the full heterogeneous-model feature, including per-layer config plus modeling/runtime changes, whereas #45333 is a narrower subset focused on config support only.", + "best_issue_reason": null, + "best_pr_reason": "#45332 is the best representative of the cluster because it subsumes the shared config work and adds the broader modeling integration, making it the most complete instance of the underlying feature.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": true, + "reason": "Same underlying heterogeneous configuration feature; #45333 is a subset of the broader implementation in #45332, with substantial overlap in config/docs/tests and no conflicting scope." + } + ] + }, + "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": 36, + "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-45346-2", + "nodes": [ + "pull_request:45346", + "pull_request:46818" + ], + "soft_pairs": [ + "pull_request:45346|pull_request:46818" + ], + "prepared_review_unit_hash": "f66b95f83c9e0c960f83091dc23bcc8b96b4951ae1d813928e306d50d3eeb342", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5591, + "estimated_input_tokens": 1398, + "estimated_eval_tokens": 3052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5591, + "estimated_input_tokens": 1398, + "estimated_eval_tokens": 3052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two pull requests are not duplicates: one is a narrow regression test for MoE router logits being raw logits instead of softmaxed probabilities, while the other is a broad expert-parallel/DeepGEMM change about router contract handling and FP8 scale format across many models.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46818 is the better canonical representative because it contains the substantive implementation changes and spans the actual code paths involved, whereas PR 45346 is a small test-focused regression fix for a different bug.", + "best_issue_reason": null, + "best_pr_reason": "PR 46818 is the best overall representative of the cluster because it is the larger code change and more central architectural fix; PR 45346 is too narrow and addresses a distinct MoE router-logits regression.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45346", + "right": "pull_request:46818", + "accept": false, + "reason": "Different underlying bugs: 45346 tests for double-softmaxed router logits in Mixtral/Qwen2 MoE, while 46818 changes EP router contracts and FP8/DeepGEMM scale handling across many models. They do not look 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": 21, + "component_count": 36, + "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": "d090635122b2ae75c0ea70f4041755a82a229a55b06d5ded368a2697f71ccb2f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5513, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5513, + "estimated_input_tokens": 1379, + "estimated_eval_tokens": 3014 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs in the same typing cleanup area, but they address different slices of the codebase and are not the same concrete fix/change.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45415 is the broader, more central type-checking sweep across many core `src/transformers/*.py` files, so it best represents the main work in this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 45415 is the best single representative PR because it covers the widest set of typing changes and touches core runtime modules, whereas PR 45425 is a narrower follow-up typing cleanup centered around `_typing.py`/`modeling_utils` integration.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Both are typing-related and share some files, but the diffs target different concrete changes: 45415 is a broad type-checking pass over many modules, while 45425 adds typing protocol aliases/helpers and related adjustments. They are related cleanup work, not the same underlying bug or mergeable single PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 36, + "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": "8155fdec04f5cb941da3a77331c3d4f98ba87e1f4dbed3bb77d40d4e6eee7ac0", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4948, + "estimated_input_tokens": 1237, + "estimated_eval_tokens": 2730 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4948, + "estimated_input_tokens": 1237, + "estimated_eval_tokens": 2730 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs fix cache-related multi-token decode bugs, but they affect different model families and code paths (Qwen3.5 linear attention vs. Mamba2 CUDA kernels), so they are not duplicates and should not be merged into one PR.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46084 is the clearest representative: it has an explicit issue target (#46032), a narrowly scoped fix for the seq_len>1 cache path in Mamba2, and accompanying tests.", + "best_issue_reason": null, + "best_pr_reason": "PR #46084 is the best single PR to keep as the cluster representative because it is tightly scoped, explicitly linked to an issue, and addresses one concrete cached-forward bug path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45513", + "right": "pull_request:46084", + "accept": false, + "reason": "Both concern cached multi-token forwarding, but they fix different architectures and implementations, so they are not the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 23, + "component_count": 36, + "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": "7e56a455d57f3a01088ed44223a30e38638c9bef8b199fe07dfccf834b13ff63", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4508, + "estimated_input_tokens": 1127, + "estimated_eval_tokens": 2510 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs are EP-related, but they fix different failure modes in different loading paths: #45548 focuses on preserving distributed_config/has_ep for DeepSpeed ZeRO-3 loading via accelerate launch, while #45662 fixes expert weights being overwritten during FSDP2 loading/broadcast. They are adjacent, not duplicates.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "#45662 is the stronger representative of the cluster because it addresses a concrete corruption bug in the expert-loading path and includes the broader set of code changes tied to that failure mode.", + "best_issue_reason": null, + "best_pr_reason": "#45662 is the best single PR to represent the cluster because it is the most specific, technically grounded fix and the more substantial end-to-end change around EP loading behavior.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Different underlying bugs and code paths: #45548 preserves distributed config state for ZeRO-3/accelerate launch, whereas #45662 prevents experts from being overwritten during FSDP2 loading. Shared EP terminology and one common file are not enough to treat them as the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 24, + "component_count": 36, + "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": "a9fd8bfc920f0594f6a6b8dae31d8ae951837c857dbcb21f497ed75d7a5338ba", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4672, + "estimated_input_tokens": 1168, + "estimated_eval_tokens": 2592 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same xdist collision around captured_info/debug logging in `testing_utils`, with 45645 expanding the fix to CI artifact collection as well. 45645 is the better canonical PR because it is the more complete end-to-end change.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the broader, more complete fix: it includes the xdist-safe output handling in `src/transformers/testing_utils.py` and also updates the workflow to read multiple `captured_info*.txt` files, preserving CI debug logs. That makes it the best representative of the change set.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the strongest single PR for this cluster because it covers both the underlying collision fix and the downstream CI log collection behavior. PR 45639 appears to be the narrower precursor/subset focused only on making the write path xdist-safe.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Same underlying xdist collision/captured_info bug in `testing_utils`; 45645 looks like a superset that adds the workflow-side handling needed to preserve logs, so they could plausibly be merged as one 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": 36, + "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": "da8025eed121ea1f7dd0ebd60011307245cfb48d7051e91d0d914154cb0f8eea", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7234, + "estimated_input_tokens": 1809, + "estimated_eval_tokens": 3874 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 7234, + "estimated_input_tokens": 1809, + "estimated_eval_tokens": 3874 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two different linter-rule PRs in the same general area, but they address unrelated changes: TRF003 adds a return-tuple decorator to model forward methods, while TRF018 changes weight-initialization logic across model classes. They are not duplicates and should not be merged together.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:45720", + "right": "pull_request:46259", + "accept": false, + "reason": "Different underlying changes and code paths: 45720 is about forward/return_dict handling via can_return_tuple, while 46259 is about _init_weights initialization behavior. Similar linter framing and overlapping model files are not enough to treat them as the same PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 26, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45737-2", + "nodes": [ + "pull_request:45737", + "pull_request:45738" + ], + "soft_pairs": [ + "pull_request:45737|pull_request:45738" + ], + "prepared_review_unit_hash": "cc2ade73a49ec6a8883a7a24c433988645588bcd049e8aad2bce56b1b24daa1c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4131, + "estimated_input_tokens": 1033, + "estimated_eval_tokens": 2322 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same MusicgenMelody cache initialization bug in `forward()`: replacing `EncoderDecoderCache(DynamicCache, DynamicCache)` with a single `DynamicCache`. PR #45737 is the more complete duplicate because it bundles the fix with a regression test, while #45738 is the fix-only variant.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45737 is the best canonical PR because it contains the same code fix as #45738 plus a regression test, making it the most complete representation of the bug fix.", + "best_issue_reason": null, + "best_pr_reason": "PR #45737 is the strongest overall candidate: it fixes the concrete cache bug and adds a targeted regression test, whereas #45738 only changes the implementation.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45737", + "right": "pull_request:45738", + "accept": true, + "reason": "Same underlying code-path bug in `musicgen_melody` cache setup; the diffs match on the core fix and the test-only addition in #45737 could plausibly live in the same PR as #45738." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet: both PRs make the same `musicgen_melody` cache initialization change, and #45737 only adds a regression test on top of that fix. The soft-edge accept is conservative enough because the diffs are mergeable into one PR for the same bugfix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 27, + "component_count": 36, + "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": "6be53d2cde71f79b2bea5956de30138a64ee1a9c3fad034ad2e359a67dd710de", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5161, + "estimated_input_tokens": 1291, + "estimated_eval_tokens": 2838 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5161, + "estimated_input_tokens": 1291, + "estimated_eval_tokens": 2838 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests are not duplicates: one is a broad model/code-path and CI-secret change set, while the other is a test migration with new kernel/model tests. They should remain separate.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45932 is the more substantial representative PR: it changes model code and workflow configuration, whereas 46100 is primarily a test migration and test additions.", + "best_issue_reason": null, + "best_pr_reason": "PR 45932 is the best single representative of this cluster because it is the broader functional/code change; 46100 is focused on test infrastructure and does not cover the same underlying change.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45932", + "right": "pull_request:46100", + "accept": false, + "reason": "They do not appear to fix the same concrete bug or implement the same change. 45932 modifies model implementation/mappings and CI workflow secrets, while 46100 adds kernel tests and a DDP test config for test migration. The overlap is only broad testing/model areas, not one mergeable code-path 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": 28, + "component_count": 36, + "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": "8ba0719448ac7deea64b2bf7c14be2bbb9bf9b9eaf34ee5720f4b98ba38295f4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5525, + "estimated_input_tokens": 1382, + "estimated_eval_tokens": 3020 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5525, + "estimated_input_tokens": 1382, + "estimated_eval_tokens": 3020 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are not duplicate PRs: one is a focused EP + Trainer integration change, while the other is a broad distributed branch base touching many unrelated areas. The overlap in distributed files looks incidental to the shared branch rather than the same concrete fix.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46126 is the more canonical change because it is narrowly scoped and centers on a single feature/fix set (EP + Trainer integration around DistributedConfig).", + "best_issue_reason": null, + "best_pr_reason": "PR #46126 is the best representative artifact: it is coherent, specific, and directly describes the underlying change, whereas #46269 is too broad and base-like to serve as a good canonical PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "The overlap is only on broad distributed files; #46126 implements EP/Trainer integration, while #46269 is a large distributed branch base spanning CI and multiple subsystems. They do not look like the same concrete code-path fix and would not plausibly merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 29, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46138-10", + "nodes": [ + "pull_request:46138", + "pull_request:46432", + "pull_request:46501", + "pull_request:46502", + "pull_request:46503", + "pull_request:46504", + "pull_request:46554", + "pull_request:46558", + "pull_request:46563", + "pull_request:46716" + ], + "soft_pairs": [ + "pull_request:46138|pull_request:46554", + "pull_request:46138|pull_request:46558", + "pull_request:46138|pull_request:46563", + "pull_request:46138|pull_request:46716", + "pull_request:46138|pull_request:46504", + "pull_request:46432|pull_request:46503", + "pull_request:46432|pull_request:46501", + "pull_request:46432|pull_request:46502", + "pull_request:46432|pull_request:46554", + "pull_request:46432|pull_request:46558", + "pull_request:46501|pull_request:46503", + "pull_request:46501|pull_request:46502", + "pull_request:46501|pull_request:46504", + "pull_request:46501|pull_request:46554", + "pull_request:46502|pull_request:46503", + "pull_request:46502|pull_request:46504", + "pull_request:46502|pull_request:46554", + "pull_request:46503|pull_request:46504", + "pull_request:46503|pull_request:46554", + "pull_request:46504|pull_request:46554", + "pull_request:46504|pull_request:46558", + "pull_request:46554|pull_request:46558", + "pull_request:46554|pull_request:46563", + "pull_request:46554|pull_request:46716", + "pull_request:46558|pull_request:46563", + "pull_request:46558|pull_request:46716", + "pull_request:46563|pull_request:46716", + "pull_request:46504|pull_request:46563", + "pull_request:46504|pull_request:46716" + ], + "prepared_review_unit_hash": "62102420565375cbce5f562fac75c3482f70d5787129751ae2fa28053f97e30c", + "component_budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 29, + "serialized_chars": 25920, + "estimated_input_tokens": 6480, + "estimated_eval_tokens": 13216 + }, + "budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 29, + "serialized_chars": 25920, + "estimated_input_tokens": 6480, + "estimated_eval_tokens": 13216 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is a set of unrelated workflow-maintenance PRs. The only apparent mini-cluster is pr-ci-security-gate.yml, but those PRs make different security/permission changes, so they are not duplicates. No soft edges should be merged.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single true canonical PR across the whole cluster; pull_request:46554 is the closest representative because it is the most substantial and central pr-ci-security-gate hardening change, but the surrounding PRs are separate follow-up edits rather than the same fix.", + "best_issue_reason": null, + "best_pr_reason": "pull_request:46554 is the best representative of the tightest subcluster by breadth and connectivity, though it is not a true duplicate hub because the other pr-ci-security-gate PRs change different behaviors.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46138", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow and bug: self-comment-ci switches PR comment handling to stdin, while 46554 hardens pr-ci-security-gate. Not the same underlying change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflow and code path; one fixes comment piping, the other changes security-gate permissions/approval flow." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow maintenance edits in different code paths; not the same bug or change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46716", + "accept": false, + "reason": "Self-comment-ci stdin fix is unrelated to the pr-ci-security-gate secrets-scope change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46504", + "accept": false, + "reason": "Different workflows and fixes: comment stdin handling versus COMMIT_MESSAGE quoting in build-ci-docker-images." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46503", + "accept": false, + "reason": "SSH-runner hardening is unrelated to benchmark_v2 dispatch-trigger changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46501", + "accept": false, + "reason": "Different workflow files and unrelated changes; ssh-runner hardening is not the same as anti-slop permission reshuffling." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46502", + "accept": false, + "reason": "Different workflow and fix: ssh-runner secret/memory-monitor hardening versus assign-reviewers checkout ref adjustment." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow and bug; ssh-runner changes are unrelated to pr-ci-security-gate behavior." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflow and concern: ssh-runner security/hardening versus pr-ci-security-gate permission tweaks." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46503", + "accept": false, + "reason": "Different workflow files and different logic; anti-slop permission change is not the benchmark_v2 trigger update." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46502", + "accept": false, + "reason": "Different workflows and unrelated fixes; not the same underlying change." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46504", + "accept": false, + "reason": "anti-slop.yml permission adjustment and build-ci-docker-images message quoting are separate changes." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow files and unrelated concerns; not a duplicate." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46503", + "accept": false, + "reason": "assign-reviewers checkout ref fix and benchmark_v2 dispatch support are different workflow changes." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46504", + "accept": false, + "reason": "Different workflows and fixes; no shared concrete bug." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow files and unrelated behavior changes." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46504", + "accept": false, + "reason": "Benchmark dispatch trigger and build-ci-docker-images grep/quoting fix are unrelated." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow and code path; benchmark_v2 dispatch support is not pr-ci-security-gate hardening." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and fixes; build-ci-docker-images COMMIT_MESSAGE handling is unrelated to pr-ci-security-gate changes." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflow files and different behavior; not the same underlying bug/change." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46558", + "accept": false, + "reason": "Same file, but different concrete changes: 46554 adds checkout and diff-based checks, while 46558 changes permissions and approval gating." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46563", + "accept": false, + "reason": "Same workflow file, but distinct changes: security checks versus reusable-workflow review note/comments." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46716", + "accept": false, + "reason": "Same file, but different fixes: added security checks versus changing secrets inheritance to an empty map." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46563", + "accept": false, + "reason": "Same workflow, but one changes permissions/if-guard and the other adds documentation about the reusable workflow. Not the same change." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46716", + "accept": false, + "reason": "Same file, but distinct security-gate adjustments: permission/approval logic versus secrets scoping." + }, + { + "left": "pull_request:46563", + "right": "pull_request:46716", + "accept": false, + "reason": "Same workflow file, but different concrete edits: explanatory comments versus secrets inheritance change." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflows and unrelated fixes; build-ci-docker-images message handling is not the pr-ci-security-gate note/comment change." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46716", + "accept": false, + "reason": "Different workflows and code paths; no shared underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 30, + "component_count": 36, + "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": "b9e92902365bd7141da431b6fa238c57bb046543f8757e31e5561db4e86457e4", + "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 brittle source-file open() usage with safer source inspection, but they fix different code paths: import_utils resource resolution vs modeling_utils attention-implementation detection.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the changes are thematically similar but target different modules and different behaviors, so they are not interchangeable duplicates.", + "best_issue_reason": null, + "best_pr_reason": "PR 46207 is the cleaner representative of the theme because it addresses one concrete source-reading heuristic and includes tests, but it should remain separate from 46152.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46152", + "right": "pull_request:46207", + "accept": false, + "reason": "Shared 'avoid open()' theme only; one is resource-path traversal in import_utils, the other is source introspection in modeling_utils. Different bugs and code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 31, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "cache", + "cache_hit": true, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46174-2", + "nodes": [ + "pull_request:46174", + "pull_request:46780" + ], + "soft_pairs": [ + "pull_request:46174|pull_request:46780" + ], + "prepared_review_unit_hash": "a48ce4ca162f555682f79411d7229ac1ccc08d7822dfab654bb881639fccddbc", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6030, + "estimated_input_tokens": 1508, + "estimated_eval_tokens": 3272 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6030, + "estimated_input_tokens": 1508, + "estimated_eval_tokens": 3272 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two PRs target the same reported MPS crash in the sinusoidal position embedding helper, but they implement different fixes and are not safe to treat as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46174 is the broader, more complete fix for the float64-on-MPS crash: it updates the helper across all affected model files and keeps the internal math on CPU float64 before casting back.", + "best_issue_reason": null, + "best_pr_reason": "PR 46174 is the best representative of the cluster because it is the original, more comprehensive fix and has more discussion/review context than the narrower follow-up.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46174", + "right": "pull_request:46780", + "accept": false, + "reason": "They share the same issue target and helper, but the patches are different solutions: 46174 avoids float64 on MPS by doing the arithmetic on CPU first, while 46780 switches the helper to use dtype for internal arithmetic. Those are alternative implementations, not one mergeable duplicate PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 32, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46310-2", + "nodes": [ + "pull_request:46310", + "pull_request:46546" + ], + "soft_pairs": [ + "pull_request:46310|pull_request:46546" + ], + "prepared_review_unit_hash": "44cd8039da96fc3c1b6692d3daf1d927165a2ef4025832317d84adde1f4d5a40", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4850, + "estimated_input_tokens": 1213, + "estimated_eval_tokens": 2682 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4850, + "estimated_input_tokens": 1213, + "estimated_eval_tokens": 2682 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs fix the same Moonshine streaming sliding-window boundary bug (`-dist <= right_window_size` vs `<`) in the same code path; PR #46546 is the broader, more complete version because it includes the Moonshine fix plus tests and a parallel Gemma4 boundary fix.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46546 is the better canonical representative: it fixes the Moonshine streaming boundary bug, adds tests, and is the more complete patch even though it also includes an additional Gemma4 change.", + "best_issue_reason": null, + "best_pr_reason": "PR #46546 is the strongest overall candidate because it addresses the same concrete Moonshine streaming bug as #46310 and includes test coverage, making it the more complete fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46310", + "right": "pull_request:46546", + "accept": true, + "reason": "Both PRs change the same Moonshine streaming right-window boundary check in the same function family; #46546 is a superset that could plausibly absorb #46310." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 33, + "component_count": 36, + "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-46376-2", + "nodes": [ + "pull_request:46376", + "pull_request:46417" + ], + "soft_pairs": [ + "pull_request:46376|pull_request:46417" + ], + "prepared_review_unit_hash": "1ec740b1c5682b5aa57701693d50a8baed39ea95d271efbd895ce89e01e39c7b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5939, + "estimated_input_tokens": 1485, + "estimated_eval_tokens": 3226 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5939, + "estimated_input_tokens": 1485, + "estimated_eval_tokens": 3226 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two pull requests are unrelated: one adds multimodal continuous batching infrastructure, while the other adds the HunYuan VL model package, docs, and tests. They should not be clustered as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR 46417 is the more self-contained and complete change, with a coherent model addition plus docs/tests. PR 46376 is a draft infrastructure enhancement and not a duplicate of it.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46376", + "right": "pull_request:46417", + "accept": false, + "reason": "Different underlying changes and code paths: continuous batching/multimodal runtime support vs. a new HunYuan VL model integration. No shared filenames or concrete fix target, so they are not mergeable as one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 34, + "component_count": 36, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-46574-2", + "nodes": [ + "pull_request:46574", + "pull_request:46580" + ], + "soft_pairs": [ + "pull_request:46574|pull_request:46580" + ], + "prepared_review_unit_hash": "fe627fd5c8f3ace331bb4121eafdb455e1d7d9ab708453999acf6c8604bc6e73", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4410, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4410, + "estimated_input_tokens": 1103, + "estimated_eval_tokens": 2462 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two unrelated pull requests: one fixes CodeLlama decode stripping behavior, the other fixes DeepSeek-Coder/DeepSeek-LLM tokenizer-class selection in AutoTokenizer. They do not describe the same underlying bug or change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR #46574 is the cleaner representative if one must be picked: it is a focused single-bug tokenizer fix with a narrow code path and explicit issue target.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46574", + "right": "pull_request:46580", + "accept": false, + "reason": "Different models, different failure modes, and different code paths: CodeLlama decode stripping vs DeepSeek checkpoint-to-tokenizer-class routing. They are not plausibly mergeable into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 35, + "component_count": 36, + "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-46603-2", + "nodes": [ + "pull_request:46603", + "pull_request:46604" + ], + "soft_pairs": [ + "pull_request:46603|pull_request:46604" + ], + "prepared_review_unit_hash": "7bb1685d8c8f84a2cf7367398f51bbe9d3a1a0da9684fd7271d67d0f8ada431f", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5495, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5495, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs touch the Mistral tokenizer stack and share many files, but they address different changes: 46603 is a refactor moving MistralConverter into the integrations package, while 46604 adds tekken tokenizer detection/conversion/save utilities. They are related but not duplicates.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46604 is the more substantive end-state feature PR for Mistral tokenizer support; PR 46603 is a narrower packaging/refactor change that does not cover the tekken detection/conversion work.", + "best_issue_reason": null, + "best_pr_reason": "PR 46604 best represents the main functional change in the cluster because it introduces new tokenizer behavior and utilities, whereas 46603 is only a code move/refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46603", + "right": "pull_request:46604", + "accept": false, + "reason": "Shared Mistral files reflect the same subsystem, but the PRs fix/add different concrete behaviors and would not be merged as one duplicate 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": 36, + "component_count": 36, + "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-46741-4", + "nodes": [ + "pull_request:46741", + "pull_request:46773", + "pull_request:46781", + "pull_request:46831" + ], + "soft_pairs": [ + "pull_request:46741|pull_request:46781", + "pull_request:46741|pull_request:46773", + "pull_request:46741|pull_request:46831" + ], + "prepared_review_unit_hash": "55a74069bc9ac1db2e56ba33f3470a06e10154c9a663ea740c755ea0c5100afd", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 10052, + "estimated_input_tokens": 2513, + "estimated_eval_tokens": 5282 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 3, + "serialized_chars": 10052, + "estimated_input_tokens": 2513, + "estimated_eval_tokens": 5282 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These four PRs are all about model-runtime bugs, but they target different code paths and failure modes: cached decoding in SSM mixers, padding-mask handling, quantized cache construction, and CPU-offload hook behavior. They are not duplicates of one another.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46741 is the best representative of the set because it fixes a narrow, concrete decoding bug with a focused model-specific patch and test coverage; the others address unrelated runtime issues in different subsystems.", + "best_issue_reason": null, + "best_pr_reason": "PR 46741 is the strongest standalone fix: it targets a specific cached single-token decoding path and is comparatively self-contained, while the other PRs are broader or solve different problems entirely.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46741", + "right": "pull_request:46781", + "accept": false, + "reason": "Different bugs: 46741 fixes cached speculative/chunked decoding in Zamba2/Nemotron-H mixers, while 46781 fixes QuantizedCache construction for hybrid SSM+attention models. Separate code paths and not mergeable as one PR." + }, + { + "left": "pull_request:46741", + "right": "pull_request:46773", + "accept": false, + "reason": "Different issues: 46741 is about single-token cached decode logic in Mamba2-style mixers; 46773 changes padding-mask application when batch size is 1. Same broad family, but not the same underlying bug." + }, + { + "left": "pull_request:46741", + "right": "pull_request:46831", + "accept": false, + "reason": "Different runtime problem: 46741 addresses cached decode state handling, whereas 46831 fixes garbage generation under device_map CPU offload by triggering a conv1d hook. Not the same code-path defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/manifest.json b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..68c7d09e3fbfef949d216482bc653ca076c7aadf --- /dev/null +++ b/snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260623t120018z", + "artifacts": { + "hybrid": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-06-23T12:12:35Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260623T120018Z", + "variant": "hybrid" +} diff --git a/snapshots/20260623T120018Z/manifest.json b/snapshots/20260623T120018Z/manifest.json index 69d7f2c6c8bb7fb9b5d56db1a6d22e17fe8e732d..2cb0f0322a6c6da30a44c1c0add5f25de651078c 100644 --- a/snapshots/20260623T120018Z/manifest.json +++ b/snapshots/20260623T120018Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 1105 }, "extracted_at": "2026-06-23T12:00:18Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260623t120018z", + "runs": { + "hybrid-model-20260623t120018z": { + "analysis_id": "hybrid-model-20260623t120018z", + "artifacts": { + "hybrid": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260623T120018Z/analysis-runs/hybrid-model-20260623t120018z/manifest.json", + "model": null, + "published_at": "2026-06-23T12:12:35Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260623T120018Z", "watermark": {