diff --git a/analysis/current/analysis-report-hybrid.json b/analysis/current/analysis-report-hybrid.json index 37e30ad524781cf93546c5a285ae241d29005495..eec1171f783b643d2095924114fbb687e17a714f 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": "20260626T120015Z", - "generated_at": "2026-06-26T12:13:37Z", + "snapshot_id": "20260626T180014Z", + "generated_at": "2026-06-26T18:10:11Z", "evidence_quality": "full", "llm_enrichment": true, "meta_bugs": [ @@ -1050,21 +1050,118 @@ ] }, { - "cluster_id": "cluster-46894-7", - "summary": "Cluster of 3 issues and 4 PRs centered on issue #46894.", + "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-46894-5", + "summary": "Cluster of 2 issues and 3 PRs centered on issue #46897.", "status": "open", "confidence": 0.8, - "canonical_issue_number": 46894, - "canonical_pr_number": 46903, + "canonical_issue_number": 46897, + "canonical_pr_number": 46895, "issue_numbers": [ 46894, - 46897, - 46901 + 46897 ], "pr_numbers": [ 46895, 46898, - 46903, 46916 ], "evidence_types": [ @@ -1084,17 +1181,6 @@ "shared_filenames": [], "shared_file_areas": [] }, - { - "left_pr_number": 46895, - "right_pr_number": 46903, - "code_similarity": 0.257, - "size_similarity": 0.851, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.582, - "shared_filenames": [], - "shared_file_areas": [] - }, { "left_pr_number": 46895, "right_pr_number": 46916, @@ -1148,17 +1234,6 @@ } ] }, - { - "left_pr_number": 46898, - "right_pr_number": 46903, - "code_similarity": 0.127, - "size_similarity": 0.568, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.09, - "shared_filenames": [], - "shared_file_areas": [] - }, { "left_pr_number": 46898, "right_pr_number": 46916, @@ -1219,116 +1294,6 @@ ] } ] - }, - { - "left_pr_number": 46903, - "right_pr_number": 46916, - "code_similarity": 0.181, - "size_similarity": 0.757, - "file_overlap": 0.0, - "area_overlap": 0.0, - "patch_similarity": 0.194, - "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 - ] - ] - } - ] } ] }, @@ -1753,6 +1718,60 @@ } ] }, + { + "cluster_id": "cluster-46869-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46869.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 46869, + "canonical_pr_number": 46886, + "issue_numbers": [ + 46869 + ], + "pr_numbers": [ + 46886, + 46922 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 46886, + "right_pr_number": 46922, + "code_similarity": 0.732, + "size_similarity": 0.857, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.571, + "shared_filenames": [ + "src/transformers/models/yolos/modeling_yolos.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/yolos/modeling_yolos.py", + "left_ranges": [ + [ + 434, + 440 + ] + ], + "right_ranges": [ + [ + 434, + 440 + ], + [ + 446, + 452 + ] + ] + } + ] + } + ] + }, { "cluster_id": "cluster-45561-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #45561.", @@ -1970,6 +1989,75 @@ } ] }, + { + "cluster_id": "cluster-46909-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46909.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 46909, + "canonical_pr_number": 46923, + "issue_numbers": [ + 46909 + ], + "pr_numbers": [ + 46915, + 46923 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 46915, + "right_pr_number": 46923, + "code_similarity": 0.622, + "size_similarity": 0.477, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.341, + "shared_filenames": [ + "docs/source/en/model_doc/dia.md" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/dia.md", + "left_ranges": [ + [ + 42, + 62 + ], + [ + 76, + 92 + ], + [ + 106, + 122 + ] + ], + "right_ranges": [ + [ + 47, + 56 + ], + [ + 75, + 84 + ], + [ + 102, + 110 + ], + [ + 114, + 119 + ] + ] + } + ] + } + ] + }, { "cluster_id": "cluster-43698-3", "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", @@ -2306,13 +2394,12 @@ ], "duplicate_issues": [ { - "cluster_id": "cluster-46894-7", - "canonical_issue_number": 46894, + "cluster_id": "cluster-46894-5", + "canonical_issue_number": 46897, "duplicate_issue_numbers": [ - 46897, - 46901 + 46894 ], - "reason": "Issues in cluster-46894-7 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." + "reason": "Issues in cluster-46894-5 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." } ], "duplicate_prs": [ @@ -2415,22 +2502,39 @@ "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-46894-7", - "canonical_pr_number": 46903, + "cluster_id": "cluster-46869-3", + "canonical_pr_number": 46886, + "duplicate_pr_numbers": [ + 46922 + ], + "target_issue_number": 46869, + "reason": "PRs in cluster-46869-3 are treated as duplicates because they converge on issue #46869 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46894-5", + "canonical_pr_number": 46895, "duplicate_pr_numbers": [ - 46895, 46898, 46916 ], - "target_issue_number": 46894, - "reason": "PRs in cluster-46894-7 are treated as duplicates because they converge on issue #46894 with closing_reference, shared_issue_target, soft_similarity evidence." + "target_issue_number": 46897, + "reason": "PRs in cluster-46894-5 are treated as duplicates because they converge on issue #46897 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-46909-3", + "canonical_pr_number": 46923, + "duplicate_pr_numbers": [ + 46915 + ], + "target_issue_number": 46909, + "reason": "PRs in cluster-46909-3 are treated as duplicates because they converge on issue #46909 with closing_reference evidence." } ], "best_issue": { - "cluster_id": "cluster-46894-7", - "issue_number": 46894, - "reason": "Issue #46894 is the strongest global issue candidate because it is open, belongs to a cluster with 7 artifacts, and carries 0 discussion comments plus 7 inbound references.", - "score": 177.0 + "cluster_id": "cluster-46894-5", + "issue_number": 46897, + "reason": "Issue #46897 is the strongest global issue candidate because it is open, belongs to a cluster with 5 artifacts, and carries 0 discussion comments plus 9 inbound references.", + "score": 159.0 }, "best_pr": { "cluster_id": "cluster-43979-11", diff --git a/analysis/current/analysis-report-hybrid.llm-reviews.json b/analysis/current/analysis-report-hybrid.llm-reviews.json index 969800ff5b750afdebe143f9f5d711b47d426c5d..1c7c8fc5f2658f564744253dbac8734af8c539ec 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": "20260626T120015Z", - "generated_at": "2026-06-26T12:13:37Z", + "snapshot_id": "20260626T180014Z", + "generated_at": "2026-06-26T18:10:11Z", "model": "gpt-5.4-mini?service_tier=flex", "reviews": [ { @@ -132,382 +132,382 @@ "issue:45237|issue:46015", "issue:43039|issue:44805" ], - "prepared_review_unit_hash": "eb8345a4309c419768dace533694d297f1bf5aaf2a278b73cb2cdb6651e3854b", + "prepared_review_unit_hash": "e57caac5f04946b685b34c61db3aa01ef85311216039204de458db7e02c077ad", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 63, - "serialized_chars": 46566, - "estimated_input_tokens": 11642, - "estimated_eval_tokens": 23540 + "serialized_chars": 46576, + "estimated_input_tokens": 11644, + "estimated_eval_tokens": 23544 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous issues, with only a few true duplicate pairs: DeepSpeed ZeRO-3 index errors, v5 docs cleanup, and torch 2.4.1 incompatibility. Most other soft matches are broad subsystem similarity rather than the same concrete bug.", - "confidence": 0.8, - "canonical_issue_reason": "issue:43596 is the clearest duplicate cluster anchor: it has the same ZeRO-3 / index-out-of-bounds failure mode as issue:43638, with a concrete reproducible loading path and a specific model class.", + "summary": "This is a heterogeneous similarity cluster with one clear duplicate family around DeepSpeed ZeRO-3/BERT index-out-of-bounds errors, while most other pairs are false positives across unrelated model, docs, and hardware issues.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 43638 is the broadest and clearest representative of the one obvious duplicate family: the DeepSpeed ZeRO-3 index-out-of-bounds failure during BERT loading.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43596 is the best representative issue in this set because it maps to the strongest duplicate pair and describes a concrete runtime failure that is easy to triage and fix.", + "best_issue_reason": "Issue 43638 is the best issue representative here because it describes the shared concrete failure mode most directly and is the closest true duplicate hub in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44322", "right": "issue:45464", "accept": false, - "reason": "Different bugs: Qwen3.5 config attribute error vs streaming chat/completions failure." + "reason": "Different bug classes: Qwen3.5 config attribute missing vs streaming chat/completions failure." }, { "left": "issue:45237", "right": "issue:45685", "accept": false, - "reason": "Different hardware/runtime paths: AMD GPU failure vs MPS histogram int limitation." + "reason": "Unrelated symptoms and platforms: AMD GPU failure vs MPS histogram int error." }, { "left": "issue:43716", "right": "issue:45237", "accept": false, - "reason": "Different model/backend issues: Gemma/Mistral dtype mismatch vs GPT-OSS failing on AMD GPUs." + "reason": "Different model/runtime problems: image dtype mismatch vs AMD GPU incompatibility." }, { "left": "issue:43828", "right": "issue:45237", "accept": false, - "reason": "Different symptoms and code paths: autocast dtype mismatch vs AMD GPU incompatibility." + "reason": "Autocast dtype mismatch and AMD GPU failure are not the same underlying bug." }, { "left": "issue:45237", "right": "issue:46468", "accept": false, - "reason": "Different models and failures; no shared concrete bug." + "reason": "Different models and failures; no concrete shared code-path." }, { "left": "issue:45237", "right": "issue:45507", "accept": false, - "reason": "Different issues: AMD GPU failure vs GraniteMoEHybrid invalid method call." + "reason": "AMD GPU issue vs GraniteMoEHybrid invalid method; unrelated." }, { "left": "issue:30277", "right": "issue:35545", "accept": false, - "reason": "Both are failures, but unrelated code paths: DeepSpeed collective mismatch vs ONNX export error." + "reason": "DeepSpeed collective mismatch and ONNX export error are different problems." }, { "left": "issue:35545", "right": "issue:42915", "accept": false, - "reason": "Different model/export bug vs Qwen3 MoE FP8 failure." + "reason": "ONNX export vs FineGrainedFP8 Qwen3Moe failure; unrelated." }, { "left": "issue:35545", "right": "issue:41720", "accept": false, - "reason": "Different problems: ONNX export vs Qwen3 auto device map CUDA assert." + "reason": "Different subsystem and symptom: export bug vs device-map cuda assert." }, { "left": "issue:45901", "right": "issue:46015", "accept": false, - "reason": "Different bugs: table QA crash vs outdated RAG examples documentation." + "reason": "Table QA crash and outdated RAG examples are separate issues; only broadly docs-related." }, { "left": "issue:44509", "right": "issue:44512", - "accept": true, - "reason": "Both are v5 documentation cleanup issues about removed functionality; same underlying docs update can cover both." + "accept": false, + "reason": "Both are docs cleanup, but they reference different removed content and would not be one PR." }, { "left": "issue:35545", "right": "issue:36010", "accept": false, - "reason": "Different failures: ModernBERT ONNX export vs GenerationMixin import regression." + "reason": "ONNX export error and GenerationMixin import failure are unrelated." }, { "left": "issue:44918", "right": "issue:46165", "accept": false, - "reason": "Different code paths: TRL SFT embedding unpacking vs resuming training from local model with trust_remote_code." + "reason": "Different topics: Qwen3.5 embedding unpacking vs resume-training remote-code problem." }, { "left": "issue:46165", "right": "issue:46869", "accept": false, - "reason": "Different issues: training resume/local model handling vs YOLO multi-GPU device map failure." + "reason": "Resume-training remote-code issue and YOLO multi-GPU auto device-map issue are unrelated." }, { "left": "issue:44805", "right": "issue:45823", "accept": false, - "reason": "Different runtime errors and subsystems: mask-shape index error vs Gemma4 PLE device mismatch." + "reason": "Mask-shape index error and Gemma4 PLE device mismatch are different bugs." }, { "left": "issue:43638", "right": "issue:44805", "accept": false, - "reason": "Both mention index/shape errors, but the concrete failures and code paths differ (ZeRO-3 Bert loading vs mask indexing)." + "reason": "Both are index-style runtime errors, but the error modes and model contexts differ too much to be the same bug." }, { - "left": "issue:43638", + "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "Unrelated: ZeRO-3 Bert loading index error vs training resume/remote-code issue." + "reason": "DeepSpeed ZeRO-3 BERT loading issue vs remote-code resume training issue; unrelated." }, { "left": "issue:43596", "right": "issue:44805", "accept": false, - "reason": "Similar error wording, but different concrete bug: ZeRO-3 Bert loading vs mask/tensor shape mismatch." + "reason": "Different failure signatures and contexts; not the same underlying bug." }, { "left": "issue:44805", "right": "issue:46294", "accept": false, - "reason": "Different model and failure mode: mask-shape mismatch vs Qwen3.5 video token/feature mismatch." + "reason": "Mask/index error and Qwen3.5 video token mismatch are unrelated." }, { "left": "issue:44805", "right": "issue:45161", "accept": false, - "reason": "Different failures: tensor mask shape mismatch vs GPT-OSS MoE TP-only problem." + "reason": "Different models and failures; no concrete shared fix." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Remote-code resume issue and Nemotron example breakage are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Video StopIteration and training resume are unrelated." }, { "left": "issue:34634", "right": "issue:35545", "accept": false, - "reason": "Unrelated: BarkProcessor voice preset bug vs ModernBERT ONNX export." + "reason": "BarkProcessor voice preset bug is unrelated to ModernBERT ONNX export." }, { "left": "issue:30277", "right": "issue:43638", "accept": false, - "reason": "Both involve DeepSpeed, but the runtime collective mismatch and the Bert index error are different bugs." + "reason": "Both involve DeepSpeed, but the concrete failures differ: collective mismatch vs zero-3 index-out-of-bounds." }, { "left": "issue:43596", "right": "issue:43638", "accept": true, - "reason": "Same underlying ZeRO-3 index-out-of-bounds bug during Bert loading; same failure class and plausible single fix." + "reason": "Same underlying DeepSpeed ZeRO-3 empty-tensor index-out-of-bounds failure during BERT loading; one is just a narrower variant of the other." }, { "left": "issue:43039", "right": "issue:45161", "accept": false, - "reason": "Different issues: Liger Kernel cross-entropy path vs GPT-OSS MoE tensor-parallel failure." + "reason": "Liger Kernel cross-entropy call vs GPT-OSS MoE TP failure are unrelated." }, { "left": "issue:43975", "right": "issue:44457", "accept": false, - "reason": "Different symptoms: detokenization regression vs LoRA merge/save output mismatch." + "reason": "DeepSeek detokenization issue and LoRA merge/save mismatch are different bugs." }, { "left": "issue:44292", "right": "issue:44457", "accept": false, - "reason": "Unrelated: Qwen NVFP4 runtime error vs LoRA merge/save inconsistency." + "reason": "Qwen NVFP4 runtime error vs LoRA merge inconsistency are unrelated." }, { "left": "issue:44841", "right": "issue:46661", "accept": false, - "reason": "Different problems: processor failure for Voxtral vs remote-code config_class AttributeError." + "reason": "Voxtral processor failure and remote-code model_class config_class error are unrelated." }, { "left": "issue:43716", "right": "issue:46869", "accept": false, - "reason": "Different model/hardware problems; not the same concrete bug." + "reason": "Different model families and failure modes; no shared bug." }, { "left": "issue:24643", "right": "issue:30277", "accept": false, - "reason": "Both are DeepSpeed-related, but one is a weight shape error and the other is a collective mismatch; not the same bug." + "reason": "Both are DeepSpeed-related, but one is a weight dimensionality error and the other is a collective mismatch; not the same bug." }, { "left": "issue:34928", "right": "issue:44805", "accept": false, - "reason": "Different training-time failures: FSDP/checkpoint recomputation mismatch vs mask shape index error." + "reason": "Activation-checkpointing/FSDP recompute mismatch is different from the mask index error." }, { "left": "issue:45859", "right": "issue:45902", "accept": false, - "reason": "Same module family, but different concrete contexts (CPU offload vs meta->to_empty materialization); not enough to call them the same bug." + "reason": "Same component family, but CPU offload incompatibility and uninitialized memory after to_empty() are distinct bugs." }, { "left": "issue:45685", "right": "issue:46869", "accept": false, - "reason": "Different hardware/runtime issues: MPS histogram op limitation vs multi-GPU device map failure." + "reason": "MPS MoE histogram issue and YOLO multi-GPU device-map issue are unrelated." }, { "left": "issue:43638", "right": "issue:45161", "accept": false, - "reason": "Different bugs: ZeRO-3 Bert index error vs GPT-OSS tensor-parallel failure." + "reason": "Zero-3 BERT index error and GPT-OSS TP failure are different problems." }, { "left": "issue:34928", "right": "issue:36331", "accept": false, - "reason": "Different areas: FSDP activation checkpointing mismatch vs CustomTrainer API signature regression." + "reason": "FSDP checkpointing mismatch vs custom trainer signature change are unrelated." }, { "left": "issue:43596", "right": "issue:45161", "accept": false, - "reason": "Different issues: ZeRO-3 Bert loading vs GPT-OSS MoE tensor-parallel behavior." + "reason": "DeepSpeed ZeRO-3 BERT loading bug vs GPT-OSS MoE TP issue are unrelated." }, { "left": "issue:43716", "right": "issue:45507", "accept": false, - "reason": "Different bugs: dtype mismatch in Mistral-3 preprocessing/model vs GraniteMoEHybrid invalid method." + "reason": "Different model paths and symptoms; not the same bug." }, { "left": "issue:44292", "right": "issue:45237", "accept": false, - "reason": "Different failures and models; no shared underlying code-path." + "reason": "Different models, hardware, and failure contexts; no shared fix." }, { "left": "issue:45874", "right": "issue:46294", "accept": false, - "reason": "Different model features: reverse embedding lookup vs video token/feature mismatch." + "reason": "Gemma4 reverse-embedding issue and Qwen3.5 video token mismatch are unrelated." }, { "left": "issue:39290", "right": "issue:45081", "accept": false, - "reason": "Different model-specific regressions: Gemma3 config attribute vs Mistral tokenizer regex patch crash." + "reason": "Gemma3 config attribute error and Mistral regex crash are different bugs." }, { "left": "issue:45800", "right": "issue:46515", - "accept": true, - "reason": "Near-identical torch 2.4.1 incompatibility reports; same underlying compatibility issue." + "accept": false, + "reason": "Both mention torch 2.4.1 incompatibility, but one is tied to transformers 5.8.0 and the underlying failure is not specific enough to treat as the same bug." }, { "left": "issue:45823", "right": "issue:45874", "accept": false, - "reason": "Different Gemma4 problems: device mismatch vs expensive reverse embedding lookup." + "reason": "Both are Gemma4-related, but one is a device mismatch and the other is an expensive reverse embedding lookup; different code paths." }, { "left": "issue:43425", "right": "issue:44292", "accept": false, - "reason": "Different issues: Torch 2.10 incompatibility vs Qwen-3 NVFP4 runtime error." + "reason": "Generic Torch 2.10 incompatibility vs Qwen NVFP4 runtime error; unrelated." }, { "left": "issue:34928", "right": "issue:41720", "accept": false, - "reason": "Different bugs: FSDP/checkpoint tensor recomputation vs Qwen3 auto device-map CUDA assert." + "reason": "FSDP activation-checkpointing mismatch and Qwen3 device-map cuda assert are different issues." }, { "left": "issue:43828", "right": "issue:45685", "accept": false, - "reason": "Different runtime paths: autocast dtype mismatch vs MPS histogram int limitation." + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." }, { - "left": "issue:43716", - "right": "issue:46468", + "left": "issue:43541", + "right": "issue:43716", "accept": false, - "reason": "Different model failures; no same concrete bug." + "reason": "Torch dynamo grouped_mm error and image-preprocessor dtype mismatch are different bugs." }, { - "left": "issue:44292", - "right": "issue:44918", + "left": "issue:43828", + "right": "issue:44292", "accept": false, - "reason": "Different problems: Qwen NVFP4 runtime error vs TRL SFT embedding unpacking." + "reason": "Autocast dtype mismatch vs Qwen NVFP4 error is not the same bug." }, { - "left": "issue:45507", - "right": "issue:45685", + "left": "issue:43716", + "right": "issue:46468", "accept": false, - "reason": "Different bugs in different backends; not mergeable as one fix." + "reason": "Mistral-3 dtype mismatch and Nemotron example breakage are unrelated." }, { - "left": "issue:43716", - "right": "issue:44292", + "left": "issue:44292", + "right": "issue:44918", "accept": false, - "reason": "Different model/runtime failures; only broad similarity." + "reason": "Both are Qwen-related, but one is an NVFP4 runtime error and the other is a TRL embedding unpacking issue." }, { "left": "issue:44918", "right": "issue:45161", "accept": false, - "reason": "Different issues: TRL SFT embedding handling vs GPT-OSS tensor-parallel behavior." + "reason": "Different models and failure modes; not a duplicate." + }, + { + "left": "issue:43716", + "right": "issue:45685", + "accept": false, + "reason": "Dtype mismatch and MPS histogram error are unrelated." }, { "left": "issue:43039", "right": "issue:44918", "accept": false, - "reason": "Different bugs: Liger Kernel cross_entropy dispatch vs TRL SFT embedding unpacking." + "reason": "Liger Kernel cross-entropy call and TRL/Qwen embedding unpacking are unrelated." }, { "left": "issue:41762", "right": "issue:43638", "accept": false, - "reason": "Both are ZeRO-3 index errors, but the model/loading paths differ enough that this is not clearly the same bug." + "reason": "Same broad ZeRO-3 loading area, but different models and error context; not clearly the same underlying bug." }, { "left": "issue:43827", "right": "issue:44512", "accept": false, - "reason": "Both are v5 docs issues, but they cover different removed features and different documentation locations." + "reason": "Both are docs regressions from v5 removals, but they cover different removed references and should not be merged as one duplicate." }, { "left": "issue:43828", - "right": "issue:45161", - "accept": false, - "reason": "Different bugs: autocast dtype mismatch vs GPT-OSS MoE tensor-parallel failure." - }, - { - "left": "issue:30277", - "right": "issue:36331", - "accept": false, - "reason": "Unrelated DeepSpeed collective mismatch and trainer signature regression." - }, - { - "left": "issue:34928", - "right": "issue:36010", - "accept": false, - "reason": "Different failures: FSDP/checkpoint mismatch vs GenerationMixin import error." - }, - { - "left": "issue:45237", - "right": "issue:46015", + "right": "issue:45685", "accept": false, - "reason": "Different topics: GPT-OSS on AMD GPUs vs outdated RAG docs/examples." + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." }, { "left": "issue:43039", "right": "issue:44805", "accept": false, - "reason": "Different issues: Liger Kernel cross_entropy path vs mask-shape mismatch." + "reason": "Liger Kernel cross-entropy issue and mask shape IndexError are unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted pairs are well-supported by the titles/body context: 43596\u219443638 and 45800\u219446515 are clear duplicate-class matches, and 44509\u219444512 is a reasonable docs-cleanup grouping. The remaining soft matches are mostly rejected with appropriately cautious reasoning, and the summary does not overclaim broader duplication beyond the evidence." + "feedback": "Grounded overall. The summary stays conservative, and the one accepted soft-edge pair (43596 \u2194 43638) is well supported by the packet: both are near-identical DeepSpeed ZeRO-3/BERT index-out-of-bounds failures. The remaining rejections are mostly appropriately cautious given the differing models, subsystems, or hardware contexts." }, "evaluator_used": true, "retried": false, @@ -621,14 +621,14 @@ "issue:43572|issue:45070", "issue:45800|issue:45823" ], - "prepared_review_unit_hash": "169078e0c2da027a8823e15e5b93c034466b15ab36c6623b63f44c78b30a18d2", + "prepared_review_unit_hash": "e106e8db2be6abc313510784bc8854f6951a540f6e9b415493e756b25bfbce50", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, @@ -643,270 +643,270 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous. I only see two strong duplicate pairs: a broad AutoTokenizer v5 regression and a repeated ZeRO-3 loading IndexError on model init. The rest look like separate model- or subsystem-specific failures.", - "confidence": 0.8, - "canonical_issue_reason": "issue:45920 is the best canonical issue for the tokenizer regression because it states the broad v5 AutoTokenizer wrong-token-ID bug across many models, making it more representative than the narrower Granite-only report in issue:45812.", + "summary": "Mostly a loose cluster of unrelated bug reports. The only clear duplicate-like soft edge is the AutoTokenizer v5 regression pair; the rest differ by model, API path, or failure mode and should stay separate.", + "confidence": 0.81, + "canonical_issue_reason": "issue:45920 is the best canonical issue because it describes the broader AutoTokenizer v5 token-ID regression across multiple model families, subsuming the narrower Granite-only report in issue:45812.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the strongest representative issue in this set: it captures the broadest, clearest root problem and has the most general applicability among the related reports.", + "best_issue_reason": "issue:45920 is the strongest global representative: broad scope, clear regression framing, and likely shared root cause with related tokenizer reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43866", "right": "issue:46402", "accept": false, - "reason": "Different models and failure modes: corrupted Ovis2 checkpoint vs missing Sam3 text encoder weights." + "reason": "Different problems: corrupted checkpoint vs missing-loaded weights in a different model family." }, { "left": "issue:45800", "right": "issue:46304", "accept": false, - "reason": "Both mention version compatibility, but one is a general torch/transformers incompatibility claim while the other is a specific USE_DISTRIBUTED=0 regression." + "reason": "Both mention version compatibility, but they describe different concrete failures and release contexts." }, { "left": "issue:43296", "right": "issue:46304", "accept": false, - "reason": "Unrelated: PaddleOCR-VL loading in vLLM versus a PyTorch distributed-build regression." + "reason": "Different stack and symptom: vLLM/PaddleOCR-VL load failure vs PyTorch distributed support regression." }, { "left": "issue:41762", "right": "issue:43596", - "accept": true, - "reason": "Both report the same ZeRO-3 model-loading IndexError path, just on different model classes." + "accept": false, + "reason": "Both are index errors under DeepSpeed ZeRO-3, but they affect different models and likely different code paths." }, { "left": "issue:39290", "right": "issue:41720", "accept": false, - "reason": "Different subsystems and symptoms: missing Gemma3 config attribute in vLLM versus CUDA assert during Qwen3 auto device mapping." + "reason": "Gemma3 config attribute error and Qwen3 auto device-map CUDA assert are unrelated failures." }, { "left": "issue:43901", "right": "issue:44512", "accept": false, - "reason": "Both are docs issues, but they concern different outdated examples and no shared underlying bug." + "reason": "Both are documentation-related, but they cover different docs and different stale references." }, { "left": "issue:44589", "right": "issue:46304", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to the PyTorch distributed support breakage." + "reason": "Unrelated runtime storage lookup error vs PyTorch distributed build/runtime incompatibility." }, { "left": "issue:45237", "right": "issue:45901", "accept": false, - "reason": "Completely different areas: AMD GPU model support vs table-question-answering task crash." + "reason": "Different model/runtime issues with no shared concrete bug." }, { "left": "issue:43975", "right": "issue:46015", "accept": false, - "reason": "Different problems: detokenization regression versus outdated RAG documentation." + "reason": "Wrong detokenization and outdated RAG examples are unrelated." }, { "left": "issue:43039", "right": "issue:44560", "accept": false, - "reason": "Liger kernel cross-entropy dispatch and Qwen3-VL video StopIteration are unrelated." + "reason": "Liger Kernel cross-entropy routing and Qwen3-vl video stop-iteration are different paths." }, { "left": "issue:45823", "right": "issue:46294", "accept": false, - "reason": "Different models and code paths: device-map forward mismatch versus chat-template video token mismatch." + "reason": "Device-map mismatch in Gemma4 PLE is unrelated to video token/feature mismatch in Qwen3.5." }, { "left": "issue:44841", "right": "issue:45823", "accept": false, - "reason": "Processor failure for Voxtral is unrelated to Gemma4 device mismatch." + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different bugs." }, { "left": "issue:43866", "right": "issue:46772", "accept": false, - "reason": "Checkpoint corruption and NVFP4 compatibility are distinct issues." + "reason": "Checkpoint corruption and ModelOpt/NVFP4 compatibility are not the same issue." }, { "left": "issue:44863", "right": "issue:46015", "accept": false, - "reason": "NemotronH checkpoint loading is unrelated to RAG documentation." + "reason": "NemotronH checkpoint loading and outdated RAG docs are unrelated." }, { "left": "issue:45020", "right": "issue:46304", "accept": false, - "reason": "Remote-code model breakage is not the same as the PyTorch USE_DISTRIBUTED=0 regression." + "reason": "Remote-code breakage and PyTorch USE_DISTRIBUTED=0 support regression are different compatibility problems." }, { "left": "issue:45812", "right": "issue:45920", "accept": true, - "reason": "Same underlying AutoTokenizer v5 regression causing wrong token IDs; 45920 is the broader report." + "reason": "Same underlying AutoTokenizer v5 regression producing wrong token IDs; 45920 is the broader multi-model manifestation and 45812 is a narrower Granite-specific report." }, { "left": "issue:43866", "right": "issue:44863", "accept": false, - "reason": "Both are load failures, but on different model families and likely different root causes." + "reason": "Corrupted checkpoint vs failed checkpoint loading for a different architecture." }, { "left": "issue:43872", "right": "issue:46304", "accept": false, - "reason": "Bitsandbytes Int8Params constructor mismatch is unrelated to distributed support." + "reason": "bitsandbytes parameter incompatibility is unrelated to distributed build support." }, { "left": "issue:44292", "right": "issue:46015", "accept": false, - "reason": "NVFP4 runtime failure is unrelated to RAG docs." + "reason": "Qwen-3 NVFP4 runtime failure and RAG docs issues are unrelated." }, { "left": "issue:39290", "right": "issue:43531", "accept": false, - "reason": "Both mention sliding_window, but the concrete failures and model families differ too much to treat as one bug." + "reason": "Different Qwen/Gemma sliding-window and device-map 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 empty-index failure." + "reason": "Both are IndexErrors, but the affected model/load setup differs enough that they are not the same bug." }, { "left": "issue:43792", "right": "issue:45901", "accept": false, - "reason": "Whisper runtime failure and table-question-answering crash are unrelated." + "reason": "Whisper model load failure and table-question-answering crash are unrelated." }, { "left": "issue:43854", "right": "issue:43866", "accept": false, - "reason": "Different model-loading failures with no clear shared code path." + "reason": "Unit test load failure for GLM-4.7 and Ovis2 checkpoint corruption are different issues." }, { "left": "issue:43425", "right": "issue:45070", "accept": false, - "reason": "Torch-version incompatibility and a pydantic PretrainedConfig regression are unrelated." + "reason": "Torch version incompatibility and pydantic PretrainedConfig regression are different breakages." }, { "left": "issue:43975", "right": "issue:45901", "accept": false, - "reason": "Detokenization regression is unrelated to table QA crashing." + "reason": "Detokenization bug and table QA crash do not share a concrete code path." }, { "left": "issue:43792", "right": "issue:45070", "accept": false, - "reason": "Whisper loading/can't-run issue has nothing to do with PretrainedConfig pydantic fields." + "reason": "Whisper load failure is unrelated to the pydantic model-field regression." }, { "left": "issue:43596", "right": "issue:45800", "accept": false, - "reason": "The ZeRO-3 model-loading bug is not the same as a general torch/transformers compatibility complaint." + "reason": "ZeRO-3 index error with BertModel is not the same as the Torch 2.4.1 compatibility report." }, { "left": "issue:45362", "right": "issue:45464", "accept": false, - "reason": "Both involve Qwen3.5, but one is a transformers chat crash and the other is a streaming API failure; too different to merge." + "reason": "Both mention Qwen3.5 and chat, but one is a model crash and the other is a streaming chat/completions API failure." }, { "left": "issue:43541", - "right": "issue:45685", + "right": "issue:46869", "accept": false, - "reason": "Different MoE/runtime failures: torch dynamo grouped_mm tracing versus an MPS histogram kernel gap." + "reason": "Mixtral torch-dynamo grouped_mm tracing failure and YOLO multi-GPU device-map failure are different code paths." }, { "left": "issue:44292", "right": "issue:45901", "accept": false, - "reason": "NVFP4 model execution and table QA crashing are unrelated." + "reason": "NVFP4 model runtime failure and table QA task crash are unrelated." }, { "left": "issue:39290", "right": "issue:45823", "accept": false, - "reason": "Gemma3 config attribute failure and Gemma4 device mismatch do not share the same concrete bug." + "reason": "Gemma3 config regression and Gemma4 auto device-map mismatch are different bugs." }, { "left": "issue:45464", "right": "issue:45685", "accept": false, - "reason": "Qwen3.5 streaming API failure is unrelated to the MPS histogram kernel issue." + "reason": "Chat/completions streaming failure and MPS histogram implementation error are unrelated." }, { "left": "issue:36010", "right": "issue:39290", "accept": false, - "reason": "ImportError for GenerationMixin and Gemma3 sliding_window_pattern are unrelated." + "reason": "Importing GenerationMixin and Gemma3TextConfig attribute errors are unrelated." }, { "left": "issue:43541", "right": "issue:45685", "accept": false, - "reason": "Different code paths and backends; no shared underlying defect." + "reason": "Torch dynamo tracing failure and MPS histogram type error are different issues." }, { "left": "issue:43541", "right": "issue:43828", "accept": false, - "reason": "Both involve MoE models, but grouped_mm tracing and autocast dtype mismatch are distinct bugs." + "reason": "Grouped_mm tracing error and autocast dtype mismatch are distinct runtime problems." }, { "left": "issue:44841", "right": "issue:45084", "accept": false, - "reason": "Processor loading and template-node compilation are unrelated failures." + "reason": "Processor failure for Voxtral and non-template-node compilation error are unrelated." }, { "left": "issue:43404", "right": "issue:43866", "accept": false, - "reason": "lm_head tying bug and checkpoint corruption are separate issues." + "reason": "Weight-tying bug in Mistral3 and Ovis2 checkpoint corruption are different model-specific issues." }, { "left": "issue:45507", "right": "issue:46869", "accept": false, - "reason": "GraniteMoEHybrid invalid method calls and YOLO multi-GPU device-map failure are unrelated." + "reason": "Invalid method call in GraniteMoEHybrid and YOLO multi-GPU device-map failure are unrelated." }, { "left": "issue:41093", "right": "issue:41762", "accept": false, - "reason": "Both are IndexErrors, but they arise from different shapes and different loading scenarios." + "reason": "Both are index-related errors, but the models and failure contexts differ." }, { "left": "issue:45399", "right": "issue:46661", "accept": false, - "reason": "Flash-attn fallback logic and remote-code config_class lookup are unrelated." + "reason": "Flash-attn fallback gating and remote-code config_class handling are different bugs." }, { "left": "issue:43572", "right": "issue:45070", "accept": false, - "reason": "Missing pad_token_idx and broken PretrainedConfig pydantic field are different configuration regressions." + "reason": "Missing pad_token_idx in StableLmConfig and pydantic PretrainedConfig regression are unrelated." }, { "left": "issue:45800", "right": "issue:45823", "accept": false, - "reason": "Torch compatibility and Gemma4 device placement are unrelated." + "reason": "Torch/transformers compatibility and Gemma4 device placement are different problems." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded overall. The two accepted pairs are supported by the packet titles: 45812/45920 are the same AutoTokenizer v5 regression, and 41762/43596 are both ZeRO-3 index errors during model loading. The remaining soft-edge rejections stay conservative and the cluster summary doesn\u2019t overstate shared root causes." + "feedback": "The clustering is grounded in the packet and stays conservative. The only clearly supported duplicate-like edge is issue:45812 \u2194 issue:45920, which both describe the AutoTokenizer v5 wrong-token-ID regression, with 45920 being the broader multi-model report. The remaining rejected pairs are appropriately treated as separate bugs based on differing model families, APIs, or failure modes." }, "evaluator_used": true, "retried": false, @@ -1009,20 +1009,20 @@ "issue:45183|issue:46463", "issue:46489|issue:46491" ], - "prepared_review_unit_hash": "08e6edde253c752b2bb70ba4086da9050ead3311602728bba59766af36f1eb25", + "prepared_review_unit_hash": "db2c189a91c1be95a2761d79df723dc3b41b003e196acf962b3712354668e83c", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36888, + "serialized_chars": 36886, "estimated_input_tokens": 9222, "estimated_eval_tokens": 18700 }, @@ -1031,198 +1031,198 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is mostly heterogeneous: it mixes unrelated regressions across model loading, multimodal/video handling, tokenizers, quantization, and distributed training. None of the soft pairs look like safe duplicate merges, so I rejected them all.", - "confidence": 0.61, - "canonical_issue_reason": "issue:45020 is the broadest, most representative report in the set: it captures recent transformers-version breakage affecting remote-code models and serves as the best umbrella issue for this mixed cluster.", + "summary": "These items are only superficially similar; they span unrelated Transformers bugs across loading, quantization, MoE, tokenization, video, and docs. None of the soft pairs look like the same concrete underlying bug/change, so all should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45005 is the broadest hub in this set and has the most similarity links, but it still does not truly subsume the others.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45020 is the best global issue candidate because it is clearly scoped, widely applicable, and more central than the other mostly model-specific regressions in this cluster.", + "best_issue_reason": "Issue 45005 is the most plausible representative by breadth and centrality among the provided issues, even though the cluster is not actually cohesive.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43296", "right": "issue:43366", "accept": false, - "reason": "Both involve model support/load failures, but one is PaddleOCR-VL loading in vLLM and the other is GGUF gpt-oss architecture support; different code paths and fixes." + "reason": "Different model families and failure modes: PaddleOCR-VL load failure vs GGUF/gpt-oss architecture support." }, { "left": "issue:44387", "right": "issue:45005", "accept": false, - "reason": "One is int4 quantization memory/OOM behavior, the other is tied-weights handling on translation models; same broad version theme, not the same bug." + "reason": "Int4 quantization OOM/reserved-memory growth is not the same bug as tied-weights handling on translation models." }, { "left": "issue:44805", "right": "issue:44918", "accept": false, - "reason": "Both are shape-related runtime failures, but one is a mask/index mismatch and the other is unpacking Qwen3.5 embeddings with TRL SFT; not the same underlying defect." + "reason": "Both mention shape-related errors, but one is a generic mask indexing mismatch and the other is Qwen3.5 embedding unpacking with TRL SFT." }, { "left": "issue:43541", "right": "issue:45161", "accept": false, - "reason": "Grouped_mm tracing failure in Mixtral is a different issue from GPT-OSS tensor-parallel support; different model and failure mechanism." + "reason": "Grouped_mm torch-dynamo tracing failure for Mixtral is a different code path from GPT-OSS tensor-parallel support." }, { "left": "issue:43828", "right": "issue:45464", "accept": false, - "reason": "Phi-tiny-MoE autocast dtype mismatch is unrelated to Qwen3.5 streaming chat/completions failures." + "reason": "Autocast dtype mismatch on Phi-tiny-MoE is unrelated to Qwen3.5 streaming chat/completions failure." }, { "left": "issue:43366", "right": "issue:45464", "accept": false, - "reason": "GGUF gpt-oss architecture support and Qwen3.5 streaming inference are distinct model/runtime problems." + "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference are unrelated bugs." }, { "left": "issue:44291", "right": "issue:45070", "accept": false, - "reason": "`init_empty_weights` argument handling is a separate initialization bug from pydantic `PretrainedConfig` field compatibility." + "reason": "Unexpected _is_hf_initialized argument in init_empty_weights is unrelated to the pydantic PretrainedConfig field regression." }, { "left": "issue:43531", "right": "issue:45070", "accept": false, - "reason": "Qwen3-MoE sliding-window behavior is unrelated to the `PretrainedConfig` pydantic regression." + "reason": "Qwen3-MoE sliding_window behavior and pydantic PretrainedConfig serialization are different issues." }, { "left": "issue:41093", "right": "issue:43596", "accept": false, - "reason": "Both are IndexErrors, but one is a mask/tensor length mismatch and the other is zero-sized parameters with DeepSpeed ZeRO-3 and BertModel." + "reason": "Both are IndexErrors, but one is a mask-length mismatch and the other is zero-sized indexing during deepspeed zero3/BertModel init." }, { "left": "issue:45020", "right": "issue:45800", "accept": false, - "reason": "Both are version-compatibility complaints, but one concerns remote_code model breakage while the other is a torch/transformers incompatibility; too broad to merge." + "reason": "Remote-code model breakage and torch/transformers version incompatibility are not the same concrete bug." }, { "left": "issue:43716", "right": "issue:46294", "accept": false, - "reason": "Mistral-3 image preprocessor dtype mismatch is not the same as Qwen3.5 video token/feature mismatch in apply_chat_template." + "reason": "Image preprocessor/model dtype mismatch is distinct from Qwen3.5 video token/feature count mismatch." }, { "left": "issue:43054", "right": "issue:43994", "accept": false, - "reason": "Both mention SigLIP2, but one is a quality/regression comparison and the other is a bad AutoModel/pipeline output bug; not the same concrete issue." + "reason": "Siglip2 poor text embeddings vs nonsensical AutoModel/pipeline output are related only at a very high level." }, { "left": "issue:44479", "right": "issue:46294", "accept": false, - "reason": "Both are video-input regressions, but the affected models and described failure modes differ; not enough evidence for one underlying bug." + "reason": "A broad v5.3 video regression across multiple Qwen models is not the same as the specific apply_chat_template video mismatch in Qwen3.5-9B." }, { "left": "issue:46468", "right": "issue:46869", "accept": false, - "reason": "An official Nemotron example breakage and YOLO multi-GPU device-map failure are unrelated." + "reason": "Nemotron example breakage and YOLO multi-GPU device-map failure are unrelated subsystems." }, { "left": "issue:45759", "right": "issue:45902", "accept": false, - "reason": "Unwrapping `text_config` for composite Qwen multimodal configs is a different bug from uninitialized `inv_freq` after `to_empty(cuda)` materialization." + "reason": "Composite Qwen config unwrapping and uninitialized inv_freq after meta->to_empty are different bugs." }, { "left": "issue:43596", "right": "issue:46165", "accept": false, - "reason": "DeepSpeed ZeRO-3 initialization and resuming local trust_remote_code training are separate loading/training flows with different failure points." + "reason": "Zero3/BertModel initialization failure is unrelated to resuming training with local models and trust_remote_code." }, { "left": "issue:46640", "right": "issue:46661", "accept": false, - "reason": "Flash Attention 2 integration issues are not the same as the remote-code `config_class` AttributeError." + "reason": "Flash Attention 2 problems and remote-code model_class/config_class AttributeError are different code paths." }, { "left": "issue:43716", "right": "issue:44918", "accept": false, - "reason": "Image-preprocessor dtype mismatch and TRL embedding unpacking failure are different multimodal/training bugs." + "reason": "Dtype mismatch in image preprocessing is not the same as Qwen3.5 input-embedding unpacking with TRL SFT." }, { "left": "issue:39186", "right": "issue:44918", "accept": false, - "reason": "FSDP `'weight' must be 2-D` is unrelated to Qwen3.5 embedding unpacking in TRL." + "reason": "FSDP 2-D weight error and Qwen3.5 embedding unpacking are unrelated." }, { "left": "issue:40990", "right": "issue:44960", "accept": false, - "reason": "High perplexity on gpt-oss-20b and the GLM5 issue are different model-specific reports with no shared failure mode." + "reason": "High perplexity on gpt-oss-20b and a GLM5 issue are not the same bug." }, { "left": "issue:43645", "right": "issue:45399", "accept": false, - "reason": "Custom-model Jupyter initialization breakage is unrelated to the FA2 fallback gating issue." + "reason": "Custom-model initialization in Jupyter and flash-attn2 fallback blocking are unrelated." }, { "left": "issue:45632", "right": "issue:45698", "accept": false, - "reason": "Both touch trust_remote_code/custom-module loading, but one is a cache-path collision and the other is loading the wrong module after save_pretrained; related area, not clearly the same bug." + "reason": "Both concern trust_remote_code/custom modules, but cache-path collision and loading the wrong module after save_pretrained are distinct bugs." }, { "left": "issue:43404", "right": "issue:45005", "accept": false, - "reason": "Both mention tied weights, but they affect different model families and likely different tie-handling code paths." + "reason": "Tied lm_head in Mistral3 image-text generation is not the same concrete issue as translation-model tied weights in v5." }, { "left": "issue:42915", "right": "issue:45399", "accept": false, - "reason": "Qwen3-MoE FP8Config failure is unrelated to FA2 fallback blocking checks." + "reason": "Qwen3Moe FineGrainedFP8Config failure and flash-attn2 fallback gating are unrelated." }, { "left": "issue:43873", "right": "issue:45005", "accept": false, - "reason": "Quantization offloading behavior is not the same as tied-weights handling on translation models." + "reason": "Quantization/offloading behavior is a different problem from tied-weights handling." }, { "left": "issue:44530", "right": "issue:45005", "accept": false, - "reason": "PagedAttentionCache linear_attention crash is a separate Qwen3.5 cache/kernel issue, not tied-weights handling." + "reason": "PagedAttentionCache linear_attention crash on Qwen3.5 is unrelated to translation-model tied weights." }, { "left": "issue:44360", "right": "issue:44512", "accept": false, - "reason": "A DSA indexer math issue and a docs typo are clearly unrelated." + "reason": "DSA indexer/ReLU discussion and docs mentioning a removed command are unrelated." }, { "left": "issue:44492", "right": "issue:44512", "accept": false, - "reason": "Cache-strategy typo documentation and removed run-command documentation are different docs issues, not duplicates." + "reason": "A typo in cache strategies is not the same bug as outdated run-command documentation." }, { "left": "issue:44863", "right": "issue:45005", "accept": false, - "reason": "NemotronH checkpoint loading failure is unrelated to tied-weights regression in translation models." + "reason": "NemotronH checkpoint loading failure is unrelated to tied-weights regressions." }, { "left": "issue:45183", "right": "issue:46463", "accept": false, - "reason": "Both concern input validation error messages, but one is generic XOR validation across models and the other is T5 decoder-input validation; different code paths." + "reason": "These are both validation-message bugs, but one is XOR input_ids/inputs_embeds handling and the other is T5 decoder-input validation; not the same concrete change." }, { "left": "issue:46489", "right": "issue:46491", "accept": false, - "reason": "DeepSeek-Coder tokenizer output regression and CodeLlama leading-space round-trip regression are distinct tokenizer bugs." + "reason": "Both are tokenizer regressions, but for different models and different symptoms." } ] }, @@ -1318,8 +1318,8 @@ "issue:44403|issue:44451", "issue:43975|issue:45520", "issue:44403|issue:44488", - "issue:42947|issue:43856", "issue:24643|issue:28530", + "issue:42947|issue:43856", "issue:43643|issue:45020", "issue:41762|issue:42915", "issue:44991|issue:45399", @@ -1328,20 +1328,20 @@ "issue:43881|issue:43883", "issue:45685|issue:45823" ], - "prepared_review_unit_hash": "de3e24e06f8a3f169cd52577e28b2cc3354ac1ba8329e5cd824c5e0f81e8d0d8", + "prepared_review_unit_hash": "4d493b836e887b56bf853a2d28bd12de0a921099cb962288c3f130507c7b136d", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37031, + "serialized_chars": 37032, "estimated_input_tokens": 9258, "estimated_eval_tokens": 18772 }, @@ -1350,204 +1350,204 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous, but it contains a few clear duplicate subgroups: the non-persistent-buffer regression around `from_pretrained`, the `cache.get_seq_length()` return-type bug, and a tokenizer regression affecting DeepSeek-Coder v1 / multiple v5 tokenizers. No PRs are present.", - "confidence": 0.89, - "canonical_issue_reason": "#43950 is the strongest representative of the clearest duplicate subgroup: the Transformers v5 non-persistent-buffer regression. It is phrased broadly and matches the related buffer-corruption reports best.", + "summary": "This cluster is mostly heterogeneous, but there is one clear duplicate family around Transformers 5.x non-persistent buffer corruption, plus one around `cache.get_seq_length()` returning a tensor instead of an int. Most other pairs are only superficially similar by model/subsystem.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 43950 is the broadest/root-cause report for the non-persistent-buffer regression and best covers the related duplicate reports.", "canonical_pr_reason": null, - "best_issue_reason": "#43950 is the best issue to anchor the cluster because it describes a broad, concrete regression and has the most direct duplicate matches among the provided issues.", + "best_issue_reason": "Issue 43950 is the most representative issue: it states the underlying regression clearly, is not overly model-specific, and matches the strongest duplicate candidates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43975", "right": "issue:44292", "accept": false, - "reason": "Different failures: DeepSeek-Coder detokenization vs Qwen-3 NVFP4 runtime error." + "reason": "Both are tokenizer/output complaints, but they involve different models and different failure modes; not the same concrete bug." }, { "left": "issue:44918", "right": "issue:45507", "accept": false, - "reason": "Unrelated problems: Qwen3.5 embedding unpacking in TRL vs GraniteMoEHybrid invalid method call." + "reason": "Different models and different code paths: TRL SFT unpacking vs an invalid model method call." }, { "left": "issue:43950", "right": "issue:44534", "accept": true, - "reason": "Same regression: both report Transformers v5 corrupting/filling non-persistent buffers with junk." + "reason": "Same underlying regression: Transformers 5.x mishandling non-persistent buffers leading to junk/uninitialized values." }, { "left": "issue:39692", "right": "issue:43054", "accept": false, - "reason": "Different issues: docs/model-processor mismatch and quantization failure vs poor SigLIP2 text embedding quality." + "reason": "Doc-example errors and a model quality/performance complaint are not the same bug." }, { "left": "issue:43644", "right": "issue:43950", "accept": true, - "reason": "Same underlying bug: both describe Transformers v5 corrupting non-persistent buffers during loading." + "reason": "Same non-persistent-buffer corruption regression in Transformers 5.x, described from the reverse angle." }, { "left": "issue:36331", "right": "issue:41762", "accept": false, - "reason": "Different code paths: custom loss signature change vs DeepSpeed ZeRO-3 loading index error." + "reason": "Unrelated training API mismatch vs DeepSpeed loading failure." }, { "left": "issue:43872", "right": "issue:44589", "accept": false, - "reason": "Both are quantization-related, but the concrete errors and libraries differ (bitsandbytes Int8Params vs missing Float8 storage)." + "reason": "Both are quantization/storage related, but the concrete errors and code paths differ." }, { "left": "issue:43828", "right": "issue:44928", "accept": false, - "reason": "Different symptoms and models: autocast dtype mismatch vs Qwen3.5 RLHF gradient explosion from 3D position IDs." + "reason": "Different model-specific dtype/autocast vs RLHF/SDPA behavior; not the same bug." }, { "left": "issue:45987", "right": "issue:46858", "accept": true, - "reason": "Same cache bug: `cache.get_seq_length()` returns a tensor instead of an int, breaking downstream arithmetic." + "reason": "Same API contract break: `cache.get_seq_length()` returns a tensor instead of an int, causing downstream failures." }, { "left": "issue:45920", "right": "issue:46489", - "accept": true, - "reason": "Same tokenizer regression family: both report incorrect tokenization/output in Transformers v5+ after the tokenizer changes." + "accept": false, + "reason": "Both are tokenizer regressions, but the reports target different model families and specific breakages." }, { "left": "issue:43054", "right": "issue:45701", "accept": false, - "reason": "Different scope: SigLIP2 embedding quality vs a general tokenization change regression." + "reason": "A model embedding-quality complaint is not the same as a tokenizer-version change issue." }, { "left": "issue:38175", "right": "issue:43054", "accept": false, - "reason": "Related model family, but one is zero probabilities and the other is broader embedding degradation; not clearly the same bug." + "reason": "Both mention SigLIP2, but the observed symptoms and likely causes are different." }, { "left": "issue:43950", "right": "issue:45902", - "accept": false, - "reason": "Both mention bad buffer values, but one is a general `from_pretrained` regression while the other is a specific Qwen3_5MoE `to_empty`/materialization issue." + "accept": true, + "reason": "Same family of non-persistent-buffer corruption/uninitialized-memory regression after materialization/loading." }, { "left": "issue:41553", "right": "issue:43883", "accept": false, - "reason": "Different models and errors: Voxtral tokenizer error message vs Molmo missing `all_tied_weights_keys` attribute." + "reason": "Bad error message for tokenizer loading is unrelated to missing model attribute behavior." }, { "left": "issue:43054", "right": "issue:43493", "accept": false, - "reason": "Both are SigLIP2-related, but one is a quality complaint and the other is an implementation discrepancy with JAX; not the same bug." + "reason": "Both are SigLIP2-related but describe different discrepancies and likely different causes." }, { "left": "issue:44910", "right": "issue:45859", "accept": false, - "reason": "Different specific failures: Flash Attention illegal memory access vs CPU-offload incompatibility in Qwen3_5MoE rotary embedding." + "reason": "Both involve Qwen3.5, but one is flash-attn packed-sequence handling and the other is CPU-offload rotary embedding compatibility." }, { "left": "issue:42947", "right": "issue:44387", "accept": false, - "reason": "Different performance issues: ineffective gradient checkpointing with LoRA vs increased CUDA reserved memory under int4 quantization." + "reason": "LoRA gradient checkpointing and int4 quantization OOM are different optimization issues." }, { "left": "issue:45663", "right": "issue:45902", "accept": false, - "reason": "Different model-specific bugs: Gemma-4 FSDP2 key error vs Qwen3_5MoE uninitialized rotary embedding memory." + "reason": "FSDP2 per-layer state rebuild vs non-persistent-buffer materialization bug are different code paths." }, { "left": "issue:41553", "right": "issue:43881", "accept": false, - "reason": "Different loading failures for different models; no shared underlying bug is evident." + "reason": "Different models and different loading failures." }, { "left": "issue:44403", "right": "issue:44451", "accept": false, - "reason": "One is about noisy logging, the other is a model loading failure; not the same issue." + "reason": "General loading noise vs a specific model loading failure are not duplicates." }, { "left": "issue:43975", "right": "issue:45520", "accept": false, - "reason": "Unrelated: DeepSeek detokenization vs Python 3.13 flash_attn import lookup." + "reason": "Tokenizer detokenization issue and import error on Python 3.13 are unrelated." }, { "left": "issue:44403", "right": "issue:44488", "accept": false, - "reason": "Noisy loading output vs a specific model not loading; not the same underlying problem." + "reason": "Both are load-related, but the symptoms and likely fixes differ." }, { - "left": "issue:42947", - "right": "issue:43856", + "left": "issue:24643", + "right": "issue:28530", "accept": false, - "reason": "Different problems in different training paths: LoRA checkpointing vs Qwen3 MoE memory usage." + "reason": "DeepSpeed weight-dimension runtime error is unrelated to early stopping metric selection." }, { - "left": "issue:24643", - "right": "issue:28530", + "left": "issue:42947", + "right": "issue:43856", "accept": false, - "reason": "Completely different training issues: DeepSpeed weight shape error vs early stopping metric configuration." + "reason": "Training-memory inefficiency and gradient-checkpointing behavior are not the same bug." }, { "left": "issue:43643", "right": "issue:45020", "accept": false, - "reason": "Related to remote-code handling, but one is missing fields in AutoConfig and the other is a broader v5 remote_code breakage." + "reason": "Remote-code config omission is related to, but not the same as, the broader remote-code breakage report." }, { "left": "issue:41762", "right": "issue:42915", "accept": false, - "reason": "Different models and mechanisms: Gemma3 ZeRO-3 loading index error vs Qwen3Moe FP8 config failure." + "reason": "Different model/loading failures with different quantization and distributed settings." }, { "left": "issue:44991", "right": "issue:45399", "accept": false, - "reason": "Unrelated: tokenizer loading regression vs flash-attn2 fallback gating." + "reason": "Tokenizer loading failure and flash-attn2 fallback logic are unrelated." }, { "left": "issue:43824", "right": "issue:43872", "accept": false, - "reason": "Different incompatibilities: missing Qwen2_5_VL symbol vs bitsandbytes Int8Params constructor change." + "reason": "Model import failure and bitsandbytes parameter incompatibility are different problems." }, { "left": "issue:43782", "right": "issue:44960", "accept": false, - "reason": "Different models and failures: Qwen3VL weight_only load error vs GLM5 issue." + "reason": "Different models and different failure modes." }, { "left": "issue:43881", "right": "issue:43883", "accept": false, - "reason": "Different bugs: GLM4V loading failure vs Molmo missing attribute." + "reason": "A model-loading failure and a missing attribute on another model are not the same underlying bug." }, { "left": "issue:45685", "right": "issue:45823", "accept": false, - "reason": "Different device/backend errors: MPS histogram on int vs Gemma4 PLE device mismatch on auto device map." + "reason": "MPS histogram limitation and Gemma4 device-mismatch are distinct issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The accepted soft edges are well supported by the issue titles/descriptions, especially the non-persistent-buffer regression and the cache.get_seq_length() bug. No mergeability problem is present because there are no PRs." + "feedback": "The assessment is grounded in the packet. The summary stays conservative, and the accepted soft-edge pairs are plausibly the same bug families from the titles: the Transformers 5.x non-persistent-buffer regression cluster and the `cache.get_seq_length()` contract regression. The rejected pairs are also cautious and generally supported by the issue titles." }, "evaluator_used": true, "retried": false, @@ -1648,14 +1648,14 @@ "issue:44530|issue:44863", "issue:44863|issue:46402" ], - "prepared_review_unit_hash": "62b48b9dbeec934177f3a3de422c4f76914f45dfbfd0037b04d2e1f8389a9d53", + "prepared_review_unit_hash": "bb65905f513d32be23c654c2367a945f8d6d3412137818a2d1fe690d9ac710d0", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, @@ -1670,192 +1670,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly a loose bundle of unrelated transformers regression reports; only a few pairs are true duplicates or near-direct follow-ups. Clear matches are the `_is_hf_initialized` constructor regression, the v5 docs pipeline cleanup, the Qwen3 training-loss double-shift reports, and the Qwen ZeRO-3 weight-loading failures.", - "confidence": 0.87, - "canonical_issue_reason": "issue:43872 is the strongest canonical issue: it\u2019s a concrete, code-path-specific compatibility regression with a clear near-identical follow-up (issue:44291) and an actionable fix surface.", + "summary": "This set is mostly unrelated single-issue regressions. The clearest duplicate clusters are the v5 docs cleanup, the `_is_hf_initialized`/`init_empty_weights` compatibility regression, and the repeated training-loss double-shift reports; most other soft edges are only loosely similar and should stay separate.", + "confidence": 0.72, + "canonical_issue_reason": "Issue 43872 is the strongest anchor: it is a concrete, widely-impacting runtime compatibility regression with a crisp failure signature, and it has the clearest close duplicate-style follow-up in 44291. It is more actionable and representative than the broader or more speculative matches in the set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43872 is the best overall anchor for the cluster because it is specific, high-signal, and has the cleanest true duplicate among the candidates; it is a better dedupe representative than the broader umbrella-style reports.", + "best_issue_reason": "43872 is the best issue to keep as the representative canonical bug in this batch because the symptom is specific, reproducible, and clearly tied to a single compatibility break.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45081", "right": "issue:45399", "accept": false, - "reason": "Different bugs: tokenizer regex patch crash vs flash-attn fallback gating." + "reason": "Both touch tokenizer/attention-related code, but one is a Mistral regex patch crash and the other is flash-attn fallback blocking; different code paths and fixes." }, { "left": "issue:45020", "right": "issue:45823", "accept": false, - "reason": "Different root causes: remote_code loading regressions vs a device-mismatch forward bug." + "reason": "Remote-code loading regressions and a device_map auto PLE device mismatch are different failure modes." }, { "left": "issue:44661", "right": "issue:45020", "accept": false, - "reason": "Different failure modes in model/tokenizer registration vs remote_code loading." + "reason": "`add-new-model-like` registry logic is unrelated to the broader remote-code breakage described here." }, { "left": "issue:42491", "right": "issue:44368", "accept": false, - "reason": "Same broad Qwen/LoRA area, but not the same concrete bug or fix." + "reason": "Both mention Qwen3/LoRA, but one is a model-compatibility/loading break and the other is a warning about `tie_word_embeddings`; not the same bug." }, { "left": "issue:43828", "right": "issue:45823", "accept": false, - "reason": "Different model/runtime issues: autocast dtype mismatch vs device placement mismatch." + "reason": "Autocast dtype mismatch and PLE device mismatch are distinct runtime errors." }, { "left": "issue:45020", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but the underlying bugs differ." + "reason": "Remote-code model breakage and Kimi tokenizer codec/regex regression are separate issues." }, { "left": "issue:45715", "right": "issue:46315", "accept": false, - "reason": "Both are tokenizer performance regressions, but they hit different APIs and code paths." + "reason": "These are opposite-direction tokenizer performance regressions (`convert_ids_to_tokens` vs `convert_tokens_to_ids`); related area, but not the same bug." }, { "left": "issue:45488", "right": "issue:46489", "accept": false, - "reason": "Related tokenizer families, but the reported breakages are not clearly the same bug." + "reason": "Both concern DeepSeek tokenizer behavior, but they target different model families and different symptom descriptions; too broad to merge." }, { "left": "issue:43827", "right": "issue:44509", "accept": true, - "reason": "Same docs cleanup: v5 removed pipeline tasks and both issues point to stale references." + "reason": "Both report stale docs still referencing pipeline tasks removed in v5; same documentation cleanup." }, { "left": "issue:46894", "right": "issue:46897", "accept": true, - "reason": "Same training-loss double-shift bug pattern; one generic fix could plausibly cover both." + "reason": "Both are the same training-loss double-shift bug pattern reported for different model variants; one shared fix could plausibly cover both." }, { "left": "issue:43872", "right": "issue:46869", "accept": false, - "reason": "Different subsystems: bitsandbytes/init-arg incompatibility vs multi-GPU device-map failure." + "reason": "The `_is_hf_initialized` bitsandbytes compatibility issue is unrelated to a YOLO multi-GPU device_map failure." }, { "left": "issue:43404", "right": "issue:43950", "accept": false, - "reason": "Different bugs: lm_head tying vs non-persistent buffer corruption." + "reason": "A tied lm_head regression and silent non-persistent buffer corruption are different model-loading bugs." }, { "left": "issue:45759", "right": "issue:46661", "accept": false, - "reason": "Both involve model loading, but the concrete failures are unrelated." + "reason": "Composite multimodal config unwrapping and remote-code `config_class` lookup failures are separate loader problems." }, { "left": "issue:43872", "right": "issue:44291", "accept": true, - "reason": "Same `_is_hf_initialized` unexpected-argument regression during model init/loading." + "reason": "Both describe the same unexpected `_is_hf_initialized` kwarg regression during model initialization/loading." }, { "left": "issue:45020", "right": "issue:46661", "accept": false, - "reason": "Remote-code model breakage and missing config_class are different loader problems." + "reason": "Generic remote-code breakage and the specific missing `config_class` attribute error are not the same underlying bug." }, { "left": "issue:43792", "right": "issue:44220", "accept": false, - "reason": "Likely related to Whisper audio loading, but not clearly the same bug from the titles alone." + "reason": "Both are audio-related, but one is a model that cannot run at all and the other is a specific `_torch_extract_fbank_features()` issue; not enough evidence of one root cause." }, { "left": "issue:43257", "right": "issue:45305", "accept": false, - "reason": "Both touch DeepSpeed, but one is weight loading and the other is gradient averaging." + "reason": "DeepSpeed weight-loading failure and gradient averaging with GAS are different training-path bugs." }, { "left": "issue:41553", "right": "issue:44991", "accept": false, - "reason": "Different tokenizer-loading failures for different models." + "reason": "Bad tokenizer error messaging for Voxtral and a tokenizer-loading regression for EMBEDDIA/est-roberta are unrelated." }, { "left": "issue:42947", "right": "issue:45923", "accept": false, - "reason": "Both mention gradient checkpointing, but one is PEFT LoRA behavior and the other is a missing capability flag." + "reason": "Gradient checkpointing being ineffective under PEFT LoRA is not the same as a missing `supports_gradient_checkpointing` flag in a remote-code variant." }, { "left": "issue:36331", "right": "issue:41093", "accept": false, - "reason": "Unrelated errors: Trainer signature mismatch vs mask/tensor shape mismatch." + "reason": "An unexpected `num_items_in_batch` kwarg and a mask/tensor shape mismatch are different failures." }, { "left": "issue:43854", "right": "issue:44863", "accept": false, - "reason": "Different model-loading bugs for different architectures." + "reason": "Different model-support issues: one is GLM-4.7 Flash loading in tests, the other is NemotronH checkpoint loading." }, { "left": "issue:43940", "right": "issue:45313", - "accept": true, - "reason": "Same ZeRO-3 weight-loading failure pattern in Qwen-family models; likely one loader bug." + "accept": false, + "reason": "Both are Qwen3-family ZeRO-3 load failures, but the model variants and missing-weight manifestations differ enough that they should stay separate." }, { "left": "issue:41553", "right": "issue:44488", "accept": false, - "reason": "Both are model/tokenizer loading complaints, but the specific failures differ." + "reason": "Voxtral tokenizer error messaging and `cjvt/sleng-bert` loading failure are unrelated." }, { "left": "issue:44928", "right": "issue:45588", "accept": false, - "reason": "Different runtime bugs: RLHF gradient explosion vs flash-attention attribute error." + "reason": "Gradient explosion from 3D position_ids/SDPA fallback and a flash_attention `s_aux=None` AttributeError are different code-path problems." }, { "left": "issue:41093", "right": "issue:44805", "accept": false, - "reason": "Same error shape family, but not enough evidence that it is the same underlying bug." + "reason": "Both are shape-mismatch IndexErrors, but there is no evidence they share the same root cause or fix." }, { "left": "issue:44479", "right": "issue:44560", "accept": false, - "reason": "Video-input regressions for different model families; not clearly the same defect." + "reason": "Both are video regressions, but they affect different model families and have different symptoms." }, { "left": "issue:24643", "right": "issue:30064", "accept": false, - "reason": "Completely different training and image-segmentation issues." + "reason": "DeepSpeed training weight shape error and void segmentation-map preprocessing are unrelated." }, { "left": "issue:44530", "right": "issue:44863", "accept": false, - "reason": "Different problems: PagedAttention cache crash vs checkpoint loading failure." + "reason": "PagedAttentionCache linear_attention crash and NemotronH checkpoint loading are separate model/runtime issues." }, { - "left": "issue:44863", + "left": "issue:43872", "right": "issue:46402", "accept": false, - "reason": "Both are loading failures, but for different model classes and causes." + "reason": "The `_is_hf_initialized` compatibility regression is unrelated to Sam3Model text-encoder weights not loading." } ] }, "evaluator_result": { "accept": true, - "feedback": "The cluster summary is grounded in the packet and stays appropriately conservative. The accepted soft pairs are well supported by the titles alone: 43827/44509, 43872/44291, 43940/45313, and 46894/46897 all look like true duplicate or near-direct follow-up patterns. The rejected pairs are phrased cautiously and do not overclaim beyond the available evidence." + "feedback": "The summary is grounded in the packet, and the accepted soft edges are well-supported by near-identical titles or clearly shared cleanup patterns. The rest of the pair verdicts stay appropriately conservative." }, "evaluator_used": true, "retried": false, @@ -1962,244 +1962,244 @@ "issue:38175|issue:39692", "issue:43644|issue:44534" ], - "prepared_review_unit_hash": "c2fac74873782cfb09222a8b4acfcc9a5f148d883f4dbaffd16d7713028ffe74", + "prepared_review_unit_hash": "603d013f23123eb51c09e7cad2d49fa886bfefcf1b71dfbbd513f58ddbe1d77c", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 35, - "serialized_chars": 38342, - "estimated_input_tokens": 9586, - "estimated_eval_tokens": 19428 + "serialized_chars": 38348, + "estimated_input_tokens": 9587, + "estimated_eval_tokens": 19430 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it mixes several unrelated Transformers v5 regressions (buffer initialization, custom model loading, meta/device-map loading), plus separate model-specific issues for SigLIP2, Gemma4, Qwen, DeepSeek, and others. I only accepted a few soft edges where the titles point to the same underlying bug; most are merely same-subsystem or same-model-family overlaps.", - "confidence": 0.82, - "canonical_issue_reason": "No single issue safely canonicalizes the entire cluster because the items split across multiple unrelated bugs rather than one duplicate set.", + "summary": "This is a heterogeneous issue cluster with a few clear duplicate subclusters, but no single theme spans everything. The strongest duplicate groups are the Transformers v5 non-persistent-buffer regression, SigLIP2 bad-output reports, Gemma4 4-bit dtype-cast failures, and a custom-model initialization regression.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 43950 is the clearest anchor for the strongest duplicate subcluster: it precisely describes the v5 `from_pretrained()` corruption of `register_buffer(persistent=False)`, which also subsumes the shorter reports in 43644/44534.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43950 is the clearest representative of the repeated Transformers v5 non-persistent-buffer regression; it is concrete, broadly relevant, and matches the strongest duplicate pattern in the set.", + "best_issue_reason": "43950 is the best representative issue overall because it states the underlying regression most explicitly and is the most actionable/diagnostic of the closely related duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43366", "right": "issue:43828", "accept": false, - "reason": "Different bugs: GGUF support for a model architecture vs an autocast dtype mismatch in Phi-tiny-MoE." + "reason": "Different problems: GGUF support for gpt-oss vs a Phi-tiny-MoE autocast dtype mismatch." }, { "left": "issue:43596", "right": "issue:46515", "accept": false, - "reason": "Unrelated failures: deepspeed ZeRO3/Bert indexing vs a torch 2.4.1 compatibility complaint." + "reason": "Unrelated failures: DeepSpeed Zero3/Bert index error vs a broad torch-version incompatibility report." }, { "left": "issue:45325", "right": "issue:45902", "accept": false, - "reason": "Different model families and code paths: Qwen2.5-VL rope scaling vs Qwen3.5 MoE rotary embedding uninitialized memory." + "reason": "Different code paths: Qwen2.5-VL rope/position-id scaling vs Qwen3-MoE meta materialization of `inv_freq`." }, { "left": "issue:43366", "right": "issue:43531", "accept": false, - "reason": "GGUF architecture support and Qwen3-MoE sliding-window behavior are not the same bug." + "reason": "No overlap beyond model family names; GGUF support request vs a sliding-window bug in Qwen3-MoE." }, { "left": "issue:43638", "right": "issue:46515", "accept": false, - "reason": "ZeRO3/Bert indexing and torch-version incompatibility are separate issues." + "reason": "Different underlying bugs: Zero3 Bert training index error vs torch 2.4.1 incompatibility." }, { "left": "issue:45663", "right": "issue:45823", "accept": false, - "reason": "Both are Gemma4, but one is an FSDP2 shared_kv_states bug and the other is a device-map mismatch in forward." + "reason": "Same model family, but distinct issues: per-layer `shared_kv_states` rebuild vs `device_map=auto` device mismatch." }, { "left": "issue:44155", "right": "issue:45584", "accept": false, - "reason": "Audio track leakage in batched inference is unrelated to Whisper empty-transcription generation." + "reason": "Different subsystems: AudioFlamingo3 batched-track leakage vs Whisper empty-transcription handling." }, { "left": "issue:43366", "right": "issue:44589", "accept": false, - "reason": "GGUF model support and Float8 storage lookup errors are different problems." + "reason": "Unrelated: gpt-oss GGUF support vs a Float8 storage lookup error." }, { "left": "issue:43824", "right": "issue:43957", "accept": false, - "reason": "Importing a missing class is not the same as meta-device loading/offload failure." + "reason": "Import error for a specific Qwen class vs a generic meta-device loading regression." }, { "left": "issue:43950", "right": "issue:45834", "accept": false, - "reason": "Non-persistent buffer corruption and a long-OCR index error do not share 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": "Qwen3-MoE FP8 config failure and meta-loading failures are distinct." + "reason": "Different failures: Qwen3-MoE FineGrainedFP8Config issue vs meta-loading breakage across some models." }, { "left": "issue:44910", "right": "issue:46858", "accept": false, - "reason": "Qwen3.5 flash-attn illegal memory access is different from cache seq-length tensor arithmetic on non-CUDA backends." + "reason": "Both touch cache/position logic, but the concrete bugs are different and not mergeable as one fix." }, { "left": "issue:43716", "right": "issue:44560", "accept": false, - "reason": "Multimodal dtype mismatch for Mistral-3 is unrelated to Qwen3-vl video StopIteration." + "reason": "Different multimodal bugs: image preprocessor dtype mismatch vs video StopIteration." }, { "left": "issue:43645", "right": "issue:43646", "accept": true, - "reason": "Same underlying Transformers v5 custom-model initialization regression; the Jupyter wording looks incidental, not a separate code-path." + "reason": "These describe the same custom-model initialization regression; one is just notebook-specific wording of the same breakage." }, { "left": "issue:44560", "right": "issue:45685", "accept": false, - "reason": "Video parsing StopIteration and MPS histogram-on-Int are unrelated." + "reason": "Unrelated: Qwen3-VL video input failure vs an MPS histogram kernel missing for Int tensors." }, { "left": "issue:43550", "right": "issue:44910", "accept": false, - "reason": "Bamba SDPA torch.compile failure and Qwen3.5 flash-attn memory access are different concrete failures." + "reason": "Different failures in different models/backends: Bamba torch.compile+SDPA vs Qwen3.5 flash-attention position-id handling." }, { "left": "issue:42617", "right": "issue:43366", "accept": false, - "reason": "A failing example script is not the same as GGUF architecture support." + "reason": "No shared bug: a 3d_parallel.py execution problem vs gpt-oss GGUF support." }, { "left": "issue:44683", "right": "issue:46515", "accept": false, - "reason": "Torch>=2.9 flex_attention regression and torch 2.4.1 incompatibility are opposite-version, unrelated issues." + "reason": "Both mention torch versions, but one is a compiled flex_attention regression on torch>=2.9 and the other is a compatibility report for torch 2.4.1." }, { "left": "issue:44560", "right": "issue:44918", "accept": false, - "reason": "Video StopIteration and trl SFT embedding unpacking are distinct bugs." + "reason": "Different Qwen3-related bugs: video StopIteration vs SFT trainer embedding unpacking." }, { "left": "issue:36010", "right": "issue:42915", "accept": false, - "reason": "GenerationMixin import-path breakage is unrelated to Qwen3-MoE FineGrainedFP8Config." + "reason": "Unrelated import-path issue for `GenerationMixin` vs a Qwen3-MoE FP8 failure." }, { "left": "issue:45812", "right": "issue:46489", "accept": false, - "reason": "Different tokenizer regressions: Granite token IDs vs DeepSeek-Coder v1 output with a PR gap." + "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths; not the same concrete bug." }, { "left": "issue:45823", "right": "issue:46515", "accept": false, - "reason": "Gemma4 PLE device mismatch is not the same as generic torch 2.4.1 incompatibility." + "reason": "Gemma4 forward-device mismatch is unrelated to the torch 2.4.1 incompatibility report." }, { "left": "issue:44661", "right": "issue:45823", "accept": false, - "reason": "Tokenizer-mapping failure and Gemma4 PLE device placement are unrelated." + "reason": "Tokenizer-mapping edge case vs Gemma4 PLE device placement bug." }, { "left": "issue:38175", "right": "issue:43994", "accept": true, - "reason": "Both report the same SigLIP2 base model producing wrong/degenerate outputs; the second is a fuller symptom description of the same bug." + "reason": "Same SigLIP2 model/processor failure mode: one report shows zero probabilities and the other nonsensical outputs from the same model family." }, { "left": "issue:44560", "right": "issue:44805", "accept": false, - "reason": "Video StopIteration and a mask-shape IndexError are different runtime errors." + "reason": "Different bugs: video StopIteration vs mask-shape index mismatch." }, { "left": "issue:45672", "right": "issue:46899", "accept": true, - "reason": "Same Gemma4 4-bit quantization / unsafe dtype-cast problem in multimodal embedders; one report is a narrower symptom of the same bug." + "reason": "Same Gemma4 4-bit multimodal quantization regression; both point to unsafe dtype handling in audio/vision embedders." }, { "left": "issue:44877", "right": "issue:45030", "accept": false, - "reason": "Strict config rejecting granite_speech and glm4v config validation changes are not the same issue." + "reason": "Both involve config validation, but they are for different model families and different validation failures." }, { "left": "issue:43632", "right": "issue:45834", "accept": false, - "reason": "_is_hf_initialized regression and Kosmos2.5 OCR index error do not match." + "reason": "Unrelated: `_is_hf_initialized` regression vs Kosmos2.5 OCR indexing." }, { "left": "issue:43638", "right": "issue:44661", "accept": false, - "reason": "ZeRO3 Bert index error and add-new-model-like tokenizer mapping failure are unrelated." + "reason": "Different issues: Zero3/Bert index error vs tokenizer-mapping behavior in add-new-model-like." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Different code paths and symptoms; these are not the same underlying bug." }, { "left": "issue:42491", "right": "issue:43720", "accept": false, - "reason": "Qwen3-MoE LoRA hf4.x/hf5.x incompatibility and BitNet packed-weight loading are different problems." + "reason": "Distinct loading regressions in different model stacks; no clear shared root cause." }, { "left": "issue:44661", "right": "issue:44805", "accept": false, - "reason": "Tokenizer-mapping logic and mask/tensor shape mismatch are separate bugs." + "reason": "Tokenizer registration edge case vs a tensor shape/indexing bug." }, { "left": "issue:43317", "right": "issue:44756", "accept": false, - "reason": "GPU+CPU offload loading failure and Strix Halo mmap OOM are not the same root cause." + "reason": "Different memory/loading issues: gpu+cpu offload load failure vs disabling mmap to avoid OOM on Strix Halo." }, { "left": "issue:38175", "right": "issue:39692", - "accept": false, - "reason": "Same model family, but one is incorrect outputs while the other is a documentation example with model/processor mismatch and quantization failure." - }, - { - "left": "issue:43644", - "right": "issue:44534", "accept": true, - "reason": "Same Transformers v5 non-persistent-buffer corruption regression; the later issue is essentially a renamed duplicate." + "reason": "Same underlying SigLIP2 regression around model/processor mismatch; 39692 adds docs-example failures on top of the same bad setup." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: it correctly treats the cluster as heterogeneous, and the accepted soft edges are conservative duplicate calls based on the same model/bug pattern. No pair is overstated, and the chosen representative issue fits the strongest repeated regression in the set." + "feedback": "The summary is grounded in the packet and stays conservative. The identified subclusters are supported by the titles/body cues, and the accepted soft-edge pairs are reasonably mergeable as the same underlying bugfix rather than just thematically similar issues." }, "evaluator_used": true, "retried": false, @@ -2291,8 +2291,8 @@ "issue:44960|issue:46050", "issue:46897|issue:46901", "issue:45488|issue:45812", - "issue:45362|issue:45406", "issue:45276|issue:45335", + "issue:45362|issue:45406", "issue:44164|issue:44291", "issue:44910|issue:45588", "issue:44402|issue:44488", @@ -2300,224 +2300,224 @@ "issue:43425|issue:43957", "issue:41720|issue:44155" ], - "prepared_review_unit_hash": "5d240dedc7038a7e273d00218bde6c9c1d4f982bc2e763ed133a50d85b77ebeb", + "prepared_review_unit_hash": "fac7adefd08e96300735fdaf953459089415767200680769ed268cba6ba686d0", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 36101, - "estimated_input_tokens": 9026, - "estimated_eval_tokens": 18308 + "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 set is mostly heterogeneous. The only clear duplicate-like soft edge is the `resize_token_embeddings` regression across model variants; the rest are different models, different APIs, or different failure modes.", - "confidence": 0.81, - "canonical_issue_reason": "issue:45276 is the cleanest representative of the shared `resize_token_embeddings` bug: it states the concrete API regression and has a near-identical counterpart in 45335.", + "summary": "The cluster is mostly heterogeneous, with several unrelated model-loading, quantization, tokenizer, and training bugs. The 44451 \u2194 44488 pair is the closest match, but the evidence only supports \u201csimilar BERT-loading failures,\u201d not a confident duplicate/root-cause match.", + "confidence": 0.61, + "canonical_issue_reason": "No strong cluster-wide canonical issue: the packet spans many unrelated regressions. If a representative anchor is needed, 44488 is the broadest BERT-loading report, but the evidence is not strong enough to treat it as a confirmed duplicate target for 44451.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45276 is the best overall issue anchor because it is specific, reproducible, and describes the underlying behavior change rather than a model-specific symptom.", + "best_issue_reason": "44488 is the best representative issue here because it is the broader, later BERT-loading report and has the stronger discussion footprint, but only as a loose anchor rather than a proven duplicate.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42491", "right": "issue:45276", "accept": false, - "reason": "Different problems: Qwen3 MoE LoRA compatibility across versions vs Gemma4 token-embedding resize behavior." + "reason": "Different bugs: Qwen3 MoE LoRA compatibility vs Gemma4 token-embedding resize behavior." }, { "left": "issue:41720", "right": "issue:42915", "accept": false, - "reason": "Both are Qwen3-related, but one is an auto-device-mapping CUDA assert and the other is a FineGrainedFP8Config failure." + "reason": "Same model family, but different failures: device-mapping cuda assert vs FineGrainedFP8Config loading." }, { "left": "issue:42915", "right": "issue:44756", "accept": false, - "reason": "Completely different subsystems: Qwen3MoE FP8 config vs Strix Halo mmap/OOM." + "reason": "Unrelated problems: Qwen3 MoE FP8 config vs Strix Halo mmap/OOM behavior." }, { "left": "issue:44451", "right": "issue:44488", "accept": false, - "reason": "Both are model-load failures, but the titles point to different models and no shared concrete bug is evident from the metadata." + "reason": "Very similar BERT-loading symptoms, but the packet does not \u0581\u0578\u0582\u0575\u0581 a shared root cause or deterministic duplicate signal." }, { "left": "issue:44960", "right": "issue:44977", "accept": false, - "reason": "Different models and failure modes: GLM5 is not the same issue as Qwen3.5 flash-attention generation breakage." + "reason": "Different models and failure modes: GLM5 vs Qwen3.5 flash-attention generation." }, { "left": "issue:44863", "right": "issue:46846", "accept": false, - "reason": "Unrelated models and code paths: NemotronH checkpoint loading vs Qwen3.5 tensor-parallel plan entries." + "reason": "Different code paths: NemotronH checkpoint loading vs Qwen3.5 tensor-parallel plan/OOM." }, { "left": "issue:46459", "right": "issue:46512", "accept": false, - "reason": "Both involve dtype handling, but one is AutoModelForCausalLM loading and the other is from_config initialization; not the same bug." + "reason": "Both involve dtype handling, but one is composite-checkpoint loading and the other is config state consistency." }, { "left": "issue:45313", "right": "issue:45357", "accept": false, - "reason": "Different regressions: ZeRO-3 weight loading vs save_pretrained visual-encoder key serialization." + "reason": "Same model family, but separate bugs: ZeRO-3 loading vs save_pretrained visual-key serialization." }, { "left": "issue:45507", "right": "issue:46468", "accept": false, - "reason": "Different models and errors; no concrete shared code-path is implied." + "reason": "Different artifacts and symptoms: GraniteMoEHybrid invalid method vs broken Nemotron example." }, { "left": "issue:44488", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/model loading issues, but they affect different models and the failure descriptions do not match." + "reason": "Both are loading-related, but 44991 is a distinct tokenizer-loading regression for another model family." }, { "left": "issue:43644", "right": "issue:43645", "accept": false, - "reason": "Both mention Transformers 5.0, but one is non-persistent buffer initialization and the other is notebook-defined custom model construction." + "reason": "Both are Transformers 5.x issues, but one is buffer initialization and the other is custom-model init in Jupyter." }, { "left": "issue:42491", "right": "issue:45292", "accept": false, - "reason": "Different issues: version compatibility for Qwen3 MoE LoRA vs `resize_token_embeddings` not updating output embeddings." + "reason": "Same API surface, but different bugs and model paths; not the same concrete issue." }, { "left": "issue:39692", "right": "issue:43994", "accept": false, - "reason": "Same model family, but one is a docs/example mismatch plus quantization failure and the other is wrong outputs with AutoModel/pipeline." + "reason": "Different SigLIP2 failures: docs example mismatch/quantization vs nonsensical AutoModel and pipeline results." }, { "left": "issue:43257", "right": "issue:43854", "accept": false, - "reason": "Different problems: Qwen3 MoE weight conversion with accelerate/deepspeed vs GLM-4.7-Flash unit-test loading." + "reason": "Different Qwen3-related failures: accelerate+deepspeed MoE conversion vs unit-test loading error." }, { "left": "issue:43856", "right": "issue:43873", "accept": false, - "reason": "Both touch memory/quantization, but one is training-memory inefficiency and the other is offloading behavior during quantized loading." + "reason": "Both mention memory, but the underlying issues are different: training efficiency vs quantization offloading behavior." }, { "left": "issue:43856", "right": "issue:44387", "accept": false, - "reason": "Both are memory-related, but one is Qwen3 MoE training memory and the other is a Transformers 5.x int4 reserved-memory regression." + "reason": "Different memory regressions: Qwen3 MoE training memory vs int4 quantization reserved-memory OOM." }, { "left": "issue:43854", "right": "issue:43931", "accept": false, - "reason": "Different loading failures on different models; no shared underlying bug is apparent." + "reason": "Different model-loading failures on different models; no shared code-path evidence." }, { "left": "issue:43708", "right": "issue:44075", "accept": false, - "reason": "Trainer checkpoint-resume step calculation is unrelated to optimizer argument handling." + "reason": "Trainer checkpoint-resume/max_steps bug vs optimizer args being ignored." }, { "left": "issue:42491", "right": "issue:45335", "accept": false, - "reason": "Qwen3 MoE version-compatibility issue is unrelated to the token-embedding resize regression." + "reason": "Different model families and embedding paths; too model-specific to merge." }, { "left": "issue:44960", "right": "issue:46050", "accept": false, - "reason": "GLM5 is unrelated to quantized-model save/load corruption after sinq." + "reason": "Different features: flash-attention generation failure vs quantization save/load corruption." }, { "left": "issue:46897", "right": "issue:46901", "accept": false, - "reason": "Both describe a double-shift training-loss pattern, but the titles indicate different model implementations with no proof of the same code path." + "reason": "Same broad symptom pattern, but different model implementations and no evidence of a shared code path." }, { "left": "issue:45488", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different model families and the specific tokenization bug differs." + "reason": "Different tokenizer regressions in different stacks: pre-tokenizer override vs Granite token ID mismatch." }, { - "left": "issue:45362", - "right": "issue:45406", + "left": "issue:45276", + "right": "issue:45335", "accept": false, - "reason": "Qwen3.5 chat-generation crash and Gemma4Processor serve crash are different bugs." + "reason": "Same API, but different model families and embedding wiring; not the same bug." }, { - "left": "issue:45276", - "right": "issue:45335", - "accept": true, - "reason": "Same underlying regression: `resize_token_embeddings` fails to propagate updated embeddings/output heads across model variants." + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Unrelated crashes: Qwen3.5 chat generation vs Gemma4Processor missing _tokenizer in serve." }, { "left": "issue:44164", "right": "issue:44291", "accept": false, - "reason": "Different state/init bugs: extra_state save/load handling vs unexpected `_is_hf_initialized` during init_empty_weights." + "reason": "Different save/load bugs: extra_state handling vs unexpected _is_hf_initialized argument." }, { "left": "issue:44910", "right": "issue:45588", "accept": false, - "reason": "Both involve flash attention, but the concrete failures are unrelated: packed-sequence misinterpretation vs `s_aux=None` AttributeError." + "reason": "Both touch flash attention, but they fail on different inputs and distinct code paths." }, { "left": "issue:44402", "right": "issue:44488", "accept": false, - "reason": "Tokenizer vocab-size mismatch is a different bug from model loading failure on another BERT model." + "reason": "Pad/unk vocab-size mismatch is a different tokenizer/model-config issue from the BERT loading failure." }, { "left": "issue:39692", "right": "issue:44960", "accept": false, - "reason": "SigLIP2 doc/example issues are unrelated to the GLM5 issue." + "reason": "Completely different models and symptoms: SigLIP2 docs/example issues vs GLM5." }, { "left": "issue:43425", "right": "issue:43957", "accept": false, - "reason": "Torch version incompatibility is not the same as `torch.device('meta')` loading breakage." + "reason": "Different compatibility failures: Torch version incompatibility vs torch.device('meta') loading breakage." }, { "left": "issue:41720", "right": "issue:44155", "accept": false, - "reason": "Qwen3 auto-device-map CUDA assert and AudioFlamingo3 embedding/token leak are unrelated." + "reason": "Different failures: Qwen3 device-mapping assert vs AudioFlamingo3 batched inference leak." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative. The only accepted soft edge, issue:45276 vs issue:45335, is well-supported by the near-identical `resize_token_embeddings` regression pattern across model variants. The remaining rejects are appropriately cautious given the titles and metadata. No overstatement stands out." + "accept": false, + "feedback": "The summary overstates the evidence around 44451 \u2194 44488. The packet only shows two similar BERT-loading issues with no deterministic duplicate signal and no explicit shared root cause; calling it the \"only clear duplicate pair\" and the same underlying regression is stronger than the evidence supports. The rest of the pairwise rejections are mostly conservative, but the canonical/summary claim is not sufficiently grounded." }, "evaluator_used": true, - "retried": false, - "accepted_nontrivial_soft_edge": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null }, @@ -2585,10 +2585,10 @@ ], "soft_pairs": [ "issue:45663|issue:45759", - "issue:43404|issue:43454", "issue:44079|issue:46829", - "issue:43873|issue:44387", + "issue:43404|issue:43454", "issue:44291|issue:44756", + "issue:43873|issue:44387", "issue:42947|issue:45704", "issue:44292|issue:44912", "issue:36010|issue:39404", @@ -2604,238 +2604,238 @@ "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:45588|issue:45663", "issue:45292|issue:45335", "issue:46894|issue:46901", - "issue:44402|issue:44451", "issue:43828|issue:43957", + "issue:44402|issue:44451", "issue:43646|issue:43708", "issue:43645|issue:44164", "issue:44589|issue:44756", "issue:43425|issue:43828" ], - "prepared_review_unit_hash": "076000a185757a5ebd2b443d31fb8985769c5eb038d6a4de47ffdc84ffb6d745", + "prepared_review_unit_hash": "1a6de54c84c3c3d05b6695eded916a1100d706169f6f29423b063768ab07639d", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37128, - "estimated_input_tokens": 9282, - "estimated_eval_tokens": 18820 + "serialized_chars": 37131, + "estimated_input_tokens": 9283, + "estimated_eval_tokens": 18822 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly unrelated transformer bug reports, but it does contain two real duplicate subclusters: resize_token_embeddings embedding-tie regressions and training-loss double-shift reports.", - "confidence": 0.81, - "canonical_issue_reason": "issue:45276 is the broadest and clearest representative of the resize_token_embeddings regression family, covering both embed_tokens_per_layer and output_embeddings; it can subsume the narrower follow-ups.", + "summary": "This cluster is mostly heterogeneous; the only strong duplicate-like family is the resize_token_embeddings propagation bug around output embeddings / decoder embeddings. Most other pairs are same subsystem at best, but different concrete bugs and should stay separate.", + "confidence": 0.74, + "canonical_issue_reason": "Issue 45292 is the cleanest generic statement of the resize_token_embeddings/output_embeddings bug and best serves as the umbrella for the closely related 45276 and 45335 reports. The rest of the cluster is unrelated enough that no single issue can represent the whole set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45276 is the best single issue representative in this cluster because it states the shared embedding-resize failure most completely and has the cleanest path to absorb nearby variants.", + "best_issue_reason": "45292 is the best overall issue candidate because it is the broadest, least model-specific description of the shared embedding-resize failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45663", "right": "issue:45759", "accept": false, - "reason": "Different bugs: Gemma4 FSDP2 shared_kv_states KeyError vs Qwen multimodal text_config unwrapping." - }, - { - "left": "issue:43404", - "right": "issue:43454", - "accept": false, - "reason": "Both mention untied lm_head weights, but they are model-specific regressions in different code paths; too broad to merge confidently." + "reason": "Different bugs: FSDP2 KeyError in Gemma4 attention vs from_config not unwrapping text_config for Qwen multimodal configs." }, { "left": "issue:44079", "right": "issue:46829", "accept": false, - "reason": "Unrelated output-shape bugs: ModelOutput key assignment vs VideoPrism hidden_states wiring." + "reason": "Both involve outputs, but one is a generic ModelOutput key assignment bug and the other is a specific model returning the wrong tensor in hidden_states." }, { - "left": "issue:43873", - "right": "issue:44387", + "left": "issue:43404", + "right": "issue:43454", "accept": false, - "reason": "Both involve quantization/memory, but one is offloading behavior and the other is int4 OOM from reserved memory; different failures." + "reason": "Both are lm_head tying problems, but in different model classes with different likely code paths and fixes." }, { "left": "issue:44291", "right": "issue:44756", "accept": false, - "reason": "Different problem domains: init_empty_weights argument handling vs Strix Halo mmap OOM." + "reason": "Unrelated: init_empty_weights argument handling vs mmap-related OOM on Strix Halo." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both mention quantization and memory, but one is offloading behavior and the other is int4 reserved-memory OOM; not the same bug." }, { "left": "issue:42947", "right": "issue:45704", "accept": false, - "reason": "PEFT/gradient-checkpointing behavior is unrelated to the T5 apex RMSNorm memory leak." + "reason": "Gradient checkpointing/LoRA issue is unrelated to the apex.FusedRMSNorm memory leak in T5." }, { "left": "issue:44292", "right": "issue:44912", "accept": false, - "reason": "Loading with init_empty_weights/_is_hf_initialized is unrelated to MXFP4 quantization fallback on git-oss-20b." + "reason": "Different quantization failures: NVFP4 loading error vs MXFP4 load fallback to bf16." }, { "left": "issue:36010", "right": "issue:39404", "accept": false, - "reason": "ImportError for GenerationMixin and Whisper return_language pipeline regression are separate issues." + "reason": "Import error for GenerationMixin is unrelated to Whisper pipeline return_language regression." }, { "left": "issue:45663", "right": "issue:45859", "accept": false, - "reason": "Gemma4 attention KeyError and Qwen3_5Moe CPU-offload rotary embedding are unrelated." + "reason": "Different model families and different offload/device behavior bugs." }, { "left": "issue:44387", "right": "issue:44756", "accept": false, - "reason": "Both are OOM-related, but one is int4 quantization reserved-memory growth and the other is mmap policy on Strix Halo." + "reason": "Both can lead to OOM, but the causes are different: int4 reserved memory vs mmap on Strix Halo." }, { "left": "issue:42947", "right": "issue:43381", "accept": false, - "reason": "Gradient checkpointing misuse in eval mode is not the same as LoRA checkpointing being ineffective." + "reason": "Both mention gradient checkpointing, but one is ineffective with PEFT LoRA while the other forbids checkpointing in eval mode." }, { "left": "issue:44155", "right": "issue:45823", "accept": false, - "reason": "AudioFlamingo3 track leakage and Gemma4 device mismatch are unrelated inference/training bugs." + "reason": "AudioFlamingo batching leak is unrelated to Gemma4 device mismatch under device_map=auto." }, { "left": "issue:43957", "right": "issue:44756", "accept": false, - "reason": "Model loading with torch.device('meta') and disabling mmap on Strix Halo are different loading/memory issues." + "reason": "Meta-device model loading regression is unrelated to Strix Halo mmap OOM." }, { "left": "issue:43335", "right": "issue:45663", "accept": false, - "reason": "SwitchTransformers sparse-layer creation and Gemma4 FSDP2 attention KeyError are unrelated." + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to Gemma4 FSDP2 KeyError." }, { "left": "issue:45276", "right": "issue:45292", "accept": true, - "reason": "Same resize_token_embeddings bug family; 45276 is the broader report and 45292 is a narrower manifestation." + "reason": "Same resize_token_embeddings propagation bug: both report output embeddings not being updated, with 45276 adding a Gemma4-specific manifestation." }, { "left": "issue:44863", "right": "issue:46468", "accept": false, - "reason": "NemotronH checkpoint loading and a broken official example may be related by ecosystem, but not clearly the same concrete bug." + "reason": "NemotronH checkpoint loading failure is separate from the broken official Nemotron example." }, { "left": "issue:44479", "right": "issue:44877", "accept": false, - "reason": "Qwen video-input regression and granite_speech config strictness are unrelated." + "reason": "Video-input regression for Qwen VL models is unrelated to strict config blocking granite_speech." }, { "left": "issue:43381", "right": "issue:44928", "accept": false, - "reason": "Gradient checkpointing in eval mode is unrelated to Qwen3.5 RLHF NaN explosion from 3D position_ids/SDPA fallback." + "reason": "Checkpointing-in-eval-mode and Qwen3.5 RLHF NaN explosion are different problems." }, { "left": "issue:44164", "right": "issue:44756", "accept": false, - "reason": "extra_state save/load handling is unrelated to mmap OOM behavior." + "reason": "save/from_pretrained extra_state handling is unrelated to mmap OOM behavior." }, { "left": "issue:45684", "right": "issue:45698", "accept": false, - "reason": "Both involve custom-model save/load, but one is permissions propagation and the other is wrong module resolution; different code paths." + "reason": "Both involve custom-model save/load, but one is read-only permission propagation and the other is loading the wrong custom module." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention crash on s_aux=None is unrelated to Gemma4 FSDP2 KeyError." }, { "left": "issue:43854", "right": "issue:45216", "accept": false, - "reason": "GLM-4.7-Flash loading in tests and Qwen3.5 save_pretrained regression are separate model-specific problems." + "reason": "Different serialization/load regressions affecting different models and APIs." }, { "left": "issue:42796", "right": "issue:46326", "accept": false, - "reason": "Both are tokenizer-related, but one is BPE marker handling and the other is continue_final_message logic." - }, - { - "left": "issue:45588", - "right": "issue:45663", - "accept": false, - "reason": "Flash-attention s_aux=None crash and Gemma4 FSDP2 KeyError are unrelated." + "reason": "Tokenizer BPE marker handling is not the same as continue_final_message behavior for mistral_common tokenizers." }, { "left": "issue:45292", "right": "issue:45335", "accept": true, - "reason": "Same resize_token_embeddings failure to update downstream embeddings; 45335 is a model-specific instance of the same bug." + "reason": "Same resize_token_embeddings bug family: both say resized token embeddings do not propagate to model output/input embeddings." }, { "left": "issue:46894", "right": "issue:46901", - "accept": true, - "reason": "Same training-loss double-shift bug: both describe training against already-shifted labels." - }, - { - "left": "issue:44402", - "right": "issue:44451", "accept": false, - "reason": "Tokenizer vocab-size mismatch and ScandiBERT loading failure are not the same bug." + "reason": "Same broad double-shift pattern, but different model-specific training-loss implementations and not a single concrete code-path." }, { "left": "issue:43828", "right": "issue:43957", "accept": false, - "reason": "Autocast dtype mismatch and meta-device model-loading failures are unrelated." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to meta-device model loading." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab-size mismatch is unrelated to inability to load ScandiBERT." }, { "left": "issue:43646", "right": "issue:43708", "accept": false, - "reason": "Custom-model initialization breakage and Trainer resume_from_checkpoint max_steps miscalculation are distinct issues." + "reason": "Custom model initialization breakage and Trainer max_steps resume miscalculation are unrelated." }, { "left": "issue:43645", "right": "issue:44164", "accept": false, - "reason": "Jupyter custom-model initialization and extra_state save/load handling are different failure modes." + "reason": "Jupyter custom-model initialization bug is unrelated to save/from_pretrained extra_state handling." }, { "left": "issue:44589", "right": "issue:44756", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to the Strix Halo mmap OOM issue." + "reason": "Float8 storage lookup failure is unrelated to Strix Halo mmap OOM." }, { "left": "issue:43425", "right": "issue:43828", "accept": false, - "reason": "Torch 2.10 incompatibility is broader and different from the Phi-tiny-MoE autocast dtype mismatch." + "reason": "Torch 2.10 compatibility and autocast dtype mismatch are separate issues." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet, and the soft-edge verdicts stay conservative. The accepted duplicates are well supported by the titles, and the rejected pairs are framed narrowly enough to avoid overclaiming sameness." + "feedback": "The summary is grounded in the packet and stays conservative. The only clearly duplicate-like family supported by the titles is the resize_token_embeddings/output_embeddings group around 45292, 45276, and 45335, and the rejected soft pairs are framed as likely different bugs rather than asserted duplicates. No overstatement stands out." }, "evaluator_used": true, "retried": false, @@ -2863,6 +2863,7 @@ "issue:43452", "issue:43596", "issue:43611", + "issue:43644", "issue:43645", "issue:43716", "issue:43856", @@ -2915,9 +2916,9 @@ "issue:43425|issue:46869", "issue:43425|issue:43716", "issue:43425|issue:45823", - "issue:43906|issue:44530", - "issue:44479|issue:44805", "issue:44589|issue:45084", + "issue:44479|issue:44805", + "issue:43906|issue:44530", "issue:43901|issue:44509", "issue:44671|issue:44991", "issue:43994|issue:44079", @@ -2927,237 +2928,244 @@ "issue:46050|issue:46326", "issue:44164|issue:45759", "issue:43906|issue:46326", - "issue:43299|issue:43931", "issue:44991|issue:45081", + "issue:43299|issue:43931", "issue:45823|issue:46899", "issue:45479|issue:46894", "issue:43856|issue:44756", - "issue:45479|issue:46897", "issue:43596|issue:44683", + "issue:45479|issue:46897", "issue:44164|issue:46821", "issue:44488|issue:44987", - "issue:44488|issue:45081" + "issue:44488|issue:45081", + "issue:43611|issue:43644" ], - "prepared_review_unit_hash": "42d96cd555882e9d5e7618551ba107144420157d4e6dc9d0e4f1e14b0ff63295", + "prepared_review_unit_hash": "5dd70508918d855533bf2ae4e6203be8da2b3e0b961a689c686194101e55d7c9", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { - "node_count": 47, - "item_count": 47, - "soft_pair_count": 32, - "serialized_chars": 36878, - "estimated_input_tokens": 9220, - "estimated_eval_tokens": 18696 + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37737, + "estimated_input_tokens": 9435, + "estimated_eval_tokens": 19126 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a collection of unrelated Transformers v5 regressions across tokenizers, model loading, quantization, and multimodal paths. I found no soft-edge pair that clearly refers to the same underlying bug/change, so all candidate merges are rejected.", + "summary": "This cluster is a grab-bag of mostly unrelated Transformers v5 regressions across tokenizers, model loading, multimodal inputs, quantization, and docs. No soft-edge pair looks like the same concrete bug/change, so all candidates are rejected. The broadest representative issue is the Granite tokenizer regression.", "confidence": 0.84, - "canonical_issue_reason": "Issue 45812 is the strongest representative issue: it describes a concrete, high-impact v4\u2192v5 regression with broad tokenizer corruption across all Granite models and clear user-visible fallout.", + "canonical_issue_reason": "Issue 45812 is the strongest representative: it describes a broad, user-visible tokenizer regression across all Granite models, has inbound references, and is more central than the many highly model-specific failures.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45812 is the best overall issue candidate because it is specific, reproducible, and a broad silent regression rather than a narrow edge case.", + "best_issue_reason": "Best overall issue candidate because it is the clearest high-impact regression with broad scope and likely duplicate potential, rather than a narrow one-off model bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46315", "right": "issue:46491", "accept": false, - "reason": "Both involve tokenizers, but one is a slowdown in convert_tokens_to_ids and the other is a CodeLlama round-trip whitespace regression; different bugs and code paths." + "reason": "One is a tokenizer speed regression in convert_tokens_to_ids; the other is an encode/decode space-roundtrip correctness bug in CodeLlama. Different symptoms and code paths." }, { "left": "issue:45812", "right": "issue:46315", "accept": false, - "reason": "Granite token ID corruption is unrelated to convert_tokens_to_ids performance; same subsystem, not the same bug." + "reason": "Granite token IDs are wrong vs slow convert_tokens_to_ids for slow tokenizers. Different failure mode: correctness regression vs performance regression." }, { "left": "issue:39404", "right": "issue:45701", "accept": false, - "reason": "Whisper return_language pipeline behavior and generic tokenization changes are too broad and do not match the same failure." + "reason": "Whisper pipeline return_language broke; the other is a broad tokenization version-change complaint. Too general and not the same concrete bug." }, { "left": "issue:39404", "right": "issue:46665", "accept": false, - "reason": "Pipeline return_language regression is separate from inconsistent from_pretrained behavior inside Python classes." + "reason": "Whisper pipeline behavior and from_pretrained inside Python classes are unrelated issues." }, { "left": "issue:44530", "right": "issue:45084", "accept": false, - "reason": "Different failures: Qwen3.5 PagedAttentionCache linear_attention crash vs Mistral tokenizer regex patch compile error." + "reason": "Qwen3.5 PagedAttentionCache crashes on linear_attention; the other is a template compilation error. Different subsystems and symptoms." }, { "left": "issue:43611", "right": "issue:43645", "accept": false, - "reason": "Both are loading-related, but base_model_prefix loading and Jupyter custom-model initialization are distinct issues." + "reason": "Both involve loading/custom models, but one is base_model_prefix loading breakage and the other is Jupyter custom-model initialization. Not the same bug." }, { "left": "issue:44683", "right": "issue:45823", "accept": false, - "reason": "flex_attention compilation on torch>=2.9 is unrelated to Gemma4 device-map forward mismatches." + "reason": "Torch>=2.9 flex_attention compilation failure is unrelated to Gemma4 device-map forward mismatches." }, { "left": "issue:43425", "right": "issue:46869", "accept": false, - "reason": "Torch version compatibility and YOLO multi-GPU device-map forwarding are not the same underlying bug." + "reason": "Torch 2.10 incompatibility is not the same as a YOLO multi-GPU device_map auto failure." }, { "left": "issue:43425", "right": "issue:43716", "accept": false, - "reason": "Torch 2.10 incompatibility is not the same as Mistral-3 image dtype mismatch." + "reason": "Torch version incompatibility and Mistral image-preprocessor dtype mismatch are different problems." }, { "left": "issue:43425", "right": "issue:45823", "accept": false, - "reason": "Global torch compatibility and Gemma4 PLE device mismatch are different code-path problems." + "reason": "General torch incompatibility vs Gemma4 device mismatch during forward; not the same underlying bug." }, { - "left": "issue:43906", - "right": "issue:44530", + "left": "issue:44589", + "right": "issue:45084", "accept": false, - "reason": "An isolated reproduction of another issue is not enough to identify the same bug as the Qwen3.5 cache crash." + "reason": "Float8 storage lookup failure is unrelated to the template-node compilation error." }, { "left": "issue:44479", "right": "issue:44805", "accept": false, - "reason": "Video-input regression for Qwen models is different from a mask-shape IndexError." + "reason": "Video-input regression for Qwen VL models vs a tensor mask shape mismatch. Different code paths." }, { - "left": "issue:44589", - "right": "issue:45084", + "left": "issue:43906", + "right": "issue:44530", "accept": false, - "reason": "Float8 storage lookup failure and non-template node compilation failure are unrelated." + "reason": "An isolated reproduction of an earlier issue is not the same as a Qwen3.5 cache crash." }, { "left": "issue:43901", "right": "issue:44509", "accept": false, - "reason": "Both are docs-related, but one is about return_all_scores and the other about removed pipeline tasks." + "reason": "Both are docs-related, but they cover different pipeline-task deprecations and behavior mismatches." }, { "left": "issue:44671", "right": "issue:44991", "accept": false, - "reason": "CamemBERT masked-LM prediction regression is not the same as EMBEDDIA tokenizer loading failure." + "reason": "CamemBERT masked-LM prediction error and tokenizer load failure for est-roberta are distinct." }, { "left": "issue:43994", "right": "issue:44079", "accept": false, - "reason": "SigLIP2 nonsensical outputs and ModelOutput key assignment are different defects." + "reason": "Siglip2 produces bad outputs; ModelOutput key assignment is a separate data-structure bug." }, { "left": "issue:44265", "right": "issue:44898", "accept": false, - "reason": "torch.export failures with torch_compilable_check are unrelated to Perceiver interpolation failures." + "reason": "torch.export failure with torch_compilable_check is unrelated to Perceiver interpolation failures." }, { "left": "issue:36032", "right": "issue:43452", "accept": false, - "reason": "Tokenizer add_special_tokens conflict and gguf_file loading breakage are distinct loading bugs." + "reason": "T5 tokenizer add_special_tokens conflict and gguf_file loading breakage are different loading issues." }, { "left": "issue:39404", "right": "issue:44912", "accept": false, - "reason": "Whisper return_language and MXFP4 fallback on git-oss-20b are separate regressions." + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback on git-oss-20b." }, { "left": "issue:46050", "right": "issue:46326", "accept": false, - "reason": "sinq save/load breakage and mistral_common continue_final_message behavior do not share the same root cause." + "reason": "Quantized-model save/load breakage via sinq is not the same as continue_final_message for mistral_common tokenizers." }, { "left": "issue:44164", "right": "issue:45759", "accept": false, - "reason": "extra_state save/load handling and text_config unwrapping in composite Qwen configs are different mechanisms." + "reason": "extra_state save/from_pretrained handling and unwrapping text_config in from_config are different model-construction problems." }, { "left": "issue:43906", "right": "issue:46326", "accept": false, - "reason": "A reproduction issue and a mistral_common tokenizer bug are not the same concrete defect." + "reason": "Repro of a prior issue and a mistral_common tokenizer behavior bug are unrelated." }, { - "left": "issue:43299", - "right": "issue:43931", + "left": "issue:44991", + "right": "issue:45081", "accept": false, - "reason": "Both concern Qwen3-VL loading, but one is a broader MoE loading regression and the other is a specific weight-shape mismatch; not enough to merge." + "reason": "Both are tokenizer loading regressions, but for different models and different code paths (generic tokenizer load vs Mistral regex patching)." }, { - "left": "issue:44991", - "right": "issue:45081", + "left": "issue:43299", + "right": "issue:43931", "accept": false, - "reason": "Two tokenizer loading failures in different models and different code paths; not the same bug." + "reason": "Both are Qwen3-VL-family loading failures, but one is a MoE-model regression and the other is a specific weight-shape mismatch. Not enough to treat as one bug." }, { "left": "issue:45823", "right": "issue:46899", "accept": false, - "reason": "Gemma4 device-map forward mismatch differs from the 4-bit dtype-cast problem in multimodal embedders." + "reason": "Gemma4 device-map mismatch during forward and Gemma4 4-bit multimodal dtype-cast breakage are separate issues." }, { "left": "issue:45479", "right": "issue:46894", "accept": false, - "reason": "Sequence-classification zero-loss bug and CohereASR training-loss double-shift are unrelated." + "reason": "Single-label classification zero-loss bug and CohereASR training-loss double-shift bug are unrelated model-loss errors." }, { "left": "issue:43856", "right": "issue:44756", "accept": false, - "reason": "Qwen3 MoE training memory inefficiency is unrelated to disabling mmap on Strix Halo." + "reason": "Qwen3 MoE memory usage and Strix Halo mmap OOM are different memory-related problems." }, { - "left": "issue:45479", - "right": "issue:46897", + "left": "issue:43596", + "right": "issue:44683", "accept": false, - "reason": "Sequence-classification zero-loss bug is not the same as the Florence2 training-loss double-shift regression." + "reason": "Deepspeed zero3/BertModel index error is unrelated to compiled flex_attention on torch>=2.9." }, { - "left": "issue:43596", - "right": "issue:44683", + "left": "issue:45479", + "right": "issue:46897", "accept": false, - "reason": "Deepspeed zero3/BertModel index error is unrelated to compiled flex_attention on newer torch." + "reason": "Different training-loss bugs in different model families; not the same underlying defect." }, { "left": "issue:44164", "right": "issue:46821", "accept": false, - "reason": "extra_state handling and TimesFM window_size AttributeError are different loading/runtime issues." + "reason": "Handling extra_state in save/load is unrelated to a window_size AttributeError in TimesFM 2.5." }, { "left": "issue:44488", "right": "issue:44987", "accept": false, - "reason": "Loading cjvt/sleng-bert and loading physical-intelligence/fast are separate model-specific failures." + "reason": "Both are model-loading failures, but for different models and likely different tokenizer/config issues." }, { "left": "issue:44488", "right": "issue:45081", "accept": false, - "reason": "Sleng-bert load failure and Mistral regex patch crash are unrelated tokenization/loading issues." + "reason": "Different tokenizer/model load failures; no evidence they share the same root cause." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "base_model_prefix loading breakage and non-persistent buffer junk are separate Transformers v5 regressions." } ] }, @@ -3199,7 +3207,6 @@ "issue:43482", "issue:43541", "issue:43582", - "issue:43611", "issue:43632", "issue:43643", "issue:43644", @@ -3214,6 +3221,7 @@ "issue:44462", "issue:44488", "issue:44530", + "issue:44534", "issue:44661", "issue:44683", "issue:44737", @@ -3232,7 +3240,6 @@ ], "soft_pairs": [ "issue:44462|issue:46489", - "issue:43611|issue:43644", "issue:42796|issue:43404", "issue:46285|issue:46402", "issue:36032|issue:42915", @@ -3260,226 +3267,224 @@ "issue:45084|issue:45507", "issue:43262|issue:43377", "issue:43746|issue:46326", - "issue:43883|issue:44991" + "issue:43883|issue:44991", + "issue:43646|issue:44534" ], - "prepared_review_unit_hash": "1262685ef501325432e541ae7f372418c4dac55ef07b306d1ab9fb1a4950dfd2", + "prepared_review_unit_hash": "fe7f134db625d6a87dca485f09b5b689c309f53f627b47ed4db8a4a3307a3057", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36717, - "estimated_input_tokens": 9180, - "estimated_eval_tokens": 18616 + "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": "Nearly all candidate edges are false positives: they share broad themes like tokenizers, v5 regressions, or model loading, but not the same concrete bug. The only plausible duplicate is the audio feature-extraction pair around `_torch_extract_fbank_features()` / `LasrFeatureExtractor` center-arg handling.", - "confidence": 0.88, - "canonical_issue_reason": "issue:44206 is the clearest representative of the one likely duplicate pair because it states the specific regression and failure mode; issue:44220 looks like the same audio fbank extraction path but is phrased more generically.", + "summary": "This cluster is a loose bucket of Transformers v5 regression reports across many unrelated subsystems; the suggested soft pairs mostly share broad theme similarity but not the same concrete bug, so I reject all of them.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the broadest, most representative report here: a generic Transformers 5.0.0 custom-model initialization regression that fits the cluster\u2019s overall v5-breakage theme better than the more niche issues.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44206 is the best issue to keep as canonical: it is specific, user-facing, and points to the likely root cause in the feature extractor rather than the downstream crash wrapper.", + "best_issue_reason": "issue:43646 is the best single anchor for this cluster because it is general enough to represent the repeated v5 initialization regressions, even though the cluster itself is too heterogeneous to treat as duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44462", "right": "issue:46489", "accept": false, - "reason": "Both concern tokenizer behavior, but one is about `AutoTokenizer` ignoring `tokenizer.json` while the other is a DeepSeek-Coder v1 output regression on v5+; different failures and likely different fixes." - }, - { - "left": "issue:43611", - "right": "issue:43644", - "accept": false, - "reason": "Both are v5 loading regressions, but one is about `base_model_prefix` and the other about non-persistent buffers being filled with junk; different code paths." + "reason": "Both are tokenizer-loading regressions, but one is AutoTokenizer ignoring tokenizer.json while the other is a DeepSeek-Coder v1 output bug tied to PR #44801; different failure modes." }, { "left": "issue:42796", "right": "issue:43404", "accept": false, - "reason": "Tokenizer BPE-marker handling and Mistral3 `lm_head` weight tying are unrelated bugs." + "reason": "Ministral-3 tokenizer BPE-marker handling and Mistral3 lm_head weight tying are unrelated bugs in different code paths." }, { "left": "issue:46285", "right": "issue:46402", "accept": false, - "reason": "CLIP structure inconsistency and Sam3 text-encoder weight loading are different model-specific loading issues." + "reason": "CLIP text-model structure inconsistency and Sam3 text-encoder weight loading are different model-specific regressions." }, { "left": "issue:36032", "right": "issue:42915", "accept": false, - "reason": "Tokenizer `add_special_tokens` conflict and Qwen3Moe FP8 config failure are unrelated errors." + "reason": "T5 tokenizer method/name conflict and Qwen3Moe FineGrainedFP8Config failure are unrelated issues." }, { "left": "issue:32090", "right": "issue:39404", "accept": false, - "reason": "Trainer GPU broadcast `NoneType` error and Whisper `return_language` pipeline behavior are separate subsystems and bugs." + "reason": "Trainer GPU broadcast NoneType error and Whisper return_language pipeline regression are different bugs." }, { "left": "issue:43716", "right": "issue:45084", "accept": false, - "reason": "Mistral3 image preprocessor dtype mismatch is unrelated to a template-node compilation error." + "reason": "Mistral-3 image preprocessor dtype mismatch and template-node compilation failure are unrelated." }, { "left": "issue:44737", "right": "issue:45859", "accept": false, - "reason": "Both involve device placement, but XLNet positional encoding on CPU and Qwen rotary embedding CPU offload are different model/code-path problems." + "reason": "XLNet CPU arange device placement and Qwen3_5Moe rotary embedding CPU-offload incompatibility are distinct model/runtime problems." }, { "left": "issue:44661", "right": "issue:44683", "accept": false, - "reason": "`add-new-model-like` mapping logic and compiled `flex_attention` on torch >=2.9 are unrelated failures." + "reason": "Tokenizer-mapping failure in add-new-model-like and compiled flex_attention failure on torch 2.9 are unrelated." }, { "left": "issue:43643", "right": "issue:44530", "accept": false, - "reason": "Missing fields from `AutoConfig.from_pretrained(trust_remote_code=True)` is not the same bug as PagedAttentionCache rejecting `linear_attention` groups." + "reason": "AutoConfig trust_remote_code field loss and PagedAttentionCache linear_attention crash are different layers and bugs." }, { "left": "issue:44451", "right": "issue:44987", "accept": false, - "reason": "Both are loading regressions, but they target different models and symptoms; no clear shared concrete code defect." + "reason": "Both are model-loading failures, but they affect different repositories and appear to have different root causes." }, { "left": "issue:40990", "right": "issue:43421", "accept": false, - "reason": "High perplexity on one model and runtime tokenizer post-processor updates are not the same issue." + "reason": "High perplexity on gpt-oss-20b and runtime post-processor refresh for special tokens are not the same underlying bug." }, { "left": "issue:42915", "right": "issue:44530", "accept": false, - "reason": "Qwen3Moe FP8 config failure and PagedAttentionCache group-type crash are different model/runtime problems." + "reason": "Qwen3Moe FP8 config failure and PagedAttentionCache linear_attention crash are separate issues." }, { "left": "issue:43541", "right": "issue:45084", "accept": false, - "reason": "Torch dynamo tracing failure in Mixtral and template-node compilation failure are unrelated." + "reason": "Mixtral grouped_mm torch-dynamo tracing and non-template-node compilation errors are unrelated failures." }, { "left": "issue:43644", "right": "issue:43646", "accept": false, - "reason": "Non-persistent buffer junk and custom model initialization regressions are related to v5 but not the same concrete bug." + "reason": "Both are v5 regressions, but one is junk non-persistent buffers and the other is custom model initialization; same theme, different bug." }, { "left": "issue:43582", "right": "issue:43632", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and `_is_hf_initialized` flag breakage are distinct regressions." + "reason": "AppleSilicon caching_allocator_warmup TypeError and the _is_hf_initialized flag regression are unrelated." }, { "left": "issue:34567", "right": "issue:40444", "accept": false, - "reason": "Trainer token-count tracking and Qwen2.5-VL iterable-dataset multi-image failure are unrelated." + "reason": "TrainerState token counter not updating and Qwen2.5-VL multi-image iterable finetuning failure are different bugs." }, { "left": "issue:42175", "right": "issue:46178", "accept": false, - "reason": "Missing TensorFlow in the `.[torch]` install extra and missing `tensorflow_text` in backend mapping are packaging issues, but not the same bug." + "reason": "Missing tensorflow in a torch install and an unknown tensorflow_text backend mapping error are separate packaging/backend issues." }, { "left": "issue:43883", "right": "issue:44488", "accept": false, - "reason": "Missing `all_tied_weights_keys` on Molmo and failure to load `cjvt/sleng-bert` are separate model-loading issues." + "reason": "Missing all_tied_weights_keys and loading cjvt/sleng-bert are different problems." }, { "left": "issue:40444", "right": "issue:46326", "accept": false, - "reason": "Multi-image iterable dataset failure and `continue_final_message` on mistral_common tokenizers are different features and failures." + "reason": "Qwen2.5-VL iterable multi-image finetuning and mistral_common continue_final_message behavior are unrelated." }, { "left": "issue:44206", "right": "issue:44220", - "accept": true, - "reason": "Both point to the same audio feature-extraction path: a `center`/fbank handling regression that crashes in `_torch_extract_fbank_features()` and surfaces through `LasrFeatureExtractor`." + "accept": false, + "reason": "Both are audio-related, but one is unsupported center arg in LasrFeatureExtractor and the other is _torch_extract_fbank_features; different API layers and failures." }, { "left": "issue:44811", "right": "issue:45584", "accept": false, - "reason": "Whisper `batch_decode` skipping special tokens and empty-transcription generation after `align_special_tokens` are different decoder behaviors." + "reason": "Whisper batch_decode skip_special_tokens handling and empty-transcription failure after align_special_tokens are related only by model area, not by the same bug." }, { "left": "issue:43452", "right": "issue:43482", "accept": false, - "reason": "GGUF loading breakage for auto loaders and Qwen2.5-GGUF failures in v5 are both loader regressions, but not the same concrete defect." + "reason": "GGUF from_pretrained breakage and Qwen2.5-GGUF loading failure in v5 are both loading issues but not the same concrete defect." }, { "left": "issue:36032", "right": "issue:43645", "accept": false, - "reason": "Tokenizer special-token method conflicts and custom-model initialization in notebooks are unrelated." + "reason": "T5 tokenizer add_special_tokens conflict and notebook custom model initialization breakage are unrelated." }, { "left": "issue:44987", "right": "issue:44991", "accept": false, - "reason": "Both involve model/tokenizer loading on v5+, but the affected repos and failure modes differ and there is no clear shared root cause." + "reason": "Loading physical-intelligence/fast and loading EMBEDDIA/est-roberta are separate repository-specific failures." }, { "left": "issue:43317", "right": "issue:43482", "accept": false, - "reason": "Dequantized model offload failure and GGUF loading failure are distinct loading paths." + "reason": "Dequantized model device_map offload failure and Qwen2.5-GGUF loading failure are different loading paths and bugs." }, { "left": "issue:45084", "right": "issue:45507", "accept": false, - "reason": "Template-node compilation errors and GraniteMoEHybrid invalid method calls are different bugs." + "reason": "Non-template-node compilation error and GraniteMoEHybrid invalid method call are unrelated." }, { "left": "issue:43262", "right": "issue:43377", "accept": false, - "reason": "Audio chat-template sampling-rate defaulting and MIMI batching/padding-mask mismatch are different issues." + "reason": "Audio chat-template sampling-rate defaulting and MIMI batch padding-mask behavior are different issues." }, { "left": "issue:43746", "right": "issue:46326", "accept": false, - "reason": "PEFT adapter loading in GraniteSpeech and mistral_common `continue_final_message` behavior are unrelated." + "reason": "GraniteSpeech PEFT checkpoint loading and mistral_common continue_final_message are unrelated." }, { "left": "issue:43883", "right": "issue:44991", "accept": false, - "reason": "Missing tied-weights metadata and tokenizer loading failures for a specific model are not the same bug." + "reason": "Missing all_tied_weights_keys and tokenizer loading for EMBEDDIA/est-roberta are not the same bug." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Both are Transformers v5 regressions, but they concern different internals: custom model init versus non-persistent buffer initialization." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative overall. The only accepted soft edge, issue:44206 vs issue:44220, is a plausible same-bug audio feature-extraction duplicate based on the titles. The remaining rejections are appropriately cautious and do not overstate 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 }, @@ -3512,13 +3517,13 @@ "issue:43611", "issue:43632", "issue:43645", - "issue:43646", "issue:43742", "issue:43792", "issue:43819", "issue:43994", "issue:44008", "issue:44062", + "issue:44220", "issue:44265", "issue:44291", "issue:44403", @@ -3529,6 +3534,7 @@ "issue:44534", "issue:44589", "issue:44610", + "issue:44683", "issue:44849", "issue:44877", "issue:44987", @@ -3545,7 +3551,6 @@ "issue:46899" ], "soft_pairs": [ - "issue:43646|issue:44534", "issue:45084|issue:46869", "issue:43792|issue:46869", "issue:44403|issue:45081", @@ -3554,9 +3559,9 @@ "issue:43377|issue:43819", "issue:43404|issue:44466", "issue:44464|issue:45987", + "issue:44987|issue:45081", "issue:44008|issue:46899", "issue:43994|issue:44610", - "issue:44987|issue:45081", "issue:44451|issue:45081", "issue:28530|issue:34567", "issue:42915|issue:44291", @@ -3574,223 +3579,227 @@ "issue:43531|issue:44877", "issue:43425|issue:44403", "issue:43296|issue:43531", + "issue:44220|issue:44683", "issue:43611|issue:44534" ], - "prepared_review_unit_hash": "075524ff8a534b513eee92c62c4faae68d6adde75cc8eeb55c4a5644712149cc", + "prepared_review_unit_hash": "eff9080efc3371a2363bd51c3798ef32f2311ae5cccbd5867d3ec4504a2ad979", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36206, - "estimated_input_tokens": 9052, - "estimated_eval_tokens": 18360 + "serialized_chars": 36746, + "estimated_input_tokens": 9187, + "estimated_eval_tokens": 18630 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is a mixed bag of largely unrelated Transformers bugs. None of the soft pairs look like true duplicates or the same concrete fix, so I reject them all. If one issue must be used as the representative center, 45081 is the most concrete and actionable candidate.", - "confidence": 0.44, - "canonical_issue_reason": "Issue 45081 is the strongest representative: it has a concrete tokenizer crash, explicit traceback context, and multiple inbound references, making it the most actionable issue in the set.", + "summary": "This cluster is mostly a loose semantic group of unrelated Transformers regression reports (model loading, multimodal inputs, quantization, generation, tokenizer issues). Only one soft edge looks like a true duplicate: the Qwen3.5 `save_pretrained` regression pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45216 is the broader, earlier report about Qwen3.5 `save_pretrained` producing an incorrect checkpoint; issue:45357 reads like a narrower follow-up about one specific symptom (wrong visual encoder keys) of the same save-path bug.", "canonical_pr_reason": null, - "best_issue_reason": "45081 is the best overall tracking issue because it is specific, reproducible, and likely to anchor discussion/fixes better than the more vague or model-specific reports.", + "best_issue_reason": "issue:45216 is the best representative because it describes the general regression, while issue:45357 is a more specific manifestation of that same checkpoint-saving problem.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43646", - "right": "issue:44534", - "accept": false, - "reason": "Different bugs: custom model init in notebooks vs non-persistent buffer initialization. Same major version context, not the same underlying issue." - }, { "left": "issue:45084", "right": "issue:46869", "accept": false, - "reason": "Unrelated failures: template-node compilation error vs YOLO auto device-map multi-GPU failure." + "reason": "Different failures and code paths: template-node compilation vs YOLO multi-GPU auto device mapping." }, { "left": "issue:43792", "right": "issue:46869", "accept": false, - "reason": "Different model paths and symptoms; Whisper runtime failure is not the same bug as YOLO multi-GPU device mapping." + "reason": "Unrelated model/runtime failures (Whisper inference vs YOLO device-map auto)." }, { "left": "issue:44403", "right": "issue:45081", "accept": false, - "reason": "One is about noisy loading logs; the other is a tokenizer regex crash. Not the same code-path problem." + "reason": "Generic loading noise report vs a specific Mistral tokenizer regex crash; not the same bug." }, { "left": "issue:43645", "right": "issue:44534", "accept": false, - "reason": "Notebook custom-model initialization issue vs buffer corruption on load; related to v5 regressions, but not the same bug." + "reason": "Custom model init in notebooks vs non-persistent buffer serialization; different underlying issues." }, { "left": "issue:44849", "right": "issue:45325", "accept": false, - "reason": "Different regressions in Qwen family: hidden-states handling vs RoPE temporal \u0564\u056b\u0580 position scaling." + "reason": "Qwen3.5 hidden-states bug vs Qwen2.5-VL temporal position-id scaling; related family, different bug." }, { "left": "issue:43377", "right": "issue:43819", "accept": false, - "reason": "Both are audio-model bugs, but they affect different models and different mechanisms (padding-mask batching vs STE/latents mismatch)." + "reason": "MIMI batching/padding-mask mismatch is unrelated to DAC latent/STE behavior." }, { "left": "issue:43404", "right": "issue:44466", "accept": false, - "reason": "Both mention weight tying, but one is a model-definition bug and the other is a device-dependent serialization regression." + "reason": "One is missing weight tying in a model class; the other is serialization inconsistency of tied weights. Same area, not the same bug." }, { "left": "issue:44464", "right": "issue:45987", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is not the same as StaticCache returning the wrong type for sequence length." + "reason": "Chunked generation with compiled forward vs cache length return type; unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different failure modes: loading a specific model vs tokenizer regex patching." }, { "left": "issue:44008", "right": "issue:46899", "accept": false, - "reason": "Different multimodal model failures: variable name collision in Gemma 3n vs unsafe dtype casting in Gemma4 quantized embedders." + "reason": "Different Gemma models and different defects: variable-name collision vs unsafe dtype casting." }, { "left": "issue:43994", "right": "issue:44610", "accept": false, - "reason": "Different symptoms and model stacks; bad SigLIP2 outputs are not the same underlying problem as OmDet-Turbo input-size mismatch." - }, - { - "left": "issue:44987", - "right": "issue:45081", - "accept": false, - "reason": "Loading a specific checkpoint fails for different reasons than the Mistral tokenizer regex crash." + "reason": "Both are multimodal inference bugs, but one is SigLIP output quality and the other is processor/model input-size mismatch." }, { "left": "issue:44451", "right": "issue:45081", "accept": false, - "reason": "ScandiBERT load failure is unrelated to the Mistral regex backend-tokenizer crash." + "reason": "Different model loading issue vs Mistral tokenizer patch crash." }, { "left": "issue:28530", "right": "issue:34567", "accept": false, - "reason": "Training early-stopping metric resolution and token-count tracking are unrelated Trainer issues." + "reason": "Early stopping metric selection is unrelated to token-count tracking." }, { "left": "issue:42915", "right": "issue:44291", "accept": false, - "reason": "Different failures: FineGrainedFP8Config handling vs init_empty_weights/_is_hf_initialized argument incompatibility." + "reason": "FP8 config failure vs `_is_hf_initialized` init-empty-weights incompatibility; different problems." }, { "left": "issue:36032", "right": "issue:43742", "accept": false, - "reason": "Tokenizer special-token method conflict and MobileLLM key error are distinct loader issues." + "reason": "Tokenizer method name conflict vs model key error on load; not the same bug." }, { "left": "issue:44062", "right": "issue:44987", "accept": false, - "reason": "AddedToken constructor conflict is a tokenizer API bug, not the same as loading a particular model checkpoint." + "reason": "AddedToken constructor argument duplication is unrelated to model loading failure for a specific checkpoint." }, { "left": "issue:44265", "right": "issue:44610", "accept": false, - "reason": "torch.export compatibility with torch_compilable_check is a different problem from a processor producing the wrong image size." + "reason": "torch.export/torch_compilable_check failure is unrelated to OmDet processor image-size regression." }, { "left": "issue:42915", "right": "issue:44589", "accept": false, - "reason": "Both involve model/runtime compatibility, but one is FP8 config failure and the other is missing storage-object support in deserialization." + "reason": "Both involve FP8/quantization, but one is a model/config issue and the other is a storage deserialization issue." }, { "left": "issue:43089", "right": "issue:45987", "accept": false, - "reason": "Generation-performance overhead and cache-sequence-length return-type mismatch are not the same bug." + "reason": "Generation performance overhead vs cache-length return type bug; different code paths." }, { "left": "issue:43322", "right": "issue:43540", "accept": false, - "reason": "Different multimodal crashes: Llava Next loading segfault vs Qwen3OmniMoe video-processing ValueError." + "reason": "LLaVA model load segfault vs Qwen3OmniMoe video-input ValueError; both multimodal, not same underlying bug." }, { "left": "issue:44479", "right": "issue:45325", "accept": false, - "reason": "Both concern Qwen VL variants, but one is a video-input regression and the other is a still-image RoPE scaling bug." + "reason": "Both concern Qwen vision/video inputs, but one is a broad video-input regression and the other is a specific still-image temporal-position bug; not clearly the same defect." }, { "left": "issue:43632", "right": "issue:46821", "accept": false, - "reason": "_is_hf_initialized compatibility break and TimesFM window_size AttributeError are unrelated." + "reason": "`_is_hf_initialized` regression is unrelated to TimesFM `window_size` attribute handling." }, { "left": "issue:36032", "right": "issue:41720", "accept": false, - "reason": "Tokenizer load conflict and Qwen3 auto device-map CUDA assert are different failure classes." + "reason": "Tokenizer class conflict vs Qwen3 auto device-map CUDA assert; unrelated." }, { "left": "issue:45216", "right": "issue:45357", - "accept": false, - "reason": "These are related to Qwen3.5 save_pretrained regressions, but the reports focus on different incorrect artifacts and are not clearly the same concrete bug." + "accept": true, + "reason": "Both report the same Qwen3.5 `save_pretrained` regression; 45357 is a more specific symptom of the broken checkpoint save path." }, { "left": "issue:44991", "right": "issue:46315", "accept": false, - "reason": "Tokenizer load failure for a specific model and slow convert_tokens_to_ids performance are separate issues." + "reason": "Tokenizer loading failure vs slow `convert_tokens_to_ids` performance regression; not the same issue." }, { "left": "issue:43531", "right": "issue:44877", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and granite_speech strict-config loading are unrelated." + "reason": "Qwen3-MoE sliding-window behavior vs granite_speech strict-config loading; different bugs." }, { "left": "issue:43425", "right": "issue:44403", "accept": false, - "reason": "Torch version incompatibility and loading-noise cleanup do not describe the same problem." + "reason": "Torch version incompatibility is unrelated to generic loading noise." }, { "left": "issue:43296", "right": "issue:43531", "accept": false, - "reason": "PaddleOCR-VL load failure and Qwen3-MoE sliding_window bug are different model-specific regressions." + "reason": "PaddleOCR-VL load failure vs Qwen3-MoE sliding-window issue; different model families and failures." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Audio feature extraction bug vs compiled flex_attention torch-version failure; unrelated." }, { "left": "issue:43611", "right": "issue:44534", "accept": false, - "reason": "base_model_prefix compatibility and non-persistent buffer corruption are different v5 regressions with different code paths." + "reason": "Base-model-prefix loading breakage vs non-persistent buffer junk values; different mechanisms." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The only accepted pair, issue:45216 vs issue:45357, is a plausible duplicate/follow-up on the same Qwen3.5 save_pretrained checkpoint bug. The rejected pairs are framed narrowly enough and do not overclaim equivalence." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -3815,6 +3824,7 @@ "issue:41720", "issue:42175", "issue:42796", + "issue:43296", "issue:43317", "issue:43377", "issue:43381", @@ -3830,7 +3840,6 @@ "issue:43792", "issue:43856", "issue:44079", - "issue:44220", "issue:44291", "issue:44403", "issue:44462", @@ -3838,7 +3847,6 @@ "issue:44534", "issue:44568", "issue:44589", - "issue:44683", "issue:44756", "issue:44811", "issue:44991", @@ -3852,11 +3860,11 @@ "issue:45923", "issue:46491", "issue:46710", + "issue:46821", "issue:46899", "issue:46909" ], "soft_pairs": [ - "issue:44220|issue:44683", "issue:43716|issue:46899", "issue:40444|issue:45325", "issue:45812|issue:46491", @@ -3879,215 +3887,216 @@ "issue:43632|issue:44534", "issue:43531|issue:44291", "issue:42175|issue:45542", - "issue:43317|issue:43856", "issue:44462|issue:45812", "issue:41720|issue:45081", + "issue:43317|issue:43856", "issue:44079|issue:44534", "issue:44509|issue:46909", + "issue:43296|issue:46821", "issue:43377|issue:45491" ], - "prepared_review_unit_hash": "394466011fd6bf4651953099007f5ed85e51580bf7d7e0f1ac82c85479519cde", + "prepared_review_unit_hash": "e7b9db8ac4eb059a164ceb5cec662fc6315bc7abdbf91ed52d0a74c7ea841209", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35859, - "estimated_input_tokens": 8965, - "estimated_eval_tokens": 18186 + "serialized_chars": 35906, + "estimated_input_tokens": 8977, + "estimated_eval_tokens": 18210 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a loose collection of transformers regressions and model-specific bugs, especially v5-related tokenizer/model-loading issues, plus a few multimodal training/runtime reports. I don\u2019t see any true duplicate pair among the soft edges; they are related by broad subsystem or version, not the same concrete bug.", - "confidence": 0.61, - "canonical_issue_reason": "issue:43632 is the broadest, most central transformers-v5 regression in the set: it describes a core initialization flag breakage that can affect many model-loading paths, making it the best umbrella representative for this noisy cluster.", + "summary": "This cluster is a grab bag of mostly unrelated Transformers v5-era bugs: tokenizer regressions, multimodal/model-loading issues, config/API breakages, and a few training/runtime edge cases. The soft pairs only show surface similarity, not the same underlying defect, so none should be merged as duplicates.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 45812 is the best representative anchor: it describes a broad, high-impact tokenizer regression affecting all Granite models, with inbound references and a clear v4\u2192v5 breakage pattern.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43632 is the best representative issue because it is model-agnostic and version-wide, whereas most other reports are narrower model-specific tokenizer or multimodal failures.", + "best_issue_reason": "45812 is the strongest global issue candidate because it is cross-model, user-visible, and regression-shaped; among these issues it is the most canonical-sounding and broadly impactful.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44220", - "right": "issue:44683", - "accept": false, - "reason": "Different bugs: `_torch_extract_fbank_features()` is an audio feature extraction issue, while compiled flex_attention on torch>=2.9 is an attention compiler/runtime regression." - }, { "left": "issue:43716", "right": "issue:46899", "accept": false, - "reason": "Both mention multimodal dtype problems, but they affect different models and paths: Mistral image preprocessing vs Gemma4 audio/vision 4-bit embedding casts." + "reason": "Different bugs: Mistral-3 image preprocessor/model dtype mismatch vs Gemma4 4-bit quantization unsafe cast in multimodal embedders." }, { "left": "issue:40444", "right": "issue:45325", "accept": false, - "reason": "Qwen2.5-VL finetuning with multi-image iterable batches is a data/collation failure; the rope-index issue is an inference-time position-id scaling bug." + "reason": "One is multi-image finetuning with IterableDataset; the other is Qwen2.5-VL rope/index scaling for still-image positions." }, { "left": "issue:45812", "right": "issue:46491", "accept": false, - "reason": "These are both tokenizer regressions, but they are different concrete failures: Granite token IDs vs CodeLlama leading-space round-trip behavior." + "reason": "Granite-wide wrong token IDs regression and CodeLlama leading-space round-trip loss are distinct tokenizer bugs." }, { "left": "issue:44403", "right": "issue:44991", "accept": false, - "reason": "One is generic noisy logging on transformer load; the other is a tokenizer-loading regression for a specific repository/model." + "reason": "Loading noise vs tokenizer loading failure for a specific model; same subsystem, different failure modes and fixes." }, { "left": "issue:43381", "right": "issue:45923", "accept": false, - "reason": "Gradient checkpointing in eval mode is a training-mode misuse/guard issue, while the other is a missing capability flag on a trust_remote_code model." + "reason": "Eval-mode gradient checkpointing restriction is not the same as a missing supports_gradient_checkpointing flag on a trust_remote_code model." }, { "left": "issue:43632", "right": "issue:44756", "accept": false, - "reason": "The former is a transformers v5 initialization-flag regression; the latter is a platform-specific mmap/OOM workaround on Strix Halo." + "reason": "_is_hf_initialized compatibility break and Strix Halo mmap OOM mitigation are unrelated problems." }, { "left": "issue:42796", "right": "issue:43746", "accept": false, - "reason": "Different tokenizer/model-loading bugs: BPE marker handling in Ministral tokenizer vs PEFT-adapter local checkpoint loading in GraniteSpeech." + "reason": "Tokenizer BPE-marker handling and PEFT local-checkpoint loading are separate code paths." }, { "left": "issue:43792", "right": "issue:44589", "accept": false, - "reason": "Whisper model execution failure is not the same as the Float8 storage lookup TypeError; they involve different failure points and causes." + "reason": "Whisper runtime failure and Float8 storage lookup TypeError do not share the same root cause." }, { "left": "issue:33666", "right": "issue:40444", "accept": false, - "reason": "Both are Qwen2-VL/Qwen2.5-VL training topics, but one is generic multi-GPU training and the other is a specific multi-image iterable dataset failure." + "reason": "Multi-GPU training support and iterable-dataset multi-image failures are different training issues." }, { "left": "issue:43404", "right": "issue:46899", "accept": false, - "reason": "lm_head weight tying in Mistral3 is a model architecture bug; Gemma4 quantization dtype casting is a separate multimodal runtime issue." + "reason": "Mistral3 lm_head weight tying and Gemma4 quantized multimodal casting are unrelated model bugs." }, { "left": "issue:40990", "right": "issue:44811", "accept": false, - "reason": "High perplexity on a language model and Whisper batch_decode skipping special tokens are unrelated bugs in different models and code paths." + "reason": "High perplexity on gpt-oss-20b and Whisper batch_decode skip_special_tokens behavior are unrelated." }, { "left": "issue:44291", "right": "issue:44589", "accept": false, - "reason": "Both are loading/runtime TypeErrors, but one is an unexpected init_empty_weights argument and the other is missing Float8 storage registration." + "reason": "init_empty_weights/_is_hf_initialized argument breakage is distinct from Float8 storage resolution." }, { "left": "issue:42796", "right": "issue:46710", "accept": false, - "reason": "Tokenizer BPE-marker handling for Ministral is unrelated to an accuracy regression in DeepSeek after a Transformers upgrade." + "reason": "Tokenizer BPE marker handling and DeepSeek accuracy regression are not the same underlying issue." }, { "left": "issue:43482", "right": "issue:43525", "accept": false, - "reason": "Qwen2.5-GGUF loading failure in v5 and missing pad_token_id on Llama4Config are different model/config regressions." + "reason": "GGUF loading failure in transformers v5 and missing pad_token_id on Llama4Config are separate compatibility bugs." }, { "left": "issue:35707", "right": "issue:40444", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is an autoregressive cache bug, not the same as finetuning failures on multi-image iterable datasets." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image IterableDataset training failures." }, { "left": "issue:44079", "right": "issue:45663", "accept": false, - "reason": "ModelOutput key assignment and Gemma4 FSDP2 shared_kv_states rebuilding are separate internal-state issues with different code paths." + "reason": "ModelOutput key assignment and Gemma4 FSDP shared_kv_states rebuilding are different implementation bugs." }, { "left": "issue:41553", "right": "issue:43582", "accept": false, - "reason": "Bad AutoTokenizer error for Voxtral and a TypeError in caching_allocator_warmup on Apple Silicon are unrelated failures." + "reason": "Bad tokenizer error for Voxtral and Apple Silicon caching warmup TypeError are unrelated." }, { "left": "issue:43482", "right": "issue:45538", "accept": false, - "reason": "GGUF loading regression for Qwen2.5 and CLIPTokenizer model_max_length behavior are different tokenizer/loading issues." + "reason": "GGUF loading failure and CLIPTokenizer max_length misconfiguration are different issues." }, { "left": "issue:43421", "right": "issue:44568", "accept": false, - "reason": "Runtime special-token post-processor updates are not the same as add_special_tokens failing to inject BOS/EOS for mDeBERTa-v3 in v5." + "reason": "Runtime post-processor updates for special tokens and add_special_tokens BOS/EOS regression are separate tokenizer behaviors." }, { "left": "issue:43632", "right": "issue:44534", "accept": false, - "reason": "Both are v5 regressions, but one is init-flag handling and the other is non-persistent buffers being filled with junk; distinct internal bugs." + "reason": "_is_hf_initialized flag breakage and non-persistent buffer junk are different Transformers v5 regressions." }, { "left": "issue:43531", "right": "issue:44291", "accept": false, - "reason": "Sliding-window issues in Qwen3-MoE and unexpected _is_hf_initialized argument in init_empty_weights are unrelated and affect different paths." + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument incompatibility do not match." }, { "left": "issue:42175", "right": "issue:45542", "accept": false, - "reason": "Missing TensorFlow in the torch extra and tensorboard-installed-without-TF backend errors are packaging/setup problems, not the same bug." - }, - { - "left": "issue:43317", - "right": "issue:43856", - "accept": false, - "reason": "Dequantized model loading with device_map=auto offload is a loading/offload issue, while Qwen3 MoE training memory usage is a training efficiency problem." + "reason": "Missing TensorFlow in torch-only install and undefined tf backend from tensorboard-only install are related only at a broad packaging level, not the same bug." }, { "left": "issue:44462", "right": "issue:45812", "accept": false, - "reason": "AutoTokenizer ignoring tokenizer.json and Granite token ID regression both involve tokenizers, but they are different failure modes and models." + "reason": "Ignoring tokenizer.json and wrong token IDs for Granite models are different tokenizer-loading regressions." }, { "left": "issue:41720", "right": "issue:45081", "accept": false, - "reason": "Auto device mapping cudaErrorAssert on Qwen3 and Mistral regex patch crashes are unrelated model-loading/runtime bugs." + "reason": "Auto device mapping cudaErrorAssert and Mistral regex patch backend_tokenizer AttributeError are unrelated failures." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Dequantized model offload loading failure and Qwen3 MoE memory inefficiency are different performance/load issues." }, { "left": "issue:44079", "right": "issue:44534", "accept": false, - "reason": "Key assignment in ModelOutput and junk-filled non-persistent buffers are separate v5 regressions with different mechanics." + "reason": "ModelOutput key assignment and non-persistent buffer initialization are separate internal state bugs." }, { "left": "issue:44509", "right": "issue:46909", "accept": false, - "reason": "Removed pipeline docs vs broken DIA documentation example are both documentation issues, but they are not the same doc defect or change request." + "reason": "Removed pipeline docs references and a broken DIA docs example are both documentation issues but not the same artifact." + }, + { + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "PaddleOCR-VL load failure in vLLM and TimesFM window_size AttributeError are distinct model/runtime bugs." }, { "left": "issue:43377", "right": "issue:45491", "accept": false, - "reason": "MIMI encoder batch-vs-single mismatch from missing padding mask support is a different bug than Gemma3 NaNs from sliding-window attention/all-padding windows." + "reason": "Missing padding-mask support in MIMI encoder and Gemma3 NaN embeddings from sliding-window all-padding windows are different root causes." } ] }, @@ -4160,15 +4169,14 @@ "issue:46821" ], "soft_pairs": [ - "issue:43296|issue:46821", "issue:43611|issue:43646", - "issue:44829|issue:45479", "issue:43818|issue:46402", + "issue:44829|issue:45479", "issue:43644|issue:44079", "issue:42175|issue:43296", "issue:43818|issue:45357", - "issue:43493|issue:43994", "issue:44265|issue:45072", + "issue:43493|issue:43994", "issue:40444|issue:45381", "issue:45325|issue:45381", "issue:45325|issue:45834", @@ -4192,220 +4200,214 @@ "issue:44186|issue:44265", "issue:41720|issue:45823" ], - "prepared_review_unit_hash": "abfaf09bac5384ed88b82a469107761894c1f28a51a766fd87ede734edafa126", + "prepared_review_unit_hash": "58ffa82905bf794f62ac6f3e620bef82027f29e1ba24c880f88e4b2d6cf63884", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 36523, - "estimated_input_tokens": 9131, - "estimated_eval_tokens": 18518 + "soft_pair_count": 30, + "serialized_chars": 36224, + "estimated_input_tokens": 9056, + "estimated_eval_tokens": 18368 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: it contains many unrelated model-loading, multimodal, tokenizer, and runtime-regression bug reports. None of the soft pairs look like true duplicates or mergeable PR-equivalents, so I would not collapse any of them together.", - "confidence": 0.89, - "canonical_issue_reason": "issue:43873 is the broadest and most active bug report in the set (open, high discussion/inbound references), so it is the best anchor for this mixed cluster, even though the cluster is not a true duplicate group.", + "summary": "The cluster is mostly heterogeneous: the soft pairs share broad model families or regression themes, but not the same concrete bug/change. I would not merge any of the suggested links as duplicates.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45325 is the most concrete, actionable report in the set and the closest anchor to one of the few semantically adjacent Qwen2.5-VL position-id reports, even though the overall cluster is too mixed to treat it as a true duplicate hub.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43873 has the strongest standalone triage signal and the widest wording around offloading/quantization, making it the most useful issue to surface from this set.", + "best_issue_reason": "issue:45325 is the best issue to anchor triage because it names a specific helper-path regression with a clear symptom and is narrowly scoped compared with the broader, less directly comparable reports.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:43296", - "right": "issue:46821", - "accept": false, - "reason": "Different products and failure modes: PaddleOCR-VL load failure vs TimesFM window_size AttributeError." - }, { "left": "issue:43611", "right": "issue:43646", "accept": false, - "reason": "Both are Transformers 5 regressions, but one is base_model_prefix loading and the other is custom initialization; different underlying bugs." + "reason": "Both are Transformers 5.0 regressions, but one is loading with base_model_prefix and the other is custom initialization; different concrete failure paths." }, { - "left": "issue:44829", - "right": "issue:45479", + "left": "issue:43818", + "right": "issue:46402", "accept": false, - "reason": "Both concern degenerate training, but the triggers differ: flash_attention_3 regression vs num_labels=1 / problem_type loss bug." + "reason": "Completely different models and bugs: Video-LLaVA temporal attention/weights vs Sam3Model text encoder loading." }, { - "left": "issue:43818", - "right": "issue:46402", + "left": "issue:44829", + "right": "issue:45479", "accept": false, - "reason": "Unrelated models and code paths: Video-LLaVA temporal attention/weight sharing vs Sam3Model text encoder weights not loading." + "reason": "Unrelated symptoms: flash_attention_3 training regression vs num_labels=1 single-label loss degeneracy." }, { "left": "issue:43644", "right": "issue:44079", "accept": false, - "reason": "Different issues entirely: non-persistent buffer junk in v5 vs ModelOutput key assignment with previously None values." + "reason": "One is a 5.0 buffer serialization/loading issue; the other is ModelOutput key assignment behavior. Same framework, different bug." }, { "left": "issue:42175", "right": "issue:43296", "accept": false, - "reason": "Package/backend installation issue vs model loading failure; no shared concrete bug." + "reason": "TensorFlow extra/install packaging problem vs PaddleOCR-VL model load failure in vLLM; no shared code path." }, { "left": "issue:43818", "right": "issue:45357", "accept": false, - "reason": "Different model families and symptoms; Video-LLaVA architecture bug vs Qwen3.5 visual encoder save_pretrained regression." + "reason": "Both touch multimodal model internals, but one is Video-LLaVA temporal attention and the other is Qwen3.5 visual encoder save_pretrained keys." }, { - "left": "issue:43493", - "right": "issue:43994", + "left": "issue:44265", + "right": "issue:45072", "accept": false, - "reason": "Both are SigLIP2-related, but one is HF/JAX implementation discrepancy and the other is nonsensical AutoModel/pipeline results; not the same bug." + "reason": "torch.export with torch_compilable_check is unrelated to dtype mismatches in bfloat16 inference." }, { - "left": "issue:44265", - "right": "issue:45072", + "left": "issue:43493", + "right": "issue:43994", "accept": false, - "reason": "torch.export/torch_compilable_check failure is unrelated to bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." + "reason": "Both mention SigLIP2, but one is an implementation-vs-JAX discrepancy and the other is bad AutoModel/pipeline outputs; not clearly the same defect." }, { "left": "issue:40444", "right": "issue:45381", "accept": false, - "reason": "Different Qwen multimodal failures: iterable dataset with multiple images vs video input position-id regression." + "reason": "Qwen2.5-VL multi-image finetuning failure vs Qwen2.5-VL video position-id regression; related family, different concrete bug." }, { "left": "issue:45325", "right": "issue:45381", "accept": false, - "reason": "Both hit Qwen2.5-VL position ids, but one is still-image temporal scaling and the other is video vision_position_ids; not the same concrete bug." + "reason": "Both are Qwen2.5-VL position-id/rope issues, but one is still-image temporal scaling and the other is video vision_position_ids; not enough evidence they are the same bug." }, { "left": "issue:45325", "right": "issue:45834", "accept": false, - "reason": "Qwen2.5-VL rope/position-id regression is unrelated to Kosmos2.5 OCR long-input index errors." + "reason": "Different problems in the same model family: rope/position-index logic vs OCR long-input index error." }, { "left": "issue:43454", "right": "issue:44466", "accept": false, - "reason": "Both mention lm_head/tied weights, but one is missing weight tying causing bad generation and the other is inconsistent serialization by device." + "reason": "One is missing lm_head weight tying causing bad generation; the other is inconsistent serialization depending on device. Related area, different bug." }, { "left": "issue:42915", "right": "issue:43482", "accept": false, - "reason": "Distinct loading failures: FineGrainedFP8Config on Qwen3Moe vs Qwen2.5-GGUF with transformers v5." + "reason": "Different models and failure modes: FineGrainedFP8Config on Qwen3Moe vs GGUF loading on Qwen2.5." }, { "left": "issue:44291", "right": "issue:44877", "accept": false, - "reason": "init_empty_weights/_is_hf_initialized TypeError is not the same as strict config blocking granite_speech config loading." + "reason": "init_empty_weights/_is_hf_initialized TypeError is unrelated to strict config loading for granite_speech." }, { "left": "issue:43701", "right": "issue:45325", "accept": false, - "reason": "resume_from_checkpoint key mismatch is unrelated to Qwen2.5-VL position-id handling." + "reason": "resume_from_checkpoint key mismatch is a training checkpoint issue, not a Qwen2.5-VL position-id regression." }, { "left": "issue:43873", "right": "issue:45701", "accept": false, - "reason": "Quantization/offloading behavior is different from a general tokenization-version regression." + "reason": "Quantization/offloading behavior vs tokenization changes across versions; no concrete shared bug." }, { "left": "issue:42796", "right": "issue:45381", "accept": false, - "reason": "Tokenizer BPE-marker handling and video position-id generation are separate bugs." + "reason": "Tokenizer BPE marker handling is unrelated to Qwen2.5-VL video vision_position_ids." }, { "left": "issue:43606", "right": "issue:44265", "accept": false, - "reason": "CPU offload device mismatch in bark-small is unrelated to torch.export failures on torch_compilable_check models." + "reason": "CPU offload device mismatch in bark-small is unrelated to torch.export failures with torch_compilable_check." }, { "left": "issue:40990", "right": "issue:45584", "accept": false, - "reason": "High perplexity on WikiText-2 and Whisper empty-transcription failure after align_special_tokens are unrelated symptoms." + "reason": "High perplexity on gpt-oss-20b is unrelated to Whisper empty-transcription generation failure." }, { "left": "issue:43262", "right": "issue:43746", "accept": false, - "reason": "Audio processor chat-template sampling-rate defaulting is unrelated to PEFT adapter checkpoint loading." + "reason": "Audio processor chat-template sample-rate defaulting is unrelated to PEFT/local checkpoint loading." }, { "left": "issue:44479", "right": "issue:45823", "accept": false, - "reason": "Qwen video-input regression and Gemma4 PLE device mismatch are different models and failure modes." + "reason": "Qwen video-input regression and Gemma4 PLE device_map mismatch are different model-specific forward-path bugs." }, { "left": "issue:43881", "right": "issue:44488", "accept": false, - "reason": "glm-4v-9b loading failure is not the same as failing to load cjvt/sleng-bert." + "reason": "Different model-loading failures for unrelated models; no evidence of a shared root cause." }, { "left": "issue:35707", "right": "issue:41720", "accept": false, - "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device mapping cudaErrorAssert." + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping cudaErrorAssert." }, { "left": "issue:43828", "right": "issue:46821", "accept": false, - "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to a TimesFM window_size AttributeError." + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to TimesFM window_size AttributeError." }, { "left": "issue:34634", "right": "issue:35707", "accept": false, - "reason": "BarkProcessor voice_preset failure and progressive generation inputs_embeds/past_key_values are different bugs." + "reason": "BarkProcessor voice_preset and progressive generation with embeddings/past KV are unrelated bugs." }, { "left": "issue:39692", "right": "issue:40444", "accept": false, - "reason": "SigLIP2 doc-example errors and Qwen2.5-VL multi-image fine-tuning failures are separate issues." + "reason": "SigLIP2 doc-example errors are unrelated to Qwen2.5-VL multi-image finetuning failures." }, { "left": "issue:43454", "right": "issue:43550", "accept": false, - "reason": "AyaVision lm_head tying bug and Bamba-9B-v2 torch.compile SDPA failure do not share a concrete code-path problem." + "reason": "lm_head weight tying/serialization issues are unrelated to Bamba-9B-v2 torch.compile+SDPA failures." }, { "left": "issue:43606", "right": "issue:43819", "accept": false, - "reason": "Both are model-specific runtime bugs, but device offload mismatch and DAC.from_latents/STE mismatch are different code paths." + "reason": "CPU offload device mismatch and DAC.from_latents/STE mismatch are different code paths and symptoms." }, { "left": "issue:44186", "right": "issue:44265", "accept": false, - "reason": "LayoutLMv2Tokenizer NER/padding crash is unrelated to torch.export and torch_compilable_check." + "reason": "Tokenizer padding/truncation crashes are unrelated to torch.export with torch_compilable_check." }, { "left": "issue:41720", "right": "issue:45823", "accept": false, - "reason": "Qwen3 cudaErrorAssert on device mapping and Gemma4 PLE forward device mismatch are different failures." + "reason": "Qwen3 auto device-map assert and Gemma4 PLE device mismatch are separate forward/device-placement bugs." } ] }, @@ -4490,8 +4492,8 @@ "issue:43454|issue:43720", "issue:40990|issue:42947", "issue:42617|issue:42915", - "issue:43317|issue:43873", "issue:43749|issue:44534", + "issue:43317|issue:43873", "issue:43572|issue:45759", "issue:43526|issue:43883", "issue:43322|issue:46050", @@ -4500,225 +4502,231 @@ "issue:36331|issue:42617", "issue:44387|issue:45704", "issue:41720|issue:44977", - "issue:44466|issue:45357", "issue:36331|issue:42915", + "issue:44466|issue:45357", "issue:43450|issue:45701", + "issue:44877|issue:44991", "issue:43065|issue:46402", "issue:40990|issue:45701", - "issue:44877|issue:44991", - "issue:26413|issue:28530", "issue:44186|issue:45072", + "issue:26413|issue:28530", "issue:41720|issue:42617", "issue:43873|issue:44756" ], - "prepared_review_unit_hash": "eb4ee95898a1602b7cf8d939687649b6a0c30a69e4576d1c954c4297498de385", + "prepared_review_unit_hash": "541e3372519f8cc48f4ed5e75a599af9febdd5213a6d29fa6f3af20ea1ecf38e", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36987, - "estimated_input_tokens": 9247, - "estimated_eval_tokens": 18750 + "serialized_chars": 36984, + "estimated_input_tokens": 9246, + "estimated_eval_tokens": 18748 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated Transformers regressions across tokenization, loading/offloading, multimodal configs, and quantization. I did not find any soft pair that looks like the same underlying bug or change.", - "confidence": 0.71, - "canonical_issue_reason": "No clear canonical duplicate for the whole cluster; the issues are heterogeneous and touch different code paths.", + "summary": "This cluster is highly heterogeneous: the issues span unrelated regressions in tokenizers, model loading, quantization, offloading, training, and model-specific bugs. None of the soft pairs look like the same underlying bug/change, so all are rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No strong canonical issue: the cluster does not form a single duplicate family, and the items cover many different subsystems and failure modes.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45920 is the broadest representative because it describes a version-wide tokenizer regression affecting multiple model families, and it has broader applicability than the more model-specific reports.", + "best_issue_reason": "Issue 43873 is the most broadly framed and active representative among the set, but only as a loose cluster anchor\u2014not as a true duplicate target.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43377", "right": "issue:44610", "accept": false, - "reason": "Different models and failures: MIMI batching/padding-mask behavior vs OmDet image-size mismatch." + "reason": "Both mention batched-input behavior, but they affect different models and different code paths: MIMI encoder padding masks vs OmDet-Turbo processor image sizing." }, { "left": "issue:46305", "right": "issue:46315", "accept": false, - "reason": "Unrelated regressions: sliding-window mask off-by-one vs tokenizer lookup performance." + "reason": "Different regressions: a sliding-window mask off-by-one in streaming audio vs a tokenizer performance regression in convert_tokens_to_ids." }, { "left": "issue:43257", "right": "issue:45923", "accept": false, - "reason": "Different code paths: MoE weight conversion/loading vs missing gradient-checkpointing flag." + "reason": "Unrelated bugs: Qwen3 MoE weight conversion during accelerate/deepspeed loading vs a missing gradient-checkpointing flag on a trust_remote_code variant." }, { "left": "issue:44568", "right": "issue:45920", "accept": false, - "reason": "Both are tokenizer regressions, but one is add_special_tokens behavior for a specific tokenizer while the other is wrong token IDs across many models." + "reason": "Both are tokenizer regressions, but one is a specific BOS/EOS special-token bug while the other is wrong token IDs across several models." }, { "left": "issue:42796", "right": "issue:46209", "accept": false, - "reason": "Tokenizer BPE-marker handling and FP8 inference on B200 are unrelated bugs." + "reason": "Completely different failures: tokenizer BPE-marker handling vs FP8 inference failing on B200 hardware." }, { "left": "issue:43454", "right": "issue:45072", "accept": false, - "reason": "Weight-tying serialization bug vs bfloat16 dtype mismatch are different problems." + "reason": "Different model paths and symptoms: tied lm_head weights for AyaVision vs dtype mismatches in SwitchTransformers/TimmWrapperModel bf16 inference." }, { "left": "issue:45823", "right": "issue:46821", "accept": false, - "reason": "Gemma4 device mismatch and TimesFM window_size AttributeError do not share the same underlying bug." + "reason": "Unrelated issues: Gemma4 device placement during forward vs a TimesFM window_size AttributeError." }, { "left": "issue:43377", "right": "issue:44898", "accept": false, - "reason": "Both are batching/shape-related, but they affect different models and different mechanisms." + "reason": "Both are batching/resolution-related, but they affect different vision models and different bugs: missing padding-mask support vs Perceiver resolution handling." }, { "left": "issue:43454", "right": "issue:43720", "accept": false, - "reason": "lm_head tying/serialization is unrelated to BitNet packed-weight unpacking during accelerate loading." + "reason": "Different loading bugs in different model families: weight tying vs packed-weight unpacking during accelerate loading." }, { "left": "issue:40990", "right": "issue:42947", "accept": false, - "reason": "Perplexity regression and ineffective gradient checkpointing with LoRA are distinct issues." + "reason": "Different areas entirely: perplexity on a specific model/dataset vs gradient checkpointing behavior with PEFT LoRA." }, { "left": "issue:42617", "right": "issue:42915", "accept": false, - "reason": "3d_parallel.py failure and FineGrainedFP8Config failure are different runtime paths." + "reason": "No shared underlying defect: a 3d_parallel.py execution problem vs a Qwen3Moe FineGrainedFP8Config failure." }, { - "left": "issue:43317", - "right": "issue:43873", + "left": "issue:43749", + "right": "issue:44534", "accept": false, - "reason": "Both mention offloading/quantization, but one is a specific dequantized-model load failure and the other is a broader offloading complaint; not enough to treat as the same bug." + "reason": "Not the same bug: FSDP CPU RAM-efficient loading breakage vs non-persistent buffers being filled with junk in v5." }, { - "left": "issue:43749", - "right": "issue:44534", + "left": "issue:43317", + "right": "issue:43873", "accept": false, - "reason": "FSDP CPU RAM efficient loading and non-persistent buffer corruption are separate regressions." + "reason": "Both involve offloading/quantization, but one is a device_map load failure for dequantized models and the other is a broader quantization offloading issue." }, { "left": "issue:43572", "right": "issue:45759", "accept": false, - "reason": "Missing pad_token_idx in one config and text_config unwrapping for composite multimodal configs are different fixes." + "reason": "Different config/model-resolution problems: missing pad_token_idx in StableLmConfig vs failing to unwrap text_config for composite Qwen multimodal configs." }, { "left": "issue:43526", "right": "issue:43883", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug is unrelated to missing all_tied_weights_keys." + "reason": "Different bugs: BeitImageProcessorFast reduce_labels returning one label vs Molmo missing all_tied_weights_keys during loading." }, { "left": "issue:43322", "right": "issue:46050", "accept": false, - "reason": "Llava Next loading segfault and sinq save/load breakage are different failure modes." + "reason": "Different model/load paths: Llava Next segfault on load vs sinq quantized models breaking after save_pretrained/from_pretrained." }, { "left": "issue:41720", "right": "issue:43322", "accept": false, - "reason": "Different models and symptoms: Qwen3 device-map cuda assert vs Llava Next segfault." + "reason": "Both are model loading failures, but one is Qwen3 auto device mapping cuda assert and the other is Llava Next segfault; no concrete shared code-path." }, { "left": "issue:43881", "right": "issue:44991", "accept": false, - "reason": "glm-4v-9b load failure and tokenizer loading failure for est-roberta are unrelated." + "reason": "Different components: a glm-4v-9b loading failure vs tokenizer loading regression for EMBEDDIA/est-roberta." }, { "left": "issue:36331", "right": "issue:42617", "accept": false, - "reason": "Unexpected compute_loss keyword and 3d_parallel.py execution failure are not the same issue." + "reason": "Unrelated: CustomTrainer compute_loss signature change vs failure to run 3d_parallel.py." }, { "left": "issue:44387", "right": "issue:45704", "accept": false, - "reason": "Int4 CUDA memory regression and apex RMSNorm memory leak are separate memory issues." + "reason": "Different performance/memory issues in different subsystems: int4 quantization OOM vs an apex FusedRMSNorm memory leak." }, { "left": "issue:41720", "right": "issue:44977", "accept": false, - "reason": "Different Qwen3.5 problems: device-map assert vs flash-attention generation issue." + "reason": "Both mention Qwen, but the failures differ: device-mapping cuda assert vs flash-attention generation malfunction." }, { - "left": "issue:44466", - "right": "issue:45357", + "left": "issue:36331", + "right": "issue:42915", "accept": false, - "reason": "Serialization of tied lm_head weights and incorrect visual encoder keys are different save_pretrained regressions." + "reason": "No relation: trainer compute_loss API break vs Qwen3Moe FP8 configuration failure." }, { - "left": "issue:36331", - "right": "issue:42915", + "left": "issue:44466", + "right": "issue:45357", "accept": false, - "reason": "compute_loss signature mismatch and Qwen3Moe FineGrainedFP8Config failure are unrelated." + "reason": "Both involve save_pretrained regressions, but they affect different serialized artifacts and different model families." }, { "left": "issue:43450", "right": "issue:45701", "accept": false, - "reason": "Video processor batched shape bug and tokenizer version-dependent tokenization differences are distinct." + "reason": "Different areas: batched video processor output shape vs a tokenization behavior change across versions." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Completely different bugs: strict config blocking granite_speech loading vs tokenizer loading failure for est-roberta." }, { "left": "issue:43065", "right": "issue:46402", "accept": false, - "reason": "Dummy Conv2d in Sam3PixelDecoder and missing Sam3Model text encoder weights are different model-loading issues." + "reason": "Same broad family (Sam3) but different failures and code paths: dummy Conv2d in pixel decoder vs text encoder weights not loading." }, { "left": "issue:40990", "right": "issue:45701", "accept": false, - "reason": "High perplexity on GPT-OSS and tokenization changes across versions are not the same bug." + "reason": "Different regression classes: perplexity evaluation anomaly vs tokenization output changes between versions." }, { - "left": "issue:44877", - "right": "issue:44991", + "left": "issue:44186", + "right": "issue:45072", "accept": false, - "reason": "Strict config rejection for granite_speech and tokenizer loading failure for est-roberta are unrelated." + "reason": "Unrelated: LayoutLMv2Tokenizer crashes with NER/batched padding vs bf16 dtype mismatches in model inference." }, { - "left": "issue:44186", - "right": "issue:45072", + "left": "issue:26413", + "right": "issue:28530", "accept": false, - "reason": "LayoutLMv2 tokenizer crashes and bfloat16 dtype mismatches do not share an underlying cause." + "reason": "Different training loop issues: resume_from_checkpoint epoch iterator failure vs early stopping/metric_for_best_model disabling." }, { "left": "issue:41720", "right": "issue:42617", "accept": false, - "reason": "Qwen3 device-map assert and inability to run 3d_parallel.py are different problems." + "reason": "Different failures in different workflows: auto device mapping cuda assert vs inability to run 3d_parallel.py." }, { "left": "issue:43873", "right": "issue:44756", "accept": false, - "reason": "Quantization/offloading issues on one side and mmap OOM on Strix Halo on the other are distinct." + "reason": "Not the same bug: quantization/offloading behavior vs mmap being disabled on Strix Halo to avoid OOM." } ] }, @@ -4822,14 +4830,14 @@ "issue:43425|issue:46821", "issue:42915|issue:43322" ], - "prepared_review_unit_hash": "c690b8895385066b54a0f848b7e6417a17bd328e270562e4cbfd902c502135a0", + "prepared_review_unit_hash": "0f49742dd20ee0fa60a3c50c051a8e4a133cd0ad8af820cbc79c4479ff2ce332", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, @@ -4844,186 +4852,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is broad and mostly non-duplicate: trainer/config regressions, tokenizer bugs, multimodal loading failures, device-map/offload issues, and several generation/cache edge cases. None of the soft pairs look like the same underlying bug/change, so they should remain separate.", - "confidence": 0.83, - "canonical_issue_reason": "Issue 46858 is the closest representative of the repeated generation/cache regression theme and has the strongest conceptual overlap with nearby cache-related reports, but the cluster overall is heterogeneous.", + "summary": "This is not a true duplicate cluster: the items span unrelated training, loading, tokenizer, processor, and generation bugs. All soft edges are false positives. If a single issue must anchor the set, 46858 is the most concrete low-level generation/cache bug, but it does not represent the rest well.", + "confidence": 0.97, + "canonical_issue_reason": "46858 is the clearest, most actionable low-level bug report in the set: it has a specific cache/arithmetic failure surface and a concrete generation code path. It is still only a weak anchor because the other issues are largely unrelated.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46858 is the best single issue to anchor the cluster because it is concrete, technical, and nearer to the recurring cache/generation failures than the other candidates.", + "best_issue_reason": "46858 has the strongest problem statement and most explicit fix surface among these otherwise heterogeneous reports, so it is the best representative issue only in a very loose sense.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43550", "right": "issue:43720", "accept": false, - "reason": "Both are runtime failures around model execution, but one is torch.compile/SDPA on Bamba and the other is packed-weight unpacking during accelerate loading for BitNet; different code paths and fixes." + "reason": "Different model families and bugs: torch.compile/SDPA vs packed-weight loading." }, { "left": "issue:46612", "right": "issue:46858", "accept": false, - "reason": "Both involve generation/cache behavior, but one is beam-search cache reorder omissions for specific recurrent models while the other is cache length/device-tensor arithmetic in prefill with static cache and torch.compile." + "reason": "Both are cache/generation issues, but one is missing cache reorder support and the other is a seq-length/device-tensor arithmetic bug." }, { "left": "issue:43577", "right": "issue:43708", "accept": false, - "reason": "Blip2 dtype propagation and Trainer resume/max_steps calculation are unrelated bugs in different subsystems." + "reason": "Unrelated: dtype propagation on load vs resume_from_checkpoint max_steps." }, { "left": "issue:43295", "right": "issue:45701", "accept": false, - "reason": "Both mention regressions, but one is a processor/tokenizer API break for custom multimodal code and the other is a tokenization output change across versions; not the same failure." + "reason": "Both touch tokenization, but one is a regression in processor.tokenizer/images handling and the other is a version-dependent tokenization change." }, { "left": "issue:45685", "right": "issue:46821", "accept": false, - "reason": "A MoE MPS histogram kernel error and a TimesFM window_size AttributeError are unrelated model-specific issues." + "reason": "Different subsystems: MPS histogram on Int vs TimesFM window_size AttributeError." }, { "left": "issue:43550", "right": "issue:43606", "accept": false, - "reason": "Both are device/compile-related, but they affect different models and different failure mechanisms (SDPA compile vs CPU offload mismatch)." + "reason": "Both are device/compile-related, but the failure modes and code paths differ." }, { "left": "issue:44186", "right": "issue:44610", "accept": false, - "reason": "Tokenizer padding/truncation crashes for LayoutLMv2 vs image-size mismatch for OmDet-Turbo; different modalities and root causes." + "reason": "Both are preprocessing bugs, but one is tokenizer/NER padding and the other is image-size mismatch." }, { "left": "issue:43646", "right": "issue:46665", "accept": false, - "reason": "Both concern custom model loading/initialization, but one is a Transformers 5.0 breaking change and the other is inconsistent from_pretrained behavior inside Python classes." + "reason": "Both involve model loading, but one is custom initialization breaking in v5 and the other is from_pretrained semantics in Python classes." }, { "left": "issue:43404", "right": "issue:45357", "accept": false, - "reason": "Weight tying in Mistral3 and wrong visual encoder keys in Qwen3.5 save_pretrained are distinct multimodal/model-state bugs." + "reason": "Different models and distinct state-management bugs: weight tying vs save_pretrained key mapping." }, { "left": "issue:44734", "right": "issue:46858", "accept": false, - "reason": "Both touch cache handling, but one is KV-cache continuation indexing in serve and the other is get_seq_length returning a device tensor during prefill; different code paths." + "reason": "Both mention cache continuation, but one is wrong tensor indexing in serve and the other is static-cache slice arithmetic on non-CUDA backends." }, { "left": "issue:44610", "right": "issue:44898", "accept": false, - "reason": "Processor resolution mismatch and Perceiver interpolate_pos_encoding failure are both vision-input bugs, but not the same underlying issue." + "reason": "Both are image input-size issues, but model/processor expectations and failures are unrelated." }, { "left": "issue:42915", "right": "issue:43317", "accept": false, - "reason": "FineGrainedFP8Config failure for Qwen3Moe and device_map auto offload failure for a dequantized model are separate quantization/offload problems." + "reason": "Different loading failures: FineGrainedFP8Config vs dequantized device_map offload." }, { "left": "issue:36010", "right": "issue:42617", "accept": false, - "reason": "GenerationMixin import error and failure to run 3d_parallel.py are unrelated." + "reason": "Completely unrelated: import error vs 3d_parallel.py runtime failure." }, { "left": "issue:44423", "right": "issue:45406", "accept": false, - "reason": "Both involve serve/multimodal processor handling, but one is a string-to-device crash in continuous batching and the other is a missing _tokenizer attribute in Gemma4Processor." + "reason": "Both are serve/multimodal crashes, but different processors and attribute failures." }, { "left": "issue:44008", "right": "issue:45823", "accept": false, - "reason": "Gemma3n variable-name collision and Gemma4 PLE device mismatch are different Gemma-family bugs with different symptoms." + "reason": "Gemma-family only by name; one is a variable collision, the other is a device-mapping mismatch." }, { "left": "issue:46821", "right": "issue:46869", "accept": false, - "reason": "TimesFM window_size AttributeError and YOLO multi-GPU device_map failure are unrelated." + "reason": "Unrelated: TimesFM config attribute vs YOLO multi-GPU device mapping." }, { "left": "issue:45632", "right": "issue:45684", "accept": false, - "reason": "Cache-path collisions for trust_remote_code and read-only permissions during save_pretrained are different filesystem issues." + "reason": "Both involve save/load paths, but one is cache-path collision and the other is file permission propagation." }, { "left": "issue:44464", "right": "issue:46858", "accept": false, - "reason": "Both involve compiled/generation behavior, but one is inconsistent outputs from compiled forward and the other is a cache-length device-tensor arithmetic bug; not the same defect." + "reason": "Both involve compiled generation, but one is inconsistent outputs and the other is a seq-length tensor arithmetic bug." }, { "left": "issue:41720", "right": "issue:45161", "accept": false, - "reason": "A800 cudaErrorAssert with auto device mapping and GPT-OSS TP failure are different parallelism/device-placement issues." + "reason": "Different distributed/device-map failures on different model families." }, { "left": "issue:28530", "right": "issue:33666", "accept": false, - "reason": "Early stopping metric selection and Qwen2-VL multi-GPU training are unrelated." + "reason": "Unrelated training UX issue vs Qwen2-VL multi-GPU training request." }, { "left": "issue:44671", "right": "issue:44743", "accept": false, - "reason": "CamemBERT masked-LM prediction errors and Qwen3.5 recurrent-state reset with cache are different model-specific regressions." + "reason": "Different models and code paths; masked LM predictions vs recurrent state reset under cache." }, { "left": "issue:43531", "right": "issue:43572", "accept": false, - "reason": "Qwen3-MoE sliding_window and StableLmConfig pad_token_idx are separate configuration bugs." + "reason": "Unrelated model-specific sliding_window bug vs config default missing after 5.0 update." }, { "left": "issue:42915", "right": "issue:46291", "accept": false, - "reason": "A model-specific FP8 failure and a pipelines import dependency/version error are unrelated." + "reason": "Different failures entirely: FP8 model loading vs pipelines import dependency check." }, { "left": "issue:36331", "right": "issue:41720", "accept": false, - "reason": "Trainer compute_loss signature breakage and Qwen3 auto device mapping cudaErrorAssert are unrelated." + "reason": "Unrelated trainer API break vs CUDA assert during device mapping." }, { "left": "issue:44186", "right": "issue:44361", "accept": false, - "reason": "Different tokenizer crashes with different models and causes; not the same underlying bug." + "reason": "Both are tokenizer/task AttributeErrors, but they affect different tokenizers and inputs." }, { "left": "issue:43526", "right": "issue:43881", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels behavior and glm-4v-9b loading failure are unrelated." + "reason": "Unrelated: image processor reduce_labels bug vs glm-4v-9b loading failure." }, { "left": "issue:43316", "right": "issue:44913", "accept": false, - "reason": "Config API discrepancy for Gemma3TextConfig and GPTNeoX rotary_pct reload behavior are separate serialization/config issues." + "reason": "Both are config consistency issues, but one is Gemma3TextConfig API disparity and the other is GPTNeoX rotary_pct reload." }, { "left": "issue:43425", "right": "issue:46821", "accept": false, - "reason": "Torch 2.10 incompatibility and TimesFM window_size AttributeError are unrelated." + "reason": "Unrelated Torch version compatibility vs TimesFM window_size AttributeError." }, { "left": "issue:42915", "right": "issue:43322", "accept": false, - "reason": "Qwen3Moe FineGrainedFP8Config failure and Llava Next segmentation fault are different model/runtime bugs." + "reason": "Different bug classes: fp8 config load failure vs Llava Next segmentation fault." } ] }, @@ -5079,13 +5087,14 @@ "issue:44589", "issue:44625", "issue:44898", + "issue:44987", "issue:44991", + "issue:45042", "issue:45059", "issue:45072", "issue:45357", "issue:45406", "issue:45464", - "issue:45479", "issue:45701", "issue:45704", "issue:45759", @@ -5115,70 +5124,70 @@ "issue:43606|issue:43653", "issue:34567|issue:35707", "issue:43531|issue:46821", - "issue:43540|issue:46858", "issue:34634|issue:41720", - "issue:45759|issue:45859", + "issue:43540|issue:46858", "issue:44451|issue:44991", + "issue:45759|issue:45859", "issue:45357|issue:46402", "issue:43531|issue:44589", "issue:44077|issue:44625", "issue:43644|issue:46665", "issue:44062|issue:44451", - "issue:44625|issue:45479", - "issue:44898|issue:45072" + "issue:44898|issue:45072", + "issue:44987|issue:45042" ], - "prepared_review_unit_hash": "7334591a9103e8cd7b23af48819cb229f2c8e63b46494ee90e2925f8b7aca6bc", + "prepared_review_unit_hash": "ced30994b2e1895074275556351f15d8a97e6757c639817b25d946a9290b52c6", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { - "node_count": 47, - "item_count": 47, + "node_count": 48, + "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36304, - "estimated_input_tokens": 9076, - "estimated_eval_tokens": 18408 + "serialized_chars": 36794, + "estimated_input_tokens": 9199, + "estimated_eval_tokens": 18654 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab bag of unrelated issues across tokenizers, multimodal models, offloading, and docs. The strongest duplicate-like theme is the Transformers v5 non-persistent-buffer regression, so issue 43644 is the best representative issue here.", - "confidence": 0.36, - "canonical_issue_reason": "issue:43644 is the clearest broad regression report and the best representative of the only strong duplicate-style theme in the set (non-persistent buffer corruption in Transformers v5).", + "summary": "Loose cluster of mostly unrelated issue reports spanning tokenization, model loading, multimodal processors, offload/quantization, and training-state bugs; none of the soft pairs look like the same underlying defect.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45701 is the broadest umbrella report in this otherwise heterogeneous set and best represents the cluster\u2019s recurring version-regression/tokenization-loading theme, even though it is not a true duplicate hub.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43644 is the most general and central bug report among these issues, with wording that matches the broad regression theme better than the more model-specific reports.", + "best_issue_reason": "issue:45701 is the most representative issue artifact because it is broad and central rather than model-specific, making it the closest thing to a canonical issue for this mixed cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:28530", "right": "issue:30064", "accept": false, - "reason": "Early stopping metric selection and void segmentation-map processing are unrelated subsystems and failure modes." + "reason": "Early-stopping metric selection and void segmentation-map processing are unrelated bugs." }, { "left": "issue:33357", "right": "issue:35707", "accept": false, - "reason": "MacOS bus error with a CLIP model is unrelated to progressive generation with inputs_embeds/past_key_values." + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values affect different code paths." }, { "left": "issue:29159", "right": "issue:36032", "accept": false, - "reason": "Both are tokenizer-related, but one is slow/fast inconsistency and the other is a T5 special-tokens import conflict; different bugs." + "reason": "Both are tokenizer-related, but one is fast/slow behavioral inconsistency and the other is a T5 special-method load error." }, { "left": "issue:43540", "right": "issue:43782", "accept": false, - "reason": "Video-processing ValueError in Qwen3OmniMoe and Qwen3VL weight_only load error are different code paths and symptoms." + "reason": "Both touch Qwen multimodal loading, but video input validation and weight_only loading failure are distinct problems." }, { "left": "issue:43550", @@ -5190,151 +5199,145 @@ "left": "issue:44387", "right": "issue:45701", "accept": false, - "reason": "CUDA memory growth under int4 quantization is not the same bug as tokenization changes across versions." + "reason": "int4 memory growth under quantization is a runtime/memory bug, not the tokenization change reported in 45701." }, { "left": "issue:43873", "right": "issue:45704", "accept": false, - "reason": "Quantization/offloading behavior and a T5 apex RMSNorm memory leak are distinct problems." + "reason": "Quantization offloading and a T5 RMSNorm memory leak are different failure modes." }, { "left": "issue:30064", "right": "issue:34634", "accept": false, - "reason": "Segmentation-map processing and Bark voice_preset handling are unrelated model/processor issues." + "reason": "Image processor segmentation-map handling and Bark voice_preset loading are unrelated." }, { "left": "issue:44492", "right": "issue:44509", "accept": false, - "reason": "A cache-strategy typo is not the same underlying issue as stale docs mentioning removed pipeline tasks." + "reason": "A cache-strategy typo and removed pipeline-task docs are both docs-ish, but not the same change or bug." }, { "left": "issue:43606", "right": "issue:43720", "accept": false, - "reason": "CPU offload device mismatch in Bark-small and BitNet packed-weight unloading are different accelerate/loading bugs." + "reason": "CPU offload device mismatch in Bark is separate from BitNet packed-weight unpacking during accelerate loading." }, { "left": "issue:44534", "right": "issue:46665", "accept": false, - "reason": "Non-persistent buffer junk in v5 and inconsistent from_pretrained behavior inside Python classes are not the same code-path issue." + "reason": "Non-persistent buffer initialization and from_pretrained behavior inside Python classes are different loading issues." }, { "left": "issue:43531", "right": "issue:45759", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and unwrapping text_config in composite Qwen multimodal configs are distinct issues." + "reason": "Qwen3-MoE sliding_window behavior and composite Qwen text_config unwrapping are different config bugs." }, { "left": "issue:41720", "right": "issue:43746", "accept": false, - "reason": "Qwen3 auto device-mapping CUDA assert and GraniteSpeech PEFT local checkpoint loading are unrelated." + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to GraniteSpeech PEFT checkpoint loading." }, { "left": "issue:42490", "right": "issue:43329", "accept": false, - "reason": "Trailing-slash import warning and undefined video-token helper variables are different loading/path vs multimodal-tokenization bugs." + "reason": "Trailing-slash local model path warnings and undefined multimodal token counting helpers are unrelated." }, { "left": "issue:43388", "right": "issue:45059", "accept": false, - "reason": "Metric gathering drops tuple labels in the last batch, which is unrelated to SAM3 text+box behavior." + "reason": "Batch gathering label truncation and SAM3 text/box behavior are different evaluation vs inference bugs." }, { "left": "issue:45406", "right": "issue:45464", "accept": false, - "reason": "Missing _tokenizer in Gemma4Processor and streaming inference failure for Qwen3.5 are different failures." + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming chat/completions failures are separate service/runtime issues." }, { "left": "issue:43606", "right": "issue:43653", "accept": false, - "reason": "CPU offload device mismatch and tokenizer special-token registration are not the same underlying problem." + "reason": "Offload device mismatch in a model is not the same as tokenizer special-token registration." }, { "left": "issue:34567", "right": "issue:35707", "accept": false, - "reason": "TrainerState token-count tracking and progressive generation with past_key_values are unrelated." + "reason": "Trainer token-count bookkeeping and generation with cached key values are unrelated training/runtime bugs." }, { "left": "issue:43531", "right": "issue:46821", "accept": false, - "reason": "These both mention windowing, but one is Qwen3-MoE sliding window and the other is a TimesFM window_size AttributeError; different models and bugs." - }, - { - "left": "issue:43540", - "right": "issue:46858", - "accept": false, - "reason": "Video-input processing in Qwen3OmniMoe and static-cache slice arithmetic on non-CUDA backends are different issues." + "reason": "Qwen3-MoE sliding_window and TimesFM window_size AttributeError do not share a code path." }, { "left": "issue:34634", "right": "issue:41720", "accept": false, - "reason": "BarkProcessor voice_preset and Qwen3 auto device mapping failures are unrelated." + "reason": "Bark voice_preset loading and Qwen3 device-mapping failures are unrelated model-specific bugs." }, { - "left": "issue:45759", - "right": "issue:45859", + "left": "issue:43540", + "right": "issue:46858", "accept": false, - "reason": "Unwrapping text_config in composite multimodal configs and CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding are different fixes." + "reason": "Qwen3OmniMoe video-input processing and cache.get_seq_length tensor arithmetic are different failures." }, { "left": "issue:44451", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer/loading regressions, but they affect different models and appear to stem from different serialization/tokenizer issues." + "reason": "Both are loading regressions for specific models, but the reported tokenizer/model failures are not the same underlying bug." }, { - "left": "issue:45357", - "right": "issue:46402", + "left": "issue:45759", + "right": "issue:45859", "accept": false, - "reason": "Incorrect visual encoder keys on save_pretrained and missing text-encoder weights on load are not the same bug." + "reason": "Composite Qwen config unwrapping and Qwen3_5Moe rotary embedding CPU-offload compatibility are distinct issues." }, { - "left": "issue:43531", - "right": "issue:44589", + "left": "issue:45357", + "right": "issue:46402", "accept": false, - "reason": "Qwen3-MoE sliding window and Float8 storage lookup failure are unrelated." + "reason": "Incorrect visual encoder save keys and missing text encoder load weights are separate save/load problems." }, { "left": "issue:44077", "right": "issue:44625", "accept": false, - "reason": "patchtsmixer post_init allowance and Qwen3.5 num_labels propagation are different config-validation issues." + "reason": "PatchTSMixer post_init policy and Qwen3.5 num_labels propagation are unrelated model-config bugs." }, { "left": "issue:43644", "right": "issue:46665", "accept": false, - "reason": "The non-persistent-buffer junk regression is not the same as inconsistent from_pretrained behavior inside Python classes, even though both touch loading." + "reason": "Non-persistent buffer junk values and inconsistent from_pretrained-in-class behavior are different initialization/loading defects." }, { "left": "issue:44062", "right": "issue:44451", "accept": false, - "reason": "AddedToken special-argument duplication and ScandiBERT loading failure are different tokenizer errors." + "reason": "AddedToken special-argument conflict is a tokenizer constructor bug, not the ScandiBERT loading failure." }, { - "left": "issue:44625", - "right": "issue:45479", + "left": "issue:44898", + "right": "issue:45072", "accept": false, - "reason": "num_labels propagation in Qwen3.5 is not the same underlying cause as degenerate zero-loss single-label classification across models." + "reason": "Perceiver interpolation at non-default resolution and bfloat16 dtype mismatches in inference are unrelated." }, { - "left": "issue:44898", - "right": "issue:45072", + "left": "issue:44987", + "right": "issue:45042", "accept": false, - "reason": "Perceiver image-resolution failure and dtype mismatches in bfloat16 inference are unrelated." + "reason": "A model-loading failure and PIL image processors requiring torchvision are separate v5 regressions." } ] }, @@ -5381,17 +5384,17 @@ "issue:44514", "issue:44545", "issue:44610", + "issue:44625", "issue:44671", "issue:44843", "issue:44918", "issue:44936", - "issue:44987", "issue:44995", - "issue:45042", "issue:45081", "issue:45356", "issue:45357", "issue:45381", + "issue:45479", "issue:45704", "issue:45759", "issue:45834", @@ -5408,14 +5411,14 @@ "issue:46858" ], "soft_pairs": [ - "issue:44987|issue:45042", + "issue:44625|issue:45479", "issue:43381|issue:43844", "issue:44368|issue:45357", "issue:39692|issue:44610", "issue:46655|issue:46693", "issue:43421|issue:43927", - "issue:43746|issue:46050", "issue:41720|issue:45704", + "issue:43746|issue:46050", "issue:43381|issue:45704", "issue:42886|issue:44843", "issue:45704|issue:45923", @@ -5423,14 +5426,14 @@ "issue:44514|issue:44545", "issue:45759|issue:45923", "issue:43746|issue:43873", - "issue:43994|issue:44671", "issue:43421|issue:43653", + "issue:43994|issue:44671", "issue:39692|issue:43493", "issue:46032|issue:46463", "issue:43421|issue:44186", "issue:43782|issue:45381", - "issue:45081|issue:45356", "issue:44995|issue:46858", + "issue:45081|issue:45356", "issue:30064|issue:36010", "issue:45356|issue:45907", "issue:39692|issue:43873", @@ -5440,226 +5443,226 @@ "issue:43653|issue:43720", "issue:43295|issue:45834" ], - "prepared_review_unit_hash": "1da9f207072d1ec119832421fe5bcede2d70afacf005244710612af335ebdc41", + "prepared_review_unit_hash": "894dfd93e6fb41971280b3a19c1e385e2a4503280e9eb5fa7023fcdd4a341460", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 37484, - "estimated_input_tokens": 9371, - "estimated_eval_tokens": 18998 + "serialized_chars": 37562, + "estimated_input_tokens": 9391, + "estimated_eval_tokens": 19038 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a collection of unrelated Transformers issues. I found one true duplicate pair: the Qwen2_5_VLProcessor batched `padding=False` crash (44514/44545). All other soft-similarity candidates look like neighboring subsystem bugs, not the same underlying issue.", - "confidence": 0.94, - "canonical_issue_reason": "Issue 44514 is the best canonical anchor for the only clear duplicate pair: it is the earlier report, and 44545 describes the same `Qwen2_5_VLProcessor.apply_chat_template` batched `padding=False` crash.", + "summary": "This cluster is mostly a grab-bag of unrelated Hugging Face issues; the only clear duplicate pair is the Qwen2_5_VLProcessor batched padding=False crash. Most other soft links are broad topical similarities, not the same concrete bug.", + "confidence": 0.91, + "canonical_issue_reason": "issue 44514 is the best canonical issue: it is the earlier, more detailed report and is an exact duplicate of 44545.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44514 is the strongest overall issue candidate in this set because it is a clean, concrete bug report and has a near-exact duplicate in 44545; the rest are mostly distinct bugs across different code paths.", + "best_issue_reason": "issue 44514 is the strongest cluster representative because it cleanly captures the only confirmed duplicate theme and has the fuller problem description.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44987", - "right": "issue:45042", + "left": "issue:44625", + "right": "issue:45479", "accept": false, - "reason": "Different failures: one is a model-loading regression for physical-intelligence/fast, the other is a PIL backend torchvision dependency issue." + "reason": "Different sequence-classification bugs: one is config propagation, the other is a single-label zero-loss regression." }, { "left": "issue:43381", "right": "issue:43844", "accept": false, - "reason": "Both mention training mechanics, but one is eval-mode gradient checkpointing and the other is ZeRO-3 gradient explosion; not the same bug." + "reason": "Both mention gradients/training behavior, but the concrete failure modes and code paths are different." }, { "left": "issue:44368", "right": "issue:45357", "accept": false, - "reason": "Both involve Qwen3.5 warnings/regressions, but one is a LoRA tie_word_embeddings warning and the other is incorrect visual encoder keys on save_pretrained." + "reason": "One is a warning about tie_word_embeddings, the other is a save_pretrained key regression for a multimodal model." }, { "left": "issue:39692", "right": "issue:44610", "accept": false, - "reason": "Both are multimodal processor/model mismatches, but they affect different models and different concrete bugs (SigLIP2 docs vs OmDet-Turbo image size)." + "reason": "Different models and failures: documentation/example mismatch vs processor/model input-size mismatch." }, { "left": "issue:46655", "right": "issue:46693", "accept": false, - "reason": "Both are performance/processing complaints, but one is a video padding calculation bug and the other is an FA2 long-context performance regression." + "reason": "Video padding bug versus FA2 long-context performance regression; not the same underlying change." }, { "left": "issue:43421", "right": "issue:43927", "accept": false, - "reason": "Tokenizer-related, but one is runtime special-token post-processing and the other is save/load custom token IDs causing generation IndexError." + "reason": "Both involve tokenizer internals, but one is runtime post-processor updates and the other is save/load config corruption." }, { - "left": "issue:43746", - "right": "issue:46050", + "left": "issue:41720", + "right": "issue:45704", "accept": false, - "reason": "Both involve model loading/quantization, but one is PEFT adapter local checkpoint loading and the other is sinq quantized save/load breakage." + "reason": "Unrelated issues: Qwen3 device-mapping cuda assert vs T5 RMSNorm memory leak." }, { - "left": "issue:41720", - "right": "issue:45704", + "left": "issue:43746", + "right": "issue:46050", "accept": false, - "reason": "Different subsystems and symptoms: Qwen3 auto device mapping cuda assert vs T5 silently using apex FusedRMSNorm." + "reason": "Local checkpoint loading with PEFT adapters is a different path from quantized sinq save/load breakage." }, { "left": "issue:43381", "right": "issue:45704", "accept": false, - "reason": "Unrelated bugs: gradient checkpointing in eval mode vs a T5/apex memory leak." + "reason": "Eval-mode gradient checkpointing and T5 apex leak are separate defects." }, { "left": "issue:42886", "right": "issue:44843", "accept": false, - "reason": "Both are offline/tokenizer loading related, but one is cache loading under HF_HUB_OFFLINE and the other is an unconditional model_info() call in `_patch_mistral_regex`." + "reason": "Both are offline-related tokenizer problems, but 44843 is specifically a _patch_mistral_regex model_info call and 42886 is broader cache-loading failure." }, { "left": "issue:45704", "right": "issue:45923", "accept": false, - "reason": "Different problems: T5/apex memory leak vs missing supports_gradient_checkpointing on a trust_remote_code model." + "reason": "T5 apex memory leak and missing supports_gradient_checkpointing are unrelated." }, { "left": "issue:44918", "right": "issue:44936", "accept": false, - "reason": "Different failure modes: TRL SFT input embedding unpacking vs trainer.evaluate() after trainer.train()." + "reason": "One is an embedding-unpacking failure in TRL SFT, the other is trainer.evaluate breaking after trainer.train()." }, { "left": "issue:44514", "right": "issue:44545", "accept": true, - "reason": "Same underlying bug: identical `Qwen2_5_VLProcessor.apply_chat_template` batched `padding=False` crash, with the later report effectively duplicating the earlier one." + "reason": "Same exact crash: Qwen2_5_VLProcessor.apply_chat_template fails on batched input when padding=False." }, { "left": "issue:45759", "right": "issue:45923", "accept": false, - "reason": "Composite config handling vs checkpointing flag propagation; these are distinct model-loading issues." + "reason": "Different multimodal/config issue versus a missing gradient-checkpointing flag." }, { "left": "issue:43746", "right": "issue:43873", "accept": false, - "reason": "Both mention loading/offloading, but one is PEFT local checkpoint loading and the other is quantization/offloading behavior." + "reason": "Local PEFT checkpoint loading and quantization/offloading behavior are not the same bug." }, { - "left": "issue:43994", - "right": "issue:44671", + "left": "issue:43421", + "right": "issue:43653", "accept": false, - "reason": "Both are model output quality regressions, but for different models and different causes (SigLIP2 inference mismatch vs CamemBERT MLM predictions)." + "reason": "Related to special tokens, but one is a backend post-processor update issue and the other is a BigBirdTokenizer registration bug." }, { - "left": "issue:43421", - "right": "issue:43653", + "left": "issue:43994", + "right": "issue:44671", "accept": false, - "reason": "Tokenizer special-token handling is the broad area, but these are different bugs: post-processor update vs BigBird mask token registration." + "reason": "Different model families and symptoms: SigLIP2 inference weirdness vs CamemBERT MLM regression." }, { "left": "issue:39692", "right": "issue:43493", "accept": false, - "reason": "Same model family (SigLIP2) but not the same bug: docs/example mismatch and quantization failure vs HF/JAX implementation discrepancy." + "reason": "Both mention SigLIP2, but one is docs/example errors and the other is an implementation discrepancy against JAX." }, { "left": "issue:46032", "right": "issue:46463", "accept": false, - "reason": "Both are runtime correctness issues, but Mamba2 cache/use_cache corruption and T5 decoder input validation error message are unrelated." + "reason": "Cache/use_with_seq_len bug and a T5 error-message bug are unrelated." }, { "left": "issue:43421", "right": "issue:44186", "accept": false, - "reason": "Tokenizer issues in different code paths: runtime post-processor updates vs LayoutLMv2 NER/batched padding crashes." + "reason": "Special-token post-processor updates versus LayoutLMv2 tokenizer crashes are different tokenizer bugs." }, { "left": "issue:43782", "right": "issue:45381", "accept": false, - "reason": "Both are Qwen/VL related, but one is from_pretrained weight_only loading and the other is video input vision_position_ids in qwen2.5-vl." + "reason": "Different Qwen multimodal failures: weight_only loading error versus incorrect video vision_position_ids." }, { - "left": "issue:45081", - "right": "issue:45356", + "left": "issue:44995", + "right": "issue:46858", "accept": false, - "reason": "Both involve Mistral-regex-related tokenizer handling, but one is an AttributeError crash and the other is a Kimi-K2.5 codec/warning regression; not the same concrete bug." + "reason": "Both involve cache, but the affected models and failure mechanisms are different." }, { - "left": "issue:44995", - "right": "issue:46858", + "left": "issue:45081", + "right": "issue:45356", "accept": false, - "reason": "Different cache bugs in different code paths: stale indexer cache on second forward vs `_prefill` static-cache sequence length tensor arithmetic." + "reason": "Both touch fix_mistral_regex, but one is an AttributeError in the patch function and the other is a tokenizer regression with codec handling." }, { "left": "issue:30064", "right": "issue:36010", "accept": false, - "reason": "Completely unrelated: image processor void segmentation maps vs an ImportError for `GenerationMixin`." + "reason": "Completely unrelated: image segmentation map processing versus a GenerationMixin import error." }, { "left": "issue:45356", "right": "issue:45907", "accept": false, - "reason": "Tokenizer regression vs repository file listing API forwarding; different modules and symptoms." + "reason": "Tokenizer regression and repository listing API forwarding are unrelated." }, { "left": "issue:39692", "right": "issue:43873", "accept": false, - "reason": "Both touch quantization-ish workflows, but one is SigLIP2 example mismatch and the other is offloading behavior with quantization." + "reason": "SigLIP2 docs/example errors are unrelated to quantization offloading behavior." }, { "left": "issue:43720", "right": "issue:43819", "accept": false, - "reason": "Different model internals: BitNet packed-weight loading vs DAC `from_latents` / missing STE mismatch." + "reason": "BitNet accelerate-loading failure and DAC from_latents/STE mismatch are different code paths." }, { "left": "issue:45987", "right": "issue:46829", "accept": false, - "reason": "Cache length type bug vs video classification output field mismatch; unrelated code paths." + "reason": "StaticCache seq-length typing bug and VideoPrism output-field bug are unrelated." }, { "left": "issue:46829", "right": "issue:46832", "accept": false, - "reason": "Different models and bugs: output object field mismatch in VideoPrism vs RTDetr forward failure when feature levels exceed backbone outputs." + "reason": "Different model implementations and failure modes; no shared concrete bug." }, { "left": "issue:43653", "right": "issue:43720", "accept": false, - "reason": "Both are CI/runtime model issues, but BigBird mask-token registration and BitNet packed-weight unpacking are unrelated." + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight loading are unrelated." }, { "left": "issue:43295", "right": "issue:45834", "accept": false, - "reason": "Both are processor/model loading regressions, but one is processor.tokenizer/image handling in v4.57.5 and the other is Kosmos2.5 long OCR index error." + "reason": "Custom checkpoint loading with processor/tokenizer access and Kosmos2.5 OCR index errors are different bugs." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it identifies the near-exact 44514/44545 duplicate pair and treats the remaining soft matches as distinct bugs without over-claiming shared root causes. No PR-pair mergeability issue is present here." + "feedback": "The cluster summary is well grounded in the packet: 44514/44545 is an exact duplicate pair by title, and the canonical choice for 44514 is conservative and supported by created_at/body length. The rest of the soft-edge rejections stay cautious and are consistent with the titles, without overstating duplicate evidence." }, "evaluator_used": true, "retried": false, @@ -5730,8 +5733,8 @@ ], "soft_pairs": [ "issue:43550|issue:44361", - "issue:44062|issue:44488", "issue:43550|issue:43927", + "issue:44062|issue:44488", "issue:43596|issue:44220", "issue:45381|issue:46693", "issue:43540|issue:46294", @@ -5758,20 +5761,20 @@ "issue:45183|issue:45479", "issue:45584|issue:45834" ], - "prepared_review_unit_hash": "40eeecea5cdf190e6032104e6eae2cb331ac30701e9195398e6a7aaf6d4858d0", + "prepared_review_unit_hash": "4d7e57b94e0a09dede32c7b6abae0c620240f17d8b8dba413f44c221721559f3", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35676, + "serialized_chars": 35675, "estimated_input_tokens": 8919, "estimated_eval_tokens": 18094 }, @@ -5780,180 +5783,180 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is highly heterogeneous: the issues span unrelated tokenizer, model-loading, training, video/audio, config, and performance bugs. None of the soft-pair candidates appear to be the same underlying bug, so all soft edges should be rejected.", - "confidence": 0.98, - "canonical_issue_reason": "No single canonical issue fits this cluster; the items are not duplicates of one another and cover many unrelated code paths and models.", + "summary": "The set is highly heterogeneous, with no clear duplicate cluster and no PRs present. All soft edges are rejected as they point to different concrete bugs or failure modes, even when they touch similar subsystems.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "No best issue for the cluster overall, because there is no shared root problem tying these issues together.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43550", "right": "issue:44361", "accept": false, - "reason": "Different failures and subsystems: Bamba-9B-v2 torch.compile/SDPA vs MLukeTokenizer AttributeError on tasks." + "reason": "Different models and failures: Bamba torch.compile/SDPA vs MLuke tokenizer AttributeError." }, { - "left": "issue:44062", - "right": "issue:44488", + "left": "issue:43550", + "right": "issue:43927", "accept": false, - "reason": "Both are tokenizer-related, but one is an AddedToken construction error and the other is a model/tokenizer load failure for a specific checkpoint." + "reason": "Different code paths: torch.compile/SDPA crash vs DiaConfig save/load token ID loss." }, { - "left": "issue:43550", - "right": "issue:43927", + "left": "issue:44062", + "right": "issue:44488", "accept": false, - "reason": "Unrelated bugs: torch.compile with SDPA vs DiaConfig save/load losing custom token IDs." + "reason": "AddedToken special-arg duplication is unrelated to loading EMBEDDIA/est-roberta tokenizer failure." }, { "left": "issue:43596", "right": "issue:44220", "accept": false, - "reason": "Different code paths and model families: Deepspeed zero3 init/BertModel indexing vs _torch_extract_fbank_features." + "reason": "Deepspeed ZeRO3/BertModel index error is unrelated to `_torch_extract_fbank_features()`." }, { "left": "issue:45381", "right": "issue:46693", "accept": false, - "reason": "Distinct problems: qwen2.5-vl video position IDs vs FA2 long-context performance regression." + "reason": "Qwen2.5-VL video position IDs bug is unrelated to FA2 long-context performance regression." }, { "left": "issue:43540", "right": "issue:46294", "accept": false, - "reason": "Both involve video/Qwen, but one is Qwen3OmniMoe input processing and the other is Qwen3.5-9B chat-template token/feature mismatch." + "reason": "Both are video-related Qwen bugs, but one is a Qwen3OmniMoe processing ValueError and the other is a chat-template token/feature mismatch." }, { "left": "issue:44008", "right": "issue:46858", "accept": false, - "reason": "Completely different bugs: Gemma3n variable-name collision in forward vs cache.get_seq_length() returning a device tensor in static cache." + "reason": "Gemma3n variable-name collision is unrelated to `_prefill` cache seq-length device tensor arithmetic." }, { "left": "issue:43531", "right": "issue:43643", "accept": false, - "reason": "Different areas: Qwen3-MoE sliding_window behavior vs trust_remote_code AutoConfig returning missing fields." + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to missing fields from `trust_remote_code=True` config loading." }, { "left": "issue:43611", "right": "issue:46665", "accept": false, - "reason": "Both concern loading, but one is a Transformers 5 base_model_prefix regression and the other is inconsistent from_pretrained behavior inside Python classes." + "reason": "Transformers 5 base_model_prefix breakage is unrelated to inconsistent `from_pretrained` behavior inside Python classes." }, { "left": "issue:43653", "right": "issue:43819", "accept": false, - "reason": "Tokenizer special-token registration bug vs DAC.from_latents not matching forward pass are unrelated." + "reason": "BigBirdTokenizer special-token registration is unrelated to DAC `from_latents` vs forward mismatch." }, { "left": "issue:44220", "right": "issue:44661", "accept": false, - "reason": "Different failures: fbank feature extraction issue vs add-new-model-like failing due to TOKENIZER_MAPPING_NAMES placement." + "reason": "Audio fbank extraction bug is unrelated to `add-new-model-like` failing on `TOKENIZER_MAPPING_NAMES`." }, { "left": "issue:44186", "right": "issue:44898", "accept": false, - "reason": "Different models and symptoms: LayoutLMv2Tokenizer NER/padding crash vs Perceiver non-default resolution classification failure." + "reason": "LayoutLMv2 tokenizer crashes on NER padding/truncation are unrelated to Perceiver image classification with non-default resolution." }, { "left": "issue:43618", "right": "issue:43697", "accept": false, - "reason": "Both are output discrepancies, but they affect different models and different outputs; not the same concrete bug." + "reason": "CLIPOutput attentions regression is unrelated to RTDetr producing different outputs in v5." }, { "left": "issue:41720", "right": "issue:44991", "accept": false, - "reason": "Unrelated: Qwen3 auto device mapping cudaErrorAssert vs tokenizer loading failure for EMBEDDIA/est-roberta." + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to loading the EMBEDDIA/est-roberta tokenizer." }, { "left": "issue:30064", "right": "issue:36331", "accept": false, - "reason": "Different domains entirely: image processor void segmentation maps vs CustomTrainer.compute_loss signature mismatch." + "reason": "Void segmentation-map processing is unrelated to `compute_loss(num_items_in_batch)` signature mismatch." }, { "left": "issue:43643", "right": "issue:44877", "accept": false, - "reason": "Both are config-loading related, but one is missing fields under trust_remote_code and the other is strict config blocking granite_speech loading." + "reason": "Both are config-loading issues, but one is missing fields from `trust_remote_code` while the other is strict config rejection for `granite_speech`." }, { "left": "issue:43550", "right": "issue:43819", "accept": false, - "reason": "Different models and failure modes: torch.compile with SDPA vs DAC latent decoding mismatch." + "reason": "Bamba torch.compile/SDPA crash and DAC `from_latents` mismatch are unrelated bugs." }, { "left": "issue:43335", "right": "issue:43756", "accept": false, - "reason": "Separate model/config bugs: SwitchTransformers sparse-layer creation vs Smollm3 RoPE layer dropping." + "reason": "Sparse-layer creation with zero encoder layers is unrelated to Smollm3 RoPE layer dropping." }, { "left": "issue:44843", "right": "issue:45081", "accept": false, - "reason": "Same helper name, but different bugs: unconditional model_info() call in offline mode vs backend_tokenizer AttributeError in regex patching." + "reason": "Same helper name, but different root causes: offline `model_info()` call vs missing `backend_tokenizer` attribute in `_patch_mistral_regex`." }, { "left": "issue:44821", "right": "issue:45084", "accept": false, - "reason": "Unrelated issues: loading AutoImageProcessor from URL vs TypeError compiling non-template nodes." + "reason": "AutoImageProcessor URL loading is unrelated to Jinja/template compilation errors." }, { "left": "issue:44186", "right": "issue:44442", "accept": false, - "reason": "Different tokenizer problems: LayoutLMv2 NER/padding crash vs AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + "reason": "LayoutLMv2 NER/padding crash is unrelated to AutoTokenizer failing to load FastSpeech2ConformerTokenizer." }, { "left": "issue:39692", "right": "issue:45834", "accept": false, - "reason": "Docs-example errors for SigLIP2 vs long-OCR index error in Kosmos2.5 are not the same bug." + "reason": "SigLIP2 doc example errors are unrelated to Kosmos2.5 long OCR index errors." }, { "left": "issue:44448", "right": "issue:44671", "accept": false, - "reason": "Different model regressions: Pegasus v4/v5 output difference vs CamemBERT masked LM predictions." + "reason": "Pegasus v4/v5 output drift is unrelated to CamemBERT masked LM prediction changes." }, { "left": "issue:43540", "right": "issue:44008", "accept": false, - "reason": "Different multimodal code paths: video input processing in Qwen3OmniMoe vs Gemma3n audio mask attribute collision." + "reason": "Qwen3OmniMoe video processing ValueError is unrelated to Gemma3n audio_mel_mask variable collision." }, { "left": "issue:43644", "right": "issue:43749", "accept": false, - "reason": "Both involve v5/loading behavior, but non-persistent buffer junk and FSDP CPU RAM efficient loading are separate bugs." + "reason": "Non-persistent buffer initialization junk is unrelated to broken FSDP CPU RAM-efficient loading." }, { "left": "issue:34634", "right": "issue:36010", "accept": false, - "reason": "Unrelated: BarkProcessor voice_preset issue vs GenerationMixin import error." + "reason": "Bark voice_preset and GenerationMixin import errors are different components and failures." }, { "left": "issue:45183", "right": "issue:45479", "accept": false, - "reason": "Different issues: validation error messaging for input_ids/inputs_embeds vs degenerate zero loss with num_labels=1." + "reason": "Input/inputs_embeds validation messaging bug is unrelated to degenerate single-label classification loss." }, { "left": "issue:45584", "right": "issue:45834", "accept": false, - "reason": "Whisper empty transcription after align_special_tokens and Kosmos2.5 long OCR index error are unrelated." + "reason": "Whisper empty transcription after special-token alignment is unrelated to Kosmos2.5 long OCR index errors." } ] }, @@ -6029,222 +6032,222 @@ "soft_pairs": [ "issue:43606|issue:45072", "issue:43322|issue:45685", - "issue:43825|issue:43994", "issue:43550|issue:45072", "issue:43388|issue:46032", "issue:44479|issue:45834", + "issue:43825|issue:43994", "issue:43782|issue:44977", - "issue:43577|issue:43818", "issue:43582|issue:46512", + "issue:43577|issue:43818", "issue:44991|issue:46291", "issue:43421|issue:44811", "issue:44561|issue:45020", - "issue:43716|issue:44008", "issue:43950|issue:45987", + "issue:43716|issue:44008", "issue:43931|issue:46846", "issue:46294|issue:46463", "issue:43825|issue:43827", - "issue:44805|issue:44936", "issue:28218|issue:29159", + "issue:44805|issue:44936", "issue:44610|issue:45072", "issue:43540|issue:45381", "issue:43720|issue:45072", "issue:43749|issue:44291", - "issue:43645|issue:46665", "issue:43577|issue:43582", + "issue:43645|issue:46665", "issue:44671|issue:45701", "issue:42175|issue:45081" ], - "prepared_review_unit_hash": "2d560910181fa43342b8f588ee1a65967c5ff37618e39421ab646a243989b246", + "prepared_review_unit_hash": "cb1855bd669dc602a989e98a009aeefe9d62ea8d97fdeeadfd18bd022ca1a70b", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 36223, - "estimated_input_tokens": 9056, - "estimated_eval_tokens": 18368 + "serialized_chars": 36219, + "estimated_input_tokens": 9055, + "estimated_eval_tokens": 18366 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These issues span many unrelated tokenizer, model-loading, video, and dtype regressions; none of the soft pairs are safe duplicate merges from titles alone.", - "confidence": 0.67, - "canonical_issue_reason": "issue:43749 has the strongest discussion/inbound-reference signal and is a concrete, high-impact regression in a core loading path, so it is the best issue anchor in this set.", + "summary": "This cluster is a broad mix of Transformers v5 regression reports: tokenization, remote-code loading, multimodal/video bugs, dtype/device mismatches, and trainer/cache edge cases. The soft pairs are mostly thematically similar but do not look like the same concrete bug or change.", + "confidence": 0.66, + "canonical_issue_reason": "issue:45020 is the broadest representative issue here, covering the recurring 'recent transformers versions break remote-code/models' regression theme that many other reports in this cluster loosely resemble.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 is the most triage-worthy standalone issue here because it is broadly impactful, well-discussed, and describes a clear reproducible failure mode.", + "best_issue_reason": "issue:45020 is the best global issue to anchor the cluster because it is the most umbrella-like report and can plausibly subsume several nearby v5 compatibility breakages, even though the candidates are not true duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43606", "right": "issue:45072", "accept": false, - "reason": "Different failures: CPU offload device mismatch for bark-small vs bfloat16 dtype mismatch in other models." + "reason": "Both mention inference/device mismatch, but one is a CPU offload bug for bark-small and the other is a bfloat16 dtype mismatch; different models and code paths." }, { "left": "issue:43322", "right": "issue:45685", "accept": false, - "reason": "Segmentation fault while loading Llava Next is unrelated to an MPS histogram/int implementation error." - }, - { - "left": "issue:43825", - "right": "issue:43994", - "accept": false, - "reason": "One is a pipeline error-message regression about translation support; the other is bad SigLIP2 outputs with AutoModel/pipeline." + "reason": "Segfault while loading Llava Next is unrelated to the MPS histogram Int error in moe code; no shared concrete failure path." }, { "left": "issue:43550", "right": "issue:45072", "accept": false, - "reason": "torch.compile/SDPA breakage in Bamba is a different code path than the dtype mismatch issue." + "reason": "torch.compile + SDPA failure in Bamba is a different bug from the dtype mismatch issue in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:43388", "right": "issue:46032", "accept": false, - "reason": "Metric gathering drops labels in the last batch; Mamba2 use_cache produces incorrect sequence results." + "reason": "Label truncation in gather_for_metrics and Mamba2 use_cache producing wrong results are separate training/runtime bugs." }, { "left": "issue:44479", "right": "issue:45834", "accept": false, - "reason": "Qwen VL video-input regression is not the same as Kosmos2.5 long-OCR index errors." + "reason": "Both are vision/video-ish regressions, but one is a Qwen video-input regression and the other is an OCR index error in Kosmos2.5." }, { - "left": "issue:43782", - "right": "issue:44977", + "left": "issue:43825", + "right": "issue:43994", "accept": false, - "reason": "from_pretrained weight_only loading failure is distinct from flash-attention generation behavior." + "reason": "One is an error-message/docs regression around pipeline translation support; the other is a model output quality problem with AutoModel/pipeline." }, { - "left": "issue:43577", - "right": "issue:43818", + "left": "issue:43782", + "right": "issue:44977", "accept": false, - "reason": "BLIP2 dtype propagation and Video-LLaVA temporal-attention/shared-weight issues are different bugs." + "reason": "Qwen3VL weight_only loading failure and Qwen3.5 flash-attention generation corruption are different symptoms with different triggers." }, { "left": "issue:43582", "right": "issue:46512", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError is unrelated to stale model.config.dtype after from_config." + "reason": "Apple Silicon TypeError in caching warmup is unrelated to from_config leaving config.dtype stale." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "BLIP2 dtype persistence and Video-LLaVA temporal-attention/shared-weights issues are separate model-loading/model-architecture bugs." }, { "left": "issue:44991", "right": "issue:46291", "accept": false, - "reason": "Tokenizer loading for one model is separate from pipelines import failing on a kernels dependency requirement." + "reason": "Tokenizer loading failure for a specific model is not the same as importing pipelines failing due to kernels package requirements." }, { "left": "issue:43421", "right": "issue:44811", "accept": false, - "reason": "Runtime special-token update bug is not the same as batch_decode ignoring skip_special_tokens." + "reason": "Special-token post-processor sync in TokenizersBackend and Whisper batch_decode ignoring skip_special_tokens are different tokenizer API bugs." }, { "left": "issue:44561", "right": "issue:45020", "accept": false, - "reason": "Removing is_torch_fx_available is one specific breakage; the other is a broader remote_code regression report." + "reason": "Both concern remote-code breakage, but one is specifically about removing is_torch_fx_available while the other is a broader report; not the same concrete cause." }, { - "left": "issue:43716", - "right": "issue:44008", + "left": "issue:43950", + "right": "issue:45987", "accept": false, - "reason": "Dtype mismatch between preprocessor and model is unrelated to a Gemma3n variable-name collision." + "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to StaticCache.get_seq_length returning the wrong type/shape." }, { - "left": "issue:43950", - "right": "issue:45987", + "left": "issue:43716", + "right": "issue:44008", "accept": false, - "reason": "Non-persistent buffer corruption in from_pretrained is a different failure from get_seq_length returning a shape-(1,) tensor." + "reason": "A dtype mismatch between preprocessor and model is different from a variable-name collision causing AttributeError in Gemma3nModel.forward()." }, { "left": "issue:43931", "right": "issue:46846", "accept": false, - "reason": "Qwen3-VL weight-shape mismatch is not the same as missing tensor-parallel plan entries causing OOM/shape errors." + "reason": "Weight-shape mismatch loading Qwen3-VL-30B and missing linear_attn entries in TP plan are distinct model-configuration bugs." }, { "left": "issue:46294", "right": "issue:46463", "accept": false, - "reason": "Qwen3.5 video feature/token mismatch is unrelated to T5 decoder-input validation messaging." + "reason": "Qwen3.5 video chat-template mismatch and T5 decoder-input validation error message are unrelated issues." }, { "left": "issue:43825", "right": "issue:43827", "accept": false, - "reason": "An incorrect pipeline error message and stale docs references are related but not the same underlying bug/change." + "reason": "One is a pipeline error-message regression and the other is stale documentation; same area, different artifact and fix." }, { - "left": "issue:44805", - "right": "issue:44936", + "left": "issue:28218", + "right": "issue:29159", "accept": false, - "reason": "Mask/index shape mismatch is a tensor-shape bug; trainer.evaluate() after train is a separate training-state issue." + "reason": "Both are tokenizer-related, but one is an added-token spacing bug and the other is a broad slow-vs-fast tokenizer inconsistency report." }, { - "left": "issue:28218", - "right": "issue:29159", + "left": "issue:44805", + "right": "issue:44936", "accept": false, - "reason": "Both are tokenizer-related, but one is about added-token spacing and the other about slow/fast tokenizer inconsistency; too broad to merge." + "reason": "Mask shape mismatch in indexing and trainer.evaluate failing after train are different runtime failures." }, { "left": "issue:44610", "right": "issue:45072", "accept": false, - "reason": "Processor image-size mismatch and dtype mismatches are different regressions." + "reason": "Processor output size mismatch for OmDet-Turbo is unrelated to the bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." }, { "left": "issue:43540", "right": "issue:45381", "accept": false, - "reason": "Different video-input bugs in different Qwen/VL models with different symptoms." + "reason": "Both involve video input, but they target different Qwen variants and different bugs: ValueError processing video inputs vs wrong vision_position_ids." }, { "left": "issue:43720", "right": "issue:45072", "accept": false, - "reason": "Packed-weight unpacking during accelerate loading is a different code path than a bfloat16 dtype mismatch." + "reason": "Packed-weight unpacking during accelerate loading is a different failure mode from the dtype mismatch reports." }, { "left": "issue:43749", "right": "issue:44291", "accept": false, - "reason": "FSDP CPU RAM efficient loading and init_empty_weights argument handling are separate loading regressions." + "reason": "FSDP CPU RAM efficient loading and init_empty_weights rejecting _is_hf_initialized are separate loading regressions." }, { - "left": "issue:43645", - "right": "issue:46665", + "left": "issue:43577", + "right": "issue:43582", "accept": false, - "reason": "Jupyter custom-model initialization breakage is not the same as inconsistent from_pretrained behavior inside Python classes." + "reason": "BLIP2 dtype handling and Apple Silicon caching allocator warmup are unrelated bugs." }, { - "left": "issue:43577", - "right": "issue:43582", + "left": "issue:43645", + "right": "issue:46665", "accept": false, - "reason": "BLIP2 dtype issues and Apple Silicon caching_allocator_warmup TypeError are unrelated." + "reason": "Custom models in Jupyter notebooks and inconsistent from_pretrained behavior inside Python classes are related only at a very high level." }, { "left": "issue:44671", "right": "issue:45701", "accept": false, - "reason": "CamemBERT MLM prediction regression is not the same as general tokenization changes across versions." + "reason": "CamemBERT masked LM prediction regression and version-dependent tokenization changes are not the same underlying bug." }, { "left": "issue:42175", "right": "issue:45081", "accept": false, - "reason": "Packaging/backend install behavior is unrelated to the Mistral regex patch tokenizer attribute error." + "reason": "Missing tensorflow in extras install and the Mistral regex patch crash are unrelated packaging vs tokenizer-initialization issues." } ] }, @@ -6331,10 +6334,10 @@ "issue:43550|issue:44912", "issue:43742|issue:45020", "issue:43329|issue:45600", - "issue:44060|issue:44368", "issue:45084|issue:45538", - "issue:43526|issue:43697", + "issue:44060|issue:44368", "issue:43452|issue:44991", + "issue:43526|issue:43697", "issue:43257|issue:43329", "issue:43526|issue:44554", "issue:43906|issue:46209", @@ -6347,208 +6350,202 @@ "issue:42490|issue:43257", "issue:43720|issue:44610" ], - "prepared_review_unit_hash": "9c893f05311862645eb10492946a8d58ebb1bb527fb5ac338475c6e18074a1a6", + "prepared_review_unit_hash": "3ac680d38e8e1c248cffd2aa3bd5d10754f99a689e85ceb96e22631994508261", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36526, - "estimated_input_tokens": 9132, - "estimated_eval_tokens": 18520 + "serialized_chars": 36521, + "estimated_input_tokens": 9131, + "estimated_eval_tokens": 18518 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a similarity bucket, not a duplicate set: the issues span unrelated subsystems (audio, tokenizers, training, loading, vision, and packaging). None of the soft pairs look like the same underlying bug/change or a mergeable PR target.", + "summary": "This cluster is heterogeneous: the items span unrelated bugs across audio, tokenizer loading, training, multimodal processors, quantization, and backend-specific regressions. None of the suggested soft edges look like the same concrete bug/change, so all are rejected.", "confidence": 0.92, - "canonical_issue_reason": "No clear canonical issue: the items are heterogeneous, and the soft pairs are mostly false-positive thematic matches rather than duplicates.", + "canonical_issue_reason": "No single issue clearly represents a duplicate set here; if a representative anchor is needed, issue:45020 is the broadest systemic regression report, but the cluster is too mixed to treat as one bug family.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good global representative. Issue 45020 is the broadest loading-related item, but it still doesn't cover the rest of the cluster.", + "best_issue_reason": "issue:45020 is the most broadly scoped and could serve as a weak representative anchor, but it does not actually unify the other issues into a duplicate cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:34634", "right": "issue:36331", "accept": false, - "reason": "Different subsystems and failure modes: BarkProcessor voice_preset vs CustomTrainer.compute_loss signature mismatch." + "reason": "Different subsystems and failures: Bark voice_preset handling vs CustomTrainer compute_loss signature mismatch." }, { "left": "issue:45003", "right": "issue:46285", "accept": false, - "reason": "Both touch model loading internals, but one is unsafe sys.modules access and the other is a CLIP class structure inconsistency; not the same bug." + "reason": "One is unsafe sys.modules access in modeling_utils; the other is a CLIP class structure inconsistency. Related theme only, not same bug." }, { "left": "issue:46697", "right": "issue:46728", "accept": false, - "reason": "Both are image-processor sizing bugs, but they affect different models and different logic (over-padding vs patch-count calculation)." + "reason": "Both are image-processor padding/patch-count issues, but they affect different models and different code paths." }, { "left": "issue:43653", "right": "issue:44912", "accept": false, - "reason": "Tokenizer special-token registration is unrelated to quantization fallback behavior." + "reason": "BigBirdTokenizer special-token decode bug is unrelated to MXFP4 quantization fallback on git-oss-20b." }, { "left": "issue:43262", "right": "issue:43819", "accept": false, - "reason": "Both are audio-related, but one is chat-template sampling-rate defaulting and the other is DAC forward/from_latents mismatch." + "reason": "Audio chat-template sample-rate default bug vs DAC latent/forward mismatch; no shared concrete code path." }, { "left": "issue:43782", "right": "issue:45325", "accept": false, - "reason": "Both involve Qwen VL models, but they are different code paths: weight_only loading vs rope index temporal scaling." + "reason": "Different Qwen/VL bugs: weight_only loading failure vs rope index temporal scaling." }, { "left": "issue:45081", "right": "issue:45406", "accept": false, - "reason": "Both are tokenizer/processor loading failures, but for different models and different missing attributes/root causes." + "reason": "Tokenizer regex patching crash and Gemma4Processor missing _tokenizer are separate processor-loading failures." }, { "left": "issue:45372", "right": "issue:46326", "accept": false, - "reason": "Both mention mistral_common, but one is an import error for ReasoningEffort and the other is continue_final_message behavior." + "reason": "Both involve mistral_common-adjacent processor/tokenizer behavior, but one is a missing import and the other is continue_final_message logic." }, { "left": "issue:43582", "right": "issue:44857", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and CUDA float16 loss crash are unrelated." + "reason": "AppleSilicon caching allocator warmup TypeError and LwDetrImageLoss AMP crash are unrelated runtime errors." }, { "left": "issue:44589", "right": "issue:45042", "accept": false, - "reason": "Float8 storage lookup failure is unrelated to PIL backend/torchvision dependency requirements." + "reason": "Float8 storage lookup failure is unrelated to PIL image processors incorrectly requiring torchvision." }, { "left": "issue:43550", "right": "issue:44912", "accept": false, - "reason": "torch.compile+SDPA failure in Bamba is unrelated to MXFP4 quantization fallback." + "reason": "Bamba torch.compile SDPA failure and git-oss-20b MXFP4 loading fallback are different model/runtime problems." }, { "left": "issue:43742", "right": "issue:45020", "accept": false, - "reason": "A model-specific KeyError on load is not the same as the broad remote_code regression, and they likely need different fixes." + "reason": "MobileLLM key error during load is a model-specific mapping bug, not the broad remote_code regression described in 45020." }, { "left": "issue:43329", "right": "issue:45600", "accept": false, - "reason": "One is a multimodal token counting bug; the other is an auto_mappings config reference regression." - }, - { - "left": "issue:44060", - "right": "issue:44368", - "accept": false, - "reason": "Both concern tied-weights warnings in Qwen3-family models, but the reported symptoms and likely code paths differ enough that they are not the same concrete bug." + "reason": "Video-token counting bug in multimodal processors is unrelated to stale auto_mappings.py references breaking CI." }, { "left": "issue:45084", "right": "issue:45538", "accept": false, - "reason": "Template-node compilation error and CLIPTokenizer max_length behavior are unrelated." + "reason": "Chat template compilation TypeError is unrelated to CLIPTokenizer's extreme model_max_length value." }, { - "left": "issue:43526", - "right": "issue:43697", + "left": "issue:44060", + "right": "issue:44368", "accept": false, - "reason": "BeitImageProcessorFast label reduction is unrelated to RTDetrV2 output drift in v5." + "reason": "Both mention Qwen3-Next/Qwen3.5 tie-weight warnings, but one is a specific incorrect tie mapping and the other is a user-facing warning from fine-tuning." }, { "left": "issue:43452", "right": "issue:44991", "accept": false, - "reason": "Both are loading failures, but gguf_file support and EMBEDDIA/est-roberta tokenizer loading are different problems." + "reason": "gguf_file handling and EMBEDDIA tokenizer loading failure are separate from different loader paths and symptoms." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to RTDetrV2 output drift in Transformers v5." }, { "left": "issue:43257", "right": "issue:43329", "accept": false, - "reason": "Qwen3 MoE weight conversion with accelerate/deepspeed is unrelated to multimodal token counting." + "reason": "Qwen3 MoE accelerate/deepspeed conversion issue is unrelated to multimodal video token counting." }, { "left": "issue:43526", "right": "issue:44554", "accept": false, - "reason": "ImageProcessorFast label handling is unrelated to the MPS attention correctness issue." + "reason": "reduce_labels returning one label and MPS attention correctness with mismatched heads are unrelated bugs." }, { "left": "issue:43906", "right": "issue:46209", "accept": false, - "reason": "An isolated reproduction of an old issue is unrelated to FP8 inference on B200." + "reason": "An isolated repro of issue 38071 is not the same as FP8 inference failing on B200." }, { "left": "issue:36010", "right": "issue:36331", "accept": false, - "reason": "ImportError for GenerationMixin and CustomTrainer.compute_loss signature mismatch are unrelated." + "reason": "GenerationMixin import regression and compute_loss signature change are different breakages." }, { "left": "issue:43819", "right": "issue:43927", "accept": false, - "reason": "DAC latent reconstruction mismatch and DiaConfig custom token-ID persistence are different audio/config bugs." + "reason": "DAC latent reconstruction mismatch is unrelated to DiaConfig losing custom token IDs on save/load." }, { "left": "issue:36010", "right": "issue:43645", "accept": false, - "reason": "GenerationMixin import failure and Jupyter custom-model initialization breakage are unrelated." + "reason": "GenerationMixin import error is unrelated to Jupyter custom-model initialization problems." }, { "left": "issue:43531", "right": "issue:45042", "accept": false, - "reason": "Qwen3-MoE sliding_window behavior and PIL torchvision dependency requirements are unrelated." + "reason": "Qwen3-MoE sliding_window behavior and PIL backend requiring torchvision are unrelated." }, { "left": "issue:43638", "right": "issue:44936", "accept": false, - "reason": "DeepSpeed ZeRO-3 index error on a non-pretrained Bert model is not the same as evaluate() failing after train()." + "reason": "Zero3 index error with non-pretrained Bert and trainer.evaluate() after train() are different training failures." }, { "left": "issue:42175", "right": "issue:46291", "accept": false, - "reason": "TensorFlow extras omission with pip install '.[torch]' and kernels v0.15 import requirements are packaging/dependency issues, but not the same one." + "reason": "Torch-only install missing TensorFlow backend is unrelated to pipelines import failing because kernels now needs version/revision." }, { "left": "issue:42490", "right": "issue:43257", "accept": false, - "reason": "Trailing-slash local model warning and Qwen3 MoE weight conversion are unrelated." + "reason": "Trailing-slash local model import warning and Qwen3 MOE weight conversion are different loading issues." }, { "left": "issue:43720", "right": "issue:44610", "accept": false, - "reason": "Packed-weight loading in BitNet and OmDet-Turbo processor input-size mismatch are unrelated." - }, - { - "left": "issue:44589", - "right": "issue:45042", - "accept": false, - "reason": "Float8 storage resolution and PIL backend torchvision dependency issues are unrelated." + "reason": "BitNet packed-weight unpacking during accelerate loading and OmDet-Turbo 640px-vs-224px preprocessing mismatch are unrelated." } ] }, @@ -6623,41 +6620,41 @@ ], "soft_pairs": [ "issue:39692|issue:44912", - "issue:43643|issue:45759", "issue:43746|issue:44811", "issue:43854|issue:44977", + "issue:43643|issue:45759", "issue:38175|issue:44610", "issue:43257|issue:43696", "issue:43262|issue:44811", "issue:44387|issue:44617", "issue:43257|issue:43931", - "issue:46421|issue:46424", "issue:45459|issue:45803", "issue:43329|issue:43450", + "issue:46421|issue:46424", "issue:43329|issue:44479", "issue:42915|issue:43742", "issue:43749|issue:44756", "issue:34567|issue:40990", - "issue:43650|issue:43906", "issue:44291|issue:45042", + "issue:43650|issue:43906", "issue:42915|issue:44821", "issue:29159|issue:34567", "issue:43606|issue:45859", "issue:43927|issue:44912", "issue:43540|issue:46701", "issue:28282|issue:29870", - "issue:44568|issue:45356", "issue:43454|issue:43577", - "issue:43525|issue:46291" + "issue:43525|issue:46291", + "issue:44568|issue:45356" ], - "prepared_review_unit_hash": "8295b7e2313083061cee10a851356d89195acbf5b5fee449605320f2e6da8fd5", + "prepared_review_unit_hash": "b1a06a8903c377e64e3719af326c3ba4daa05899cb244bb4db6f5d59d925b1b2", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, @@ -6672,184 +6669,181 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Only one soft edge looks like the same underlying bug: the video multimodal-token counting regression behind Qwen/Qwen3 video-input failures. The rest are related only at a broad subsystem level (tokenizers, config loading, OOMs, cache, etc.) and are too different to merge as duplicates. No PRs are present.", - "confidence": 0.91, - "canonical_issue_reason": "issue:43329 is the clearest root-cause report in the only accepted duplicate pair: it identifies the undefined video-branch helpers in `_get_num_multimodal_tokens`, which directly explains the broader video-input regression reported in issue:44479.", + "summary": "This cluster is highly heterogeneous: the soft pairs mostly share only broad themes (tokenizers, multimodal loading, quantization, offload, or video processing) but not the same underlying bug. I would not merge any of the candidate pairs.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue fits the cluster; the items cover unrelated failures across different models, processors, and code paths, so there is no true duplicate set here.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43749 is the strongest standalone anchor in the set because it has the highest discussion/inbound-reference signal and a well-scoped, impactful loading regression report, even though it does not appear duplicate to the others.", + "best_issue_reason": "No strong global best issue. If forced, the most concrete code-path bug is issue 43329, but it does not represent the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:39692", "right": "issue:44912", "accept": false, - "reason": "Both are model/example failures, but they describe different models and different failure modes; no shared code-path bug is evident." - }, - { - "left": "issue:43643", - "right": "issue:45759", - "accept": false, - "reason": "Both involve config/model-loading shape issues, but one is missing fields from `AutoConfig` with `trust_remote_code`, while the other is `from_config` not unwrapping `text_config`; related theme, different bugs." + "reason": "Both involve model/loading quality, but one is a SigLIP2 docs/example mismatch and the other is MXFP4 quantization fallback for gpt-oss-20b; different bugs." }, { "left": "issue:43746", "right": "issue:44811", "accept": false, - "reason": "PEFT local-checkpoint loading failure vs Whisper batch decode ignoring `skip_special_tokens`; unrelated code paths." + "reason": "GraniteSpeech PEFT checkpoint loading and Whisper batch_decode skip_special_tokens are unrelated code paths." }, { "left": "issue:43854", "right": "issue:44977", "accept": false, - "reason": "Different models and different symptoms; one is a unit-test loading mismatch, the other is a flash-attention generation regression." + "reason": "GLM-4.7-Flash test loading error and Qwen3.5 flash-attention generation regression are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Both concern config handling, but one is AutoConfig trust_remote_code field loss and the other is AutoModelForCausalLM text_config unwrapping for composite Qwen configs." }, { "left": "issue:38175", "right": "issue:44610", "accept": false, - "reason": "Both are vision-model correctness issues, but one is zero probabilities and the other is processor/model input-size mismatch; not the same bug." + "reason": "SigLIP2 zero probabilities and OmDet-Turbo processor/model size mismatch are distinct failures." }, { "left": "issue:43257", "right": "issue:43696", "accept": false, - "reason": "Qwen3 MoE weight-conversion/loading bug vs a plain CUDA OOM on GPT-oss; not the same underlying issue." + "reason": "Qwen3 MoE weight conversion with accelerate+deepspeed is unrelated to GPT-oss-20b CUDA OOM." }, { "left": "issue:43262", "right": "issue:44811", "accept": false, - "reason": "Audio processor chat-template sampling-rate default vs Whisper decode option handling; different functions and failure modes." + "reason": "Audio chat-template sampling-rate behavior and Whisper batch_decode token skipping are different processor bugs." }, { "left": "issue:44387", "right": "issue:44617", "accept": false, - "reason": "Both are OOM reports, but they stem from different models and likely different memory paths; too broad to merge." + "reason": "Both are OOM-related, but one is an int4 reserved-memory regression and the other is Sam3Video running out of memory; not the same underlying defect." }, { "left": "issue:43257", "right": "issue:43931", "accept": false, - "reason": "Both concern Qwen-family loading, but one is MoE weight conversion under accelerate+deepspeed and the other is a shape mismatch in Qwen3-VL loading; distinct bugs." - }, - { - "left": "issue:46421", - "right": "issue:46424", - "accept": false, - "reason": "Same cache subsystem, but one is lazy-init recompilation and the other is `max_cache_length` handling for quantized cache; not the same code-path problem." + "reason": "Qwen3 MoE conversion failure and Qwen3-VL weight-shape mismatch are separate loading problems." }, { "left": "issue:45459", "right": "issue:45803", "accept": false, - "reason": "Both mention exception handling hiding errors, but they are different helpers (`import_protobuf_decode_error` vs `FuyuBatchFeature.convert_to_tensors`) and should not be merged as duplicates." + "reason": "Both involve exception handling, but the affected functions and failure modes differ; they are not the same bug." }, { "left": "issue:43329", "right": "issue:43450", "accept": false, - "reason": "Both are video-related, but one is undefined variables in `_get_num_multimodal_tokens` while the other is incorrect batch shape in video processors; different defects." + "reason": "Undefined helpers in _get_num_multimodal_tokens versus wrong batched video processor output shape are different video-path issues." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Chunked-prefill recompilation and max_cache_length only applying to Quantized Cache are separate cache features." }, { "left": "issue:43329", "right": "issue:44479", - "accept": true, - "reason": "These point to the same underlying video-input regression: 43329 identifies the broken video-branch token counting logic, and 44479 reports the resulting failures across multiple Qwen video/VL variants." + "accept": false, + "reason": "The first is a specific undefined-variable bug in multimodal token counting; the second is a broader v5.3.0 video-input regression across several VLMs." }, { "left": "issue:42915", "right": "issue:43742", "accept": false, - "reason": "Different models and different load-time failures; no shared concrete bug is evident." + "reason": "FineGrainedFP8Config failure for Qwen3Moe and MobileLLM-125M key error are unrelated model-loading bugs." }, { "left": "issue:43749", "right": "issue:44756", "accept": false, - "reason": "Both concern memory usage, but one is broken FSDP CPU RAM efficient loading and the other is mmap on Strix Halo; unrelated fixes." + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different loading/memory issues." }, { "left": "issue:34567", "right": "issue:40990", "accept": false, - "reason": "Trainer token-count tracking vs GPT-oss perplexity; no duplicate relationship." + "reason": "TrainerState token counting and gpt-oss-20b perplexity are unrelated training/evaluation problems." }, { - "left": "issue:43650", - "right": "issue:43906", + "left": "issue:44291", + "right": "issue:45042", "accept": false, - "reason": "Issue:43650 is not a technical bug report, while 43906 is an isolated reproduction of a specific upstream issue." + "reason": "init_empty_weights argument handling and PIL image processor torchvision dependency are unrelated issues." }, { - "left": "issue:44291", - "right": "issue:45042", + "left": "issue:43650", + "right": "issue:43906", "accept": false, - "reason": "`init_empty_weights` argument mismatch vs PIL image processors requiring torchvision; unrelated code paths." + "reason": "Generic 'ADD THE DATA' report and isolated reproduction of another issue are not duplicates." }, { "left": "issue:42915", "right": "issue:44821", "accept": false, - "reason": "Qwen3Moe FP8 loading failure vs AutoImageProcessor URL loading failure; different components and failures." + "reason": "Qwen3Moe FP8 config failure and AutoImageProcessor-from-URL loading are unrelated." }, { "left": "issue:29159", "right": "issue:34567", "accept": false, - "reason": "Tokenizer slow/fast inconsistency vs trainer token counter not updating; not the same bug." + "reason": "Tokenizer slow/fast inconsistency and TrainerState num_input_tokens_seen not updating are different components and bugs." }, { "left": "issue:43606", "right": "issue:45859", "accept": false, - "reason": "Both involve CPU offload, but they affect different models and different components; too broad to treat as the same concrete defect." + "reason": "Both mention CPU offload, but they affect different models and different code paths." }, { "left": "issue:43927", "right": "issue:44912", "accept": false, - "reason": "DiaConfig save/load custom token IDs vs MXFP4 quantization fallback on gpt-oss; unrelated failures." + "reason": "DiaConfig token IDs lost on save/load and MXFP4 quantization fallback are unrelated." }, { "left": "issue:43540", "right": "issue:46701", "accept": false, - "reason": "Different multimodal pipelines and different error classes; no clear shared bug or change." + "reason": "Qwen3OmniMoe video-processing ValueError and DiffusionGemma pipeline failure are different multimodal problems." }, { "left": "issue:28282", "right": "issue:29870", "accept": false, - "reason": "ImportError for missing PyTorch vs a BART generation UnboundLocalError; completely unrelated." - }, - { - "left": "issue:44568", - "right": "issue:45356", - "accept": false, - "reason": "Both are tokenizer regressions, but they concern different models and different behaviors (special tokens vs codec/regex handling); not the same issue." + "reason": "Missing PyTorch import for AutoModel and Bart generate() decoder_attention_mask error are unrelated." }, { "left": "issue:43454", "right": "issue:43577", "accept": false, - "reason": "AyaVision weight tying bug vs BLIP2 dtype propagation bug; different model-init code paths." + "reason": "AyaVision lm_head tying and Blip2 dtype propagation are separate model-loading bugs." }, { "left": "issue:43525", "right": "issue:46291", "accept": false, - "reason": "Missing `pad_token_id` on Llama4Config vs pipelines import failing due to kernels package requirements; unrelated." + "reason": "Llama4Config missing pad_token_id and transformers.pipelines import failure with kernels v0.15 are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and different tokenizer behaviors; not the same underlying bug." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "The summary is largely grounded in the packet, and the only accepted soft edge (issue:43329 \u2194 issue:44479) is a conservative, plausible duplicate call. The rejected pairs are generally separated by model/component and symptom, and the note that no PRs are present matches the packet." - }, - "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 }, @@ -6940,207 +6934,207 @@ "issue:24643|issue:29159", "issue:44987|issue:45907", "issue:41720|issue:43540", - "issue:44568|issue:45488", "issue:42175|issue:44488", + "issue:44568|issue:45488", "issue:43540|issue:44560", "issue:44792|issue:44964" ], - "prepared_review_unit_hash": "7d07749b0f4c70ad4a130306537f5605678abffa87c1cfc3459f0e817a96984a", + "prepared_review_unit_hash": "5828c21ec554c57af9a86fe82dc79bc3b7c615ed4db46a88e3474ded5f9194ad", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36013, - "estimated_input_tokens": 9004, - "estimated_eval_tokens": 18264 + "serialized_chars": 36012, + "estimated_input_tokens": 9003, + "estimated_eval_tokens": 18262 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a mixed set of mostly unrelated Transformers issues, with the strongest shared theme being tokenizer/model-loading regressions. I did not find any soft edge that clearly rises to the level of a true duplicate or mergeable PR-equivalent.", - "confidence": 0.74, - "canonical_issue_reason": "issue:45488 is the best anchor here: it is active, high-discussion, and captures a broad tokenizer regression with cross-model impact (DeepSeek V3/R1), making it the most representative issue in this mixed cluster.", + "summary": "This cluster is highly heterogeneous: the soft edges are mostly broad thematic overlaps (tokenizers, loading, multimodal models, training) rather than true duplicates. I would not merge any of the candidate pairs.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical issue emerges from this set; the items describe unrelated bugs and regressions, so any one issue would be only an arbitrary representative.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45488 has the strongest signal for cluster-level usefulness: current, well-discussed, and broad enough to represent the tokenizer-regression side of the cluster.", + "best_issue_reason": "If a representative issue is needed, issue #45488 is the strongest candidate because it is a detailed, high-signal regression report with active discussion and broader impact on DeepSeek V3/R1 tokenization.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44987", "right": "issue:45538", "accept": false, - "reason": "Different failures: one is a model-loading regression for a specific repo, the other is a tokenizer max-length quirk; not the same underlying bug." + "reason": "Both are version-related regressions, but one is a model-loading failure and the other is a tokenizer max-length bug; different code paths and symptoms." }, { "left": "issue:28282", "right": "issue:34689", "accept": false, - "reason": "One is a missing-PyTorch ImportError, the other is a model-loading regression for a specific vision model/version; different code paths." + "reason": "These are distinct loading failures for different models and causes; one is a missing PyTorch dependency error, the other is a specific model-loading regression." }, { "left": "issue:45538", "right": "issue:46869", "accept": false, - "reason": "Tokenizer max-length behavior and YOLO multi-GPU device-map failure are unrelated." + "reason": "Unrelated problems: tokenizer max_length behavior versus multi-GPU device-map failure in YOLO." }, { "left": "issue:44857", "right": "issue:46832", "accept": false, - "reason": "Both are object-detection related, but they describe different runtime failures in different models/code paths." + "reason": "Different RT-DETR issues: one is an AMP/CUDA crash in the loss, the other is a forward-pass shape/feature-level mismatch." }, { "left": "issue:43697", "right": "issue:45357", "accept": false, - "reason": "RT-DETR output drift and Qwen3.5 save_pretrained key corruption are separate issues." + "reason": "Both involve model behavior regressions, but one is RT-DETR output differences and the other is Qwen3.5 save_pretrained key serialization." }, { "left": "issue:43425", "right": "issue:46291", "accept": false, - "reason": "Torch version incompatibility is unrelated to the kernels revision/version requirement failure." + "reason": "One is a Torch-version compatibility complaint, the other is an import-time kernels API requirement; different failure modes." }, { "left": "issue:33357", "right": "issue:34567", "accept": false, - "reason": "A MacOS bus error on CLIP loading and a TrainerState counter not updating are unrelated." + "reason": "A MacOS bus error for CLIP community models is unrelated to TrainerState token-count tracking." }, { "left": "issue:43404", "right": "issue:45127", "accept": false, - "reason": "Both mention tied weights, but one is a model-class weight-tying bug and the other is a LoRA merge/extended-vocab collapse; not the same concrete problem." + "reason": "Both mention tying weights, but they are different bugs: one is a missing lm_head tie in Mistral3, the other is collapse after LoRA merge with extended vocab." }, { "left": "issue:42915", "right": "issue:45538", "accept": false, - "reason": "Qwen3Moe FP8 config failure and CLIPTokenizer max-length behavior are unrelated." + "reason": "Fine-grained FP8 failure in Qwen3Moe is unrelated to CLIPTokenizer max_length behavior." }, { "left": "issue:45137", "right": "issue:45405", "accept": false, - "reason": "DeepSpeed ZeRO3 deque failure and an unreleased PEFT version bump are unrelated." + "reason": "DeepSpeed ZeRO3 deque error and a PEFT version pin issue are unrelated." }, { "left": "issue:42175", "right": "issue:43824", "accept": false, - "reason": "Package/backend installation issue and a missing Qwen2.5-VL import are unrelated." + "reason": "Packaging/backend selection for torch install does not match an import error for Qwen2_5_VL." }, { "left": "issue:28282", "right": "issue:29127", "accept": false, - "reason": "Missing PyTorch installation and a LayoutLMv3 error-message clarity issue are unrelated." + "reason": "A missing PyTorch dependency error is unrelated to LayoutLMv3's box-content validation/error-message issue." }, { "left": "issue:44062", "right": "issue:44991", "accept": false, - "reason": "Both concern tokenizers, but one is an AddedToken TypeError and the other is a model-specific tokenizer load failure; not enough evidence they are the same bug." + "reason": "Both are tokenizer-related, but one is an AddedToken argument collision and the other is a model-specific tokenizer loading regression." }, { "left": "issue:44479", "right": "issue:46693", "accept": false, - "reason": "Video-input regression for Qwen3 variants and FA2 long-context performance regression are different bugs." + "reason": "A video-input regression in Qwen VL models is not the same as an FA2 long-context performance regression." }, { "left": "issue:43742", "right": "issue:44291", "accept": false, - "reason": "A missing key while loading a model and an unexpected _is_hf_initialized argument in init_empty_weights are different loader failures." + "reason": "Loading MobileLLM-125M fails for a different reason than init_empty_weights rejecting an unexpected internal argument." }, { "left": "issue:44936", "right": "issue:46821", "accept": false, - "reason": "Trainer evaluate/train state handling and TimesFM window_size attribute handling are unrelated." + "reason": "Trainer evaluate-after-train behavior is unrelated to the TimesFM window_size attribute error." }, { "left": "issue:43577", "right": "issue:46512", "accept": false, - "reason": "One reports actual module dtype staying float32; the other reports stale config dtype. Related theme, but not the same bug." + "reason": "One bug is incorrect module dtypes in BLIP2 loading; the other is stale config dtype after from_config. Similar topic, different defect." }, { "left": "issue:42915", "right": "issue:43749", "accept": false, - "reason": "FP8 config failure and FSDP CPU RAM efficient loading are distinct training/loading issues." + "reason": "Qwen3Moe FP8 config failure and FSDP CPU RAM efficient loading are separate training/serialization issues." }, { "left": "issue:44849", "right": "issue:45701", "accept": false, - "reason": "Qwen3.5 hidden-state bug and a broad tokenization-change report are different issues." + "reason": "Qwen3.5 output_hidden_states regression is not the same as a tokenizer-version change issue." }, { "left": "issue:43452", "right": "issue:43582", "accept": false, - "reason": "GGUF loading breakage and AppleSilicon allocator warmup TypeError are unrelated." + "reason": "GGUF loading problems and an Apple Silicon caching_allocator_warmup TypeError do not share the same underlying bug." }, { "left": "issue:27957", "right": "issue:28282", "accept": false, - "reason": "Flash Attention 2 with XLM-Roberta and missing PyTorch installation are unrelated." + "reason": "XLMRoberta Flash Attention 2 support and a missing PyTorch import error are unrelated." }, { "left": "issue:24643", "right": "issue:29159", "accept": false, - "reason": "DeepSpeed weight shape runtime error and tokenizer slow/fast inconsistency are unrelated." + "reason": "DeepSpeed training weight-shape runtime error is unrelated to tokenizer fast/slow inconsistency." }, { "left": "issue:44987", "right": "issue:45907", "accept": false, - "reason": "A model-specific loading failure and a hub API revision-forwarding issue are not clearly the same underlying bug." + "reason": "The first is a model loading regression; the second is a Hub utility bug about forwarding revision. Same ecosystem, different code paths." }, { "left": "issue:41720", "right": "issue:43540", "accept": false, - "reason": "Auto device mapping CUDA assert and Qwen3OmniMoe video-input ValueError are different failures." + "reason": "Qwen3 auto device-map CUDA assert and Qwen3OmniMoe video processing ValueError are different failures." }, { - "left": "issue:44568", - "right": "issue:45488", + "left": "issue:42175", + "right": "issue:44488", "accept": false, - "reason": "Different tokenizer regressions affecting different models/code paths; no clear same-bug evidence." + "reason": "Backend-install packaging issue and a model-loading failure for cjvt/sleng-bert are unrelated." }, { - "left": "issue:42175", - "right": "issue:44488", + "left": "issue:44568", + "right": "issue:45488", "accept": false, - "reason": "Backend installation/package issue and a specific model-loading failure are unrelated." + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS on add_special_tokens and the other is overriding tokenizer.json pre-tokenizer behavior; not the same bug." }, { "left": "issue:43540", "right": "issue:44560", "accept": false, - "reason": "Both are Qwen3 video-input regressions, but the models and failure modes differ enough that I can't treat them as the same concrete bug." + "reason": "Different Qwen multimodal failures: one is video-input processing in Qwen3OmniMoe, the other is a StopIteration in Qwen3-vl-embedding." }, { "left": "issue:44792", "right": "issue:44964", "accept": false, - "reason": "Janus image-generation test failure and Phi-4 multimodal model loading failure are unrelated." + "reason": "Janus image-generation test failure and Phi-4 multimodal model loading failure are not the same issue." } ] }, @@ -7215,10 +7209,10 @@ "soft_pairs": [ "issue:44479|issue:45381", "issue:38175|issue:44898", + "issue:44291|issue:44821", "issue:43540|issue:44479", "issue:34567|issue:36331", "issue:43257|issue:43756", - "issue:44291|issue:44821", "issue:44062|issue:45081", "issue:45042|issue:45081", "issue:43582|issue:46821", @@ -7241,136 +7235,136 @@ "issue:45357|issue:45920", "issue:44625|issue:44877" ], - "prepared_review_unit_hash": "9602c42c36a6d253e041c23b77fb5dfb099e2a224bf80e56942e3d2c7cd296ed", + "prepared_review_unit_hash": "8aac51839931c7cf4cf902b3d3eb82ea265b88f377945f6089428636dc7b4dde", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 27, - "serialized_chars": 35600, - "estimated_input_tokens": 8900, - "estimated_eval_tokens": 18056 + "serialized_chars": 35593, + "estimated_input_tokens": 8899, + "estimated_eval_tokens": 18054 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a loose similarity cluster, not a true duplicate set. Most items are unrelated regressions across different models, tokenizers, loading paths, or trainer APIs; none of the soft edges are strong enough to merge.", - "confidence": 0.66, - "canonical_issue_reason": "issue:44479 is the closest thing to a cluster anchor because it is the broadest regression report in the only semi-coherent subtheme (Qwen video issues), but the overall cluster is too heterogeneous for a true canonical duplicate.", + "summary": "This cluster is a loose collection of semantically related but mostly distinct Transformers bugs spanning tokenizers, model/config loading, trainer behavior, dtype/device mismatches, and Qwen video regressions. None of the soft pairs look like the same concrete underlying bug or fix.", + "confidence": 0.91, + "canonical_issue_reason": "If a single issue must anchor the cluster, issue:44479 is the strongest representative: it is a concrete regression report, names the affected models and version, and has the clearest bug scope. That said, the overall set is heterogeneous rather than a true duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 is the best representative of the cluster\u2019s narrowest shared theme (Qwen video regressions), even though it is not a duplicate of the others.", + "best_issue_reason": "Issue:44479 is the best issue candidate overall because it is regression-focused, well-scoped, and actionable, with explicit affected models and version context.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44479", "right": "issue:45381", "accept": false, - "reason": "Related Qwen video regression theme, but the symptoms and affected paths differ: one is a broad video-input regression, the other is a specific rope-index scaling bug." + "reason": "Both concern Qwen video behavior, but 44479 is a broader v5.3.0 regression across several Qwen families while 45381 is a specific qwen2.5-vl position_ids bug; different concrete failures." }, { "left": "issue:38175", "right": "issue:44898", "accept": false, - "reason": "Both are model-output problems, but they involve different models and failure modes (zero probabilities vs image classification failure)." + "reason": "Different model families and failure modes: zero probabilities in SigLIP2 vs non-default-resolution Perceiver classification failure." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "One is an init_empty_weights / unexpected argument error, the other is AutoImageProcessor URL loading; unrelated code paths." }, { "left": "issue:43540", "right": "issue:44479", "accept": false, - "reason": "Both mention video inputs, but they hit different model families and different errors; not the same concrete bug." + "reason": "Both involve video inputs, but they target different model families and different bugs; not the same fixable issue." }, { "left": "issue:34567", "right": "issue:36331", "accept": false, - "reason": "Trainer state tracking and custom compute_loss API changes are separate issues with different code paths." + "reason": "Both are trainer-related, but one is token counting not updating and the other is a compute_loss signature break; distinct problems." }, { "left": "issue:43257", "right": "issue:43756", "accept": false, - "reason": "Completely different bugs: DeepSpeed weight conversion for Qwen3 MoE vs Smollm3 RoPE layer behavior." - }, - { - "left": "issue:44291", - "right": "issue:44821", - "accept": false, - "reason": "Different loading failures in different entry points; unexpected init argument vs URL-based image processor loading." + "reason": "Different domains: DeepSpeed/accelerate weight conversion vs Smollm3 RoPE layer omission." }, { "left": "issue:44062", "right": "issue:45081", "accept": false, - "reason": "Both are tokenizer-related errors, but one is AddedToken construction and the other is a Mistral regex patch/backend_tokenizer crash." + "reason": "Tokenizer AddedToken keyword collision and Mistral regex patch crash are separate tokenizer-loading bugs." }, { "left": "issue:45042", "right": "issue:45081", "accept": false, - "reason": "Image processor torchvision dependency issue and tokenizer regex crash are unrelated code paths." + "reason": "PIL backend torchvision dependency regression is unrelated to the Mistral regex AttributeError." }, { "left": "issue:43582", "right": "issue:46821", "accept": false, - "reason": "Apple Silicon allocator warmup TypeError and TimesFM window_size AttributeError are unrelated." + "reason": "Apple Silicon allocator warmup TypeError and TimesFM window_size AttributeError are different API/runtime failures." }, { "left": "issue:45672", "right": "issue:45674", "accept": false, - "reason": "Both involve 4-bit inference, but one is a Gemma4 audio dtype/torch.finfo problem and the other is a BitsAndBytesConfig skip-module regression." + "reason": "Both touch 4-bit inference, but one is Gemma4 audio-module dtype handling and the other is BitsAndBytes skip_modules clearing a default exclusion; different bugs." }, { "left": "issue:43742", "right": "issue:44589", "accept": false, - "reason": "Different model-loading failures: key error for MobileLLM-125M vs missing Float8 storage object." + "reason": "Separate model-loading failures: key error for MobileLLM-125M versus missing Float8 storage type." }, { "left": "issue:43756", "right": "issue:46399", "accept": false, - "reason": "Different models and causes: Smollm3 RoPE-layer mismatch vs Qwen3.5 kernel registration/decorator failure." + "reason": "RoPE-layer dropping in Smollm3 and Qwen3.5 missing decorator for apply_rotary_pos_emb are unrelated." }, { "left": "issue:43756", "right": "issue:45325", "accept": false, - "reason": "Both mention RoPE-related behavior, but they affect different models and different logic (layer dropping vs temporal position scaling)." + "reason": "Different Qwen branches and different defects: Smollm3 RoPE-layer count vs Qwen2.5-VL still-image temporal position scaling." }, { "left": "issue:44568", "right": "issue:46491", "accept": false, - "reason": "Tokenizer regressions, but one is special-token insertion and the other is encode/decode leading-space preservation." + "reason": "Tokenizer regressions, but not the same one: missing BOS/EOS insertion versus CodeLlama leading-space round-trip loss." }, { "left": "issue:45405", "right": "issue:45600", "accept": false, - "reason": "Both are repository maintenance/config issues, not the same bug: unavailable PEFT version vs stale auto_mappings references." + "reason": "One is a dependency version bump problem; the other is stale auto-mapping references breaking CI." }, { "left": "issue:43716", "right": "issue:46701", "accept": false, - "reason": "Different multimodal pipelines and failure modes; dtype mismatch in Mistral-3 preprocessing is not the same as DiffusionGemma pipeline breakage." + "reason": "Mistral-3 image preprocessor dtype mismatch and DiffusionGemma image-text-to-text pipeline failure are different paths and symptoms." }, { "left": "issue:43577", "right": "issue:45072", "accept": false, - "reason": "Both mention dtype mismatches, but they concern different models/components and do not appear to share the same code path." + "reason": "Both are dtype mismatches, but they affect different model architectures and loading/inference flows; not one shared bug." }, { "left": "issue:43531", @@ -7382,55 +7376,55 @@ "left": "issue:43906", "right": "issue:46468", "accept": false, - "reason": "One is an isolated reproduction of an older issue; the other is a broken Nemotron example. No shared concrete bug is evident." + "reason": "The first is an isolated reproduction of an existing issue, while the Nemotron example breakage is a different model/example failure." }, { "left": "issue:24643", "right": "issue:29870", "accept": false, - "reason": "DeepSpeed training weight-shape error and Bart generate decoder_attention_mask bug are unrelated." + "reason": "DeepSpeed training weight-shape runtime error and BART generate decoder_attention_mask initialization bug are different code paths." }, { "left": "issue:44737", "right": "issue:46463", "accept": false, - "reason": "XLNet CPU-device placement bug and T5 decoder-input validation messaging are separate issues." + "reason": "XLNet CPU device placement bug and T5 decoder-input validation error message are not the same underlying issue." }, { "left": "issue:43906", "right": "issue:46693", "accept": false, - "reason": "An isolated reproduction of an old issue is unrelated to an FA2 long-context performance regression." + "reason": "One is a reproduction of another issue, the other is a long-context FA2 performance regression; different bugs." }, { "left": "issue:45823", "right": "issue:46463", "accept": false, - "reason": "Gemma4 device-map mismatch and T5 decoder-input validation are different subsystems and errors." + "reason": "Gemma4 PLE device mismatch during forward and T5 decoder validation messaging are unrelated." }, { "left": "issue:43452", "right": "issue:44488", "accept": false, - "reason": "Both are loading-related, but the former is a gguf_file breakage for tokenizer/model loading and the latter is a specific model load failure for cjvt/sleng-bert." + "reason": "gguf_file breaking tokenizer/model.from_pretrained and a specific sleng-bert load failure are different loading issues." }, { "left": "issue:43697", "right": "issue:46829", "accept": false, - "reason": "Different detection/classification models and different output mismatches; not the same bug." + "reason": "RTDetrV2 output divergence in v5 and VideoPrism hidden_states/last_hidden_state mix-up are distinct output-shape/semantics bugs." }, { "left": "issue:45357", "right": "issue:45920", "accept": false, - "reason": "Saving visual encoder keys and wrong token IDs are unrelated regressions." + "reason": "Qwen3.5 visual encoder save_pretrained key bug and AutoTokenizer wrong token IDs across many models are unrelated regressions." }, { "left": "issue:44625", "right": "issue:44877", "accept": false, - "reason": "Both are config-loading issues, but one is num_labels propagation and the other is strict config blocking granite_speech; different concrete problems." + "reason": "num_labels propagation in Qwen3.5 config and strict config loading for granite_speech are different configuration bugs." } ] }, @@ -7531,190 +7525,190 @@ "issue:45542|issue:46178", "issue:43295|issue:45356" ], - "prepared_review_unit_hash": "b2af0d43b4e05b078db1dd6888b51669f34712deed221e9f616600c0f8962c09", + "prepared_review_unit_hash": "f6f331295973e96c9f0aea7e9449a134502b6ef46d2c345c539f115a0070e7ac", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 36193, - "estimated_input_tokens": 9049, - "estimated_eval_tokens": 18354 + "serialized_chars": 36199, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The candidate pairs are mostly unrelated bug reports across different models, loaders, tokenizers, and training paths. I would reject every soft edge; none looks like the same underlying defect or a PR that could plausibly be merged together.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "The cluster is mostly a grab bag of unrelated bug reports spanning models, tokenizers, training, loading, and backend issues. None of the soft pairs look like the same underlying defect, so all candidate duplicate edges are rejected.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43749 is the strongest standalone report here: it has the most discussion/inbound refs and a concrete, high-impact training failure, making it the best anchor if one issue must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a strong global canonical: the set spans many unrelated bugs (loading, config serialization, training, quantization, attention, and multimodal pipelines), so there is no clear representative duplicate target.", + "best_issue_reason": "issue:43749 is the clearest and most actionable issue in the set, with broad training impact and enough context to serve as the representative item.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45663", "right": "issue:46846", "accept": false, - "reason": "Different model families and failures: Gemma-4 FSDP2 attention KeyError vs Qwen3.5 TP-plan OOM/shape errors." + "reason": "Different models and different failure modes: Gemma-4 FSDP2 attention state vs Qwen3.5 TP plan entries." }, { "left": "issue:43540", "right": "issue:46693", "accept": false, - "reason": "Unrelated problems: Qwen3OmniMoe video-input ValueError vs long-context FA2 performance regression." + "reason": "One is a video-input processing ValueError; the other is a long-context FA2 performance regression." }, { "left": "issue:44913", "right": "issue:46121", "accept": false, - "reason": "Both touch rope/config handling, but one is GPTNeoX config persistence and the other is a TypeError on JSON-loaded lists in a validation helper." + "reason": "Both are rope-related, but one is config reload persistence and the other is a validation helper type error." }, { "left": "issue:44737", "right": "issue:45588", "accept": false, - "reason": "Different code paths and symptoms: XLNet CPU device bug vs flash_attention crash on sink-less models." + "reason": "XLNet CPU tensor creation and flash-attention aux-state handling are unrelated code paths." }, { "left": "issue:28282", "right": "issue:32090", "accept": false, - "reason": "Completely different errors: missing PyTorch for AutoModel import vs Trainer GPU broadcast TypeError." + "reason": "PyTorch import availability vs Trainer GPU broadcast NoneType handling are completely different failures." }, { "left": "issue:46463", "right": "issue:46899", "accept": false, - "reason": "T5 decoder-input validation message bug is unrelated to Gemma4 multimodal 4-bit dtype casting." + "reason": "Incorrect T5 error messaging is unrelated to Gemma-4 4-bit multimodal dtype casting." }, { "left": "issue:43697", "right": "issue:45216", "accept": false, - "reason": "Different regressions in different models: RTDetrV2 output differences vs Qwen3.5 save_pretrained checkpoint corruption." + "reason": "Different issues in different paths: RTDetrV2 output mismatch vs Qwen3.5 save_pretrained regression." }, { "left": "issue:43526", "right": "issue:45479", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug is not the same as sequence-classification zero-loss behavior across models." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to sequence-classification zero-loss behavior." }, { "left": "issue:43720", "right": "issue:44898", "accept": false, - "reason": "BitNet packed-weight loading bug and Perceiver interpolation failure are unrelated model-specific issues." + "reason": "Packed-weight loading in BitNet and Perceiver positional interpolation are distinct problems." }, { "left": "issue:43643", "right": "issue:45375", "accept": false, - "reason": "Both mention missing fields, but one is generic AutoConfig trust_remote_code behavior and the other is a specific Qwen3_5MoeVisionConfig field being dropped." + "reason": "A generic AutoConfig trust_remote_code field-loss issue is not the same as a specific Qwen3_5MoeVisionConfig field omission." }, { "left": "issue:43761", "right": "issue:46829", "accept": false, - "reason": "Different output-shape bugs in different vision/video models; not the same underlying defect." + "reason": "Both mention hidden states, but the models and concrete bugs differ; one omits hidden_states, the other misassigns last_hidden_state." }, { "left": "issue:45200", "right": "issue:46463", "accept": false, - "reason": "Gemma 4 mm_token_type_ids defaulting issue is unrelated to T5 decoder validation messaging." + "reason": "Gemma 4 mm_token_type_ids defaulting is unrelated to T5 decoder-input validation messaging." }, { "left": "issue:44466", "right": "issue:45005", "accept": false, - "reason": "Both involve tied weights, but one is device-dependent lm_head serialization and the other is a translation-model v5 tied-weights issue; not enough evidence of the same code-path." + "reason": "Both touch tied weights, but one is device-dependent serialization and the other is a broader translation-model tied-weights issue." }, { "left": "issue:44361", "right": "issue:44442", "accept": false, - "reason": "Different tokenizer failures affecting different tokenizers and likely different bugs." + "reason": "Different tokenizer failures affecting different models and code paths." }, { "left": "issue:44493", "right": "issue:46050", "accept": false, - "reason": "Unexpected position-id key warnings are unrelated to sinq quantized model save/load corruption." + "reason": "Unexpected position-id keys and sinq quantization save/load breakage are unrelated." }, { "left": "issue:43653", "right": "issue:44361", "accept": false, - "reason": "BigBirdTokenizer special-token registration bug is unrelated to MLukeTokenizer task AttributeError." + "reason": "BigBird special-token registration and MLukeTokenizer task AttributeError are not the same defect." }, { "left": "issue:41762", "right": "issue:44936", "accept": false, - "reason": "Gemma3 ZeRO-3 loading IndexError and trainer.evaluate() after train() are different failures." + "reason": "Gemma3 ZeRO-3 loading IndexError and trainer.evaluate() post-train failure are different training/runtime issues." }, { "left": "issue:43452", "right": "issue:44843", "accept": false, - "reason": "Both are from AutoTokenizer/from_pretrained flows, but one is gguf_file support and the other is an offline-mode model_info call in Mistral regex patching." + "reason": "Both concern from_pretrained loading, but one is gguf_file support and the other is an unconditional model_info() offline-mode bug." }, { "left": "issue:43688", "right": "issue:43749", "accept": false, - "reason": "Aux-loss normalization in MoE models is unrelated to FSDP CPU RAM efficient loading." + "reason": "Auxiliary-loss normalization in MoE models is unrelated to FSDP CPU RAM efficient loading." }, { "left": "issue:43819", "right": "issue:44811", "accept": false, - "reason": "DAC latent/forward mismatch is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "DAC latent reconstruction mismatch and Whisper batch_decode skip_special_tokens behavior are unrelated." }, { "left": "issue:38175", "right": "issue:43295", "accept": false, - "reason": "Siglip2 zero-probability outputs are unrelated to processor.tokenizer regression in v4.57.5." + "reason": "Both mention Siglip/processor flows, but one is zero probabilities from a model and the other is a processor.tokenizer regression." }, { "left": "issue:43994", "right": "issue:44898", "accept": false, - "reason": "Different models and symptoms: Siglip2 nonsensical outputs vs Perceiver interpolation failure." + "reason": "Siglip2 nonsensical outputs and Perceiver non-default-resolution failures are different model-specific bugs." }, { "left": "issue:43525", "right": "issue:46701", "accept": false, - "reason": "Llama4Config pad_token_id attribute error is unrelated to DiffusionGemmaGenerationOutput pipeline failure." + "reason": "Llama4Config pad_token_id and DiffusionGemma image-text pipeline failures are unrelated." }, { "left": "issue:45372", "right": "issue:46170", "accept": false, - "reason": "Gemma 4 mistral_common import breakage is unrelated to Windows ROCm torch.distributed dependency errors." + "reason": "Gemma 4 processor import breakage from mistral_common is unrelated to Windows ROCm torch.distributed import/type issues." }, { "left": "issue:45542", "right": "issue:46178", "accept": false, - "reason": "Both concern backend detection, but tensorboard-without-TF and tensorflow_text backend mapping are separate issues." + "reason": "Different backend resolution failures: tensorboard-without-TF vs tensorflow_text backend mapping." }, { "left": "issue:43295", "right": "issue:45356", "accept": false, - "reason": "Processor.tokenizer regression and Kimi-K2.5 codec/mistral-regex regression do not share the same bug." + "reason": "Processor.tokenizer/image passing regression and Kimi-K2.5 codec/fix_mistral_regex regression are different bugs." } ] }, @@ -7787,8 +7781,8 @@ "issue:46821" ], "soft_pairs": [ - "issue:42898|issue:46491", "issue:43531|issue:43742", + "issue:42898|issue:46491", "issue:29159|issue:33357", "issue:43653|issue:44568", "issue:43646|issue:43749", @@ -7809,149 +7803,149 @@ "issue:44734|issue:46612", "issue:44466|issue:46050", "issue:43504|issue:43720", - "issue:43504|issue:43550", "issue:44977|issue:46693", + "issue:43504|issue:43550", "issue:44464|issue:44977", "issue:41720|issue:44464", "issue:43643|issue:44561", "issue:43577|issue:44554" ], - "prepared_review_unit_hash": "e5d8a25a87fa8608755d67c334a41b81ad43d5fce8083b33de34440b379ff904", + "prepared_review_unit_hash": "7b5177315b6760faf1d22060e592e245537f8ae6e83df0876ac66755cc388465", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35543, - "estimated_input_tokens": 8886, - "estimated_eval_tokens": 18028 + "serialized_chars": 35540, + "estimated_input_tokens": 8885, + "estimated_eval_tokens": 18026 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: it mixes tokenizer regressions, model-loading/config bugs, generation issues, and performance problems. None of the soft edges look like the same concrete underlying bug/change, so I would reject them all.", - "confidence": 0.84, - "canonical_issue_reason": "issue:45701 is the best umbrella representative because its title is the broadest version/regression-oriented tokenizer report and it is the closest thematic anchor for the tokenizer-related subset of the cluster.", + "summary": "This is a grab-bag cluster of mostly unrelated issue reports; the soft-linked pairs are only superficially similar, so none should be merged as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "issue:44464 is the strongest representative by connectivity and clarity: it describes a concrete, reproducible generation inconsistency in compiled-forward chunked generation, whereas the other items are more specialized or clearly orthogonal.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45701 is the most generally phrased, cross-cutting issue in the set and has enough discussion/reference activity to serve as a representative anchor if one issue must be chosen.", + "best_issue_reason": "issue:44464 is the best issue candidate overall because it is a precise, high-signal bug report with a concrete code-path regression and the strongest soft-link presence in the set.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:42898", - "right": "issue:46491", + "left": "issue:43531", + "right": "issue:43742", "accept": false, - "reason": "Both are tokenizer regressions, but one is a broad v5 cleanup-space behavior change and the other is a CodeLlama-specific round-trip space stripping issue; not the same bug." + "reason": "Different failures and code paths: Qwen3-MoE `sliding_window` behavior vs a MobileLLM pretrained-loading key error." }, { - "left": "issue:43531", - "right": "issue:43742", + "left": "issue:42898", + "right": "issue:46491", "accept": false, - "reason": "Different models and failures: sliding-window behavior in Qwen3-MoE vs a key error while loading MobileLLM-125M." + "reason": "Both are tokenizer regressions, but one is `clean_up_tokenization_spaces` semantics in v5 and the other is CodeLlama leading-space round-trip loss." }, { "left": "issue:29159", "right": "issue:33357", "accept": false, - "reason": "Tokenizer inconsistency vs MacOS bus error on a CLIP model; no shared underlying failure mode." + "reason": "Tokenizer inconsistency vs MacOS CLIP bus error; unrelated symptoms and root causes." }, { "left": "issue:43653", "right": "issue:44568", "accept": false, - "reason": "Both involve special tokens, but one is BigBird mask-token registration and the other is mDeBERTa BOS/EOS insertion; different code paths." + "reason": "Both involve special tokens, but BigBird mask-token registration and mDeBERTa BOS/EOS insertion are different bugs." }, { "left": "issue:43646", "right": "issue:43749", "accept": false, - "reason": "Trust-remote-code config field loss is unrelated to FSDP CPU RAM-efficient loading." + "reason": "`trust_remote_code` config-field loss is unrelated to FSDP CPU RAM-efficient loading." }, { "left": "issue:43452", "right": "issue:43577", "accept": false, - "reason": "GGUF loading failures are unrelated to Blip2 dtype propagation bugs." + "reason": "GGUF/AutoTokenizer-AutoModel loading failure and BLIP2 dtype propagation are separate subsystems." }, { "left": "issue:43688", "right": "issue:45704", "accept": false, - "reason": "Aux-loss normalization in MoE models is unrelated to a T5/apex RMSNorm memory leak." + "reason": "Aux-loss normalization is unrelated to T5's `apex.FusedRMSNorm` memory leak." }, { "left": "issue:43756", "right": "issue:46402", "accept": false, - "reason": "Smollm3 RoPE-layer count mismatch and Sam3Model text-encoder weight loading are different model-initialization issues." + "reason": "Smollm3 RoPE-layer selection and Sam3 text-encoder weight loading are different model-specific issues." }, { "left": "issue:44843", "right": "issue:44991", "accept": false, - "reason": "Both are tokenizer-loading related, but one is an offline hub/model_info call bug and the other is a tokenizer load failure for a specific model." + "reason": "Both are tokenizer-loading problems, but offline `model_info()` access and EMBEDDIA tokenizer breakage have different causes." }, { "left": "issue:43653", "right": "issue:43927", "accept": false, - "reason": "Special-token registration in BigBird is not the same as DiaConfig losing custom token IDs after save/load." + "reason": "Special-token registration vs custom token IDs lost on save/load; not the same bug." }, { "left": "issue:44464", "right": "issue:44743", "accept": false, - "reason": "Both affect generation correctness, but one is chunked generation with compiled forward and the other is Qwen3.5 recurrent-state resetting with cache." + "reason": "Both affect generation/cache behavior, but compiled-forward chunked generation mismatch and recurrent-state reset are distinct code paths." }, { "left": "issue:43688", "right": "issue:45701", "accept": false, - "reason": "One is an MoE auxiliary-loss bug; the other is a broad tokenization regression report." + "reason": "Aux-loss normalization and broad tokenization-behavior change are unrelated." }, { "left": "issue:44484", "right": "issue:44704", "accept": false, - "reason": "Save-pretrained shard-size behavior is unrelated to AutoProcessor kwarg forwarding." + "reason": "Checkpoint shard-size policy is unrelated to `AutoProcessor` kwargs forwarding." }, { "left": "issue:45491", "right": "issue:46002", "accept": false, - "reason": "Both mention NaNs, but one is a Gemma3 sliding-window GPU embedding issue and the other is a classification-head init/checkpoint issue." + "reason": "Both mention NaNs, but Gemma3 sliding-window attention and classification-head init/checkout are different failures." }, { "left": "issue:44479", "right": "issue:44625", "accept": false, - "reason": "A video-input regression across Qwen VL variants is not the same as num_labels propagation in Qwen3.5 config." + "reason": "Qwen3.5 video-input regression and `num_labels` propagation are separate issues." }, { "left": "issue:44821", "right": "issue:46821", "accept": false, - "reason": "AutoImageProcessor URL loading and TimesFM window_size AttributeError are unrelated loader/API bugs." + "reason": "AutoImageProcessor URL loading and TimesFM `window_size` attribute error are unrelated." }, { "left": "issue:43688", "right": "issue:45440", "accept": false, - "reason": "Aux-loss normalization is unrelated to DeepseekV3MoE divergence from remote implementation." + "reason": "Aux-loss normalization is unrelated to Native DeepseekV3MoE divergence." }, { "left": "issue:33666", "right": "issue:36010", "accept": false, - "reason": "Qwen2-VL multi-GPU training and a GenerationMixin import error are different problems." + "reason": "Multi-GPU training and a `GenerationMixin` import error are different problems." }, { "left": "issue:32090", @@ -7963,55 +7957,61 @@ "left": "issue:44734", "right": "issue:46612", "accept": false, - "reason": "KV-cache continuation indexing in serve is unrelated to beam-search cache reorder omissions in stateful models." + "reason": "Both involve generation state, but server KV-cache continuation indexing and beam-search cache reorder are not the same bug." }, { "left": "issue:44466", "right": "issue:46050", "accept": false, - "reason": "Device-dependent lm_head serialization and sinq quantized save/load breakage are both serialization-adjacent but not the same bug." + "reason": "Device-dependent tied-weight serialization differs from quantized-model save/load corruption." }, { "left": "issue:43504", "right": "issue:43720", "accept": false, - "reason": "Legacy-field loading for Beit semantic segmentation is unrelated to BitNet packed-weight handling during accelerate loading." + "reason": "Beit legacy-field loading and BitNet packed-weight unpacking during accelerate loading are distinct issues." }, { - "left": "issue:43504", - "right": "issue:43550", + "left": "issue:44977", + "right": "issue:46693", "accept": false, - "reason": "Different model families and failures: Beit preset loading vs Bamba torch.compile with SDPA." + "reason": "Qwen3.5 flash-attention correctness and FA2 long-context performance regression are different failure modes." }, { - "left": "issue:44977", - "right": "issue:46693", + "left": "issue:43504", + "right": "issue:43550", "accept": false, - "reason": "A model-specific flash-attention generation bug is not the same as a broad FA2 long-context performance regression." + "reason": "Beit pretrained loading and Bamba torch.compile/SDPA failure are unrelated." }, { "left": "issue:44464", "right": "issue:44977", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency and Qwen3.5 flash-attention misgeneration are different code paths." + "reason": "Chunked-generation inconsistency and Qwen3.5 flash-attention regression are different code paths." }, { "left": "issue:41720", "right": "issue:44464", "accept": false, - "reason": "Auto device mapping cudaErrorAssert on A800 is unrelated to compiled-forward chunked generation differences." + "reason": "Auto device mapping cuda assert on A800 is unrelated to compiled-forward chunked generation output mismatch." }, { "left": "issue:43643", "right": "issue:44561", "accept": false, - "reason": "Both touch trust_remote_code, but one is missing fields from AutoConfig and the other is v5 removal of is_torch_fx_available breaking remote models." + "reason": "`trust_remote_code` missing-fields behavior and removal of `is_torch_fx_available` are distinct compatibility issues." }, { "left": "issue:43577", "right": "issue:44554", "accept": false, - "reason": "Blip2 dtype propagation and an MPS attention correctness issue are unrelated." + "reason": "BLIP2 dtype propagation and MPS attention correctness are unrelated." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer offline-path failure and EMBEDDIA tokenizer loading regression are not the same root cause." } ] }, @@ -8088,15 +8088,15 @@ "issue:43611|issue:46846", "issue:43720|issue:43927", "issue:43504|issue:44610", - "issue:43278|issue:43688", - "issue:43606|issue:43927", "issue:33666|issue:36331", + "issue:43606|issue:43927", + "issue:43278|issue:43688", "issue:43782|issue:46693", "issue:43262|issue:43295", "issue:43720|issue:44361", - "issue:43493|issue:43825", "issue:44514|issue:45290", "issue:44964|issue:45405", + "issue:43493|issue:43825", "issue:44554|issue:46365", "issue:43525|issue:43572", "issue:44977|issue:46209", @@ -8111,219 +8111,219 @@ "issue:43278|issue:43844", "issue:43646|issue:44466", "issue:46209|issue:46710", - "issue:44849|issue:46209", "issue:33357|issue:33666", + "issue:44849|issue:46209", "issue:44568|issue:46489", "issue:43278|issue:43701" ], - "prepared_review_unit_hash": "26cbac2755bcd99685a3f6347317a50e8ed99eb2286b4d56400470253511826d", + "prepared_review_unit_hash": "1df35fef2f8a05068eb4ff352756d858ff2d629f3851027c611600476e3fb5f9", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36790, - "estimated_input_tokens": 9198, - "estimated_eval_tokens": 18652 + "serialized_chars": 36787, + "estimated_input_tokens": 9197, + "estimated_eval_tokens": 18650 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: the items span unrelated tokenizer, model-loading, training, processor, config, and performance problems. None of the soft pairs look like true duplicates or the same fixable code-path issue.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the set is not a duplicate cluster; it mixes unrelated issues across tokenizer, model loading, training, processor, and config behavior.", + "summary": "These items are mostly distinct bug reports that are only loosely related by broad themes like tokenization, model loading, and multimodal regressions. I did not find any soft pair that looks like the same concrete underlying bug/change. Best umbrella issue candidate is the generic model-loading regression in issue:43611.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43611 is the broadest, most representative regression report here: it describes a concrete loading failure affecting models that rely on `base_model_prefix`, which makes it a good umbrella issue compared with the more model-specific reports.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is suitable as the cluster representative because there is no shared underlying bug to dedupe against.", + "best_issue_reason": "issue:43611 is the best issue candidate because it is a clearly scoped, general regression with likely wider impact and a clearer path to a shared fix than the other, more specialized reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43611", "right": "issue:46846", "accept": false, - "reason": "Different bugs: model-loading/base_model_prefix regression vs Qwen3.5 tensor-parallel planning OOM/shape error." + "reason": "Different failures: generic model-loading regression vs Qwen3.5 tensor-parallel `linear_attn` OOM/shape issue." }, { "left": "issue:43720", "right": "issue:43927", "accept": false, - "reason": "Unrelated areas: BitNet packed-weight loading vs DiaConfig token-ID persistence/generation error." + "reason": "Unrelated bugs: packed-weight loading in BitNet vs DiaConfig token-ID persistence after save/load." }, { "left": "issue:43504", "right": "issue:44610", "accept": false, - "reason": "Different code paths and models: BEiT legacy-field loading issue vs OmDet-Turbo processor size mismatch." + "reason": "Different model paths and symptoms: Beit legacy-field load failure vs OmDet-Turbo processor/model input-size mismatch." }, { - "left": "issue:43278", - "right": "issue:43688", + "left": "issue:33666", + "right": "issue:36331", "accept": false, - "reason": "Distinct failures: dtype change in evaluate vs auxiliary-loss normalization in OLMoE/GPT Oss." + "reason": "One is multi-GPU Qwen2-VL training behavior; the other is a custom trainer signature mismatch. Not the same bug." }, { "left": "issue:43606", "right": "issue:43927", "accept": false, - "reason": "CPU offload device mismatch is unrelated to DiaConfig losing token IDs after save/load." + "reason": "CPU offload device mismatch in Bark is unrelated to DiaConfig token-ID loss on serialization." }, { - "left": "issue:33666", - "right": "issue:36331", + "left": "issue:43278", + "right": "issue:43688", "accept": false, - "reason": "Qwen2-VL multi-GPU training and CustomTrainer compute_loss signature mismatch are different issues." + "reason": "Different issues: dtype drift between train/eval vs auxiliary-loss normalization." }, { "left": "issue:43782", "right": "issue:46693", "accept": false, - "reason": "Weight-only loading error for Qwen3VL is unrelated to FA2 long-context performance regression." + "reason": "Different concrete problems: Qwen3VL `weight_only` load error vs FA2 long-context performance regression." }, { "left": "issue:43262", "right": "issue:43295", "accept": false, - "reason": "Both involve processors, but one is a sampling-rate default bug and the other is a processor/tokenizer API regression." + "reason": "Both involve processors, but the bugs differ: audio sampling-rate default vs `processor.tokenizer`/image handling regression." }, { "left": "issue:43720", "right": "issue:44361", "accept": false, - "reason": "BitNet accelerate-loading bug and MLukeTokenizer AttributeError are unrelated." - }, - { - "left": "issue:43493", - "right": "issue:43825", - "accept": false, - "reason": "SigLIP2 implementation discrepancy is unrelated to pipeline translation-task messaging in v5." + "reason": "BitNet packed-weight loading and MLukeTokenizer task AttributeError are unrelated." }, { "left": "issue:44514", "right": "issue:45290", "accept": false, - "reason": "Same API family, but different failure modes: batched padding=False crash vs tool-call/no-content chat-template crash." + "reason": "Both touch chat templates, but the failure modes differ: batched padding behavior vs assistant tool-call messages with no content." }, { "left": "issue:44964", "right": "issue:45405", "accept": false, - "reason": "Model-loading failure for Phi-4 multimodal is unrelated to a PEFT version bump issue." + "reason": "Phi-4 multimodal loading and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation mismatch is not the same as the pipeline translation-task error message regression." }, { "left": "issue:44554", "right": "issue:46365", "accept": false, - "reason": "MPS attention correctness and importing Trainer affected by datasets are unrelated." + "reason": "MPS attention correctness and Trainer import failure via datasets are unrelated." }, { "left": "issue:43525", "right": "issue:43572", "accept": false, - "reason": "Similar theme, but different models and fields; not enough evidence of one shared underlying bug." + "reason": "Similar missing-field shape, but different configs and fields (`pad_token_id` vs `pad_token_idx`) and no shared code-path shown." }, { "left": "issue:44977", "right": "issue:46209", "accept": false, - "reason": "Qwen3.5 flash-attention generation bug is unrelated to FP8 inference failure on B200." + "reason": "Different hardware/model regressions: Qwen3.5 flash-attention generation vs FP8 inference on B200." }, { "left": "issue:44625", "right": "issue:45584", "accept": false, - "reason": "Qwen3.5 num_labels propagation and Whisper empty-transcription failure after align_special_tokens are different bugs." + "reason": "Qwen3.5 `num_labels` propagation bug is unrelated to Whisper empty-transcription generation failure." }, { "left": "issue:43452", "right": "issue:44451", "accept": false, - "reason": "Generic gguf_file loading breakage is not the same as failing to load one specific model." + "reason": "General `gguf_file` loading breakage and a specific ScandiBERT load failure are not the same concrete bug." }, { "left": "issue:28218", "right": "issue:33357", "accept": false, - "reason": "Tokenizer-added-space behavior and a MacOS bus error in CLIP are clearly unrelated." + "reason": "Tokenizer spacing after added tokens is unrelated to the MacOS bus error on a CLIP model." }, { "left": "issue:45823", "right": "issue:46050", "accept": false, - "reason": "Gemma4 device mismatch with device_map='auto' is unrelated to sinq save/load quantization breakage." + "reason": "Gemma4 PLE device mismatch during forward is different from sinq quantization breakage after save/load." }, { "left": "issue:42796", "right": "issue:44977", "accept": false, - "reason": "Ministral tokenizer BPE-marker handling and Qwen3.5 flash-attention generation are different problems." + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation are unrelated." }, { "left": "issue:43994", "right": "issue:46192", "accept": false, - "reason": "SigLIP2 bad outputs and return_tensors being ignored are unrelated API behaviors." + "reason": "SigLIP2 incorrect outputs and `return_tensors` being ignored are separate API bugs." }, { "left": "issue:43493", "right": "issue:43844", "accept": false, - "reason": "SigLIP2 HF/JAX discrepancy and ZeRO-3 gradient growth are different code paths." + "reason": "SigLIP2 discrepancy vs JAX and a DeepSpeed gradient regression are different problems." }, { "left": "issue:39692", "right": "issue:43295", "accept": false, - "reason": "Documentation example errors are unrelated to the processor.tokenizer regression in v4.57.5." + "reason": "Documentation/example errors for SigLIP2 do not match the processor/tokenizer regression." }, { "left": "issue:43278", "right": "issue:43844", "accept": false, - "reason": "Embedding dtype drift and abnormal gradient increase under ZeRO-3 do not share a concrete bug." + "reason": "Embedding dtype change and the ZeRO-3 gradient increase regression are distinct training issues." }, { "left": "issue:43646", "right": "issue:44466", "accept": false, - "reason": "Custom model initialization breakage and inconsistent lm_head.weight serialization are separate v5 regressions." + "reason": "Custom model initialization breakage is not the same as device-dependent `lm_head.weight` serialization." }, { "left": "issue:46209", "right": "issue:46710", "accept": false, - "reason": "FP8 inference on B200 and a DeepSeek-R1 accuracy regression are unrelated." + "reason": "FP8 inference failure on B200 and an accuracy regression after upgrade are different bugs." }, { - "left": "issue:44849", - "right": "issue:46209", + "left": "issue:33357", + "right": "issue:33666", "accept": false, - "reason": "Qwen3.5 hidden-state bug and FP8 inference failure are different issues." + "reason": "MacOS bus error in CLIP and Qwen2-VL multi-GPU training are unrelated." }, { - "left": "issue:33357", - "right": "issue:33666", + "left": "issue:44849", + "right": "issue:46209", "accept": false, - "reason": "A MacOS bus error in CLIP is unrelated to Qwen2-VL multi-GPU training." + "reason": "Qwen3.5 hidden-state output bug and FP8 inference failure are different issues." }, { "left": "issue:44568", "right": "issue:46489", "accept": false, - "reason": "Special-tokens not added for mdeberta-v3-base and DeepSeek-Coder v1 tokenizer output regression are not the same bug." + "reason": "Both are tokenizer bugs, but they affect different models and behaviors; no shared concrete code-path is evident." }, { "left": "issue:43278", "right": "issue:43701", "accept": false, - "reason": "Embedding dtype change and resume_from_checkpoint key mismatch are unrelated." + "reason": "Embedding dtype drift and checkpoint key mismatch are unrelated." } ] }, @@ -8399,8 +8399,8 @@ "soft_pairs": [ "issue:44488|issue:44843", "issue:44792|issue:45405", - "issue:44568|issue:45812", "issue:43927|issue:44361", + "issue:44568|issue:45812", "issue:43824|issue:44589", "issue:43575|issue:45405", "issue:44849|issue:45357", @@ -8408,10 +8408,10 @@ "issue:43257|issue:43299", "issue:46399|issue:46846", "issue:45245|issue:46710", - "issue:44493|issue:46285", "issue:46821|issue:46832", - "issue:43606|issue:44112", + "issue:44493|issue:46285", "issue:44589|issue:45520", + "issue:43606|issue:44112", "issue:43334|issue:43572", "issue:43334|issue:45759", "issue:45072|issue:46832", @@ -8422,206 +8422,206 @@ "issue:46832|issue:46858", "issue:43844|issue:44964", "issue:45200|issue:45229", - "issue:44869|issue:46752", "issue:43873|issue:46050", + "issue:44869|issue:46752", "issue:29942|issue:30333" ], - "prepared_review_unit_hash": "066b4dc1ecd199e53243b7cb5334efcfb628d312781106845eade6d549c48d14", + "prepared_review_unit_hash": "a56984f95d5d1043eb13f5685653846153579374271b37dfd6c2d2237f113946", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36308, - "estimated_input_tokens": 9077, - "estimated_eval_tokens": 18410 + "serialized_chars": 36302, + "estimated_input_tokens": 9076, + "estimated_eval_tokens": 18408 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous transformer issues; the soft pairs share surface similarity (same model family, loading, cache, quantization, or tokenizer wording) but not the same concrete bug or fix. I would reject all soft edges.", - "confidence": 0.9, - "canonical_issue_reason": "issue:45812 is the broadest, highest-signal regression here (all Granite tokenization, silent v4\u2192v5 break, multiple inbound refs), but there is no single issue that truly subsumes the whole cluster.", + "summary": "The cluster is not a true duplicate set: the issues span unrelated tokenizer, loading, generation, quantization, and CI failures. All soft-pair candidates are only superficially similar and should stay separate.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue is a reliable canonical because the set does not share one underlying bug; the apparent similarities are mostly broad theme overlap (model loading/tokenization/version regressions) rather than duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45812 has the widest blast radius and strongest regression profile among the candidates.", + "best_issue_reason": "No issue cleanly represents the whole cluster. If forced, a broad regression report could be a weak proxy, but none of the items subsume the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44488", "right": "issue:44843", "accept": false, - "reason": "Both are loading-related, but one is a specific model load failure and the other is an HF offline/model_info regression; different root causes." + "reason": "Both involve loading/tokenizer behavior, but one is a specific model load failure and the other is an HF Hub offline regression in Mistral regex patching." }, { "left": "issue:44792", "right": "issue:45405", "accept": false, - "reason": "Janus image-generation test failure is unrelated to the PEFT version pin issue." - }, - { - "left": "issue:44568", - "right": "issue:45812", - "accept": false, - "reason": "Both are tokenizer regressions, but they affect different models and different behavior; not one concrete bug." + "reason": "Janus image generation test failure and an unreleased PEFT version bump are unrelated bugs." }, { "left": "issue:43927", "right": "issue:44361", "accept": false, - "reason": "Different model classes and symptoms: config save/load token IDs vs tokenizer AttributeError on tasks." + "reason": "Different models and failure modes: DiaConfig save/load token IDs vs MLukeTokenizer task AttributeError." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but one is mDeBERTa special-token handling and the other is a broad Granite token-ID regression; not the same code-path bug." }, { "left": "issue:43824", "right": "issue:44589", "accept": false, - "reason": "Qwen2.5-VL import failure and Float8 storage TypeError are unrelated." + "reason": "Qwen2.5-VL import error and Float8 storage lookup failure are unrelated." }, { "left": "issue:43575", "right": "issue:45405", "accept": false, - "reason": "One is a tensor-parallel OOM, the other is a dependency version bump; no shared bug." + "reason": "Qwen2-57B-A14B-Instruct TP OOM is unrelated to a PEFT dependency version bump." }, { "left": "issue:44849", "right": "issue:45357", "accept": false, - "reason": "Same model family, but output_hidden_states behavior and save_pretrained visual-key corruption are different code paths." + "reason": "Both are Qwen3.5-related, but one is an output_hidden_states bug and the other is incorrect save_pretrained visual-encoder keys." }, { "left": "issue:43504", "right": "issue:43606", "accept": false, - "reason": "Legacy field load failure in BEiT is unrelated to Bark CPU offload device mismatch." + "reason": "BEiT legacy-field loading and bark-small CPU offload device mismatch are different subsystems and bugs." }, { "left": "issue:43257", "right": "issue:43299", "accept": false, - "reason": "Both involve Qwen MoE loading, but one is accelerate+deepspeed weight conversion and the other is a dev-version load break; too different to merge." + "reason": "Both mention Qwen MoE, but one is accelerate+deepspeed conversion and the other is a Qwen3VL loading break in dev transformers." }, { "left": "issue:46399", "right": "issue:46846", "accept": false, - "reason": "Qwen3.5 kernel registration failure and TP plan omissions are distinct bugs." + "reason": "Qwen3.5 kernel registration crash and missing TP plan entries are separate problems with different failure points." }, { "left": "issue:45245", "right": "issue:46710", "accept": false, - "reason": "Category-cardinality runtime error and an accuracy regression after upgrade are not the same issue." - }, - { - "left": "issue:44493", - "right": "issue:46285", - "accept": false, - "reason": "Unexpected position-id keys and CLIP structural inconsistency are unrelated regressions." + "reason": "Category-count runtime limit and an accuracy regression after upgrade are not the same bug." }, { "left": "issue:46821", "right": "issue:46832", "accept": false, - "reason": "TimesFM window_size AttributeError and RT-DETR feature-level mismatch are different model bugs." + "reason": "TimesFM window_size AttributeError and RT-DETR forward TypeError are unrelated model-specific failures." }, { - "left": "issue:43606", - "right": "issue:44112", + "left": "issue:44493", + "right": "issue:46285", "accept": false, - "reason": "One is a real CPU-offload device mismatch, the other is a CI test regression in a different model." + "reason": "Unexpected position-id keys across many models is not the same as the CLIP text-model structure change." }, { "left": "issue:44589", "right": "issue:45520", "accept": false, - "reason": "Float8 storage lookup failure and flash_attn import KeyError are unrelated." + "reason": "Float8 storage lookup failure and Python 3.13 flash_attn KeyError are different import/runtime issues." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/device-related, but bark offload mismatch and GraniteSpeech stale device override are not the same defect." }, { "left": "issue:43334", "right": "issue:43572", "accept": false, - "reason": "Both are missing token-id fields after v5 changes, but they affect different configs and failure modes." + "reason": "Different missing-config fields on different models: Qwen3-VL pad_token_id vs StableLm pad_token_idx." }, { "left": "issue:43334", "right": "issue:45759", "accept": false, - "reason": "Qwen3-VL pad_token_id issue and text_config unwrapping bug are separate composite-config problems." + "reason": "Qwen3-VL loading error and composite Qwen multimodal text_config unwrapping are distinct load-path issues." }, { "left": "issue:45072", "right": "issue:46832", "accept": false, - "reason": "dtype mismatch in inference and RT-DETR forward arity mismatch are unrelated." + "reason": "bfloat16 dtype mismatch in inference and RT-DETR feature-level mismatch are unrelated." }, { "left": "issue:43504", "right": "issue:43653", "accept": false, - "reason": "BEiT legacy-field load bug is unrelated to BigBirdTokenizer special-token registration." + "reason": "A legacy-field pretrained-load bug and a BigBirdTokenizer special-token registration bug are not the same issue." }, { "left": "issue:43572", "right": "issue:44291", "accept": false, - "reason": "Both are 5.0-related loading regressions, but different models and different argument mismatches." + "reason": "Missing pad_token_idx after 5.0 and unexpected _is_hf_initialized argument are different 5.0 migration problems." }, { "left": "issue:44464", "right": "issue:46424", "accept": false, - "reason": "Chunked-generation inconsistency with compiled forward and max_cache_length handling are different cache behaviors." + "reason": "Chunked generation inconsistency with compiled forward and cache.max_cache_length behavior are different cache/generation bugs." }, { "left": "issue:43937", "right": "issue:46701", "accept": false, - "reason": "GLM GenerationConfig validation and DiffusionGemma pipeline failure are unrelated." + "reason": "GLM-5 GenerationConfig validation and DiffusionGemma pipeline failure are unrelated generation-path issues." }, { "left": "issue:46832", "right": "issue:46858", "accept": false, - "reason": "RT-DETR backbone/output mismatch and static-cache slice arithmetic bug are different subsystems." + "reason": "RT-DETR forward error and static-cache slice arithmetic breakage on non-CUDA backends are different code paths." }, { "left": "issue:43844", "right": "issue:44964", "accept": false, - "reason": "ZeRO-3 gradient inflation and Phi-4 multimodal load failure are unrelated." + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading failure are unrelated." }, { "left": "issue:45200", "right": "issue:45229", "accept": false, - "reason": "Same Gemma 4 family, but text-only fine-tuning defaults and multi-GPU inference OOM are different issues." + "reason": "Gemma4 text-only fine-tuning token IDs and multi-GPU inference OOM are different problems." }, { - "left": "issue:44869", - "right": "issue:46752", + "left": "issue:43873", + "right": "issue:46050", "accept": false, - "reason": "Whisper timestamp decode crash and empty-list chat-template IndexError do not share a bug." + "reason": "Quantization offloading behavior and sinq save/load breakage are related only broadly by quantization, not the same bug." }, { - "left": "issue:43873", - "right": "issue:46050", + "left": "issue:44869", + "right": "issue:46752", "accept": false, - "reason": "Quantization/offloading behavior and sinq save/load breakage are related only broadly, not the same fix." + "reason": "Whisper word-timestamp decode crash and empty-list chat-template IndexError are unrelated." }, { "left": "issue:29942", "right": "issue:30333", "accept": false, - "reason": "Flash Attention 2 test failures and MLFlow reporting on failed jobs are unrelated." + "reason": "Flash Attention 2 test failures and MLFlow reporting on failed training jobs are unrelated CI/issues." } ] }, @@ -8648,6 +8648,8 @@ "issue:15354", "issue:28218", "issue:29159", + "issue:29870", + "issue:29942", "issue:32090", "issue:34567", "issue:37428", @@ -8659,11 +8661,9 @@ "issue:43450", "issue:43493", "issue:43540", - "issue:43550", "issue:43572", "issue:43696", "issue:43701", - "issue:43844", "issue:43873", "issue:43957", "issue:43994", @@ -8721,198 +8721,204 @@ "issue:43295|issue:44610", "issue:45823|issue:46192", "issue:45200|issue:45823", - "issue:43550|issue:43844" + "issue:29870|issue:29942" ], - "prepared_review_unit_hash": "a53d697307b113a93663f94bef2f59a90db8c524b783ee1168020e8e4e03160c", + "prepared_review_unit_hash": "8a4371ffabb603dbd12462e72d1ca407f12c5182eb7f51da02ec47d9d4934ce1", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35550, - "estimated_input_tokens": 8888, - "estimated_eval_tokens": 18032 + "serialized_chars": 35563, + "estimated_input_tokens": 8891, + "estimated_eval_tokens": 18038 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The soft pairs are all false positives: they span unrelated bugs, models, and subsystems, so there is no true duplicate cluster here.", - "confidence": 0.97, - "canonical_issue_reason": "No canonical issue: the items are heterogeneous and none is a clear duplicate target for the others.", + "summary": "This cluster is heterogeneous: the soft links are mostly title-level similarity across unrelated bugs (tokenizers, Gemma4, serving, flash-attention, quantization, device mapping, and assorted model-specific regressions). I would not merge any of the candidate soft edges.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue fits the cluster; the items span unrelated bug families rather than one underlying defect.", "canonical_pr_reason": null, - "best_issue_reason": "No best issue: this is not a coherent duplicate set, so choosing a representative issue would be arbitrary.", + "best_issue_reason": "No globally best issue: 43873 is the broadest/open item, but it does not actually unify the other issues into one duplicate set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45229", "right": "issue:45823", "accept": false, - "reason": "Same model family, but one is CUDA OOM and the other is a device-mapping forward mismatch; different failure modes." + "reason": "Both are Gemma4-related, but one is CUDA OOM in multi-GPU inference and the other is a device-mismatch forward bug; different failure modes." }, { "left": "issue:43994", "right": "issue:44493", "accept": false, - "reason": "SigLIP2 bad outputs vs a generic position-id key warning; not the same bug." + "reason": "SigLIP2 output-quality issue vs generic position-id key warning; no shared code-path bug." }, { "left": "issue:45588", "right": "issue:46050", "accept": false, - "reason": "Different problems: flash-attention crash on None aux state vs quantized model save/load corruption." + "reason": "Flash-attention sink-less crash vs sinq save/load corruption; unrelated problems." }, { "left": "issue:42490", "right": "issue:43696", "accept": false, - "reason": "Local-path import warning and GPT-oss OOM are unrelated." + "reason": "Import warning on trailing-slash model paths vs GPT-oss CUDA OOM; not the same bug." }, { "left": "issue:29159", "right": "issue:32090", "accept": false, - "reason": "Tokenizer slow/fast inconsistency vs Trainer GPU broadcast TypeError; separate code paths." + "reason": "Tokenizer fast/slow inconsistency is unrelated to the Trainer broadcast NoneType error." }, { "left": "issue:44977", "right": "issue:46701", "accept": false, - "reason": "Flash-attention generation issue vs image-text pipeline failure; not the same concrete defect." + "reason": "Qwen3.5 flash-attention generation bug vs DiffusionGemma pipeline failure; different models and code paths." }, { "left": "issue:45084", "right": "issue:45520", "accept": false, - "reason": "Template compilation error vs flash_attn import-key error; unrelated." + "reason": "Template-node compilation error vs Python 3.13 flash_attn import KeyError; different root causes." }, { "left": "issue:28218", "right": "issue:34567", "accept": false, - "reason": "Extra token spacing and TrainerState counter not updating are distinct tokenizer/training issues." + "reason": "Added-token spacing bug is unrelated to TrainerState token counting." }, { "left": "issue:43065", "right": "issue:46369", "accept": false, - "reason": "Dummy Conv2d implementation vs memory overflow in a different Sam3 model; not duplicate." + "reason": "Dummy Conv2d in Sam3PixelDecoder vs SAM3 litetext memory overflow; not duplicates." }, { "left": "issue:44554", "right": "issue:45446", "accept": false, - "reason": "MPS attention correctness bug vs PyTorch version-check import bug; different layers." + "reason": "MPS attention correctness bug vs flex_attention version-check import bug; separate issues." }, { "left": "issue:44423", "right": "issue:44734", "accept": false, - "reason": "Both are serve crashes, but different paths: continuous batching multimodal handling vs KV-cache continuation indexing." + "reason": "Both are serve crashes, but one is multimodal .to() handling and the other is KV-cache indexing; distinct bugs." }, { "left": "issue:15354", "right": "issue:38175", "accept": false, - "reason": "TorchScript GeneratorExp export limitation vs SigLIP2 zero-probability outputs; unrelated." + "reason": "TorchScript GeneratorExp export failure vs SigLIP2 zero-probabilities behavior; unrelated." }, { "left": "issue:43493", "right": "issue:43701", "accept": false, - "reason": "SigLIP2 implementation discrepancy vs resume_from_checkpoint key mismatch; no shared underlying bug." + "reason": "SigLIP2 implementation discrepancy vs resume_from_checkpoint key mismatch; no common underlying defect." }, { "left": "issue:41720", "right": "issue:46710", "accept": false, - "reason": "CUDA assert during auto device mapping vs model accuracy regression after upgrade; different symptoms and causes." + "reason": "Qwen3 auto device-mapping CUDA assert vs DeepSeek accuracy regression; different symptoms and causes." }, { "left": "issue:45406", "right": "issue:46121", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve vs rope-validation TypeError; unrelated." + "reason": "Gemma4Processor _tokenizer attribute crash vs rope-validation TypeError; unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46710", + "accept": false, + "reason": "Flash-attention generation bug and post-upgrade accuracy regression are not the same issue." }, { "left": "issue:45072", "right": "issue:45198", "accept": false, - "reason": "Inference dtype mismatch vs Wav2Vec2 save/tokenization failure; different subsystems." + "reason": "bfloat16 dtype mismatch in inference vs Wav2Vec2 save/tokenization failure; different subsystems." }, { "left": "issue:43295", "right": "issue:43450", "accept": false, - "reason": "Processor.tokenizer regression and batched video shape error are both multimodal, but not the same concrete bug." + "reason": "Processor.tokenizer regression vs batched video shape bug; both processor-related but not the same code-path." }, { "left": "issue:37428", "right": "issue:43957", "accept": false, - "reason": "Flash-attention import failure vs meta-device model-loading breakage; separate issues." + "reason": "Flash-support import failure vs meta-device model loading regression; different failures." }, { "left": "issue:45588", "right": "issue:45759", "accept": false, - "reason": "Flash-attention None aux crash vs composite Qwen text_config unwrapping bug; different code paths." + "reason": "Flash-attention sink-less crash vs composite Qwen multimodal config unwrapping; unrelated." }, { "left": "issue:43540", "right": "issue:45200", "accept": false, - "reason": "Video-input validation error vs Gemma4 text-only token-type defaulting; unrelated." + "reason": "Video-input processing error vs Gemma4 mm_token_type_ids defaulting issue; distinct multimodal bugs." }, { "left": "issue:43295", "right": "issue:43873", "accept": false, - "reason": "Processor/tokenizer regression vs quantization offloading behavior; too broad and different." + "reason": "Processor.tokenizer regression vs quantization offloading issue; no shared underlying bug." }, { "left": "issue:44186", "right": "issue:46192", "accept": false, - "reason": "LayoutLMv2 NER/batching crash vs return_tensors being ignored; different tokenizer behaviors." + "reason": "LayoutLMv2 tokenizer crash vs return_tensors being ignored when text_kwargs is passed; different tokenizer behaviors." }, { "left": "issue:43572", "right": "issue:46821", "accept": false, - "reason": "Missing pad_token_idx vs TimesFM window_size AttributeError; unrelated." + "reason": "Missing StableLmConfig pad_token_idx after update vs TimesFM window_size AttributeError; unrelated." }, { "left": "issue:43295", "right": "issue:44610", "accept": false, - "reason": "Custom processor/tokenizer regression vs OmDet-Turbo image-size mismatch; different models and bugs." + "reason": "Processor/tokenizer regression vs OmDet-Turbo image-size mismatch; separate model-specific issues." }, { "left": "issue:45823", "right": "issue:46192", "accept": false, - "reason": "Gemma4 device mismatch and return_tensors handling are unrelated." + "reason": "Gemma4 device mismatch vs return_tensors kwarg handling; unrelated." }, { "left": "issue:45200", "right": "issue:45823", "accept": false, - "reason": "Gemma4 token-type defaulting vs Gemma4 device-map forward mismatch; distinct defects." + "reason": "Text-only mm_token_type_ids defaulting bug vs Gemma4 PLE device mismatch; different concrete problems." }, { - "left": "issue:43550", - "right": "issue:43844", + "left": "issue:29870", + "right": "issue:29942", "accept": false, - "reason": "Bamba SDPA torch.compile failure vs ZeRO-3 gradient growth; different training/inference problems." + "reason": "BART decoder_attention_mask initialization bug vs Flash Attention 2 test failures; not the same defect." } ] }, @@ -8937,8 +8943,6 @@ "cluster_id": "cluster-10000-500", "nodes": [ "issue:28218", - "issue:29870", - "issue:29942", "issue:35141", "issue:37428", "issue:39692", @@ -8953,10 +8957,12 @@ "issue:43452", "issue:43493", "issue:43525", + "issue:43550", "issue:43572", "issue:43575", "issue:43653", "issue:43824", + "issue:43844", "issue:43856", "issue:43950", "issue:43975", @@ -8986,14 +8992,14 @@ "issue:46899" ], "soft_pairs": [ - "issue:29870|issue:29942", + "issue:43550|issue:43844", "issue:45520|issue:46178", "issue:43450|issue:43575", "issue:43572|issue:44987", "issue:44554|issue:45356", "issue:42490|issue:43493", - "issue:43232|issue:44849", "issue:37428|issue:43824", + "issue:43232|issue:44849", "issue:43295|issue:43493", "issue:44466|issue:45127", "issue:39692|issue:46192", @@ -9013,197 +9019,200 @@ "issue:45200|issue:46899", "issue:44792|issue:45563" ], - "prepared_review_unit_hash": "72a19dd51b56a4656ca7f31bc226c611a663313cf44643fee62d1346bf9f6dde", + "prepared_review_unit_hash": "330ac7c6775d7a7be0b42f42c798c80e8767c8e6b073cb6cbc63b2ee9177e0ca", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 26, - "serialized_chars": 35899, - "estimated_input_tokens": 8975, - "estimated_eval_tokens": 18206 + "serialized_chars": 35887, + "estimated_input_tokens": 8972, + "estimated_eval_tokens": 18200 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous, and none of the soft pairs look like the same concrete bug/change. Most are surface-similar only (e.g. tokenizer/chat-template, import errors, config attribute errors, generation bugs) but point to different code paths or different model families.", - "confidence": 0.85, - "canonical_issue_reason": "No single artifact cleanly represents this mixed cluster; if one issue must be chosen as the best standalone anchor, issue 43950 is the strongest because it is a well-scoped regression report with a clear failure mode and reproducible impact.", + "summary": "Mostly heterogeneous issues across unrelated subsystems. The only plausible true duplicate is the pair about missing `pad_token_id` on config objects during model loading; the rest are distinct bugs and should stay separate.", + "confidence": 0.66, + "canonical_issue_reason": "`issue:43525` is the broadest formulation of the only likely duplicate theme here: a load-time `AttributeError` caused by configs missing `pad_token_id`.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43950 is the best general issue candidate: it is specific, regression-oriented, and framed in a way that would make duplicate matching and follow-up debugging straightforward.", + "best_issue_reason": "`issue:44521` is the strongest standalone anchor in this mixed set: it has the richest discussion/inbound references and a clear, reproducible multimodal `apply_chat_template` bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:29870", - "right": "issue:29942", + "left": "issue:43550", + "right": "issue:43844", "accept": false, - "reason": "Tokenizer spacing bug vs Flash Attention 2 test failures; different subsystems and failure modes." + "reason": "Different failure modes in different code paths: SDPA/torch.compile on one side, ZeRO-3 gradient growth on the other." }, { "left": "issue:45520", "right": "issue:46178", "accept": false, - "reason": "Both are backend/import-related, but one is a flash_attn import path issue and the other is an unmapped tensorflow_text backend; not the same bug." + "reason": "Both are import/backend errors, but they involve different missing symbols and different backend mapping problems." }, { "left": "issue:43450", "right": "issue:43575", "accept": false, - "reason": "Video processor batching shape bug vs tensor-parallel OOM on a specific model; unrelated problems." + "reason": "One is a batched video-processor shape bug; the other is a tensor-parallel OOM when loading a model." }, { "left": "issue:43572", "right": "issue:44987", "accept": false, - "reason": "Both are load/config regressions, but one is missing pad_token_idx and the other is a model-loading failure for a different repository; no shared concrete defect." + "reason": "Both are load-time regressions, but the concrete symptoms and affected paths are different." }, { "left": "issue:44554", "right": "issue:45356", "accept": false, - "reason": "Attention correctness on MPS vs tokenizer/codec regression in Kimi-K2.5; unrelated code paths." + "reason": "MPS attention correctness and Kimi tokenizer codec handling are unrelated bugs." }, { "left": "issue:42490", "right": "issue:43493", "accept": false, - "reason": "Local-path import warning vs SigLIP2 implementation discrepancy; not the same issue." + "reason": "A trailing-slash local-path warning is unrelated to a SigLIP2 implementation discrepancy." }, { - "left": "issue:43232", - "right": "issue:44849", + "left": "issue:37428", + "right": "issue:43824", "accept": false, - "reason": "Generation kwargs update bug vs Qwen3.5 hidden-states bug; different generation paths and symptoms." + "reason": "Both are import errors, but they are for different missing names/modules and different fixes." }, { - "left": "issue:37428", - "right": "issue:43824", + "left": "issue:43232", + "right": "issue:44849", "accept": false, - "reason": "Two distinct import errors for different symbols/modules; only broad symptom overlap." + "reason": "Both touch generation, but one is a sync_gpus kwargs update bug and the other is an output_hidden_states issue." }, { "left": "issue:43295", "right": "issue:43493", "accept": false, - "reason": "Processor/tokenizer regression in custom multimodal code vs SigLIP2 discrepancy; different bugs." + "reason": "A processor.tokenizer regression is not the same as the SigLIP2 JAX-vs-HF discrepancy." }, { "left": "issue:44466", "right": "issue:45127", "accept": false, - "reason": "Both involve tied weights, but one is serialization inconsistency and the other is LoRA merge collapse after vocab extension; not mergeable as one fix." + "reason": "Both mention tied weights, but one is serialization behavior and the other is LoRA merging/model collapse." }, { "left": "issue:39692", "right": "issue:46192", "accept": false, - "reason": "Documentation example errors vs an API bug where return_tensors is ignored; unrelated." + "reason": "Documentation/example errors are not the same as `return_tensors` being ignored when `text_kwargs` is passed." }, { "left": "issue:44521", "right": "issue:45290", "accept": false, - "reason": "Both touch apply_chat_template, but one is zeroed assistant masks for multimodal inputs while the other is a crash on tool-call messages; different concrete failures." + "reason": "Same API family, but different concrete bugs: wrong assistant masks vs a crash on tool-call messages with no content." }, { "left": "issue:43653", "right": "issue:44112", "accept": false, - "reason": "Tokenizer special-token registration bug vs a stale CI test; not the same underlying defect." + "reason": "Tokenizer special-token registration and a stale CI test are unrelated." }, { "left": "issue:45381", "right": "issue:46209", "accept": false, - "reason": "Qwen2.5-VL video position-id bug vs FP8 inference failure on B200; unrelated." + "reason": "A Qwen2.5-VL position-id bug is unrelated to an FP8 inference failure on B200." }, { "left": "issue:44737", "right": "issue:46894", "accept": false, - "reason": "XLNet device-placement bug vs CohereASR training-loss shift bug; different model code paths." + "reason": "Different model families and different bugs: CPU placement in XLNet vs a training-loss shift issue." }, { "left": "issue:28218", "right": "issue:35141", "accept": false, - "reason": "Tokenizer detokenization spacing issue vs output embedding reinitialization on resize; different mechanisms." + "reason": "Tokenizer spacing after added tokens is not the same bug as embedding reinitialization after resizing." }, { "left": "issue:43278", "right": "issue:43856", "accept": false, - "reason": "Embedding dtype drift in eval vs memory inefficiency in Qwen3 MoE training; not the same bug." + "reason": "A dtype drift bug during eval is not the same as an MoE training memory inefficiency." }, { "left": "issue:42915", "right": "issue:43824", "accept": false, - "reason": "FP8 config failure in Qwen3Moe vs an import error for a Qwen2.5-VL class; unrelated." + "reason": "A FineGrainedFP8Config failure and an unrelated model import error are different issues." }, { "left": "issue:43950", "right": "issue:46285", "accept": false, - "reason": "Both are v5 regressions around model internals, but one corrupts non-persistent buffers during from_pretrained and the other changes CLIP text model structure; not the same code-path problem." + "reason": "Non-persistent buffer corruption from `from_pretrained()` is distinct from CLIPTextModel structure inconsistency." }, { "left": "issue:43824", "right": "issue:43975", "accept": false, - "reason": "ImportError for a missing class vs incorrect detokenization in a specific model; different issues." + "reason": "Missing model import and incorrect detokenization are unrelated." }, { "left": "issue:43452", "right": "issue:46512", "accept": false, - "reason": "gguf_file loading bug vs dtype bookkeeping mismatch in from_config; unrelated behaviors." + "reason": "Broken `gguf_file` loading and stale `config.dtype` after `from_config()` are different loading behaviors." }, { "left": "issue:43262", "right": "issue:43450", "accept": false, - "reason": "Audio chat-template sampling-rate default vs batched video output shape bug; different processors and failures." + "reason": "Audio chat-template sampling-rate defaults and batched video-shape bugs are unrelated modality-specific issues." }, { "left": "issue:43334", "right": "issue:43525", - "accept": false, - "reason": "Both mention missing pad_token_id, but they affect different config classes and models; no evidence of a single shared fix." + "accept": true, + "reason": "Both report the same load-time `AttributeError` for missing `pad_token_id` on config objects, likely the same regression across model configs." }, { "left": "issue:44442", "right": "issue:45198", "accept": false, - "reason": "Different tokenizer/model-loading failures for unrelated architectures." + "reason": "AutoTokenizer load failure for one speech tokenizer is not the same as save_pretrained/tokenization failure for Wav2Vec2." }, { - "left": "issue:45200", - "right": "issue:46899", + "left": "issue:44521", + "right": "issue:45290", "accept": false, - "reason": "Gemma4 token-type defaulting bug vs 4-bit quantization dtype cast issue; unrelated multimodal problems." + "reason": "Both are chat-template issues, but one is incorrect masks for multimodal inputs and the other is a crash on tool-call assistant messages." }, { "left": "issue:44792", "right": "issue:45563", "accept": false, - "reason": "Janus image-generation test failure vs a stale warning in paged generate; not the same defect." + "reason": "A failed image-generation test for Janus is unrelated to a stale paged-generate warning about `num_return_sequences`." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge verdicts stay conservative. The lone accepted duplicate pair (43334/43525) is well supported by the shared missing-`pad_token_id` load-time AttributeError pattern, while the rest are reasonably treated as distinct bugs." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -9287,12 +9296,12 @@ "issue:45759|issue:46285", "issue:46512|issue:46821", "issue:43606|issue:45083", - "issue:44206|issue:45584", "issue:38175|issue:45412", + "issue:44206|issue:45584", "issue:43335|issue:43653", "issue:37428|issue:43425", - "issue:43550|issue:44112", "issue:29942|issue:32090", + "issue:43550|issue:44112", "issue:28282|issue:33357", "issue:43611|issue:43931", "issue:44448|issue:44743", @@ -9300,20 +9309,20 @@ "issue:44855|issue:45083", "issue:43278|issue:43825" ], - "prepared_review_unit_hash": "4bdb11c0b2afd2b0d81b9a5724beddc82860a4bf92ad813902b8205a3ac0b097", + "prepared_review_unit_hash": "1bab095a14491838123eb899db599fdece0a0855d62743b8ab242d1cc0a431a5", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36745, + "serialized_chars": 36747, "estimated_input_tokens": 9187, "estimated_eval_tokens": 18630 }, @@ -9322,192 +9331,186 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous Transformers issues with only superficial textual similarity; none of the soft pairs look like true duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "No safe canonical issue: the set spans unrelated bugs and regressions, so collapsing them would over-merge.", + "summary": "This cluster is a grab-bag of unrelated Transformers bugs, regressions, and docs issues rather than a duplicate family. None of the soft pairs look like the same underlying defect or change.", + "confidence": 0.97, + "canonical_issue_reason": "No single duplicate family spans the cluster. If forced to pick an anchor, issue 43825 is the most representative concrete regression with references and a clear repro.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 43825 is the strongest standalone triage anchor: clear user-facing regression, reproducible description, and inbound references, but it still does not unify the cluster.", + "best_issue_reason": "Issue 43825 is the best representative issue because it is concrete, user-facing, and well-scoped; the rest are largely unrelated bugs or documentation regressions.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43819", "right": "issue:44112", "accept": false, - "reason": "Different subsystems: DAC latent decoding vs a GraniteSpeech CI device-override test." + "reason": "DAC latent reconstruction mismatch and a GraniteSpeech CI device-override failure are different code paths and symptoms." }, { "left": "issue:44898", "right": "issue:45412", "accept": false, - "reason": "Unrelated bugs: Perceiver resolution handling vs RT-DETR memory release behavior." + "reason": "Perceiver resolution failure and RT-DETR memory release behavior are unrelated model bugs." }, { "left": "issue:43526", "right": "issue:46192", "accept": false, - "reason": "Both mention labels, but one is image-label reduction and the other is text_kwargs/return_tensors handling." + "reason": "BeitImageProcessorFast label reduction and silent return_tensors handling are different preprocessing issues." }, { "left": "issue:45200", "right": "issue:45325", "accept": false, - "reason": "Different multimodal bugs: missing mm_token_type_ids defaults vs rope index scaling in Qwen2.5-VL." + "reason": "Gemma 4 mm_token_type_ids defaults and Qwen2.5-VL rope index scaling are separate multimodal configuration bugs." }, { "left": "issue:43827", "right": "issue:43901", "accept": false, - "reason": "Both are docs-related, but they cover different API regressions and should not be merged." + "reason": "Both are docs-related, but one is about removed pipeline support and the other about a changed text-classification flag; not the same issue." }, { "left": "issue:41720", "right": "issue:46209", "accept": false, - "reason": "Different models and failure modes: A800 auto device mapping assert vs B200 FP8 inference incompatibility." + "reason": "A800 cudaErrorAssert with auto device mapping and B200 FP8 inference failure are different hardware/model-path problems." }, { "left": "issue:43335", "right": "issue:43550", "accept": false, - "reason": "Different code paths: SwitchTransformers config construction vs Bamba torch.compile/SDPA runtime failure." + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile+SDPA failures affect different models and execution paths." }, { "left": "issue:44561", "right": "issue:46234", "accept": false, - "reason": "Both are missing utility symbols, but they are different symbols with different breakages." + "reason": "Missing is_torch_fx_available and missing is_falcon_mamba_ssm_available are different import_utils symbols, not one bug." }, { "left": "issue:43388", "right": "issue:43526", "accept": false, - "reason": "Both involve labels, but one is metric gathering in evaluation and the other is Beit image-label processing." + "reason": "gather_for_metrics truncating tuple labels and BeitImageProcessorFast reduce_labels behavior are unrelated label-processing bugs." }, { "left": "issue:43819", "right": "issue:45083", "accept": false, - "reason": "Completely different helpers and models; only the word \"output\" is broadly shared." + "reason": "DAC missing STE mismatch and qwen3_omni_moe feature-length helper behavior are different model-specific bugs." }, { "left": "issue:44857", "right": "issue:45593", "accept": false, - "reason": "Different detection bugs: AMP/CUDA crash in loss computation vs auxiliary losses not being used." + "reason": "LwDetrImageLoss AMP crash and D-FINE auxiliary-loss omission are distinct training-path problems." }, { "left": "issue:44610", "right": "issue:44857", "accept": false, - "reason": "Processor/input-size mismatch vs numeric instability in a loss; not the same defect." + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss float16 crash are unrelated components." }, { "left": "issue:46424", "right": "issue:46858", "accept": false, - "reason": "Both are cache-related, but one is config honoring and the other is device-tensor arithmetic in prefill." + "reason": "Quantized-cache max length handling and static-cache device-tensor slice arithmetic are different cache bugs." }, { "left": "issue:45520", "right": "issue:46170", "accept": false, - "reason": "Different environment/import failures: Python 3.13 flash_attn lookup vs Windows ROCm torch.distributed dependency." + "reason": "flash_attn import lookup on Python 3.13 and Windows ROCm torch.distributed dependency issues are not the same failure." }, { "left": "issue:45759", "right": "issue:46285", "accept": false, - "reason": "Different model-structure regressions affecting unrelated composite/config classes." + "reason": "Composite Qwen multimodal config unwrapping and CLIP text-model structure changes are separate model-loading regressions." }, { "left": "issue:46512", "right": "issue:46821", "accept": false, - "reason": "Unrelated: dtype bookkeeping in from_config vs TimesFM window_size attribute error." + "reason": "Stale config.dtype after from_config and TimesFM window_size AttributeError are unrelated API bugs." }, { "left": "issue:43606", "right": "issue:45083", "accept": false, - "reason": "CPU offload device mismatch is unrelated to qwen3_omni_moe feature-length computation." + "reason": "CPU offload device mismatch in bark-small and qwen3_omni_moe helper output-length behavior are different issues." }, { - "left": "issue:44206", - "right": "issue:45584", + "left": "issue:38175", + "right": "issue:45412", "accept": false, - "reason": "Different regressions: unsupported feature-extractor arg vs Whisper empty-transcription handling." + "reason": "Siglip2 zero probabilities and RT-DETR memory leak are different symptoms in different models." }, { - "left": "issue:38175", - "right": "issue:45412", + "left": "issue:44206", + "right": "issue:45584", "accept": false, - "reason": "Different symptoms and subsystems: zero probabilities vs memory leak on deletion." + "reason": "Unsupported feature-extractor arg and Whisper empty-transcription generation failure are unrelated pipeline bugs." }, { "left": "issue:43335", "right": "issue:43653", "accept": false, - "reason": "Config bug in SwitchTransformers is unrelated to BigBirdTokenizer special-token registration." + "reason": "Sparse-layer config creation and BigBirdTokenizer special-token registration are different components and failure modes." }, { "left": "issue:37428", "right": "issue:43425", "accept": false, - "reason": "Specific missing import vs broad Torch 2.10 compatibility issue; not the same bug." + "reason": "Missing flash-attention helper import and Torch 2.10 compatibility are not the same underlying problem." }, { - "left": "issue:43550", - "right": "issue:44112", + "left": "issue:29942", + "right": "issue:32090", "accept": false, - "reason": "Different areas: Bamba SDPA/compile behavior vs a GraniteSpeech CI test regression." + "reason": "Flash Attention 2 test failures and Trainer _gpu_broadcast_one NoneType errors are unrelated test/runtime problems." }, { - "left": "issue:29942", - "right": "issue:32090", + "left": "issue:43550", + "right": "issue:44112", "accept": false, - "reason": "Both are failures, but one is Flash Attention test breakage and the other is a Trainer broadcast TypeError." + "reason": "Bamba SDPA+torch.compile and GraniteSpeech stale device-override CI failures do not share a concrete code path." }, { "left": "issue:28282", "right": "issue:33357", "accept": false, - "reason": "ImportError for missing PyTorch is unrelated to a MacOS bus error in a CLIP community model." + "reason": "PyTorch missing ImportError for AutoModel and MacOS bus error loading a community CLIP model are different environment/model issues." }, { "left": "issue:43611", "right": "issue:43931", "accept": false, - "reason": "Different model-loading failures: base_model_prefix handling vs Qwen3-VL weight-shape mismatch." + "reason": "base_model_prefix loading breakage and Qwen3-VL weight-shape mismatch are distinct model-loading failures." }, { "left": "issue:44448", "right": "issue:44743", "accept": false, - "reason": "Different models and bugs: Pegasus v4/v5 output divergence vs recurrent state reset in qwen3_5." + "reason": "Pegasus v4/v5 output differences and qwen3_5 recurrent-state reset with cache are unrelated behavior changes." }, { "left": "issue:43335", "right": "issue:43504", "accept": false, - "reason": "SwitchTransformers config creation and Beit legacy-field loading are unrelated." + "reason": "SwitchTransformers sparse-layer creation and BeitForSemanticSegmentation legacy-field loading are different model/config bugs." }, { "left": "issue:44855", "right": "issue:45083", "accept": false, - "reason": "Python 3.13 JIT parsing failure is unrelated to qwen3_omni_moe length helper behavior." + "reason": "Python 3.13 IndentationError in DebertaV2Model import and qwen3_omni_moe helper behavior are unrelated." }, { "left": "issue:43278", "right": "issue:43825", "accept": false, - "reason": "Different regressions: dtype drift between train/eval vs an incorrect pipeline error message." - }, - { - "left": "issue:43611", - "right": "issue:43931", - "accept": false, - "reason": "Model-loading problems, but with different root causes and code paths." + "reason": "BF16-to-FP32 dtype drift during evaluate and pipeline() misleading translation support messaging are different regressions." } ] }, @@ -9584,26 +9587,26 @@ "issue:29127|issue:32090", "issue:43335|issue:45083", "issue:43720|issue:44112", - "issue:43335|issue:45593", + "issue:37428|issue:44351", "issue:44514|issue:44521", + "issue:43335|issue:45593", "issue:43335|issue:44112", - "issue:37428|issue:44351", "issue:45290|issue:46326", "issue:45127|issue:45245", - "issue:34689|issue:35141", "issue:45588|issue:45859", + "issue:34689|issue:35141", "issue:42490|issue:43854", "issue:46431|issue:46566", "issue:45310|issue:45478", "issue:44448|issue:44779", "issue:43819|issue:44493", - "issue:26413|issue:31501", "issue:39692|issue:43526", + "issue:26413|issue:31501", "issue:43606|issue:44857", "issue:28218|issue:32090", "issue:45459|issue:46234", - "issue:43701|issue:45083", "issue:39692|issue:44493", + "issue:43701|issue:45083", "issue:44617|issue:46697", "issue:44206|issue:45823", "issue:32090|issue:33357", @@ -9611,214 +9614,214 @@ "issue:44360|issue:44561", "issue:44857|issue:45072" ], - "prepared_review_unit_hash": "18b402c42b5eaabc6161c4076fdf37e670872f91c5019fa86b8657e26ed4a3f0", + "prepared_review_unit_hash": "ee9699a6e05df51cefcd3ff99a59772b5d85ecf2fa9c41f77f9b4ee6d0558a79", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36456, - "estimated_input_tokens": 9114, - "estimated_eval_tokens": 18484 + "serialized_chars": 36468, + "estimated_input_tokens": 9117, + "estimated_eval_tokens": 18490 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of false-positive similarity matches across unrelated bugs. The only clear duplicate pair is the repeated Qwen3.5 MoE `from_pretrained` regression; the rest differ by model, subsystem, or failure mode.", - "confidence": 0.93, - "canonical_issue_reason": "Issue 45310 is the earliest and clearest report of the only true duplicate pair (`transformers>=5.4.0` / Qwen3.5 MoE `from_pretrained` failure), so it is the best issue to anchor that duplicate set.", + "summary": "Nearly all soft pairs are false positives from broad lexical similarity. The only clear duplicate is the two identical Qwen3.5-MoE from_pretrained reports (issue:45310 and issue:45478).", + "confidence": 0.89, + "canonical_issue_reason": "issue:45310 is the better canonical: it is the earlier report of the exact same Qwen3.5-MoE from_pretrained regression, while issue:45478 appears to be a later duplicate.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45310 is the strongest standalone representative here because it describes a concrete, version-scoped regression with a reproducible model-specific failure and matches its duplicate cleanly.", + "best_issue_reason": "issue:45310 is the strongest representative because it is a clean, exact duplicate target with the earliest context and same concrete failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:29127", "right": "issue:32090", "accept": false, - "reason": "Different problems: LayoutLMv3 error-message clarity vs Trainer GPU broadcast failing on NoneType." + "reason": "Different subsystems and failures: LayoutLMv3 error clarity vs Trainer _gpu_broadcast_one(NoneType)." }, { "left": "issue:43335", "right": "issue:45083", "accept": false, - "reason": "Different code paths: SwitchTransformers config construction vs qwen3_omni_moe feature-length helper behavior." + "reason": "SwitchTransformers sparse-layer config bug vs qwen3_omni_moe feature-length helper behavior; unrelated code paths." }, { "left": "issue:43720", "right": "issue:44112", "accept": false, - "reason": "Different bugs: BitNet packed-weight loading vs a GraniteSpeech CI device-override test failure." + "reason": "BitNet accelerate-loading unpacking bug vs GraniteSpeech stale device override CI failure; not the same issue." }, { - "left": "issue:43335", - "right": "issue:45593", + "left": "issue:37428", + "right": "issue:44351", "accept": false, - "reason": "No shared underlying defect; sparse-layer config creation and D-FINE auxiliary-loss handling are unrelated." + "reason": "Missing flash-attention utility import vs missing HybridCache import; distinct API breakages." }, { "left": "issue:44514", "right": "issue:44521", "accept": false, - "reason": "Both involve `apply_chat_template`, but one is a batching/padding crash while the other is zeroed assistant masks for multimodal input; too different to merge." + "reason": "Both involve chat templates, but one is a batched padding=False crash and the other is zero assistant masks for multimodal inputs." }, { "left": "issue:43335", - "right": "issue:44112", + "right": "issue:45593", "accept": false, - "reason": "SwitchTransformers sparse-layer config bug and GraniteSpeech stale device-override CI failure are unrelated." + "reason": "SwitchTransformers sparse-layer creation vs D-FINE auxiliary-loss omission; different model families and bugs." }, { - "left": "issue:37428", - "right": "issue:44351", + "left": "issue:43335", + "right": "issue:44112", "accept": false, - "reason": "Both are import errors, but for different symbols and modules (`_flash_supports_window_size` vs `HybridCache`)." + "reason": "Sparse-layer config issue vs GraniteSpeech CI device-override failure; no shared concrete bug." }, { "left": "issue:45290", "right": "issue:46326", "accept": false, - "reason": "Both touch chat templating/tokenization, but one is tool-call handling with no content and the other is `continue_final_message` behavior in mistral_common tokenizers." + "reason": "Both are chat-template related, but the tool-call/no-content crash is not the same as continue_final_message behavior for mistral_common tokenizers." }, { "left": "issue:45127", "right": "issue:45245", "accept": false, - "reason": "Model-collapse after LoRA merge is unrelated to the 2^24 category-count limit error." + "reason": "LoRA/extended-vocab collapse vs category-cardinality runtime limit; unrelated." }, { - "left": "issue:34689", - "right": "issue:35141", + "left": "issue:45588", + "right": "issue:45859", "accept": false, - "reason": "Different failures: Llama 3.2 Vision loading regression vs embedding reinitialization after resizing tokens." + "reason": "Flash-attention s_aux=None AttributeError vs CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding; different code paths." }, { - "left": "issue:45588", - "right": "issue:45859", + "left": "issue:34689", + "right": "issue:35141", "accept": false, - "reason": "Different concrete code paths: flash_attention sinkless `s_aux=None` crash vs Qwen3.5 rotary embedding CPU-offload incompatibility." + "reason": "Model-loading regression for Llama 3.2 vision vs output embedding reinit after resize; not the same bug." }, { "left": "issue:42490", "right": "issue:43854", "accept": false, - "reason": "A trailing-slash local-path warning and a GLM-4.7-Flash load issue are not the same bug." + "reason": "Local-path trailing-slash warning vs GLM-4.7-Flash loading failure; different symptoms and causes." }, { "left": "issue:46431", "right": "issue:46566", "accept": false, - "reason": "Processor chat-template absence and DiffusionGemma meta-device loading are unrelated." + "reason": "Missing chat template error vs DiffusionGemma meta-device placement bug; unrelated." }, { "left": "issue:45310", "right": "issue:45478", "accept": true, - "reason": "Exact same reported regression: `transformers>=5.4.0` causes Qwen3.5 MoE `from_pretrained` to fail." + "reason": "Exact same title and same concrete Qwen3.5-MoE from_pretrained failure; clear duplicate pair." }, { "left": "issue:44448", "right": "issue:44779", "accept": false, - "reason": "Both are version-regression reports, but on different models/subsystems and different observable output failures." + "reason": "Pegasus v4/v5 output mismatch vs Deepseek tokenizer regression; similar area but different models and bugs." }, { "left": "issue:43819", "right": "issue:44493", "accept": false, - "reason": "DAC forward/latents mismatch is unrelated to unexpected position-id keys during model loading." + "reason": "DAC from_latents/STE mismatch vs unexpected position-id key warnings across many models; different problems." }, { - "left": "issue:26413", - "right": "issue:31501", + "left": "issue:39692", + "right": "issue:43526", "accept": false, - "reason": "Checkpoint-resume iterator failure is not the same as FineWeb training failing to start." + "reason": "SigLIP2 documentation example issues vs BeitImageProcessorFast reduce_labels bug; docs vs processor logic." }, { - "left": "issue:39692", - "right": "issue:43526", + "left": "issue:26413", + "right": "issue:31501", "accept": false, - "reason": "SigLIP2 docs errors and BeitImageProcessorFast label reduction are separate image-processing issues." + "reason": "resume_from_checkpoint epoch_iterator failure vs FineWeb training start failure; not the same root cause." }, { "left": "issue:43606", "right": "issue:44857", "accept": false, - "reason": "Device-mismatch with CPU offload and float16 AMP loss crashing are different runtime failures." + "reason": "CPU offload device mismatch in bark-small vs float16 AMP crash in LwDetrImageLoss; unrelated." }, { "left": "issue:28218", "right": "issue:32090", "accept": false, - "reason": "Tokenizer extra-space behavior and Trainer `_gpu_broadcast_one` NoneType failure are unrelated." + "reason": "Tokenizer added-space bug vs Trainer broadcast NoneType error; different components." }, { "left": "issue:45459", "right": "issue:46234", "accept": false, - "reason": "Tokenizer protobuf error masking and a missing import-utils availability helper are different import/tokenizer concerns." + "reason": "Tokenizer protobuf error masking vs missing import_utils availability helper; unrelated." }, { - "left": "issue:43701", - "right": "issue:45083", + "left": "issue:39692", + "right": "issue:44493", "accept": false, - "reason": "Checkpoint key mismatch is unrelated to qwen3_omni_moe feature-length helper behavior." + "reason": "SigLIP2 documentation example bugs vs many models showing unexpected position-id keys; not the same issue." }, { - "left": "issue:39692", - "right": "issue:44493", + "left": "issue:43701", + "right": "issue:45083", "accept": false, - "reason": "SigLIP2 documentation mistakes do not match the unexpected position-id key issue." + "reason": "resume_from_checkpoint key mismatch vs qwen3_omni_moe helper behavior; different failure modes." }, { "left": "issue:44617", "right": "issue:46697", "accept": false, - "reason": "CUDA OOM in Sam3Video and Swin2SR over-padding are completely different bugs." + "reason": "Sam3Video OOM vs Swin2SR over-padding; unrelated." }, { "left": "issue:44206", "right": "issue:45823", "accept": false, - "reason": "Unsupported `center` arg in LasrFeatureExtractor and Gemma4 PLE device mismatch are unrelated." + "reason": "Unsupported center arg in LasrFeatureExtractor vs Gemma4 PLE device mismatch; different bugs." }, { "left": "issue:32090", "right": "issue:33357", "accept": false, - "reason": "Trainer NoneType broadcast failure and MacOS bus error in CLIP loading have no shared code-path." + "reason": "Trainer broadcast NoneType vs MacOS bus error on CLIP model load; unrelated." }, { "left": "issue:42490", "right": "issue:43575", "accept": false, - "reason": "Local-path import warning and tensor-parallel OOM while loading Qwen2-57B are different issues." + "reason": "Local-path import warning vs Qwen2-57B-A14B-Instruct tensor-parallel OOM; different issues." }, { "left": "issue:44360", "right": "issue:44561", "accept": false, - "reason": "DSA indexer activation discussion and removal of `is_torch_fx_available` breaking remote-code models are unrelated." + "reason": "DSA indexer ReLU discussion vs removal of is_torch_fx_available breaking trust_remote_code models; not the same bug." }, { "left": "issue:44857", "right": "issue:45072", "accept": false, - "reason": "Both mention numeric precision, but one is an image-loss crash and the other is bfloat16 dtype mismatch in different models." + "reason": "LwDetrImageLoss AMP/CUDA crash vs SwitchTransformers/TimmWrapperModel bfloat16 dtype mismatch; different model paths." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative on the soft-edge pairs. The only accepted pair, issue:45310/issue:45478, is an exact duplicate by title and topic, and the remaining verdicts are appropriately rejected as different bugs, models, or code paths." + "feedback": "The summary is grounded in the packet: issue:45310 and issue:45478 have the same title and clearly look like the same Qwen3.5-MoE from_pretrained regression, with 45310 earlier, so the canonical choice is well supported. The soft-pair rejections are conservative and the reasons stay within the evidence in the titles/body summaries; no pair is overstated as a duplicate. No mergeability concern applies here since this is an issue-dedup/canonicalization judgment, not a PR-pair merge decision." }, "evaluator_used": true, "retried": false, @@ -9901,8 +9904,8 @@ "issue:10000|issue:15354", "issue:44977|issue:45229", "issue:43278|issue:43381", - "issue:43927|issue:45698", "issue:29870|issue:33290", + "issue:43927|issue:45698", "issue:44855|issue:46399", "issue:45290|issue:46655", "issue:40444|issue:43526", @@ -9911,217 +9914,217 @@ "issue:32090|issue:33290", "issue:46421|issue:46858", "issue:44561|issue:44779", - "issue:43873|issue:45229", "issue:44448|issue:44466", - "issue:43976|issue:45440", + "issue:43873|issue:45229", "issue:25251|issue:32090", + "issue:43976|issue:45440", "issue:42490|issue:43873", "issue:44568|issue:45491", "issue:43278|issue:43575", "issue:44871|issue:44977" ], - "prepared_review_unit_hash": "066a87a486993fa14a7e09edfde1cf61477723398f4bc54a3c7199eca45e4211", + "prepared_review_unit_hash": "f1bbd6991142d658798e50be6bd82ce3a623da2d2e5a3fb0b06e7daeb36b9ff4", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 37070, - "estimated_input_tokens": 9268, - "estimated_eval_tokens": 18792 + "serialized_chars": 37074, + "estimated_input_tokens": 9269, + "estimated_eval_tokens": 18794 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous; I don\u2019t see any true duplicate pairs. The links are mainly broad topical similarities (tokenizers, flash-attention, cache/compile, vision), but the underlying bugs differ.", - "confidence": 0.63, - "canonical_issue_reason": "Issue 45920 is the broadest concrete regression in the set: it describes a tokenizer bug affecting multiple models, with the widest impact and clearest cross-model pattern.", + "summary": "This cluster is a loose collection of mostly unrelated Transformers bug reports; none of the soft pairs look like true duplicates. The broadest representative issue is the v5 tokenizer regression in 45920.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45920 is the broadest, most central regression report here: it describes a v5 tokenizer bug affecting multiple model families with clear user impact and strong evidence.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45920 is the best overall representative because it is a high-signal, user-facing regression spanning several model families, while most other issues are narrower and model-specific.", + "best_issue_reason": "issue:45920 is the strongest standalone anchor for this set because it is broad, actionable, and tied to a concrete regression spanning several models.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43550", "right": "issue:45083", "accept": false, - "reason": "Different failures in different code paths: torch.compile/SDPA vs a multimodal feature-length helper bug." + "reason": "Different models and different failure modes: SDPA/torch.compile vs a helper-function bug in qwen3_omni_moe." }, { "left": "issue:27957", "right": "issue:34689", "accept": false, - "reason": "Both mention model loading/attention, but they are different models and different root causes." + "reason": "Both mention model/attention loading issues, but they are distinct model-specific regressions with different symptoms." }, { "left": "issue:29942", "right": "issue:35141", "accept": false, - "reason": "Failing Flash Attention tests is not the same bug as embedding reinitialization after resize." + "reason": "Flash Attention 2 test failures are not the same bug as embeddings being reinitialized after resize." }, { "left": "issue:38175", "right": "issue:44493", "accept": false, - "reason": "A SigLIP2 zero-probability issue and a position-id key warning are unrelated symptoms." + "reason": "One is zero probabilities in a specific model; the other is an unexpected position-id key regression across many models." }, { "left": "issue:29870", "right": "issue:32090", "accept": false, - "reason": "Bart generate() decoder mask bug and Trainer _gpu_broadcast_one NoneType error are distinct." + "reason": "UnboundLocalError in generate and a NoneType broadcast error in Trainer are unrelated code paths." }, { "left": "issue:43278", "right": "issue:43329", "accept": false, - "reason": "BF16->FP32 evaluation drift and undefined multimodal token variables are unrelated." + "reason": "Embedding dtype drift during evaluate is unrelated to undefined multimodal-token variables in the video branch." }, { "left": "issue:44448", "right": "issue:44977", "accept": false, - "reason": "Different models and different symptoms: Pegasus output mismatch vs Qwen3.5 flash-attention generation failure." + "reason": "Both are v5 regressions, but one is Pegasus output drift and the other is Qwen3.5 flash-attention generation failure." }, { "left": "issue:45920", "right": "issue:46192", "accept": false, - "reason": "Both are tokenizer-related, but one is wrong token IDs across models while the other is return_tensors handling." + "reason": "Tokenizer ID corruption across models is a different bug from return_tensors being ignored when text_kwargs is passed." }, { "left": "issue:45290", "right": "issue:45381", "accept": false, - "reason": "Chat-template tool-call crash and Qwen2.5-VL video position-id issue are separate bugs." + "reason": "Chat-template tool-call handling and Qwen2.5-VL video position IDs are separate features and failure modes." }, { "left": "issue:10000", "right": "issue:15354", "accept": false, - "reason": "German DistilBertModel issue and GeneratorExp torch.jit.script export failure are unrelated." + "reason": "German DistilBertModel error and GeneratorExp torch.jit.script export failure are unrelated." }, { "left": "issue:44977", "right": "issue:45229", "accept": false, - "reason": "Flash-attention generation bug and Gemma4 multi-GPU CUDA OOM are different problems." + "reason": "Flash-attention generation issues are not the same as a multi-GPU CUDA OOM on Gemma4." }, { "left": "issue:43278", "right": "issue:43381", "accept": false, - "reason": "Evaluation dtype drift is not the same as gradient checkpointing being disallowed in eval mode." + "reason": "Evaluate-time dtype change and gradient-checkpointing-in-eval are different problems." }, { - "left": "issue:43927", - "right": "issue:45698", + "left": "issue:29870", + "right": "issue:33290", "accept": false, - "reason": "Config save/load losing token IDs and loading the wrong custom module after save_pretrained are different persistence bugs." + "reason": "These are different runtime issues: generate-time decoder mask bug vs optimizer/deepspeed OOM." }, { - "left": "issue:29870", - "right": "issue:33290", + "left": "issue:43927", + "right": "issue:45698", "accept": false, - "reason": "Generation-time decoder mask error and Adafactor/DeepSpeed OOM are unrelated." + "reason": "Both involve save/load behavior, but one loses custom token IDs while the other loads the wrong custom module." }, { "left": "issue:44855", "right": "issue:46399", "accept": false, - "reason": "Python 3.13 import parsing error and Qwen3.5 missing-decorator registration error are not the same bug." + "reason": "Python 3.13 parsing/IndentationError in DebertaV2 is unrelated to the Qwen3.5 kernel registration error." }, { "left": "issue:45290", "right": "issue:46655", "accept": false, - "reason": "Chat template tool-call crash and SmolVLM video padding bug are separate code paths." + "reason": "Chat templating/tool calls and SmolVLM video padding are distinct bugs in different subsystems." }, { "left": "issue:40444", "right": "issue:43526", "accept": false, - "reason": "IterableDataset multi-image finetuning failure and BEiT label reduction bug are different processor/training issues." + "reason": "Multi-image IterableDataset finetuning failure is unrelated to BEiT image label reduction behavior." }, { "left": "issue:44857", "right": "issue:45588", "accept": false, - "reason": "AMP/CUDA crash in LwDetrImageLoss and flash_attention sink-less-model AttributeError are unrelated." + "reason": "Loss instability under float16 AMP is not the same as flash_attention.py crashing on s_aux=None." }, { "left": "issue:37428", "right": "issue:42175", "accept": false, - "reason": "Importing a missing flash-attention helper and missing TensorFlow in the torch extra are different packaging/runtime issues." + "reason": "A missing flash-attention utility import and a packaging/backend install issue are unrelated." }, { "left": "issue:32090", "right": "issue:33290", "accept": false, - "reason": "Trainer broadcast TypeError and Adafactor/DeepSpeed OOM are not the same underlying bug." + "reason": "Trainer broadcast TypeError and deepspeed/adafactor OOM are different failures." }, { "left": "issue:46421", "right": "issue:46858", "accept": false, - "reason": "Both involve cache/compile, but one is lazy-init recompilation and the other is device-tensor seq-length arithmetic; related area, distinct root cause." + "reason": "Both touch cache/prefill, but one is recompilation churn and the other is a device-tensor arithmetic bug in static cache." }, { "left": "issue:44561", "right": "issue:44779", "accept": false, - "reason": "Removal of is_torch_fx_available breaks trust_remote_code, while the other is a DeepSeek tokenizer regression." - }, - { - "left": "issue:43873", - "right": "issue:45229", - "accept": false, - "reason": "Quantization offloading behavior and Gemma4 multi-GPU OOM are different memory-management problems." + "reason": "Removing is_torch_fx_available and DeepSeek tokenizer regressions are unrelated v5 breakages." }, { "left": "issue:44448", "right": "issue:44466", "accept": false, - "reason": "Pegasus output divergence and lm_head serialization inconsistency are different bugs." + "reason": "Different concrete bugs: Pegasus output mismatch vs inconsistent lm_head.weight serialization tied to device." }, { - "left": "issue:43976", - "right": "issue:45440", + "left": "issue:43873", + "right": "issue:45229", "accept": false, - "reason": "Python version support issue and DeepseekV3MoE implementation divergence are unrelated." + "reason": "Quantization/offloading behavior and Gemma4 multi-GPU OOM are not the same underlying issue." }, { "left": "issue:25251", "right": "issue:32090", "accept": false, - "reason": "Pipeline top_k nesting change and Trainer broadcast NoneType error are unrelated." + "reason": "Pipeline top_k output shape change and Trainer NoneType broadcast error are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility and DeepseekV3MoE divergence are different classes of bugs." }, { "left": "issue:42490", "right": "issue:43873", "accept": false, - "reason": "Local-path import warning and quantization offloading issues are unrelated." + "reason": "Local-path import warning with trailing slash is unrelated to quantization offloading." }, { "left": "issue:44568", "right": "issue:45491", "accept": false, - "reason": "Tokenizer add_special_tokens behavior and Gemma3 NaN embeddings are different bugs." + "reason": "Tokenizer special-token behavior and Gemma3 NaN embeddings are distinct problems." }, { "left": "issue:44871", "right": "issue:44977", "accept": false, - "reason": "Gemma eos_token_id config mismatch and Qwen3.5 flash-attention generation failure are unrelated." + "reason": "EOS-token config inconsistency and Qwen3.5 flash-attention generation failure do not share the same underlying cause." } ] }, @@ -10197,9 +10200,9 @@ "soft_pairs": [ "issue:43381|issue:43701", "issue:45588|issue:45823", + "issue:33357|issue:39692", "issue:29127|issue:34567", "issue:41720|issue:45229", - "issue:33357|issue:39692", "issue:44977|issue:45563", "issue:45198|issue:45803", "issue:44361|issue:45198", @@ -10228,88 +10231,88 @@ "issue:45372|issue:45459", "issue:42490|issue:43381" ], - "prepared_review_unit_hash": "3305b90ecfd90f5b79c1b980b2ef08ac804c229cb05cdf59d36be4391d3cf16f", + "prepared_review_unit_hash": "e54f594dfcd80adaff1f6027f8321b58b810dbe4919e2c1dc72e818870e7c2a1", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 32, - "serialized_chars": 37294, - "estimated_input_tokens": 9324, - "estimated_eval_tokens": 18904 + "serialized_chars": 37289, + "estimated_input_tokens": 9323, + "estimated_eval_tokens": 18902 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is heterogeneous; none of the soft-similarity pairs look like the same underlying bug or change. I would not merge any pair, and there are no PRs in the cluster.", - "confidence": 0.66, - "canonical_issue_reason": "issue:45412 is the clearest self-contained bug report in the set: it describes a concrete RT-DETR memory leak with real user impact and reproducible behavior.", + "summary": "This set is heterogeneous: the soft pairs share at most a broad subsystem or model family, but none look like the same underlying bug/change. I would not merge any of the proposed pairs.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 46693 is the broadest active regression report here and has inbound references; the rest are mostly unrelated one-off bugs.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45412 is the best representative issue if one must anchor the cluster, because it is a specific runtime defect rather than a broad or documentation-only complaint.", + "best_issue_reason": "Issue 46693 has the clearest current user impact, explicit regression framing, and ongoing activity, so it is the best representative issue if one must be chosen.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43381", "right": "issue:43701", "accept": false, - "reason": "Gradient-checkpointing eval-mode behavior and resume-from-checkpoint key mismatches are different training/runtime problems." + "reason": "Different problems: eval-mode gradient checkpointing vs checkpoint resume key naming mismatch." }, { "left": "issue:45588", "right": "issue:45823", "accept": false, - "reason": "One is a flash-attention integration crash on s_aux=None; the other is a Gemma4 device-map mismatch during forward." + "reason": "Both touch Gemma/Gemma4 paths, but one is a flash-attention crash and the other is a device-mapping mismatch; different root causes." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "CLIP MacOS bus error and SigLIP2 doc/example errors are unrelated failures." }, { "left": "issue:29127", "right": "issue:34567", "accept": false, - "reason": "LayoutLMv3 error-message clarity and TrainerState token-count updates are unrelated concerns." + "reason": "LayoutLMv3 error messaging and TrainerState token counting are separate features/bugs." }, { "left": "issue:41720", "right": "issue:45229", "accept": false, - "reason": "Both mention CUDA failures, but one is an auto-device-map assert and the other is a multi-GPU OOM; different root causes." - }, - { - "left": "issue:33357", - "right": "issue:39692", - "accept": false, - "reason": "A MacOS CLIP bus error is not the same bug as a SigLIP2 documentation example/model-processor mismatch." + "reason": "Both are memory/runtime failures, but Qwen3 device-map cuda assert and Gemma4 multi-GPU OOM are different code paths." }, { "left": "issue:44977", "right": "issue:45563", "accept": false, - "reason": "Flash-attention generation failure and a stale paged-generate warning are different issues." + "reason": "Flash-attention generation failure and stale warning text are unrelated issues." }, { "left": "issue:45198", "right": "issue:45803", "accept": false, - "reason": "Wav2Vec2 save/tokenization failures are unrelated to a bare-except in FuyuBatchFeature.convert_to_tensors()." + "reason": "Wav2Vec2 save/tokenization failure is unrelated to FuyuBatchFeature's bare except swallowing errors." }, { "left": "issue:44361", "right": "issue:45198", "accept": false, - "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failures affect different code paths." + "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure do not share the same underlying bug." }, { "left": "issue:44186", "right": "issue:45198", "accept": false, - "reason": "LayoutLMv2 NER/padding crashes do not match the Wav2Vec2 save/tokenization bug." + "reason": "LayoutLMv2 tokenizer padding/NER crashes and Wav2Vec2 serialization/tokenization failures are different defects." }, { "left": "issue:43653", @@ -10321,133 +10324,133 @@ "left": "issue:43611", "right": "issue:46121", "accept": false, - "reason": "base_model_prefix loading regressions and rope-validation TypeErrors are different loader/config bugs." + "reason": "base_model_prefix loading behavior and rope-parameter dict conversion are separate model/config issues." }, { "left": "issue:29870", "right": "issue:34689", "accept": false, - "reason": "BART generate() decoder mask initialization and Llama 3.2 vision model loading are separate failures." + "reason": "Bart generate decoder mask bug and Llama 3.2 vision model-loading regression are not the same issue." }, { "left": "issue:45372", "right": "issue:45588", "accept": false, - "reason": "Gemma4 processor import failure is not the same as the flash-attention s_aux=None crash." + "reason": "Gemma4 processor import dependency failure and flash-attention crash on s_aux=None are different failure points." }, { "left": "issue:42915", "right": "issue:42994", "accept": false, - "reason": "FineGrainedFP8Config failure and quantized-model saving failure involve different quantization stages." + "reason": "Both are quantization-related, but one is FP8 config loading and the other is quantized saving; not the same bug." }, { "left": "issue:43756", "right": "issue:45440", "accept": false, - "reason": "Smollm3 RoPE-layer handling and DeepseekV3MoE divergence are distinct architecture/implementation issues." + "reason": "Smollm3 RoPE-layer behavior and DeepseekV3 implementation divergence are unrelated model issues." }, { "left": "issue:43653", "right": "issue:44442", "accept": false, - "reason": "BigBirdTokenizer special-token registration has no overlap with AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + "reason": "BigBird tokenizer special-token bug and AutoTokenizer loading FastSpeech2ConformerTokenizer are different tokenizer failures." }, { "left": "issue:29942", "right": "issue:30064", "accept": false, - "reason": "Flash Attention 2 test failures and void-segmentation-map processing are unrelated." + "reason": "Flash Attention 2 test failures and void segmentation map processing are unrelated." }, { "left": "issue:45372", "right": "issue:45406", "accept": false, - "reason": "Gemma4 processor import errors are not the same as transformers serve crashing on missing _tokenizer." + "reason": "Gemma4 processor import failure and Gemma4Processor missing _tokenizer in serve are distinct bugs." }, { "left": "issue:44206", "right": "issue:44479", "accept": false, - "reason": "Unsupported center-arg feature extraction and Qwen video-input regression are different code paths." + "reason": "Unsupported center argument in LasrFeatureExtractor and Qwen video-input regression are separate regressions." }, { "left": "issue:43412", "right": "issue:45588", "accept": false, - "reason": "ExecuTorch export guard/erfinv problems are unrelated to the flash-attention sink-less-model crash." + "reason": "Gemma3n executorch export issues and flash-attention s_aux=None crash are unrelated." }, { "left": "issue:45459", "right": "issue:45712", "accept": false, - "reason": "Masking tokenizer errors when protobuf is missing is a different defect from leftover dummy classes in dummy_pt_objects.py." + "reason": "Tokenizer error masking from protobuf import handling and leftover dummy classes in dummy_pt_objects.py are unrelated maintenance bugs." }, { "left": "issue:45491", "right": "issue:46693", "accept": false, - "reason": "Gemma3 NaN embeddings and FA2 long-context performance regression are not the same bug." + "reason": "Both involve Gemma/FA2 generation behavior, but one is NaN embeddings from sliding-window padding and the other is a performance regression; different symptoms and likely fixes." }, { "left": "issue:44265", "right": "issue:45198", "accept": false, - "reason": "torch.export failures from torch_compilable_check are unrelated to Wav2Vec2 save/tokenization failures." + "reason": "torch.export.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." }, { "left": "issue:29127", "right": "issue:33357", "accept": false, - "reason": "LayoutLMv3 error messaging and CLIP Mac bus errors have no shared concrete failure mode." + "reason": "LayoutLMv3 box-content error messaging and CLIP MacOS bus error are unrelated." }, { "left": "issue:43065", "right": "issue:43994", "accept": false, - "reason": "A dummy Conv2d in Sam3PixelDecoder is unrelated to SigLIP2 AutoModel/pipeline nonsense outputs." + "reason": "Sam3PixelDecoder dummy Conv2d cleanup and SigLIP2 nonsensical outputs are different models and bugs." }, { "left": "issue:43065", "right": "issue:45412", "accept": false, - "reason": "A placeholder convolution layer and an RT-DETR memory leak are different bugs." + "reason": "Dummy class cleanup in Sam3PixelDecoder and RT-DETR memory retention are unrelated." }, { "left": "issue:45245", "right": "issue:45563", "accept": false, - "reason": "Category-count runtime limits and stale generation warnings do not describe the same issue." + "reason": "Category-cardinality runtime error and stale warning text are unrelated." }, { "left": "issue:43575", "right": "issue:43701", "accept": false, - "reason": "Tensor-parallel OOM on Qwen2-57B and resume-from-checkpoint key mismatches are unrelated." + "reason": "Tensor-parallel OOM on Qwen2-57B and checkpoint resume key mismatch are different problems." }, { "left": "issue:43066", "right": "issue:44462", "accept": false, - "reason": "Tokenizer decoder-type changes in v5 and AutoTokenizer ignoring tokenizer.json are distinct loading behaviors." + "reason": "Wrong tokenizer decoder type in v5 and AutoTokenizer ignoring tokenizer.json are both tokenizer-related but not the same root cause." }, { "left": "issue:45440", "right": "issue:45593", "accept": false, - "reason": "DeepseekV3MoE divergence and D-FINE auxiliary-loss handling are separate model bugs." + "reason": "DeepseekV3 implementation divergence and D-FINE auxiliary-loss behavior are unrelated model bugs." }, { "left": "issue:45372", "right": "issue:45459", "accept": false, - "reason": "Gemma4 processor import failure and protobuf-related tokenizer error masking are different root causes." + "reason": "Gemma4 processor import failure and protobuf exception masking for tokenizers are unrelated." }, { "left": "issue:42490", "right": "issue:43381", "accept": false, - "reason": "A local-path trailing-slash import warning is unrelated to gradient checkpointing being unusable in eval mode." + "reason": "Local-path import warning and eval-mode gradient checkpointing are unrelated." } ] }, @@ -10544,222 +10547,222 @@ "issue:45588|issue:45600", "issue:43381|issue:43873", "issue:44514|issue:45381", - "issue:43761|issue:45859", "issue:43646|issue:44077", + "issue:43761|issue:45859", "issue:44977|issue:45245", "issue:44743|issue:44811", "issue:44442|issue:44898", "issue:43262|issue:44206", "issue:42490|issue:44743" ], - "prepared_review_unit_hash": "4e0532bb25d565a10b69740d6768dc40549c090d2347b24f907179a17b5a2ab2", + "prepared_review_unit_hash": "612aa02ee5b7363d4ded75780d6f1193c3fc26bbda03b66a304ae0aa5f71ffcb", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36563, - "estimated_input_tokens": 9141, - "estimated_eval_tokens": 18538 + "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 mostly share only broad wording (memory, Flash Attention, Qwen, load failures) but describe different models, APIs, or failure modes. I would reject all candidate duplicates; none look like the same concrete bug/change.", - "confidence": 0.93, - "canonical_issue_reason": "issue:43761 is the clearest, most self-contained bug report in the set: it names the exact regression, the affected API, and the observable wrong result (`hidden_states=None` despite `output_hidden_states=True`).", + "summary": "The soft edges are mostly false positives driven by broad keyword overlap (Flash Attention, OOM, Qwen, loading, tokenizer, etc.), but they refer to different concrete bugs or features. I do not see a true duplicate cluster here.", + "confidence": 0.94, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:43761 is the strongest issue to anchor triage because it is specific, reproducible, and narrowly scoped, which makes duplicate matching less ambiguous than the broader OOM/loading reports.", + "best_issue_reason": null, "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:29127", "right": "issue:35141", "accept": false, - "reason": "Different subsystems and symptoms: LayoutLMv3 error clarity vs token embedding reinitialization after resize." + "reason": "Different problems: LayoutLMv3 error-message clarity vs token embedding resize/post_init reinitialization." }, { "left": "issue:43421", "right": "issue:45491", "accept": false, - "reason": "Tokenizer post-processor updates at runtime is unrelated to Gemma3 NaNs from sliding-window/all-padding behavior." + "reason": "Tokenizers post-processor sync issue is unrelated to Gemma3 NaN embeddings from sliding-window/all-padding behavior." }, { "left": "issue:43994", "right": "issue:45412", "accept": false, - "reason": "SigLIP2 nonsensical inference and RT-DETR memory release are different bug classes and code paths." + "reason": "SIGLIP2 bad outputs and RT-DETR memory release are unrelated model failures." }, { "left": "issue:45412", "right": "issue:46369", "accept": false, - "reason": "Both mention memory, but one is GC/release behavior and the other is runtime overflow in sam3-litetext." + "reason": "Both mention memory, but one is a deletion/GC leak and the other is a model-call memory overflow; different code paths." }, { "left": "issue:43575", "right": "issue:44387", "accept": false, - "reason": "Both are OOM reports, but one is TP model loading and the other is int4 quantization memory growth in 5.x." + "reason": "Both are OOMs, but one is tensor-parallel load OOM and the other is int4/Transformers 5.x reserved-memory regression." }, { "left": "issue:44442", "right": "issue:45072", "accept": false, - "reason": "AutoTokenizer loading failure for one tokenizer is unrelated to dtype mismatch inference bugs in another subsystem." + "reason": "Tokenizer loading failure and dtype mismatch in inference are unrelated." }, { "left": "issue:44464", "right": "issue:45584", "accept": false, - "reason": "Compiled-forward chunked generation inconsistency is unrelated to Whisper empty-transcription failure after special-token alignment." + "reason": "Compiled-forward chunked generation mismatch is not the same as Whisper empty-transcription handling." }, { "left": "issue:42915", "right": "issue:43931", "accept": false, - "reason": "Qwen3Moe FP8 config failure and Qwen3-VL weight-shape mismatch are different model-loading problems." + "reason": "Different Qwen model families and different failure modes: FineGrainedFP8Config vs weight shape mismatch." }, { "left": "issue:43526", "right": "issue:45834", "accept": false, - "reason": "BeitImageProcessorFast label reduction bug and Kosmos2.5 long OCR index error do not share the same failure path." + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Kosmos2.5 long-OCR indexing." }, { "left": "issue:43526", "right": "issue:46693", "accept": false, - "reason": "Label-array reduction bug is unrelated to FA2 long-context performance regression." + "reason": "Different processors/models and different bug classes; only superficial image/model overlap." }, { "left": "issue:43824", "right": "issue:44279", "accept": false, - "reason": "A missing import for a specific class is too different from a vague dependency issue to treat as the same bug." + "reason": "ImportError for a missing symbol is not the same as a generic dependency issue." }, { "left": "issue:27957", "right": "issue:29127", "accept": false, - "reason": "Flash Attention support for XLM-Roberta is unrelated to a LayoutLMv3 error-message clarity issue." + "reason": "Flash Attention support for XLM-Roberta and LayoutLMv3 error-message clarity are unrelated." }, { "left": "issue:27957", "right": "issue:33290", "accept": false, - "reason": "FA2 support and adafactor/deepspeed OOM are different problems with different triggers." + "reason": "Model support for FA2 and Deepspeed Adafactor OOM are different issues." }, { "left": "issue:45250", "right": "issue:46640", "accept": false, - "reason": "Both are generic FA2 titles with no concrete evidence they refer to the same underlying bug." + "reason": "Both mention Flash Attention 2, but the titles do not indicate the same concrete bug or fix." }, { "left": "issue:43994", "right": "issue:46369", "accept": false, - "reason": "Different models and symptoms: inference quality issue vs memory overflow on model call." + "reason": "SIGLIP2 inference correctness and sam3-litetext memory overflow are unrelated." }, { "left": "issue:33290", "right": "issue:34689", "accept": false, - "reason": "Optimizer-related OOM and Llama 3.2 Vision loading regression are unrelated." + "reason": "Deepspeed Adafactor OOM and Llama 3.2 Vision model loading regression are different failures." }, { "left": "issue:43526", "right": "issue:44186", "accept": false, - "reason": "ImageProcessor label handling and LayoutLMv2 tokenizer crashes are separate preprocessing bugs." + "reason": "Different components: BeitImageProcessorFast label reduction vs LayoutLMv2Tokenizer crashes." }, { "left": "issue:33290", "right": "issue:35141", "accept": false, - "reason": "OOM in deepspeed/adafactor is not the same as embedding reinitialization after resize." + "reason": "OOM in optimizer/offload path is unrelated to embedding resize reinitialization." }, { "left": "issue:43334", "right": "issue:43531", "accept": false, - "reason": "Qwen3-VL pad_token_id loading error and Qwen3-MoE sliding-window behavior are different code paths." + "reason": "Qwen3-VL pad_token_id loading error and Qwen3-MoE sliding_window behavior are different bugs." }, { "left": "issue:46121", "right": "issue:46399", "accept": false, - "reason": "Rope-parameter JSON list handling and missing decorator on rotary-pos-emb registration are unrelated." + "reason": "RoPE validation TypeError and missing decorator for apply_rotary_pos_emb are distinct implementation issues." }, { "left": "issue:45588", "right": "issue:45600", "accept": false, - "reason": "Flash-attention crash on `s_aux=None` and stale auto_mappings references are different failures." + "reason": "Flash-attention crash on s_aux=None is unrelated to stale auto_mappings entries." }, { "left": "issue:43381", "right": "issue:43873", "accept": false, - "reason": "Gradient checkpointing eval-mode restriction is not the same as quantization offloading behavior." + "reason": "Gradient checkpointing eval-mode restriction and quantization offloading behavior are different concerns." }, { "left": "issue:44514", "right": "issue:45381", "accept": false, - "reason": "Batched chat-template padding crash and video vision_position_ids mismatch are different Qwen2.5-VL bugs." + "reason": "Batched chat-template padding crash and wrong video vision_position_ids in Qwen2.5-VL are different bugs." }, { - "left": "issue:43761", - "right": "issue:45859", + "left": "issue:43646", + "right": "issue:44077", "accept": false, - "reason": "CLIPVisionModel hidden_states regression and Qwen3_5Moe CPU-offload incompatibility are unrelated." + "reason": "Custom model initialization breakage in v5 is not the same as a patchtsmixer post_init policy issue." }, { - "left": "issue:43646", - "right": "issue:44077", + "left": "issue:43761", + "right": "issue:45859", "accept": false, - "reason": "Custom model initialization breakage and a model-specific post_init constraint are not clearly the same concrete bug." + "reason": "CLIP hidden_states regression and Qwen3_5Moe CPU offload incompatibility are unrelated." }, { "left": "issue:44977", "right": "issue:45245", "accept": false, - "reason": "Qwen3.5 flash-attention generation issue and category-count runtime error are unrelated." + "reason": "Flash-attention generation regression and category-cardinality RuntimeError are unrelated." }, { "left": "issue:44743", "right": "issue:44811", "accept": false, - "reason": "Qwen3 recurrent-state reset with cache is unrelated to Whisper batch_decode skip_special_tokens handling." + "reason": "Qwen3.5 recurrent-state reset and Whisper batch_decode skip_special_tokens handling are different code paths." }, { "left": "issue:44442", "right": "issue:44898", "accept": false, - "reason": "Tokenizer loading failure and Perceiver resolution/interpolation failure are different issues." + "reason": "Tokenizer loading and Perceiver interpolation failure are unrelated." }, { "left": "issue:43262", "right": "issue:44206", "accept": false, - "reason": "Chat-template sampling-rate default and unsupported `center` arg in feature extraction are separate audio bugs." + "reason": "Audio chat-template sampling-rate default and LASR feature extractor center-arg crash are different issues." }, { "left": "issue:42490", "right": "issue:44743", "accept": false, - "reason": "Local-path import warning has no meaningful overlap with Qwen3 recurrent-state reset behavior." + "reason": "Trailing-slash import warning is unrelated to Qwen3 recurrent-state reset." } ] }, @@ -10850,9 +10853,9 @@ "issue:43761|issue:44079", "issue:44190|issue:44496", "issue:43761|issue:44188", + "issue:44987|issue:46178", "issue:43931|issue:45216", "issue:44442|issue:45803", - "issue:44987|issue:46178", "issue:43232|issue:43388", "issue:42907|issue:45698", "issue:43701|issue:46506", @@ -10862,20 +10865,20 @@ "issue:43575|issue:43742", "issue:43606|issue:43701" ], - "prepared_review_unit_hash": "52f686079eb4f46de4ee3879a3c692196e28d94878999707507165b989de4f21", + "prepared_review_unit_hash": "06a6a0a421fe613969915c50bc3704e520f8aa8d22dad4a9bc2ee50439379f11", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36217, + "serialized_chars": 36218, "estimated_input_tokens": 9055, "estimated_eval_tokens": 18366 }, @@ -10884,180 +10887,180 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This is a heterogeneous collection of unrelated Transformers issues; none of the soft pairs look like true duplicates or the same fix target. No soft edges should be merged.", - "confidence": 0.97, - "canonical_issue_reason": null, + "summary": "The cluster is a broad false-positive mix of unrelated issue reports spanning model loading, generation, config, backend, and docs regressions. None of the proposed soft pairs look like the same underlying bug or change, so no pair should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No strong duplicate anchor stands out; if a representative is needed, issue 45698 is the broadest and most active save/load regression here, but the set is too heterogeneous to treat any item as a true canonical.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "Issue 45698 is the closest global representative because it is open, recent, and has the most discussion/inbound references, but it still does not unify the rest of the reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45003", "right": "issue:45440", "accept": false, - "reason": "One is a sys.modules access bug in modeling_utils; the other is a DeepseekV3MoE parity issue. Different code paths and symptoms." + "reason": "One is a sys.modules safety issue in modeling_utils; the other is a DeepSeekV3MoE implementation divergence. Different code paths and bug classes." }, { "left": "issue:43720", "right": "issue:44442", "accept": false, - "reason": "BitNet accelerate-loading/packed-weights issue vs tokenizer loading failure for FastSpeech2ConformerTokenizer; unrelated subsystems." + "reason": "BitNet accelerate-loading unpacking bug vs tokenizer auto-loading failure. Not the same underlying defect." }, { "left": "issue:42175", "right": "issue:46234", "accept": false, - "reason": "Packaging/backend extra omission is unrelated to a missing import_utils availability helper." + "reason": "Package/backend dependency omission vs a missing import_utils availability helper. Different targets." }, { "left": "issue:27957", "right": "issue:29870", "accept": false, - "reason": "XLM-R Flash Attention support and BART generate decoder_attention_mask UnboundLocalError are different model/generation bugs." + "reason": "Flash Attention 2 support for XLM-Roberta vs a BART generate() decoder mask initialization error. Unrelated." }, { "left": "issue:43493", "right": "issue:43827", "accept": false, - "reason": "SigLIP2 implementation mismatch vs docs still referencing pipeline(); one is code parity, the other documentation cleanup." + "reason": "SigLIP2 implementation mismatch vs docs still referencing removed pipeline() APIs. Different issue types." }, { "left": "issue:43316", "right": "issue:43525", "accept": false, - "reason": "Gemma3TextConfig API discrepancy and Llama4Config missing pad_token_id are separate config issues." + "reason": "Both are config-field discrepancies, but for different models and fields (`Gemma3TextConfig` API vs missing `pad_token_id` on Llama4Config)." }, { "left": "issue:44062", "right": "issue:45907", "accept": false, - "reason": "AddedToken special-argument duplication vs list_repo_files not forwarding revision; no shared underlying bug." + "reason": "Tokenizer AddedToken constructor error vs Hub `list_repo_files` revision forwarding. Unrelated." }, { "left": "issue:44077", "right": "issue:44479", "accept": false, - "reason": "patchtsmixer post_init validation and Qwen video-input regression are unrelated model/config problems." + "reason": "PatchTSMixer post_init validation vs a Qwen video-input regression. Different subsystems and failures." }, { "left": "issue:45325", "right": "issue:46697", "accept": false, - "reason": "Qwen2.5-VL rope index timing bug vs Swin2SR over-padding in the image processor; different components." + "reason": "Qwen2.5-VL rope/position-id scaling bug vs Swin2SR image over-padding. Not the same preprocessing logic." }, { "left": "issue:43873", "right": "issue:44871", "accept": false, - "reason": "Quantization offloading behavior is unrelated to Gemma-3 eos_token_id configuration mismatch." + "reason": "Quantization/offloading behavior issue vs Gemma-3 eos_token_id config mismatch. Different bugs." }, { "left": "issue:44855", "right": "issue:45137", "accept": false, - "reason": "Python 3.13 torch.jit parsing error in DebertaV2Model vs DeepSpeed ZeRO3 deque IndexError; distinct failure modes." + "reason": "Python 3.13 torch.jit parsing error vs DeepSpeed ZeRO3 deque underflow. Different layers." }, { "left": "issue:43643", "right": "issue:44913", "accept": false, - "reason": "AutoConfig trust_remote_code missing fields and GPTNeoXConfig rotary_pct reload regression are different config-serialization bugs." + "reason": "AutoConfig trust_remote_code field loss vs GPTNeoX rotary_pct persistence on reload. Both config-related, but distinct failures." }, { "left": "issue:43749", "right": "issue:43931", "accept": false, - "reason": "FSDP CPU RAM efficient loading regression vs Qwen3-VL weight-shape mismatch; not the same bug." + "reason": "FSDP CPU RAM-efficient loading regression vs Qwen3-VL weight-shape mismatch. Not the same root cause." }, { "left": "issue:43316", "right": "issue:45759", "accept": false, - "reason": "Gemma3TextConfig field mismatch and composite Qwen multimodal text_config unwrapping are separate API issues." + "reason": "Both touch config handling, but one is a Gemma3TextConfig API discrepancy and the other is AutoModelForCausalLM.from_config not unwrapping text_config for composite Qwen configs." }, { "left": "issue:43761", "right": "issue:44079", "accept": false, - "reason": "CLIPVisionModel not returning hidden_states and ModelOutput key assignment with None are related at most conceptually, but not the same concrete bug." + "reason": "CLIPVisionModel hidden_states regression vs ModelOutput key assignment bug. Unrelated." }, { "left": "issue:44190", "right": "issue:44496", "accept": false, - "reason": "Local dataset loading failure in an example script is unrelated to model_type/config loading for a custom model." + "reason": "Local dataset loading in a training script vs an unrecognized model/config error. Different failure domains." }, { "left": "issue:43761", "right": "issue:44188", "accept": false, - "reason": "Hidden-states regression in CLIPVisionModel vs attention kernel divergence under torch.compile; different code paths." + "reason": "CLIPVisionModel hidden_states regression vs attention-kernel branching under torch.compile. Different modules and symptoms." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "A model-loading failure for `physical-intelligence/fast` vs a backend mapping error for `tensorflow_text`. Different backend issues." }, { "left": "issue:43931", "right": "issue:45216", "accept": false, - "reason": "Weight-shape mismatch on load and incorrect checkpoint save_pretrained output are both Qwen3-related but distinct load/save problems." + "reason": "Qwen3-VL weight-shape mismatch on load vs a Qwen3.5 save_pretrained regression. Different phases and causes." }, { "left": "issue:44442", "right": "issue:45803", "accept": false, - "reason": "Tokenizer loading failure vs bare except in FuyuBatchFeature.convert_to_tensors(); no shared underlying defect." - }, - { - "left": "issue:44987", - "right": "issue:46178", - "accept": false, - "reason": "Model loading regression for physical-intelligence/fast is unrelated to tensorflow_text backend mapping errors." + "reason": "Tokenizer loading failure vs bare-except error swallowing in FuyuBatchFeature. Different code paths." }, { "left": "issue:43232", "right": "issue:43388", "accept": false, - "reason": "Generation kwargs sync after sync_gpus and gather_for_metrics label trimming are separate distributed-training bugs." + "reason": "Generation kwargs sync after sync_gpus vs gather_for_metrics dropping tuple labels. Different training utilities." }, { "left": "issue:42907", "right": "issue:45698", "accept": false, - "reason": "Saving dequantized Ministrals/Devstrals differs from loading the wrong custom module after save_pretrained; unrelated save/load failures." + "reason": "Both involve save/load, but one is dequantized-model saving and the other reloads the wrong custom module after save_pretrained. Not mergeable." }, { "left": "issue:43701", "right": "issue:46506", "accept": false, - "reason": "resume_from_checkpoint key mismatch and CPU/CUDA divergence in PPDocLayoutV3 are not the same issue." + "reason": "Checkpoint resume key mismatch vs CPU/CUDA output divergence in PPDocLayoutV3. Unrelated." }, { "left": "issue:42617", "right": "issue:43575", "accept": false, - "reason": "3d_parallel.py execution issue vs tensor-parallel OOM when loading a large Qwen model; different problem scope." + "reason": "3d_parallel.py runtime failure vs tensor-parallel OOM when loading a large model. Different symptoms and code paths." }, { "left": "issue:27957", "right": "issue:32090", "accept": false, - "reason": "Flash Attention 2 support for XLM-R is unrelated to Trainer/_gpu_broadcast_one NoneType broadcast failures." + "reason": "XLMRoberta Flash Attention support vs Trainer GPU broadcast NoneType error. Unrelated." }, { "left": "issue:43550", "right": "issue:43701", "accept": false, - "reason": "torch.compile SDPA failure in Bamba-9B-v2 and resume_from_checkpoint key mismatch are unrelated." + "reason": "torch.compile SDPA failure in Bamba-9B-v2 vs checkpoint resume key mismatch. Different issues." }, { "left": "issue:43575", "right": "issue:43742", "accept": false, - "reason": "Qwen2-57B-A14B-Instruct tensor-parallel OOM and MobileLLM-125M key error are distinct loading issues." + "reason": "Large-model TP OOM vs MobileLLM config key error. Not the same defect." }, { "left": "issue:43606", "right": "issue:43701", "accept": false, - "reason": "CPU offload device mismatch for bark-small does not match the checkpoint key-mismatch bug." + "reason": "CPU offload device mismatch vs checkpoint resume key mismatch. Different subsystems." } ] }, @@ -11136,8 +11139,8 @@ "issue:43295|issue:43825", "issue:16998|issue:22355", "issue:43295|issue:44871", - "issue:44464|issue:46421", "issue:44871|issue:45584", + "issue:44464|issue:46421", "issue:43316|issue:43716", "issue:44077|issue:44534", "issue:42175|issue:43316", @@ -11146,209 +11149,209 @@ "issue:45803|issue:46365", "issue:44162|issue:44792", "issue:43452|issue:44496", - "issue:44242|issue:46894", "issue:43404|issue:43441", + "issue:44242|issue:46894", "issue:45632|issue:46424", "issue:45803|issue:46170", "issue:44610|issue:46369", "issue:43650|issue:46202", - "issue:43761|issue:44479", "issue:43504|issue:45698", + "issue:43761|issue:44479", "issue:45542|issue:46234", "issue:43540|issue:44514", "issue:43352|issue:46640", "issue:43572|issue:46178" ], - "prepared_review_unit_hash": "04a4ce25cecad814e8902079cc5438ab032254c172e632d7dfaf5a5b6affd537", + "prepared_review_unit_hash": "e44b95a263a5811dfa9aff9aaf49390c57dd1da3a9aec52b1c2b90e94ae988a2", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35664, - "estimated_input_tokens": 8916, - "estimated_eval_tokens": 18088 + "serialized_chars": 35666, + "estimated_input_tokens": 8917, + "estimated_eval_tokens": 18090 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These are mostly superficial similarity hits, not true duplicates. The issues span different models, backends, and code paths, so I reject every soft edge.", - "confidence": 0.96, - "canonical_issue_reason": "No canonical issue: the cluster is a grab-bag of unrelated bugs and regressions (loading, config, multimodal preprocessing, backend imports, cache behavior, FlashAttention, etc.) with only broad keyword overlap.", + "summary": "The cluster is mostly a loose thematic grouping of distinct Transformers regressions and import/load issues, not true duplicates. The closest representative issue is the custom model loading/saving regression, but none of the soft pairs look like the same underlying bug/change.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45698 is the broadest and most central report around custom module resolution during from_pretrained/save_pretrained, which overlaps with several nearby loading-related reports in this set.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good global representative; any candidate would conflate distinct root causes and different fix paths.", + "best_issue_reason": "issue:45698 is the best overall representative because it is active, high-discussion, and captures the most reusable umbrella theme in the cluster: broken custom-model loading after serialization.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:25251", "right": "issue:28282", "accept": false, - "reason": "Different bugs: pipeline top_k output shape vs missing PyTorch backend import." + "reason": "One is a pipeline output-shape bug; the other is an AutoModel backend import error. Different code paths and symptoms." }, { "left": "issue:46234", "right": "issue:46291", "accept": false, - "reason": "Unrelated import_utils helper omission vs kernels version/revision requirement." + "reason": "Both are import-related, but one is a missing utility symbol and the other is a kernels dependency/version requirement. Not the same bug." }, { "left": "issue:43295", "right": "issue:43825", "accept": false, - "reason": "Both are v5 regressions, but one is processor/tokenizer image handling and the other is a pipeline error-message issue." + "reason": "Both involve v5 pipeline/processor regressions, but one is about processor.tokenizer behavior and the other is an error-message/task-support claim. Different underlying issues." }, { "left": "issue:16998", "right": "issue:22355", "accept": false, - "reason": "Different code paths: model_max_length question vs ONNX import failure." + "reason": "DeBERTa model_max_length handling and missing transformers.onnx are unrelated." }, { "left": "issue:43295", "right": "issue:44871", "accept": false, - "reason": "Custom model processor regression vs Gemma eos_token_id mismatch; not the same defect." + "reason": "Different models and different failures: processor.tokenizer regression versus Gemma-3 eos_token_id config mismatch." }, { - "left": "issue:44464", - "right": "issue:46421", + "left": "issue:44871", + "right": "issue:45584", "accept": false, - "reason": "Chunked generation inconsistency with compiled forward is separate from cache lazy-initialization recompilation." + "reason": "Gemma-3 EOS config inconsistency and Whisper empty-transcription failure are separate model-specific bugs." }, { - "left": "issue:44871", - "right": "issue:45584", + "left": "issue:44464", + "right": "issue:46421", "accept": false, - "reason": "Gemma token config mismatch and Whisper empty transcription are unrelated." + "reason": "Both mention chunked generation/cache, but one is inconsistent outputs from compiled forward while the other is lazy cache initialization causing recompilation. Not the same defect." }, { "left": "issue:43316", "right": "issue:43716", "accept": false, - "reason": "Gemma3TextConfig API discrepancy is not the same as Mistral-3 image dtype mismatch." + "reason": "Gemma3TextConfig API inconsistency and Mistral-3 dtype mismatch are unrelated model integration issues." }, { "left": "issue:44077", "right": "issue:44534", "accept": false, - "reason": "Optional post_init policy change is different from non-persistent buffer initialization corruption." + "reason": "One is about a model class still allowing optional post_init; the other is a v5 buffer initialization bug. Different changes and failures." }, { "left": "issue:42175", "right": "issue:43316", "accept": false, - "reason": "Backend install/dependency problem vs configuration API discrepancy." + "reason": "TensorFlow backend packaging and Gemma3TextConfig API discrepancy are unrelated." }, { "left": "issue:44871", "right": "issue:46050", "accept": false, - "reason": "Gemma eos_token_id mismatch and quantized save/load breakage do not share a root cause." + "reason": "Gemma-3 EOS config mismatch and sinq quantization save/load breakage are distinct serialization/config bugs." }, { "left": "issue:43646", "right": "issue:44913", "accept": false, - "reason": "Custom model initialization breakage differs from GPTNeoX rotary_pct reload regression." + "reason": "Both are config/initialization related, but one is a broad v5 custom initialization breakage and the other is GPTNeoX rotary_pct not persisting on reload. Not the same issue." }, { "left": "issue:45803", "right": "issue:46365", "accept": false, - "reason": "Bare except swallowing errors is unrelated to Trainer import conflict with datasets." + "reason": "Bare except swallowing interrupts in Fuyu conversion and Trainer import failing because of datasets are unrelated." }, { "left": "issue:44162", "right": "issue:44792", "accept": false, - "reason": "ESM2 model breakage and janus test failure are separate model-specific issues." + "reason": "ESM2 runtime breakage and a janus image-generation test failure are different model-specific problems." }, { "left": "issue:43452", "right": "issue:44496", "accept": false, - "reason": "gguf_file/from_pretrained handling is a different problem from unrecognized model config loading." + "reason": "Both are model-loading failures, but one is a gguf/from_pretrained breakage and the other is an unrecognized custom model config.json issue. Different root causes." }, { - "left": "issue:44242", - "right": "issue:46894", + "left": "issue:43404", + "right": "issue:43441", "accept": false, - "reason": "MoE load-balancing loss omission is not the same as ASR training loss double-shift." + "reason": "Mistral3 weight tying and Ministral FlashAttention support are separate model implementation issues." }, { - "left": "issue:43404", - "right": "issue:43441", + "left": "issue:44242", + "right": "issue:46894", "accept": false, - "reason": "lm_head weight tying bug and FlashAttention unsupported error are different failures." + "reason": "Both touch training loss, but one is MoE load-balancing loss gating and the other is a CohereASR double-shift bug. Not the same code-path problem." }, { "left": "issue:45632", "right": "issue:46424", "accept": false, - "reason": "Cache path collision is unrelated to max_cache_length being honored only for quantized cache." + "reason": "Both are cache-related, but one is a trust_remote_code path collision and the other is max_cache_length handling. Different bugs." }, { "left": "issue:45803", "right": "issue:46170", "accept": false, - "reason": "Bare except in tensor conversion is unrelated to Windows ROCm torch.distributed import issues." + "reason": "One is an exception-handling bug in tensor conversion; the other is a Windows/ROCm distributed import dependency issue." }, { "left": "issue:44610", "right": "issue:46369", "accept": false, - "reason": "Processor/model resolution mismatch is not the same as memory overflow in sam3-litetext." + "reason": "Processor image-size mismatch and memory overflow in sam3-litetext are unrelated." }, { "left": "issue:43650", "right": "issue:46202", "accept": false, - "reason": "A low-quality data request is not the same as Gemma pipeline roundtrip failure." + "reason": "A data request and a Gemma pipeline roundtrip failure are unrelated." }, { - "left": "issue:43761", - "right": "issue:44479", + "left": "issue:43504", + "right": "issue:45698", "accept": false, - "reason": "CLIPVision hidden_states regression and Qwen video-input regression affect different models and paths." + "reason": "Both involve loading pretrained/custom models, but one is a legacy field problem in Beit segmentation preset loading and the other is wrong custom module resolution after save/load. Different underlying bugs." }, { - "left": "issue:43504", - "right": "issue:45698", + "left": "issue:43761", + "right": "issue:44479", "accept": false, - "reason": "Legacy-field model loading bug vs wrong custom module loaded after save_pretrained are distinct." + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are separate v5 issues." }, { "left": "issue:45542", "right": "issue:46234", "accept": false, - "reason": "TensorFlow backend installation issue is different from a missing import_utils helper." + "reason": "TensorBoard/TensorFlow backend packaging and missing is_falcon_mamba_ssm_available are unrelated." }, { "left": "issue:43540", "right": "issue:44514", "accept": false, - "reason": "Qwen3OmniMoe video ValueError and batched chat_template padding=False crash are different multimodal bugs." + "reason": "Both are Qwen multimodal issues, but one is video-input processing in Qwen3OmniMoe and the other is batched chat-template padding behavior in Qwen2_5_VLProcessor. Not the same bug." }, { "left": "issue:43352", "right": "issue:46640", "accept": false, - "reason": "Nemotron explicitly lacking FlashAttention 2 support is not the same as a generic Flash Attention 2 issue." + "reason": "One is an explicit lack of Flash Attention 2 support for NemotronHForCausalLM; the other is a generic Flash Attention issue. Different specificity and likely different fixes." }, { "left": "issue:43572", "right": "issue:46178", "accept": false, - "reason": "StableLmConfig missing pad_token_idx is unrelated to the tensorflow_text backend registration error." + "reason": "StableLmConfig pad_token_idx regression and an undefined tensorflow_text backend are unrelated." } ] }, @@ -11422,10 +11425,10 @@ "issue:46897" ], "soft_pairs": [ - "issue:44792|issue:45600", - "issue:43388|issue:43688", - "issue:44792|issue:44977", + "issue:44792|issue:45600", "issue:43065|issue:44617", + "issue:44792|issue:44977", + "issue:43388|issue:43688", "issue:45127|issue:45200", "issue:43673|issue:44743", "issue:41950|issue:43475", @@ -11439,25 +11442,25 @@ "issue:44869|issue:46832", "issue:45245|issue:46858", "issue:44855|issue:45588", - "issue:43673|issue:46421", "issue:43827|issue:43994", - "issue:44242|issue:46897", + "issue:43673|issue:46421", "issue:43931|issue:44060", + "issue:44242|issue:46897", "issue:29870|issue:35141", "issue:44617|issue:45584", - "issue:44933|issue:45362", "issue:43644|issue:44913", + "issue:44933|issue:45362", "issue:25251|issue:27957", "issue:43976|issue:44855" ], - "prepared_review_unit_hash": "01d217370d4000a1a213fd4aa84d184d47cbfd609354d41d53c7d048f68ce765", + "prepared_review_unit_hash": "715c6dab26a80eca016bb9c7427e5886c3155b0812d6a563f47a9ad5a1055063", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, @@ -11472,181 +11475,184 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is broadly heterogeneous and the proposed soft edges are mostly false positives. No pair looks like the same underlying bug/change strongly enough to merge, though there is a small near-duplicate theme around the non-persistent-buffer regression elsewhere in the set.", - "confidence": 0.79, - "canonical_issue_reason": "Issue 43644 is the cleanest representative of the only tight duplicate-like theme in this batch: the v5 non-persistent-buffer corruption report. It is earlier and more explicit than later rephrasings, so it is the best issue anchor if a canonical must be chosen.", + "summary": "The cluster is mostly noisy and spans unrelated Transformers issues. The only soft pair that looks like a true duplicate is the trust_remote_code cache-loading pair; the rest are adjacent themes but different bugs or code paths.", + "confidence": 0.58, + "canonical_issue_reason": "issue:45632 is the clearest root-cause report: it names the cache-path collision for trust_remote_code local models, which also explains the FileNotFoundError symptom in issue:46617.", "canonical_pr_reason": null, - "best_issue_reason": "43644 is the strongest issue to keep as the representative bug report: concrete symptom, clear scope, and likely the earliest report in the buffer-corruption family. The rest of the set spans unrelated model, pipeline, cache, and docs problems.", + "best_issue_reason": "issue:45632 is the best representative because it describes the concrete underlying cache bug, not just one downstream failure mode.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44792", "right": "issue:45600", "accept": false, - "reason": "Different problems: Janus image-generation test failure vs auto-mapping references to removed Sam3LiteText configs." + "reason": "Janus test failure vs removed Sam3Lite config references; different subsystems and failure modes." }, { - "left": "issue:43388", - "right": "issue:43688", + "left": "issue:43065", + "right": "issue:44617", "accept": false, - "reason": "Both involve training/eval losses, but one is a gather_for_metrics batching bug and the other is auxiliary-loss normalization; not the same defect." + "reason": "Both involve SAM3, but one is a dummy Conv2d placeholder and the other is an OOM report; not the same bug." }, { "left": "issue:44792", "right": "issue:44977", "accept": false, - "reason": "Unrelated model-specific failures: Janus image generation vs Qwen3.5 flash-attention generation." + "reason": "Unrelated models and symptoms; no shared concrete code path." }, { - "left": "issue:43065", - "right": "issue:44617", + "left": "issue:43388", + "right": "issue:43688", "accept": false, - "reason": "Same SAM3 family, but dummy Conv2d setup and CUDA OOM are different issues with different fixes." + "reason": "Metric-gathering tuple truncation vs auxiliary-loss normalization; different logic and fixes." }, { "left": "issue:45127", "right": "issue:45200", "accept": false, - "reason": "Different code paths: LoRA merge/tied embeddings bug vs Gemma 4 multimodal token IDs defaulting." + "reason": "LoRA merge/vocab collapse and Gemma4 token-type-id defaults are distinct issues." }, { "left": "issue:43673", "right": "issue:44743", "accept": false, - "reason": "Both mention cache/state handling, but one is missing GenerationMixin cache in v5.0.0 and the other is recurrent state reset in Qwen3 modular code." + "reason": "Both cache-related, but one is missing cache initialization in chunked prefill and the other is recurrent-state reset in a modular model; not the same bug." }, { "left": "issue:41950", "right": "issue:43475", "accept": false, - "reason": "Both are SAM3/video-related, but pipeline processor selection and missing fpn_position_embeddings are different failures." + "reason": "Same SAM3/video area, but pipeline processor lookup and missing vision-encoder attribute are different failures." }, { "left": "issue:44190", "right": "issue:44964", "accept": false, - "reason": "Local dataset loading in a training script is unrelated to loading a multimodal pretrained model." + "reason": "Local dataset loading in a training script vs a multimodal model loading regression; different paths." }, { "left": "issue:43525", "right": "issue:43531", "accept": false, - "reason": "Completely different bugs: missing pad_token_id on Llama4Config vs sliding_window behavior in Qwen3-MoE." + "reason": "Llama4 pad_token_id config error vs Qwen3-MoE sliding_window behavior; unrelated bugs." }, { "left": "issue:44534", "right": "issue:44913", "accept": false, - "reason": "One is the non-persistent-buffer junk regression; the other is rotary_pct not persisting on reload." + "reason": "Non-persistent buffer junk and rotary_pct reload regression are separate state/config persistence problems." }, { "left": "issue:35141", "right": "issue:42490", "accept": false, - "reason": "Resizing token embeddings reinitialization is unrelated to loading local models from paths ending with a slash." + "reason": "Embedding reinitialization after resize vs import warning on a trailing-slash local path; not the same issue." }, { "left": "issue:44734", "right": "issue:45406", "accept": false, - "reason": "KV-cache continuation tensor indexing crash in serve is unrelated to Gemma4Processor missing _tokenizer." + "reason": "KV-cache continuation indexing crash vs Gemma4Processor missing _tokenizer; different code paths." }, { "left": "issue:46082", "right": "issue:46697", "accept": false, - "reason": "Config validation for head_dim is unrelated to image processor over-padding." + "reason": "Config validation for head_dim vs image over-padding; unrelated." }, { "left": "issue:45632", "right": "issue:46617", - "accept": false, - "reason": "Both touch local/trust_remote_code loading, but cache-path collisions and FileNotFoundError from cache are distinct failures." + "accept": true, + "reason": "Both report trust_remote_code cache-loading failures; 45632 describes the underlying cache-path collision that can surface as the FileNotFoundError in 46617." }, { "left": "issue:44869", "right": "issue:46832", "accept": false, - "reason": "Whisper trailing replacement-character crash is unrelated to RT-DETR feature-level mismatch TypeError." + "reason": "Whisper timestamp decoding crash and RT-DETR feature-level TypeError are unrelated." }, { "left": "issue:45245", "right": "issue:46858", "accept": false, - "reason": "Different low-level tensor/numerical issues: category-count limit vs cache seq-length device tensor arithmetic." + "reason": "Category-cardinality limit vs device-tensor slice arithmetic in cache handling; different bugs." }, { "left": "issue:44855", "right": "issue:45588", "accept": false, - "reason": "Python 3.13 import parsing failure in DebertaV2 is unrelated to flash_attention.py crashing on s_aux=None." - }, - { - "left": "issue:43673", - "right": "issue:46421", - "accept": false, - "reason": "Both involve chunked prefill/cache, but one is a missing cache regression and the other is lazy cache init causing recompilation; not mergeable as one bug." + "reason": "Python 3.13 indentation/parsing problem vs flash-attention None-attribute crash; unrelated." }, { "left": "issue:43827", "right": "issue:43994", "accept": false, - "reason": "Docs referencing removed pipeline() and a SigLIP model producing bad outputs are unrelated." + "reason": "Docs still referencing removed pipelines vs a model producing bad outputs with AutoModel/pipeline; not the same change." }, { - "left": "issue:44242", - "right": "issue:46897", + "left": "issue:43673", + "right": "issue:46421", "accept": false, - "reason": "Load-balancing loss omission in MoE routing is unrelated to Florence2 training-loss double-shift." + "reason": "Both mention chunked prefill/cache, but one is missing cache initialization and the other is spurious recompilation from lazy init; distinct symptoms and fixes." }, { "left": "issue:43931", "right": "issue:44060", "accept": false, - "reason": "Weight-shape mismatch on load and incorrect tied-weights warning are different model-loading/tieing problems." + "reason": "Qwen3-VL weight-shape mismatch vs a tied-weights warning in Qwen3-Next; different model/config bugs." + }, + { + "left": "issue:44242", + "right": "issue:46897", + "accept": false, + "reason": "Load-balancing loss omission and Florence2 double-shift are different training-loss bugs." }, { "left": "issue:29870", "right": "issue:35141", "accept": false, - "reason": "BART generate() decoder_attention_mask error is unrelated to output embeddings being reinitialized after resizing." + "reason": "BART generate() decoder mask bug vs post_init reinitializing output embeddings; unrelated." }, { "left": "issue:44617", "right": "issue:45584", "accept": false, - "reason": "CUDA OOM in Sam3Video is unrelated to Whisper empty-transcription generation after align_special_tokens." + "reason": "SAM3 video OOM vs Whisper empty-transcription failure; no shared underlying bug." }, { - "left": "issue:44933", - "right": "issue:45362", + "left": "issue:43644", + "right": "issue:44913", "accept": false, - "reason": "Nonexistent image_utils import and Qwen3.5 chat crash are different API/model failures." + "reason": "Non-persistent buffer junk on v5 and rotary_pct resetting on reload are separate regressions." }, { - "left": "issue:43644", - "right": "issue:44913", + "left": "issue:44933", + "right": "issue:45362", "accept": false, - "reason": "Different root causes: non-persistent buffer junk vs rotary_pct not surviving config reload." + "reason": "Nonexistent import from image_utils vs Qwen3.5 chat crash; unrelated." }, { "left": "issue:25251", "right": "issue:27957", "accept": false, - "reason": "Pipeline top_k nesting bug and XLMRoberta Flash Attention 2 support are unrelated." + "reason": "Pipeline top_k nesting bug vs Flash Attention 2 support issue; different code paths." }, { "left": "issue:43976", "right": "issue:44855", "accept": false, - "reason": "Version/Python compatibility report and a TorchScript decorator parsing error are not the same issue." + "reason": "Python version compatibility problem vs DebertaV2 import parsing error; different failures." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The summary correctly treats the cluster as noisy, and the one accepted soft pair (issue:45632 / issue:46617) is plausibly the same trust_remote_code cache bug rather than an overreach. The other negative verdicts also avoid claiming duplicates where the packet only shows thematic similarity." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, @@ -11738,211 +11744,211 @@ "issue:25251|issue:29870", "issue:44521|issue:45381", "issue:43673|issue:46858", - "issue:45632|issue:46234", - "issue:43854|issue:44964" + "issue:43854|issue:44964", + "issue:45632|issue:46234" ], - "prepared_review_unit_hash": "219fa57572604f86e3f74ff0e9727e37b78d43fdbc997a923321800c155374f3", + "prepared_review_unit_hash": "057543853b638f3b7f3d6ea6b62810cba729617d34871874659a03b65ca0a2b6", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35692, - "estimated_input_tokens": 8923, - "estimated_eval_tokens": 18102 + "serialized_chars": 35688, + "estimated_input_tokens": 8922, + "estimated_eval_tokens": 18100 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous issue reports; only one soft pair looks like the same underlying bug family. The DeepSeek tokenizer regression pair is the only plausible duplicate, while the rest are distinct model, processor, cache, or CI failures.", - "confidence": 0.74, - "canonical_issue_reason": "Issue 46489 is the most specific and actionable report in the one plausible duplicate family: it names DeepSeek-Coder v1, transformers v5+, and explicitly points to the remaining gap in PR #44801\u2019s fix. Issue 44779 is broader and older.", + "summary": "Mostly unrelated closed issues with one plausible duplicate pair around the DeepSeek tokenizer regression in v5; the rest are different models, subsystems, or failure modes and should stay separate.", + "confidence": 0.79, + "canonical_issue_reason": "issue:46489 is the more specific and actionable report: it names the DeepSeek-Coder v1 tokenizer regression in transformers v5+ and points to the gap in PR #44801's fix, making it a better canonical anchor than the broader symptom report.", "canonical_pr_reason": null, - "best_issue_reason": "46489 is the best representative issue overall because it is narrowly scoped, regression-oriented, and ties directly to the underlying tokenizer fix gap; the other issues are mostly unrelated one-off bugs.", + "best_issue_reason": "issue:46489 has the clearest root-cause framing and strongest linkage to an existing fix gap, so it is the best representative issue in this set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45563", "right": "issue:45584", "accept": false, - "reason": "Both are generation-related, but one is a stale warning and the other is an empty-transcription failure in Whisper; different code paths and symptoms." + "reason": "Both are generation bugs, but one is a stale warning for num_return_sequences in paged generate while the other is Whisper empty-transcription failure after align_special_tokens; different code paths and fixes." }, { "left": "issue:43550", "right": "issue:44190", "accept": false, - "reason": "A torch.compile+SDPA model crash is unrelated to a local dataset loading failure in a training script." + "reason": "Unrelated subsystems: torch.compile/SDPA model failure vs local dataset loading in a training script." }, { "left": "issue:45381", "right": "issue:46369", "accept": false, - "reason": "Different models and failure modes: Qwen video position IDs vs SAM3-LiteText memory overflow." + "reason": "Different models and problems: Qwen2.5-VL video position ids vs Sam3 LiteText memory overflow." }, { "left": "issue:44077", "right": "issue:44877", "accept": false, - "reason": "Both involve config/loading, but one is optional post_init validation and the other is strict config loading; not the same bug." + "reason": "Both involve config handling, but one is PatchTSMixer post_init validation and the other is Granite Speech strict config loading; not the same bug." }, { "left": "issue:44479", "right": "issue:44734", "accept": false, - "reason": "Both are regressions, but one is multimodal video input handling and the other is KV-cache continuation indexing in serve." + "reason": "Separate areas: multimodal video-input regression in Qwen models vs KV-cache continuation tensor indexing in transformers serve." }, { "left": "issue:41720", "right": "issue:45245", "accept": false, - "reason": "Unrelated failures: CUDA assert with auto device mapping vs category-count overflow." + "reason": "Completely different failures: Qwen3 auto device mapping CUDA assert vs category-count limit RuntimeError." }, { "left": "issue:43335", "right": "issue:44190", "accept": false, - "reason": "Sparse-layer construction in SwitchTransformers is unrelated to a dataset script local-path issue." + "reason": "SwitchTransformers sparse-layer config bug is unrelated to local dataset loading in the example script." }, { "left": "issue:43065", "right": "issue:44610", "accept": false, - "reason": "Both are vision-model issues, but one is a dummy Conv2d in SAM3 pixel decoder and the other is a processor/model input-size mismatch in OmDet-Turbo." + "reason": "Both are vision/model-input issues, but the specific bugs differ: dummy Conv2d in Sam3PixelDecoder vs OmDet-Turbo processor/model size mismatch." }, { "left": "issue:44464", "right": "issue:45563", "accept": false, - "reason": "Chunked generation inconsistency and a stale warning are different generation behaviors, not the same defect." + "reason": "Chunked generation inconsistency with compiled forward is not the same as a stale warning about num_return_sequences." }, { "left": "issue:39692", "right": "issue:44190", "accept": false, - "reason": "A documentation example/model-processor mismatch is not the same as a local dataset loading failure." + "reason": "SigLIP2 doc example errors are unrelated to image-classification script dataset loading." }, { "left": "issue:45127", "right": "issue:45357", "accept": false, - "reason": "LoRA merge with extended vocab and incorrect visual encoder keys on save_pretrained are different model-state bugs." + "reason": "Different features and failures: LoRA merging with tied embeddings vs incorrect visual encoder keys on save_pretrained." }, { "left": "issue:45381", "right": "issue:46655", "accept": false, - "reason": "Both are video-input bugs, but they affect different models and different preprocessing logic." + "reason": "Both concern video processing, but one is wrong vision_position_ids and the other is incorrect padding calculation in SmolVLM; not the same concrete bug." }, { "left": "issue:44112", "right": "issue:45600", "accept": false, - "reason": "CI test staleness in GraniteSpeech and auto_mappings references to removed configs are separate maintenance issues." + "reason": "Both are CI-related, but the root causes differ: stale GraniteSpeech device override test vs removed Sam3LiteText configs in auto_mappings." }, { "left": "issue:44779", "right": "issue:46489", "accept": true, - "reason": "Both describe DeepSeek tokenizer outputs becoming wrong in transformers v5+, and 46489 is explicitly a gap in the earlier fix for that same regression family." + "reason": "These describe the same DeepSeek tokenizer regression in transformers v5+: incorrect output/behavior in DeepSeek tokenizer families, with 46489 explicitly identifying the gap in PR #44801's fix." }, { "left": "issue:43295", "right": "issue:45276", "accept": false, - "reason": "Processor.tokenizer regression and resize_token_embeddings not propagating to embeddings are different API breakages." + "reason": "Different regressions: processor.tokenizer/image handling breakage vs resize_token_embeddings not updating tied embeddings." }, { "left": "issue:44617", "right": "issue:46369", "accept": false, - "reason": "Both mention SAM3 memory pressure, but one is video OOM and the other is LiteText memory overflow; not clearly the same root cause." + "reason": "Both mention memory pressure in Sam3 variants, but one is Sam3Video OOM and the other is Sam3 LiteText memory overflow; distinct model paths." }, { "left": "issue:43526", "right": "issue:46697", "accept": false, - "reason": "Label reduction in BeitImageProcessorFast is unrelated to Swin2SR over-padding." + "reason": "Different image processor bugs: BEiT label reduction vs Swin2SR padding on divisible dimensions." }, { "left": "issue:45290", "right": "issue:46752", "accept": false, - "reason": "Both hit apply_chat_template edge cases, but one is tool-call messages with no content and the other is an empty list IndexError." + "reason": "Both crash in apply_chat_template, but one is tool-call assistant messages with no content and the other is an empty list edge case; separate triggers." }, { "left": "issue:44462", "right": "issue:45907", "accept": false, - "reason": "Tokenizer file resolution and list_repo_files forwarding revision are different Hub/API issues." + "reason": "AutoTokenizer ignoring tokenizer.json is unrelated to list_repo_files not forwarding revision." }, { "left": "issue:45276", "right": "issue:45584", "accept": false, - "reason": "Embedding resize propagation and Whisper empty-output generation are unrelated." + "reason": "Unrelated issues: gemma4 embedding resize bug vs Whisper empty transcription generation failure." }, { "left": "issue:44737", "right": "issue:45803", "accept": false, - "reason": "CPU tensor creation in XLNet and exception swallowing in FuyuBatchFeature are unrelated bugs." + "reason": "Different code paths: XLNet relative positional encoding device placement vs bare except in FuyuBatchFeature.convert_to_tensors." }, { "left": "issue:43540", "right": "issue:44734", "accept": false, - "reason": "A video-input ValueError in Qwen3OmniMoe is not the same as a KV-cache continuation indexing crash in serve." + "reason": "Unrelated failures: Qwen3OmniMoe video-input ValueError vs serving KV-cache continuation tensor indexing." }, { "left": "issue:44938", "right": "issue:45542", "accept": false, - "reason": "Python 3.14 import/load failure and missing TensorFlow backend handling are separate environment issues." + "reason": "Python 3.14 import/load failure is unrelated to the TensorBoard-without-TensorFlow backend error." }, { "left": "issue:25251", "right": "issue:29870", "accept": false, - "reason": "Pipeline top_k nested-list output and Bart decoder_attention_mask initialization are unrelated." + "reason": "Both are generation-related, but one is pipeline top_k nesting and the other is Bart decoder_attention_mask initialization; different bugs." }, { "left": "issue:44521", "right": "issue:45381", "accept": false, - "reason": "Both involve multimodal/chat behavior, but assistant mask generation and video position IDs are distinct problems." + "reason": "Different multimodal issues: all-zero assistant masks vs wrong vision_position_ids for video inputs." }, { "left": "issue:43673", "right": "issue:46858", "accept": false, - "reason": "Both mention cache/prefill, but one is missing cache in chunked_prefill and the other is a device-tensor seq_length arithmetic bug." + "reason": "Both concern cache/prefill, but one is missing cache in v5.0.0 chunked_prefill and the other is a device-tensor seq_length bug in static cache on non-CUDA backends." }, { - "left": "issue:45632", - "right": "issue:46234", + "left": "issue:43854", + "right": "issue:44964", "accept": false, - "reason": "Cache-path collisions for local models and a missing import_utils symbol are unrelated." + "reason": "Separate model-loading failures for different models and test contexts; no shared concrete bug." }, { - "left": "issue:43854", - "right": "issue:44964", + "left": "issue:45632", + "right": "issue:46234", "accept": false, - "reason": "Both are model-loading failures, but they concern different models and likely different causes." + "reason": "Different concerns: trust_remote_code cache-path collision vs missing import_utils helper for Falcon Mamba availability." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet and stays conservative. The DeepSeek pair (44779/46489) is the only clearly same-family duplicate, and choosing 46489 as the more specific representative is supported by the titles. The other soft-edge rejections are appropriately cautious and do not overclaim shared root causes." + "feedback": "The assessment is well grounded in the packet. The DeepSeek pair (issue:44779, issue:46489) is the only clearly plausible duplicate, and the rest of the soft-edge rejections are conservative and supported by the titles/summaries. Minor wording note: some items in the set are open, so 'mostly unrelated closed issues' could be phrased a bit more generally, but this does not affect the core judgment." }, "evaluator_used": true, "retried": false, @@ -12045,220 +12051,220 @@ "issue:43673|issue:44336", "issue:44792|issue:46424" ], - "prepared_review_unit_hash": "fcacf2be3f3cf78a89a529f0f790ed2860b9e35feb30929a6971dd07113d616e", + "prepared_review_unit_hash": "37b409e7070621fd54ae3ae64041e4097e971e8fbfecccfa20a31b90e86c2b4a", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36936, - "estimated_input_tokens": 9234, - "estimated_eval_tokens": 18724 + "serialized_chars": 36930, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is not a true duplicate set: it mixes unrelated bug reports across different models, pipelines, config-loading paths, and docs. None of the soft-similarity pairs look like the same underlying bug/change closely enough to merge.", - "confidence": 0.96, - "canonical_issue_reason": null, + "summary": "This cluster is broadly Transformer/model-loading/regression related, but the individual issues are mostly about different models, APIs, and failure modes. None of the soft pairs look like the same underlying bug/change, so they should stay separate.", + "confidence": 0.71, + "canonical_issue_reason": "issue:44964 is the broadest user-facing representative of the cluster\u2019s dominant theme (latest-transformers model loading regressions), and it has some inbound discussion.", "canonical_pr_reason": null, - "best_issue_reason": null, + "best_issue_reason": "issue:44964 is the most representative of the cluster overall, but it still covers only one slice of the heterogeneous set.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44246", "right": "issue:45003", "accept": false, - "reason": "Both touch import behavior, but one reports slow `import transformers` while the other is about unsafe `sys.modules` access; different failure modes and fixes." + "reason": "One is slow imports; the other is a sys.modules access safety bug. Same area, different problems." }, { "left": "issue:44964", "right": "issue:45600", "accept": false, - "reason": "Different models and code paths: Phi-4 multimodal loading vs stale `auto_mappings.py` references to removed Sam3LiteText configs." + "reason": "Both are load-time regressions, but they affect different models and have different root causes." }, { "left": "issue:43643", "right": "issue:44496", "accept": false, - "reason": "Both are config/loading related, but one is missing fields from `trust_remote_code` configs and the other is an unrecognized model missing `model_type`; not the same bug." + "reason": "Both involve config/model loading, but one drops fields with trust_remote_code while the other rejects an unknown model type." }, { "left": "issue:44112", "right": "issue:44964", "accept": false, - "reason": "CI test failure in GraniteSpeech vs model loading failure for Phi-4 multimodal; unrelated artifacts." + "reason": "CI test failure in GraniteSpeech vs a model-loading failure for Phi-4 multimodal; unrelated bugs." }, { "left": "issue:27957", "right": "issue:35141", "accept": false, - "reason": "Flash Attention support for XLMRoberta is unrelated to embedding reinitialization after resize when `tie_word_embedding=False`." + "reason": "Flash Attention support for XLMRoberta and embedding reinit on resize are unrelated code paths." }, { "left": "issue:42175", "right": "issue:44279", "accept": false, - "reason": "One is package extra/backend dependency composition, the other is a generic dependency issue; too broad and not the same concrete problem." + "reason": "Packaging/backend dependency complaint vs a generic dependency issue; too vague and not the same defect." }, { "left": "issue:46082", "right": "issue:46121", "accept": false, - "reason": "Both are config validation issues, but one is `LlamaConfig` head_dim handling and the other is rope-validation parsing of JSON-loaded lists; different code paths." + "reason": "Both are validation-related, but one is about head_dim acceptance and the other about JSON-loaded list handling in rope validation." }, { "left": "issue:36246", "right": "issue:41628", "accept": false, - "reason": "Different missing imports: `Qwen2_5_VLImageProcessor` vs `AutoImageProcessor`; separate module/export problems." + "reason": "Both are import errors, but for different symbols and modules; not the same missing-export bug." }, { "left": "issue:42796", "right": "issue:44521", "accept": false, - "reason": "Tokenizer BPE-marker handling and multimodal assistant-mask generation are unrelated behaviors." + "reason": "Tokenizer BPE marker handling and multimodal assistant masks are different tokenization/template bugs." }, { "left": "issue:36246", "right": "issue:38617", "accept": false, - "reason": "An import error for a model-specific processor is unrelated to missing `layer_type_validation` in `configuration_utils`." + "reason": "Different import failures for different names; no shared concrete fix." }, { "left": "issue:43404", "right": "issue:45216", "accept": false, - "reason": "Different models and symptoms: untied `lm_head` weights in Mistral3 vs incorrect checkpoint saving for Qwen3.5." + "reason": "Both touch weight tying/save behavior, but they involve different models and different failure modes." }, { "left": "issue:32090", "right": "issue:35141", "accept": false, - "reason": "Trainer broadcast `NoneType` error is unrelated to post-init reinitializing embeddings after resize." + "reason": "Trainer broadcast NoneType error and embedding reinitialization are unrelated." }, { "left": "issue:44295", "right": "issue:46506", "accept": false, - "reason": "A `position_ids` buffer read error is not the same as CPU/CUDA divergence in PPDocLayoutV3." + "reason": "Reading a registered buffer vs CPU/CUDA output divergence are different bugs." }, { "left": "issue:41950", "right": "issue:44016", "accept": false, - "reason": "Video-classification pipeline processor lookup vs a notebook syntax error; no shared underlying bug." + "reason": "Pipeline image-processor lookup bug vs a notebook syntax error; not duplicates." }, { "left": "issue:43531", "right": "issue:45600", "accept": false, - "reason": "Sliding-window behavior in Qwen3-MoE is unrelated to removed config references in `auto_mappings.py`." + "reason": "Qwen3-MoE sliding_window behavior and auto_mappings config cleanup are unrelated." }, { "left": "issue:43650", "right": "issue:44220", "accept": false, - "reason": "Placeholder/invalid issue text vs `_torch_extract_fbank_features()` bug; no meaningful overlap." + "reason": "The titles are generic, but the underlying issues are different and provide no concrete overlap." }, { "left": "issue:43673", "right": "issue:46424", "accept": false, - "reason": "Both mention cache, but one is a GenerationMixin chunked-prefill regression and the other is cache length semantics for Quantized Cache only; distinct fixes." + "reason": "Both concern cache behavior, but one is chunked_prefill GenerationMixin cache and the other is max_cache_length handling." }, { "left": "issue:43976", "right": "issue:44987", "accept": false, - "reason": "Python version compatibility vs a specific model-load regression on transformers>=5.1.0; not the same bug." + "reason": "Python version compatibility vs a model-loading failure on a specific checkpoint; different root cause." }, { "left": "issue:43295", "right": "issue:45335", "accept": false, - "reason": "Processor/tokenizer regression with images is unrelated to `t5gemma` resize not propagating to decoder embeddings." + "reason": "Both are regressions around tokenization/weights, but one is processor.tokenizer/image passthrough and the other is decoder embedding resize." }, { "left": "issue:44016", "right": "issue:44220", "accept": false, - "reason": "Notebook syntax error and feature-extraction bug are unrelated." + "reason": "Notebook syntax error and audio feature extraction issue are unrelated." }, { "left": "issue:45083", "right": "issue:46506", "accept": false, - "reason": "Helper function length calculation in qwen3_omni_moe is unrelated to CPU/CUDA output mismatch in PPDocLayoutV3." + "reason": "Helper length computation in qwen3_omni_moe vs CPU/CUDA inconsistency in layout model; not the same bug." }, { "left": "issue:42907", "right": "issue:43441", "accept": false, - "reason": "Saving dequantized Ministral/Devstral models and FlashAttention failure in Ministral-3 are different issues." + "reason": "Both mention Ministral/FlashAttention, but one is save/load of dequantized weights and the other is FlashAttention inference failure." }, { "left": "issue:44683", "right": "issue:45446", "accept": false, - "reason": "Both involve flex_attention, but one is compiled flex_attention failing on torch>=2.9 while the other is a wrong version check for AuxRequest import; not the same concrete defect." + "reason": "Both are flex_attention-related, but one is a compiled-kernel failure on torch>=2.9 and the other is an incorrect version check for AuxRequest import." }, { "left": "issue:43976", "right": "issue:44279", "accept": false, - "reason": "A Python-version-specific release breakage is not the same as a generic dependency issue." + "reason": "A Python compatibility bug and a generic dependency issue are not the same concrete defect." }, { "left": "issue:43526", "right": "issue:43901", "accept": false, - "reason": "Image processor label reduction bug vs pipeline docs mentioning `return_all_scores`; one is runtime behavior, the other docs drift." + "reason": "Image processor label reduction bug vs documentation drift for return_all_scores; one is runtime behavior, the other docs." }, { "left": "issue:43381", "right": "issue:45593", "accept": false, - "reason": "Gradient checkpointing being disallowed in eval mode is unrelated to missing auxiliary losses when denoising is off." + "reason": "Eval-mode gradient checkpointing and D-FINE auxiliary losses are unrelated training behaviors." }, { "left": "issue:43479", "right": "issue:45375", "accept": false, - "reason": "Different multimodal config bugs on different models: default config initialization vs a missing `deepstack_visual_indexes` field being dropped." + "reason": "Different config-initialization bugs in different multimodal model families." }, { "left": "issue:44496", "right": "issue:45042", "accept": false, - "reason": "Model loading without `model_type` is unrelated to PIL image processors wrongly requiring torchvision." + "reason": "Unknown model type loading failure vs PIL image processors incorrectly requiring torchvision; different loading layers." }, { "left": "issue:43901", "right": "issue:44496", "accept": false, - "reason": "Pipeline documentation mismatch vs model config loading failure; not the same underlying issue." + "reason": "Docs mismatch for a pipeline option vs an unknown-model config error; not duplicates." }, { "left": "issue:43673", "right": "issue:44336", "accept": false, - "reason": "Cache regression during generation is unrelated to ANSI codes emitted by loading report utilities." + "reason": "Generation cache bug and ANSI output formatting in loading_report are unrelated." }, { "left": "issue:44792", "right": "issue:46424", "accept": false, - "reason": "Janus image generation test failure is unrelated to cache length behavior in quantized cache." + "reason": "Janus image-generation test failure vs cache max length behavior; different subsystems and symptoms." } ] }, @@ -12333,17 +12339,17 @@ ], "soft_pairs": [ "issue:44016|issue:44062", - "issue:43441|issue:45588", "issue:43824|issue:44933", "issue:44496|issue:45759", + "issue:43441|issue:45588", "issue:46123|issue:46369", "issue:44327|issue:46752", "issue:44521|issue:46693", "issue:44279|issue:46515", - "issue:43299|issue:45216", "issue:45335|issue:45584", - "issue:46082|issue:46399", + "issue:43299|issue:45216", "issue:44279|issue:45405", + "issue:46082|issue:46399", "issue:43295|issue:44779", "issue:44075|issue:44368", "issue:39692|issue:43901", @@ -12361,202 +12367,202 @@ "issue:41950|issue:45823", "issue:44242|issue:45200" ], - "prepared_review_unit_hash": "b7fa2f97aec011a7cb0d81f10ece06a8ada2c34e1d4f652622e7472545d21ad4", + "prepared_review_unit_hash": "92a3ca9e3b90919a88f381040ba9e8881e53d1720307cb139c6d76dab4100c59", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 36211, - "estimated_input_tokens": 9053, - "estimated_eval_tokens": 18362 + "serialized_chars": 36216, + "estimated_input_tokens": 9054, + "estimated_eval_tokens": 18364 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: docs examples, tokenizer issues, pipeline regressions, model-loading bugs, and config/save_pretrained problems. I did not find any strong duplicate pairs; the only visible theme is a loose Qwen/multimodal loading/serialization area, not a single underlying bug. No PRs are present.", - "confidence": 0.57, - "canonical_issue_reason": "issue:43299 is the best anchor for the loose cluster because it is a concrete regression in the Qwen multimodal/model-loading path and is the most central of the nearby issues, even though the overall set is not a true duplicate cluster.", + "summary": "This cluster is mostly a loose set of unrelated Transformers issues with some superficial similarity around multimodal/tokenizer/load-path regressions; none of the soft pairs look like true duplicates.", + "confidence": 0.37, + "canonical_issue_reason": "Issue 43295 is the best anchor because it is a concrete, reproducible regression in multimodal processor/tokenizer handling and is broader than the other highly specific model- or docs-only reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43299 is the most representative issue to keep as the cluster anchor: specific enough to be actionable, broad enough to cover the nearby Qwen-loading/serialization reports, and more central than the unrelated docs/tokenizer/pipeline outliers.", + "best_issue_reason": "43295 is the strongest standalone issue in the set: clear regression scope, concrete failure mode, and good triage value as a representative multimodal API break.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44016", "right": "issue:44062", "accept": false, - "reason": "Notebook syntax error vs tokenizer AddedToken regression; different code paths and fixes." - }, - { - "left": "issue:43441", - "right": "issue:45588", - "accept": false, - "reason": "Both involve flash attention, but one is a Ministral-specific crash and the other is an s_aux=None AttributeError in a different integration path." + "reason": "Different failures: notebook syntax error vs AddedToken keyword-argument error; no shared code path." }, { "left": "issue:43824", "right": "issue:44933", "accept": false, - "reason": "Different missing imports in different modules; not the same underlying bug." + "reason": "Both are import errors, but they concern different missing symbols/modules and different APIs." }, { "left": "issue:44496", "right": "issue:45759", "accept": false, - "reason": "One is an unrecognized-model config issue, the other is from_config not unwrapping text_config; related area, but not the same failure." + "reason": "Generic model-config recognition failure vs composite Qwen text_config unwrapping bug; related area, not same bug." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Distinct flash-attention crashes with different triggers and stack paths." }, { "left": "issue:46123", "right": "issue:46369", "accept": false, - "reason": "Both are runtime problems, but one drops partial-batch results and the other overflows memory; unrelated bugs." + "reason": "Batch-finalization bug in MaskGenerationPipeline vs memory overflow in sam3-litetext; unrelated." }, { "left": "issue:44327", "right": "issue:46752", "accept": false, - "reason": "Both crash on edge cases, but they are different functions and different boundary conditions." + "reason": "QA span top-k bounds error vs empty-list chat-template IndexError; different pipeline code." }, { "left": "issue:44521", "right": "issue:46693", "accept": false, - "reason": "Multimodal assistant-mask logic vs FA2 long-context performance regression; not the same issue." + "reason": "Assistant-mask bug in chat templates vs long-context FA2 performance regression; not the same underlying issue." }, { "left": "issue:44279", "right": "issue:46515", "accept": false, - "reason": "Generic dependency complaint vs torch version incompatibility; too broad and different root causes." + "reason": "Dependency complaint vs torch-version incompatibility; too broad and not one concrete bug." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "resize_token_embeddings propagation bug vs Whisper empty-transcription failure after token alignment." }, { "left": "issue:43299", "right": "issue:45216", "accept": false, - "reason": "Both are Qwen-related regressions, but one is model loading and the other is save_pretrained serialization; not the same bug." + "reason": "Loading regression for Qwen3VL MoE vs save_pretrained checkpoint serialization regression; different code paths." }, { - "left": "issue:45335", - "right": "issue:45584", + "left": "issue:43577", + "right": "issue:45216", "accept": false, - "reason": "Different models and different tokenization/alignment paths; not mergeable as one fix." + "reason": "Blip2 dtype propagation bug vs Qwen3.5 save_pretrained regression; unrelated models and behavior." }, { - "left": "issue:46082", - "right": "issue:46399", + "left": "issue:44351", + "right": "issue:45278", "accept": false, - "reason": "Config validation for head_dim vs missing decorator registration for rotary embedding kernel; unrelated." + "reason": "Single missing import vs broad post-update import failures; same symptom class, not same bug." }, { - "left": "issue:44279", - "right": "issue:45405", + "left": "issue:46082", + "right": "issue:46399", "accept": false, - "reason": "One is a general dependency issue, the other is a release/version pin problem; not duplicates." + "reason": "Config validation around head_dim vs missing decorator for rotary-pos-emb kernel registration." }, { "left": "issue:43295", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they affect different APIs and model families with different root causes." + "reason": "Processor.tokenizer access/passing images regression vs Deepseek tokenizer output regression; different functionality." }, { "left": "issue:44075", "right": "issue:44368", "accept": false, - "reason": "Optimizer args ignored vs a warning about tie_word_embeddings; different behavior and fixes." + "reason": "SGD args ignored vs warning about tie_word_embeddings; not the same issue." }, { "left": "issue:39692", "right": "issue:43901", "accept": false, - "reason": "Documentation/example mismatch in SigLIP2 vs stale TextClassificationPipeline docs; unrelated docs issues." + "reason": "SigLIP2 example/model mismatch and quantization failure vs outdated docs mentioning return_all_scores." }, { "left": "issue:43976", "right": "issue:45405", "accept": false, - "reason": "Python support/version breakage vs unreleased peft minimum version pin; different compatibility problems." - }, - { - "left": "issue:43577", - "right": "issue:45216", - "accept": false, - "reason": "dtype propagation in Blip2 vs incorrect Qwen3.5 checkpoint serialization; different model families and code paths." + "reason": "Python-version compatibility issue vs unreleased PEFT version bump; unrelated." }, { - "left": "issue:44351", - "right": "issue:45278", + "left": "issue:44279", + "right": "issue:45405", "accept": false, - "reason": "Single missing import vs many import errors after upgrade; overlap is too broad to be the same bug." + "reason": "Generic dependency problem vs specific unreleased dependency version; not duplicate." }, { "left": "issue:46082", "right": "issue:46512", "accept": false, - "reason": "Config rejects explicit head_dim vs stale config.dtype after from_config; separate config-handling issues." + "reason": "Head_dim validation failure vs stale config.dtype after from_config(dtype=...); distinct config bugs." }, { "left": "issue:42796", "right": "issue:43650", "accept": false, - "reason": "Tokenizer BPE marker handling vs a vague 'ADD THE DATA' report; no identifiable shared bug." + "reason": "Tokenizer BPE-marker bug vs nonspecific request to 'ADD THE DATA'; no concrete match." }, { "left": "issue:44779", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but they concern different model families and different tokenization failures." + "reason": "Different tokenizer regressions affecting different model families and behaviors." }, { "left": "issue:41950", "right": "issue:45020", "accept": false, - "reason": "Pipeline processor selection bug vs remote_code breakage; not the same underlying change." + "reason": "Video-classification processor lookup bug vs remote_code model breakage; too broad to merge." }, { "left": "issue:43299", "right": "issue:45357", "accept": false, - "reason": "Both are Qwen3.5/Qwen multimodal regressions, but one concerns model loading and the other save_pretrained key serialization; related but distinct." + "reason": "Loading failure vs incorrect visual-encoder keys on save_pretrained; different regression surfaces." }, { "left": "issue:44466", "right": "issue:45276", "accept": false, - "reason": "Tied-weight serialization issue vs resize_token_embeddings propagation; similar area, different concrete bug." + "reason": "lm_head serialization/device-dependent tied-weights issue vs resize_token_embeddings not updating Gemma4 embeddings." }, { "left": "issue:43299", "right": "issue:43854", "accept": false, - "reason": "Different model families and different loading failures; no shared code-path evidence." + "reason": "Both involve Qwen VL models, but one is a loading regression and the other an import/export issue for a different class." }, { "left": "issue:43653", "right": "issue:45593", "accept": false, - "reason": "Tokenizer special-token decoding bug vs missing auxiliary losses in D-FINE; unrelated." + "reason": "BigBird mask-token registration bug vs D-FINE auxiliary-loss omission; unrelated." }, { "left": "issue:41950", "right": "issue:45823", "accept": false, - "reason": "Image-processor lookup bug vs Gemma4 device mismatch; different subsystems." + "reason": "Pipeline processor-type bug vs Gemma4 PLE device mismatch during forward; different subsystems." }, { "left": "issue:44242", "right": "issue:45200", "accept": false, - "reason": "MoE load-balancing loss gating vs mm_token_type_ids defaulting; different model/training semantics." + "reason": "MoE load-balancing loss gating vs Gemma4 mm_token_type_ids defaulting; not the same code path." } ] }, @@ -12636,8 +12642,8 @@ "issue:44683|issue:45542", "issue:44016|issue:44291", "issue:44361|issue:45593", - "issue:33290|issue:45656", "issue:45081|issue:45907", + "issue:33290|issue:45656", "issue:43441|issue:45834", "issue:42994|issue:43317", "issue:44623|issue:44749", @@ -12649,8 +12655,8 @@ "issue:42175|issue:43976", "issue:44623|issue:45397", "issue:44671|issue:44779", - "issue:43723|issue:44162", "issue:43901|issue:44246", + "issue:43723|issue:44162", "issue:43673|issue:43761", "issue:43824|issue:43976", "issue:45542|issue:45800", @@ -12658,196 +12664,196 @@ "issue:43525|issue:43881", "issue:46291|issue:46399" ], - "prepared_review_unit_hash": "021a55c4c227eb3296ab1a9c0781f3fbfa0c01a0430af7a5d7dd6866362b7037", + "prepared_review_unit_hash": "2e6a5c7ae4ea778c30bc60102d573343b54855644eabc32351e69a29f8ed9a62", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 27, - "serialized_chars": 35700, - "estimated_input_tokens": 8925, - "estimated_eval_tokens": 18106 + "serialized_chars": 35701, + "estimated_input_tokens": 8926, + "estimated_eval_tokens": 18108 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: most items are separate Transformers v5 regressions across tokenizers, configs, model loading, attention, and training integrations. None of the soft pairs look like true duplicates; the best representative issue is the broad tokenizer-loading regression.", - "confidence": 0.72, - "canonical_issue_reason": "issue:43723 is the broadest representative of the cluster\u2019s common theme: v5 loading/instantiation regressions around AutoTokenizer and model initialization. It is more general than the model-specific or narrowly scoped reports.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions and environment/package issues; none of the soft pairs look like true duplicates.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out for the whole set because the reports span different subsystems and failure modes. If a representative issue is needed, issue:44462 is the strongest candidate because it is a concrete, reproducible tokenizer-loading regression with broad user impact.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43723 best matches the cluster\u2019s overall center of gravity because it describes a generic v5 tokenizer-loading failure rather than a model-specific symptom.", + "best_issue_reason": "issue:44462 is the best representative issue overall: it describes a clear v5 tokenizer-loading bug, has a specific failure mode, and is the most plausible anchor for nearby tokenizer-related reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43650", "right": "issue:46209", "accept": false, - "reason": "Unrelated issues: a generic data request vs FP8 inference failure on B200 hardware." + "reason": "Unrelated: a vague data request versus an FP8 inference failure on B200." }, { "left": "issue:44297", "right": "issue:44462", "accept": false, - "reason": "Both concern tokenizer save/load behavior, but they are different concrete bugs on different code paths." + "reason": "Both involve tokenizer save/load, but one is save_pretrained metadata mismatch while the other is AutoTokenizer ignoring repository tokenizer.json; different code paths." }, { "left": "issue:43479", "right": "issue:43525", "accept": false, - "reason": "Different models and failures: multimodal config default initialization vs missing pad_token_id on Llama4Config." + "reason": "Different configs and failures: Phi4Multimodal default sub-config initialization vs Llama4 missing pad_token_id." }, { "left": "issue:44683", "right": "issue:45542", "accept": false, - "reason": "Completely different problems: torch flex_attention compatibility vs TensorFlow backend packaging." + "reason": "Different subsystems: compiled flex_attention on torch>=2.9 versus TensorFlow backend packaging with tensorboard." }, { "left": "issue:44016", "right": "issue:44291", "accept": false, - "reason": "Notebook syntax error and init_empty_weights TypeError are unrelated." + "reason": "Notebook syntax error versus init_empty_weights/_is_hf_initialized argument mismatch; not the same bug." }, { "left": "issue:44361", "right": "issue:45593", "accept": false, - "reason": "Different subsystems and symptoms: tokenizer AttributeError vs D-FINE loss computation bug." + "reason": "Different model/features: MLukeTokenizer task AttributeError versus D-FINE auxiliary losses when denoising is off." }, { - "left": "issue:33290", - "right": "issue:45656", + "left": "issue:45081", + "right": "issue:45907", "accept": false, - "reason": "Both mention deepspeed/optimizer behavior, but one is OOM with Adafactor and the other is double optimizer stepping; not the same bug." + "reason": "Both touch loading-related code, but one is a Mistral regex patch crash and the other is a hub list_repo_files revision-forwarding issue." }, { - "left": "issue:45081", - "right": "issue:45907", + "left": "issue:33290", + "right": "issue:45656", "accept": false, - "reason": "Mistral tokenizer regex crash vs list_repo_files revision forwarding; unrelated." + "reason": "Both mention DeepSpeed/optimizer behavior, but one is OOM with Adafactor and the other is double optimizer stepping; distinct root causes." }, { "left": "issue:43441", "right": "issue:45834", "accept": false, - "reason": "Different model-specific failures: FlashAttention on Ministral-3 vs OCR long-input index error in Kosmos2.5." + "reason": "Different models and failure modes: FlashAttention regression for Ministral-3 versus long-ocr index error in Kosmos2.5." }, { "left": "issue:42994", "right": "issue:43317", "accept": false, - "reason": "Both involve quantization/offload, but one fails on saving and the other on loading dequantized models; distinct code paths." + "reason": "Both are quantization/offload related, but saving a quantized model and loading a dequantized model with device_map=auto are different paths." }, { "left": "issue:44623", "right": "issue:44749", "accept": false, - "reason": "Processor save_pretrained missing files vs data filtering slowdown after upgrade are unrelated." + "reason": "Processor save_pretrained file-missing bug versus a Chinese report about slower filtering after upgrade; not the same issue." }, { "left": "issue:44261", "right": "issue:44263", "accept": false, - "reason": "Unrelated model bugs: missing rms_norm_eps in MLA vs torch.split return values in GlmMoeDsaIndexer." + "reason": "Same broad GLM family, but one is a missing rms_norm_eps causing precision drift and the other is torch.split return handling in GlmMoeDsaIndexer." }, { "left": "issue:45103", "right": "issue:45372", "accept": false, - "reason": "auto_docstring annotation crash and mistral_common ReasoningEffort import error are different issues." + "reason": "Unrelated: auto_docstring future-annotations crash versus a mistral_common import error breaking Gemma 4 processor loading." }, { "left": "issue:44336", "right": "issue:46851", "accept": false, - "reason": "ANSI color output in loading_report is unrelated to Qwen3.5 variable-length input handling." + "reason": "ANSI color handling in loading_report versus Qwen3.5 variable-length position_ids behavior; different code paths." }, { "left": "issue:45600", "right": "issue:45712", "accept": false, - "reason": "Both involve repo cleanup, but stale auto_mappings references and leftover dummy classes are different maintenance bugs." + "reason": "Both are cleanup/CI-style issues, but auto_mappings stale config references and leftover dummy classes are separate problems." }, { "left": "issue:45859", "right": "issue:46851", "accept": false, - "reason": "Same model family, but CPU offload rotary embedding incompatibility is not the same as variable-length position_ids handling." + "reason": "Same model family, but CPU offload incompatibility and variable-length position_ids are different bugs." }, { "left": "issue:42175", "right": "issue:43976", "accept": false, - "reason": "TensorFlow packaging issue and Python version compatibility issue are unrelated." + "reason": "Different packaging/version problems: missing TensorFlow backend in torch extra versus Python version compatibility for Transformers 5.1.0." }, { "left": "issue:44623", "right": "issue:45397", "accept": false, - "reason": "Processor save_pretrained missing files vs gemma-4 zero3 loading failure are different bugs." + "reason": "Processor.save_pretrained missing files versus gemma-4 zero3 from_pretrained failure; different components and symptoms." }, { "left": "issue:44671", "right": "issue:44779", "accept": false, - "reason": "Both are v5 regressions in different models, but CamemBERT masked-LM prediction drift and Deepseek tokenizer corruption are not the same defect." + "reason": "Both are model-specific v5 regressions, but CamemBERT masked LM predictions and DeepSeek tokenizer outputs are separate code paths." }, { - "left": "issue:43723", - "right": "issue:44162", + "left": "issue:43901", + "right": "issue:44246", "accept": false, - "reason": "Tokenizer loading failure in v5 and ESM2 broken are unrelated." + "reason": "Docs mismatch for TextClassificationPipeline versus intermittent import slowness; unrelated." }, { - "left": "issue:43901", - "right": "issue:44246", + "left": "issue:43723", + "right": "issue:44162", "accept": false, - "reason": "Docs mismatch about return_all_scores vs import-time slowness are unrelated." + "reason": "AutoTokenizer loading regression versus ESM2 broken workflows; different models and failure mechanisms." }, { "left": "issue:43673", "right": "issue:43761", "accept": false, - "reason": "GenerationMixin cache missing in chunked_prefill and CLIPVisionModel hidden_states regression are separate model-path bugs." + "reason": "Chunked_prefill cache regression in GenerationMixin versus CLIPVisionModel hidden_states behavior; unrelated model paths." }, { "left": "issue:43824", "right": "issue:43976", "accept": false, - "reason": "ImportError for Qwen2_5_VLForConditionalGeneration is unrelated to Python 3.9/3.10 compatibility." + "reason": "Missing Qwen2_5_VL export versus Python version support issue; no shared bug." }, { "left": "issue:45542", "right": "issue:45800", "accept": false, - "reason": "TensorBoard/TensorFlow backend packaging and torch/transformers incompatibility are different issues." + "reason": "TensorFlow backend packaging issue versus torch/transformers version incompatibility; not the same problem." }, { "left": "issue:42796", "right": "issue:43441", "accept": false, - "reason": "Both mention Ministral, but tokenizer BPE marker handling and FlashAttention model execution are distinct bugs." + "reason": "Same Ministral family, but tokenizer BPE marker handling and FlashAttention loading failure are distinct issues." }, { "left": "issue:43525", "right": "issue:43881", "accept": false, - "reason": "Missing pad_token_id on Llama4Config and glm-4v-9b loading failure are unrelated." + "reason": "Llama4Config pad_token_id AttributeError versus glm-4v-9b loading failure; unrelated models." }, { "left": "issue:46291", "right": "issue:46399", "accept": false, - "reason": "kernels version/revision import failure and Qwen3.5 apply_rotary_pos_emb registration crash are different code paths." + "reason": "Both involve kernels, but kernels v0.15 revision/version requirements and Qwen3.5 rotary registration are different failures." } ] }, @@ -12922,8 +12928,8 @@ "soft_pairs": [ "issue:44188|issue:44336", "issue:44016|issue:45823", - "issue:44336|issue:44373", "issue:44589|issue:45362", + "issue:44336|issue:44373", "issue:43525|issue:43643", "issue:46655|issue:46726", "issue:45325|issue:46851", @@ -12935,199 +12941,223 @@ "issue:45561|issue:45563", "issue:46399|issue:46851", "issue:45103|issue:46399", - "issue:46614|issue:46665", "issue:43526|issue:45003", + "issue:46614|issue:46665", "issue:43540|issue:46752", "issue:45072|issue:46829", - "issue:44295|issue:45397", "issue:43531|issue:45362", + "issue:44295|issue:45397", "issue:42491|issue:43299", - "issue:44242|issue:45593", "issue:44246|issue:44871", + "issue:44242|issue:45593", "issue:44336|issue:44485", "issue:45588|issue:46851", "issue:45003|issue:46821", "issue:42886|issue:46132", "issue:44451|issue:46752" ], - "prepared_review_unit_hash": "9f1dd7cb2d884511c652f4e3e371ae7dbc95e6a5e3878f437cb0ba2bf45ea0be", + "prepared_review_unit_hash": "e6a61b56e76e7646d01e12fe4b4342e03294878b4357cc37cb8048cae2ebda90", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35843, - "estimated_input_tokens": 8961, - "estimated_eval_tokens": 18178 + "serialized_chars": 35845, + "estimated_input_tokens": 8962, + "estimated_eval_tokens": 18180 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is heterogeneous: many unrelated issue reports with only loose topical overlap (mostly around Qwen/Gemma/model-loading/runtime bugs). None of the soft-edge pairs look like the same underlying bug or change, so they should all stay separate.", - "confidence": 0.43, - "canonical_issue_reason": "issue:46851 is the clearest, most self-contained runtime bug report in the set and is easier to use as a representative anchor than the broader or more ambiguous reports.", + "summary": "These items are mostly unrelated Transformers bugs across different models, configs, processors, and utilities. None of the soft pairs look like true duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue #46851 is the clearest concrete runtime bug in the set; the others are mostly unrelated, documentation, versioning, or different code paths.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46851 is the best standalone issue to keep as the representative item: it has a concrete failure mode, a specific model family, and a focused reproduction path.", + "best_issue_reason": "Issue #46851 is the most actionable representative issue here: it names a specific model family, a specific failure mode, and a concrete input-dependent code path.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44188", "right": "issue:44336", "accept": false, - "reason": "torch.compile attention-kernel branching bug vs ANSI codes emitted by loading_report; unrelated subsystems and symptoms." + "reason": "Different subsystems: torch.compile attention-kernel branching vs ANSI color output in loading_report." }, { "left": "issue:44016", "right": "issue:45823", "accept": false, - "reason": "notebook syntax error vs Gemma4 device-map forward mismatch; different artifact type and failure path." + "reason": "Notebook syntax error is unrelated to Gemma4 device-map forward dtype/device mismatch." }, { - "left": "issue:44336", - "right": "issue:44373", + "left": "issue:44589", + "right": "issue:45362", "accept": false, - "reason": "terminal ANSI output bug vs wrong docstring for position_ids; documentation and runtime behavior are unrelated." + "reason": "Float8 storage deserialization error and Qwen3.5 chat crash are different failure modes and code paths." }, { - "left": "issue:44589", - "right": "issue:45362", + "left": "issue:44336", + "right": "issue:44373", "accept": false, - "reason": "missing Float8 storage type vs Qwen3.5 chat crash; different errors, models, and code paths." + "reason": "Terminal ANSI leakage is unrelated to a docstring typo for position_ids." }, { "left": "issue:43525", "right": "issue:43643", "accept": false, - "reason": "missing pad_token_id on Llama4Config vs missing fields from AutoConfig with trust_remote_code; both config-related but not the same bug." + "reason": "Both mention missing fields, but one is Llama4Config pad_token_id and the other is trust_remote_code AutoConfig field loss; not the same bug." }, { "left": "issue:46655", "right": "issue:46726", "accept": false, - "reason": "SmolVLM video padding calculation bug vs AutoProcessor backend-missing error; different processor layers and failure causes." + "reason": "Incorrect video padding calculation is different from AutoProcessor\u2019s missing-vision-backend import error." }, { "left": "issue:45325", "right": "issue:46851", "accept": false, - "reason": "both involve position_ids in vision/text models, but they affect different model families and different logic (Qwen2.5-VL rope scaling vs Qwen3.5 variable-length inputs)." + "reason": "Both involve position_ids, but they affect different Qwen model families and different logic." }, { "left": "issue:41628", "right": "issue:43976", "accept": false, - "reason": "import AutoImageProcessor failure vs Python version incompatibility in Transformers 5.1.0; unrelated issues." + "reason": "ImportError for AutoImageProcessor is unrelated to the Python version support regression." }, { "left": "issue:44297", "right": "issue:45397", "accept": false, - "reason": "tokenizer save_pretrained metadata mismatch vs gemma-4 Zero3 loading issue; different load/save paths and symptoms." + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma-4 zero3 loading." }, { "left": "issue:43784", "right": "issue:45542", "accept": false, - "reason": "sentence-transformers import NameError for nn vs tensorboard/TensorFlow backend misconfiguration; unrelated dependency/import bugs." + "reason": "Sentence-transformers import NameError and undefined tf backend from tensorboard-only install are separate dependency/import issues." }, { "left": "issue:44246", - "right": "issue:44496", + "right": "issue:44871", "accept": false, - "reason": "import latency complaint vs unrecognized model/config.json missing model_type; not the same problem." + "reason": "Import latency and eos_token_id inconsistency are unrelated." }, { "left": "issue:45083", "right": "issue:46851", "accept": false, - "reason": "Qwen3_omni_moe helper length calculation issue vs Qwen3.5 variable-length position_ids crash; similar theme but distinct code paths." + "reason": "Qwen3-Omni-MoE feature-length helper behavior is different from Qwen3.5 variable-length position_ids handling." }, { "left": "issue:45561", "right": "issue:45563", "accept": false, - "reason": "pytest-xdist race on captured_info.txt vs stale warning for num_return_sequences; completely different bugs." + "reason": "A pytest-xdist race in testing utils is unrelated to a stale generate() warning." }, { "left": "issue:46399", "right": "issue:46851", "accept": false, - "reason": "Qwen3.5 rotary-position registration crash vs variable-length position_ids bug; same model family, but different concrete failures." + "reason": "Kernel registration/decorator crash in Qwen3.5 is a different concrete bug than variable-length position_ids." }, { "left": "issue:45103", "right": "issue:46399", "accept": false, - "reason": "auto_docstring annotations crash vs apply_rotary_pos_emb registration decorator bug; unrelated utilities." + "reason": "auto_docstring annotation handling and apply_rotary_pos_emb registration are unrelated code paths." }, { - "left": "issue:46614", - "right": "issue:46665", + "left": "issue:43526", + "right": "issue:45003", "accept": false, - "reason": "stale phonemizer language backend vs inconsistent from_pretrained behavior inside Python classes; different components and symptoms." + "reason": "BeitImageProcessorFast label reduction and unsafe sys.modules access are different bugs." }, { - "left": "issue:43526", - "right": "issue:45003", + "left": "issue:46614", + "right": "issue:46665", "accept": false, - "reason": "BeitImageProcessorFast reduce_labels bug vs unsafe sys.modules access in modeling_utils; unrelated code paths." + "reason": "Stale phonemizer backend state is unrelated to from_pretrained behavior inside Python classes." }, { "left": "issue:43540", "right": "issue:46752", "accept": false, - "reason": "Qwen3OmniMoe video-processing ValueError vs empty-list chat template IndexError; different inputs and failure mechanisms." + "reason": "Qwen3OmniMoe video-input ValueError and empty apply_chat_template IndexError are different failures." }, { "left": "issue:45072", "right": "issue:46829", "accept": false, - "reason": "bfloat16 dtype mismatch in inference vs VideoPrism hidden_states assignment bug; different models and output handling." + "reason": "dtype mismatch in inference and hidden_states/output field confusion are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5 chat crashes are different issues." }, { "left": "issue:44295", "right": "issue:45397", "accept": false, - "reason": "position_ids buffer read error vs gemma-4 Zero3 from_pretrained issue; unrelated failure modes." + "reason": "position_ids buffer access and Gemma-4 zero3 loading are not the same bug." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen MoE loading, but they are different version/model breakages and not the same concrete change." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import latency and eos_token_id configuration mismatch are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Load-balancing loss gating and D-FINE auxiliary-loss omission are different model-loss bugs." }, { "left": "issue:44336", "right": "issue:44485", "accept": false, - "reason": "ANSI codes in loading_report vs GLM-5 RoPE implementation discussion; no shared bug behavior." + "reason": "ANSI output formatting has no relation to GLM-5 RoPE implementation." }, { "left": "issue:45588", "right": "issue:46851", "accept": false, - "reason": "flash_attention s_aux=None crash vs Qwen3.5 variable-length position_ids bug; different attention paths and symptoms." + "reason": "flash_attention sink-less-model crash is a different code path than Qwen3.5 variable-length position_ids." }, { "left": "issue:45003", "right": "issue:46821", "accept": false, - "reason": "sys.modules access safety issue vs TimesFM window_size AttributeError; unrelated." + "reason": "sys.modules access safety and TimesFM window_size attribute handling are unrelated." }, { "left": "issue:42886", "right": "issue:46132", "accept": false, - "reason": "offline tokenizer cache load bug vs AttentionInterface.register behavior change; different APIs and root causes." + "reason": "Offline tokenizer cache loading and AttentionInterface.register behavior change are unrelated." }, { "left": "issue:44451", "right": "issue:46752", "accept": false, - "reason": "ScandiBERT loading failure vs empty chat template IndexError; unrelated model-loading vs templating bugs." + "reason": "ScandiBERT loading failure is unrelated to empty chat template handling." } ] }, @@ -13202,8 +13232,8 @@ ], "soft_pairs": [ "issue:43550|issue:46506", - "issue:36246|issue:42757", "issue:44779|issue:45292", + "issue:36246|issue:42757", "issue:36246|issue:44351", "issue:41950|issue:44220", "issue:43901|issue:45003", @@ -13219,208 +13249,220 @@ "issue:43475|issue:46369", "issue:45859|issue:46399", "issue:45137|issue:45600", - "issue:44877|issue:45375", "issue:36246|issue:43596", + "issue:44877|issue:45375", "issue:43295|issue:45292", "issue:45292|issue:45584", - "issue:44242|issue:44857", - "issue:44623|issue:46726", "issue:44466|issue:45335", + "issue:44623|issue:46726", + "issue:44242|issue:44857", "issue:43526|issue:46897", "issue:44038|issue:46506", - "issue:44987|issue:45459", - "issue:29870|issue:30333" + "issue:29870|issue:30333", + "issue:44987|issue:45459" ], - "prepared_review_unit_hash": "fcd2e1e143b3f0c69b53048acc8a25801a327c6780cd39d8c19c6dff655e40a9", + "prepared_review_unit_hash": "005cbf72b70addd76a4062be5de9e3f093ee1f56ba7bf4835d18fc6dc4b848d3", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36558, - "estimated_input_tokens": 9140, - "estimated_eval_tokens": 18536 + "serialized_chars": 36554, + "estimated_input_tokens": 9139, + "estimated_eval_tokens": 18534 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly heterogeneous: a few multimodal/video/Qwen-related reports sit alongside unrelated import, config, logging, and training issues. None of the soft pairs look like the same underlying bug or change, so I would not merge any of them.", - "confidence": 0.72, - "canonical_issue_reason": "issue:44479 is the broadest, most representative bug report in the cluster\u2019s video/Qwen regression subset, but the overall cluster is still quite mixed.", + "summary": "This is a heterogeneous collection of unrelated transformer issues. None of the soft-similarity pairs look like the same underlying bug or change, so there is no true duplicate cluster to merge.", + "confidence": 0.96, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "issue:44479 has the clearest scope and the most concrete multi-model regression framing among the issues here, making it the best anchor if one must be chosen.", + "best_issue_reason": "No single issue is a strong cluster representative: the items span unrelated models, processors, configs, and runtime errors, so any choice would be arbitrary.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43550", "right": "issue:46506", "accept": false, - "reason": "Unrelated failures: torch.compile/SDPA in Bamba vs CPU/CUDA output differences in PPDocLayoutV3." + "reason": "Different model/runtime failures: torch.compile+SDPA vs CPU/CUDA result mismatch." }, { - "left": "issue:36246", - "right": "issue:42757", + "left": "issue:44779", + "right": "issue:45292", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and different root causes." + "reason": "Tokenizer correctness regression is unrelated to resize_token_embeddings not updating outputs." }, { - "left": "issue:44779", - "right": "issue:45292", + "left": "issue:36246", + "right": "issue:42757", "accept": false, - "reason": "Tokenizer correctness regression vs embedding-resize behavior; not the same code-path problem." + "reason": "Separate import errors for different symbols and packages." }, { "left": "issue:36246", "right": "issue:44351", "accept": false, - "reason": "Different missing imports from different modules; no common underlying bug is evident." + "reason": "Different missing-import symbols; not the same bug." }, { "left": "issue:41950", "right": "issue:44220", "accept": false, - "reason": "Video-classification processor selection and audio feature extraction are separate subsystems and bugs." + "reason": "Video pipeline image-processor lookup issue is unrelated to _torch_extract_fbank_features()." }, { "left": "issue:43901", "right": "issue:45003", "accept": false, - "reason": "Docs/behavior mismatch for a pipeline flag vs unsafe sys.modules access in modeling_utils." + "reason": "Docs mismatch for a pipeline is unrelated to unsafe sys.modules access." }, { "left": "issue:43257", "right": "issue:45137", "accept": false, - "reason": "Both involve DeepSpeed, but one is a weight-conversion/loading issue and the other is a zero3 deque failure." + "reason": "Both involve Deepspeed-adjacent failures, but the concrete bugs are different." }, { "left": "issue:45003", "right": "issue:45561", "accept": false, - "reason": "sys.modules access bug and xdist file-race bug are unrelated failure modes." + "reason": "Unsafe module lookup and xdist file races are unrelated issues." }, { "left": "issue:44496", "right": "issue:45003", "accept": false, - "reason": "Config/model-type loading failure is unrelated to sys.modules lookup safety." + "reason": "Unrecognized model/config loading failure is not the same as sys.modules access." }, { "left": "issue:43540", "right": "issue:46655", "accept": false, - "reason": "Both are video-related, but they affect different models and appear to stem from different preprocessing bugs." + "reason": "Different video-processor bugs in different model families and code paths." }, { "left": "issue:44162", "right": "issue:45200", "accept": false, - "reason": "ESM2 model breakage and Gemma 4 token-type defaults are unrelated." + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are unrelated." }, { "left": "issue:45375", "right": "issue:45759", "accept": false, - "reason": "Strict-config field dropping and text_config unwrapping are different multimodal config issues." + "reason": "Strict config field handling differs from AutoModel text_config unwrapping." }, { "left": "issue:43262", "right": "issue:44246", "accept": false, - "reason": "Audio sampling-rate defaulting and slow imports are not the same bug." + "reason": "Audio chat-template sampling-rate bug is unrelated to slow import behavior." }, { "left": "issue:44479", "right": "issue:46655", "accept": false, - "reason": "Both touch video inputs, but one is a broad Qwen regression and the other is a padding-calculation bug in SmolVLM." + "reason": "Both are video-related, but they concern different model-specific regressions and fixes." }, { "left": "issue:44295", "right": "issue:44749", "accept": false, - "reason": "Position_ids buffer access and performance regression after upgrade are unrelated." + "reason": "Position_ids buffer access and slower filtering after upgrade are unrelated." }, { "left": "issue:43475", "right": "issue:46369", "accept": false, - "reason": "Same SAM3 family, but one is a missing attribute and the other is a memory overflow; different code paths." + "reason": "SAM 3 attribute error and SAM lite text memory overflow are different failures." }, { "left": "issue:45859", "right": "issue:46399", "accept": false, - "reason": "Both mention Qwen3.5 and rotary-related code, but one is CPU-offload incompatibility and the other is a decorator/registration crash." + "reason": "CPU-offload incompatibility and missing decorator registration are distinct code-path problems." }, { "left": "issue:45137", "right": "issue:45600", "accept": false, - "reason": "DeepSpeed zero3 deque underflow and stale auto_mappings config references are unrelated." + "reason": "DeepSpeed deque underflow is unrelated to stale auto_mappings CI references." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for a vision processor is unrelated to a ZeRO3 BertModel index error." }, { "left": "issue:44877", "right": "issue:45375", "accept": false, - "reason": "Strict config loading problems, but for different model configs and different failure mechanisms." + "reason": "Different strict-config problems affecting different model configs and fields." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Regression in processor.tokenizer/image handling is unrelated to embedding resize propagation." }, { "left": "issue:45292", "right": "issue:45584", "accept": false, - "reason": "Embedding resize not propagating and Whisper empty-transcription generation failure are separate issues." + "reason": "Embedding resize behavior does not match empty-transcription generation failure." }, { - "left": "issue:44242", - "right": "issue:44857", + "left": "issue:44466", + "right": "issue:45335", "accept": false, - "reason": "Router-logit load-balancing loss and AMP/CUDA loss crash are different training bugs." + "reason": "Both mention embedding serialization/resize, but the concrete issues and affected modules differ." }, { "left": "issue:44623", "right": "issue:46726", "accept": false, - "reason": "Missing files on save_pretrained vs missing vision backend during AutoProcessor loading are not the same issue." + "reason": "Missing processor save files is not the same as missing vision backend handling." }, { - "left": "issue:44466", - "right": "issue:45335", + "left": "issue:44242", + "right": "issue:44857", "accept": false, - "reason": "Both involve embedding/weight handling, but they affect different models and different serialization paths." + "reason": "MoE router-loss behavior and an AMP CUDA image-loss crash are unrelated." }, { "left": "issue:43526", "right": "issue:46897", "accept": false, - "reason": "Label reduction in BeitImageProcessorFast and Florence2 training-loss shift are unrelated." + "reason": "BEiT label reduction bug is unrelated to Florence2 training-loss shifting." }, { "left": "issue:44038", "right": "issue:46506", "accept": false, - "reason": "A Qwen3-VL-Moe regression and CPU/CUDA output divergence are not the same bug." + "reason": "Qwen3-VL-Moe regression and CPU/CUDA output divergence are different problems." }, { - "left": "issue:44987", - "right": "issue:45459", + "left": "issue:29870", + "right": "issue:30333", "accept": false, - "reason": "Model loading failure for a specific repo and tokenizer protobuf error masking are distinct import/load paths." + "reason": "BART generate() UnboundLocalError and MLflow job-status reporting are unrelated." }, { - "left": "issue:29870", - "right": "issue:30333", + "left": "issue:44987", + "right": "issue:45459", "accept": false, - "reason": "Generation-time decoder mask bug vs MLflow job-reporting bug; no shared underlying code-path." + "reason": "Model-loading failure with physical-intelligence/fast is unrelated to tokenizer protobuf error masking." } ] }, @@ -13523,20 +13565,20 @@ "issue:43065|issue:46123", "issue:44297|issue:44623" ], - "prepared_review_unit_hash": "b7842d71a505cbda596f6c94927af71335af875c9c6c1a7207bb20c28e625ece", + "prepared_review_unit_hash": "18463ae58cec961f8ad88813298466d33c3fea48c14c82405b791a4284a60626", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35967, + "serialized_chars": 35968, "estimated_input_tokens": 8992, "estimated_eval_tokens": 18240 }, @@ -13545,186 +13587,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous: tokenizer, cache, vision, generation, and model-loading bugs across different subsystems. None of the soft pairs look like the same concrete bug/change, so I would not merge any of them as duplicates. The best representative issue is the broad video-regression report because it clearly describes a multi-model, user-visible regression with a specific failure mode.", - "confidence": 0.87, - "canonical_issue_reason": "Issue 44479 is the strongest representative: it\u2019s a clear regression report, affects multiple related Qwen VL/video model families, and has a concrete, user-visible failure mode.", + "summary": "Most pairs are unrelated lookalikes and should be kept separate. The only credible duplicate is the pair about `StaticCache.get_seq_length()` returning a tensor instead of an `int`.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 45987 is the cleanest root-cause report for the `get_seq_length()` return-type bug; it states the contract violation directly and can subsume the backend-specific failure in 46858.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 44479 is the best single issue to anchor the cluster because it is broad, high-impact, and actionable compared with the more narrow or ambiguous reports.", + "best_issue_reason": "45987 is the strongest canonical issue candidate overall because it describes the underlying API bug broadly and precisely, rather than just one downstream symptom.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43352", "right": "issue:45250", "accept": false, - "reason": "Same general topic (Flash Attention 2.0), but one is a specific Nemotron support gap and the other is a generic request; not the same underlying fix." + "reason": "Different scopes: model-specific Flash Attention support error vs a generic FA2 topic/request." }, { "left": "issue:42199", "right": "issue:46851", "accept": false, - "reason": "Completely different model families and failure modes: DETR cardinality vs Qwen3.5 variable-length position_ids." + "reason": "Completely unrelated models and failures: DETR cardinality error vs Qwen3.5 variable-length position_ids." }, { "left": "issue:43526", "right": "issue:46894", "accept": false, - "reason": "Different processors and bugs; label reduction in BEiT is unrelated to CohereASR training-loss shifting." + "reason": "Different bugs in unrelated pipelines; no shared code-path or symptom." }, { "left": "issue:44060", "right": "issue:44077", "accept": false, - "reason": "One is a tied-weights warning in Qwen3-Next, the other is a patchtsmixer post_init policy issue; no shared code-path bug." + "reason": "One is a tied-weights warning in Qwen3-Next; the other is a patchtsmixer post_init policy issue." }, { "left": "issue:44368", "right": "issue:44977", "accept": false, - "reason": "Both mention Qwen3.5, but one is a config warning and the other is a flash-attention generation failure; not the same bug." + "reason": "Both mention Qwen3.5, but one is a weight-tying warning and the other is a flash-attention generation failure." }, { "left": "issue:45412", "right": "issue:46123", "accept": false, - "reason": "RT-DETR memory release and MaskGenerationPipeline batching are unrelated bugs in different subsystems." + "reason": "RT-DETR memory cleanup and MaskGenerationPipeline batch handling are unrelated." }, { "left": "issue:43653", "right": "issue:44162", "accept": false, - "reason": "BigBirdTokenizer special-token decoding is unrelated to the ESM2 regression." + "reason": "Tokenizer special-token registration vs ESM2 breakage; no common underlying bug." }, { "left": "issue:44479", "right": "issue:46170", "accept": false, - "reason": "Different problem domains: video input regression vs Windows/ROCm import dependency failure." + "reason": "Video-input regression in Qwen VL models vs a Windows/ROCm distributed import failure." }, { "left": "issue:42886", "right": "issue:45229", "accept": false, - "reason": "Tokenizer offline-cache loading and Gemma4 multi-GPU OOM are unrelated." + "reason": "Offline tokenizer cache loading and Gemma multi-GPU OOM are different failure modes." }, { "left": "issue:45397", "right": "issue:45656", "accept": false, - "reason": "Gemma-4 ZeRO3 loading and a deepspeed double-step optimizer bug are different failure modes." + "reason": "Gemma-4 loading/ZeRO-3 issue is not the same as a DeepSpeed double optimizer-step bug." }, { "left": "issue:43065", "right": "issue:43408", "accept": false, - "reason": "Both involve SAM3 naming, but one is a dummy Conv2d implementation detail and the other is a wrong model-type warning." + "reason": "Both touch SAM3, but one is a dummy conv implementation detail and the other is a model-type mismatch warning." }, { "left": "issue:43475", "right": "issue:43540", "accept": false, - "reason": "SAM3 vision encoder attribute error is unrelated to Qwen3OmniMoe video input processing." + "reason": "SAM3 video attribute error and Qwen3OmniMoe video ValueError are unrelated." }, { "left": "issue:44877", "right": "issue:46170", "accept": false, - "reason": "Strict config loading for granite_speech is unrelated to the Windows/ROCm distributed import issue." + "reason": "Strict config loading for granite_speech is unrelated to the Windows ROCm dependency error." }, { "left": "issue:46170", "right": "issue:46832", "accept": false, - "reason": "Both are runtime errors, but one is a platform import/dependency problem and the other is an RT-DETR forward shape/path issue." + "reason": "Different problems: import/dependency failure vs RT-DETR forward-shape TypeError." }, { "left": "issue:44368", "right": "issue:45563", "accept": false, - "reason": "A Qwen3.5 tied-embeddings warning is unrelated to the paged generate() stale warning." + "reason": "Weight-tying warning and stale `num_return_sequences` warning do not share the same bug." }, { "left": "issue:43723", "right": "issue:46710", "accept": false, - "reason": "Tokenizer-loading issues are not the same as the DeepSeek accuracy regression after upgrading Transformers." + "reason": "Tokenizer loading regression and model accuracy regression are different issues." }, { "left": "issue:44995", "right": "issue:45200", "accept": false, - "reason": "GlmMoeDsa stale cache crash and Gemma4 mm_token_type_ids defaulting are different code paths and symptoms." + "reason": "GlmMoeDsa cache crash vs Gemma 4 text-only fine-tuning mask-token defaults." }, { "left": "issue:43931", "right": "issue:44368", "accept": false, - "reason": "Weight-shape mismatch on model load is unrelated to the Qwen3.5 tie_word_embeddings warning." + "reason": "Model weight-shape mismatch is unrelated to a tied-embeddings warning." }, { "left": "issue:44188", "right": "issue:46202", "accept": false, - "reason": "Attention-kernel divergence under torch.compile is unrelated to Gemma pipeline roundtrip text issues." + "reason": "torch.compile attention-kernel divergence and Gemma token roundtrip failure are not the same bug." }, { "left": "issue:44483", "right": "issue:46202", "accept": false, - "reason": "API request rejection in /v1/chat/completions does not match Gemma pipeline roundtrip failure." + "reason": "API request rejection and tokenizer/pipeline roundtrip behavior are unrelated." }, { "left": "issue:46829", "right": "issue:46858", "accept": false, - "reason": "Both mention output/cache behavior, but one is a classifier output field mix-up and the other is a cache length tensor/arithmetic bug." + "reason": "VideoPrism output packaging bug vs cache-length arithmetic bug; different code paths and symptoms." }, { "left": "issue:45987", "right": "issue:46082", "accept": false, - "reason": "StaticCache return-type bug is unrelated to LlamaConfig rejecting an explicit head_dim." + "reason": "StaticCache return-type bug and LlamaConfig head_dim validation are unrelated." }, { "left": "issue:43906", "right": "issue:46202", "accept": false, - "reason": "A reproduction of a different upstream issue is not the same as the Gemma pipeline roundtrip bug." + "reason": "An isolated reproduction issue is not the same as the Gemma roundtrip failure reported here." }, { "left": "issue:44995", "right": "issue:45588", "accept": false, - "reason": "Both are model-runtime crashes, but the stale indexer cache and s_aux=None flash-attention errors are different problems." + "reason": "Second-forward cache corruption in GlmMoeDsa is unrelated to flash_attention.py crashing on `s_aux=None`." }, { "left": "issue:44188", "right": "issue:46421", "accept": false, - "reason": "Attention kernel branching and cache lazy initialization causing recompilation are not the same underlying bug." + "reason": "Both involve compile/cache words, but one is kernel divergence and the other is lazy cache init causing recompilation." }, { "left": "issue:45685", "right": "issue:46431", "accept": false, - "reason": "MPS histogram/int support and missing chat template errors are unrelated." + "reason": "MPS histogram/int issue is unrelated to missing chat-template support in a processor." }, { "left": "issue:45103", "right": "issue:46858", "accept": false, - "reason": "Docstring annotation handling and StaticCache seq-length arithmetic are distinct issues." + "reason": "Docstring annotation handling and static-cache sequence-length arithmetic are unrelated." }, { "left": "issue:43065", "right": "issue:46123", "accept": false, - "reason": "SAM3 decoder implementation detail and MaskGenerationPipeline partial-batch behavior are unrelated." + "reason": "SAM3 pixel-decoder implementation details and MaskGenerationPipeline batch termination are unrelated." }, { "left": "issue:44297", "right": "issue:44623", "accept": false, - "reason": "Tokenizer save metadata mismatch and processor save missing files are both save_pretrained problems, but not the same concrete bug." + "reason": "Tokenizer save metadata mismatch and processor save missing files are both save_pretrained-adjacent but not the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46858", + "accept": true, + "reason": "Same underlying bug: `get_seq_length()` returns a tensor instead of an `int`, and 46858 is the downstream static-cache/compile manifestation of the same contract violation." } ] }, @@ -13815,9 +13863,9 @@ "issue:41628|issue:44933", "issue:43475|issue:46821", "issue:45310|issue:45375", + "issue:44964|issue:45137", "issue:45230|issue:45397", "issue:43986|issue:46726", - "issue:44964|issue:45137", "issue:44869|issue:45356", "issue:44038|issue:44623", "issue:39692|issue:46726", @@ -13828,106 +13876,106 @@ "issue:43723|issue:46326", "issue:45406|issue:46202" ], - "prepared_review_unit_hash": "a30584cb546cc6bade2a1b6f7da111adeb72bf3b106c335aa90616741ebc64d5", + "prepared_review_unit_hash": "d8c870de3ab39edd36f51ee753c103fb92c5b7654cdcf687609afb282a083dc9", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36055, - "estimated_input_tokens": 9014, - "estimated_eval_tokens": 18284 + "serialized_chars": 36049, + "estimated_input_tokens": 9013, + "estimated_eval_tokens": 18282 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "Mostly heterogeneous cluster with a few tight duplicates around Transformers v5 config/init regressions and a couple of AutoProcessor/backend-loading issues. The broadest anchor is still a generic v5 model-initialization compatibility report, while most pairs are only loosely related by subsystem.", - "confidence": 0.58, - "canonical_issue_reason": "issue:43646 is the broadest representative of the cluster: a Transformers 5.0 model-initialization compatibility regression that can subsume several narrower load-time/config failures.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues. The only soft duplicate that looks credible is the Qwen3.5-Moe config/from_pretrained regression pair; the rest are different models, different code paths, or different symptoms.", + "confidence": 0.9, + "canonical_issue_reason": "issue:45375 is the better canonical issue for the Qwen3.5-Moe regression because it identifies the likely root cause directly: a missing `deepstack_visual_indexes` field being dropped by `@strict`.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43646 is the best anchor because it captures the widest underlying regression class in the set, whereas most other issues are more specific manifestations or unrelated bugs.", + "best_issue_reason": "issue:45310 is the best overall representative because it is the clearest user-facing report of the Qwen3.5-Moe `from_pretrained` failure and provides the broadest symptom framing.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44361", "right": "issue:44749", "accept": false, - "reason": "Different failures: MLukeTokenizer task AttributeError vs a data-filtering performance regression." + "reason": "Different problems: a tokenizer AttributeError versus a dataset/filtering slowdown regression." }, { "left": "issue:43994", "right": "issue:46726", "accept": false, - "reason": "Reject: 43994 is about nonsensical SigLIP2 outputs via AutoModel/pipeline, while 46726 is a missing-vision-backend AutoProcessor error path; same broad area, but not the same concrete bug or code path from the packet." + "reason": "Both involve vision models, but one is wrong output from AutoModel/pipeline while the other is a missing-backend AutoProcessor error; not the same bug." }, { "left": "issue:44749", "right": "issue:45656", "accept": false, - "reason": "One is a filtering slowdown; the other is a deepspeed optimizer-step duplication issue." + "reason": "Unrelated: data filtering performance versus DeepSpeed optimizer stepping twice." }, { "left": "issue:43701", "right": "issue:44749", "accept": false, - "reason": "Checkpoint key mismatch is unrelated to the filtering slowdown regression." + "reason": "Checkpoint key mismatch is unrelated to the reported slowdown regression." }, { "left": "issue:44871", "right": "issue:46132", "accept": false, - "reason": "Gemma eos-token config mismatch is a model config issue, not an AttentionInterface.register behavior change." + "reason": "Gemma eos token config mismatch and AttentionInterface behavior change are different code paths." }, { "left": "issue:44623", "right": "issue:45656", "accept": false, - "reason": "processor.save_pretrained missing files is unrelated to deepspeed optimizer-step duplication." + "reason": "Processor save files issue is unrelated to DeepSpeed optimizer stepping." }, { "left": "issue:43646", "right": "issue:44162", "accept": false, - "reason": "Both are broad v5 regressions, but they affect different model paths and concrete failures." + "reason": "Different failures: missing fields in AutoConfig with remote code versus ESM2 breakage." }, { "left": "issue:43441", "right": "issue:46752", "accept": false, - "reason": "FlashAttention failure for Ministral-3 is unrelated to empty-list chat template indexing." + "reason": "FlashAttention/Ministral regression and empty-chat-template crash are unrelated." }, { "left": "issue:45375", "right": "issue:45478", "accept": true, - "reason": "Same Qwen3.5-Moe from_pretrained/config loading regression; the missing field in 45375 plausibly explains the later generic load failure." + "reason": "Same underlying Qwen3.5-Moe loading/config regression; the missing field in one issue plausibly explains the from_pretrained error in the other." }, { "left": "issue:42994", "right": "issue:43482", "accept": false, - "reason": "Quantized model saving failure and GGUF loading failure are different code paths and formats." + "reason": "Quantized model saving and GGUF loading are different operations with different failure modes." }, { "left": "issue:43475", "right": "issue:43643", "accept": false, - "reason": "SAM3 output attribute missing and trust_remote_code config field loss are distinct bugs." + "reason": "SAM3 vision encoder attribute error is unrelated to trust_remote_code config field loss." }, { "left": "issue:43232", "right": "issue:43673", "accept": false, - "reason": "Both are generation-related, but one concerns sync_gpus/update_model_kwargs and the other chunked_prefill cache availability." + "reason": "Both are generation-related, but they describe different internal bugs and scenarios." }, { "left": "issue:44188", @@ -13939,106 +13987,106 @@ "left": "issue:28282", "right": "issue:30990", "accept": false, - "reason": "PyTorch import-missing error and Sentence Transformers loading hang are different failures." + "reason": "ImportError for missing PyTorch and a Sentence Transformers loading hang are not the same issue." }, { "left": "issue:41628", "right": "issue:44933", "accept": false, - "reason": "Both mention imports, but one is a missing top-level AutoImageProcessor export and the other is an image_utils import issue; not clearly the same bug." + "reason": "Both mention imports, but one is a missing top-level AutoImageProcessor symbol and the other is a bad/nonexistent image_utils import." }, { "left": "issue:43475", "right": "issue:46821", "accept": false, - "reason": "SAM3VisionEncoderOutput missing an attribute is unrelated to the TimesFM window_size AttributeError." + "reason": "SAM3 encoder attribute error and TimesFM window_size AttributeError are unrelated model bugs." }, { "left": "issue:45310", "right": "issue:45375", "accept": true, - "reason": "Both describe the same Qwen3.5-Moe from_pretrained/config loading regression." + "reason": "Same Qwen3.5-Moe regression: one reports the user-facing from_pretrained failure, the other identifies the missing config field root cause." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." }, { "left": "issue:45230", "right": "issue:45397", "accept": false, - "reason": "A generic bug report is not enough, and gemma-4 zero3 from_pretrained is a separate issue." + "reason": "A generic bug report is not the same as a Gemma-4 ZeRO3 loading failure." }, { "left": "issue:43986", "right": "issue:46726", - "accept": true, - "reason": "Both are AutoProcessor crashes tied to missing vision-backend support (notably torchvision), so they look like the same backend-handling bug." - }, - { - "left": "issue:44964", - "right": "issue:45137", "accept": false, - "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque exhaustion are unrelated." + "reason": "Both are about missing vision dependencies, but one is a video-model AutoProcessor crash and the other is an unrecognized image processor error; too different to merge." }, { "left": "issue:44869", "right": "issue:45356", "accept": false, - "reason": "Whisper timestamp decoding with a trailing replacement character is a different tokenizer/codec problem than Kimi-K2.5 regression handling." + "reason": "Whisper timestamp decoding crash and Kimi tokenizer regression affect different tokenization paths." }, { "left": "issue:44038", "right": "issue:44623", "accept": false, - "reason": "Qwen3-VL-Moe transformers v5 bug and processor.save_pretrained missing files are separate issues." + "reason": "Qwen3-VL-Moe from_pretrained failure and processor.save_pretrained file omission are not the same bug." }, { "left": "issue:39692", "right": "issue:46726", "accept": false, - "reason": "SigLIP2 docs example errors are documentation/repro problems, not the same runtime missing-backend processor crash." + "reason": "Documentation example problems and missing-backend AutoProcessor runtime error are different issues." }, { "left": "issue:46463", "right": "issue:46877", "accept": false, - "reason": "Incorrect T5 error messaging and auto_docstring stdout diagnostics are different concerns." + "reason": "T5 decoder-input validation messaging and auto_docstring stdout lint noise are unrelated." }, { "left": "issue:44373", "right": "issue:46851", "accept": false, - "reason": "Wrong docstring for position_ids is documentation-only; Qwen3.5 variable-length input failure is a runtime bug." + "reason": "A docstring bug about position_ids is not the same as the Qwen3.5 variable-length input failure." }, { "left": "issue:42994", "right": "issue:43525", "accept": false, - "reason": "Quantized model saving and Llama4Config missing pad_token_id are different failures." + "reason": "Quantized saving failure and missing pad_token_id on Llama4Config are different regressions." }, { "left": "issue:42371", "right": "issue:45092", "accept": false, - "reason": "TF32 API settings guidance is unrelated to remote-code checkpoint/meta-initialization incompatibility." + "reason": "TF32 API guidance and InternVL2 meta-init incompatibility are unrelated." }, { "left": "issue:43723", "right": "issue:46326", "accept": false, - "reason": "Tokenizer loading in v5 and continue_final_message for mistral_common are different tokenizer paths." + "reason": "Tokenizer loading failure and continue_final_message behavior change are different tokenizer issues." }, { "left": "issue:45406", "right": "issue:46202", "accept": false, - "reason": "Gemma4Processor missing _tokenizer in serve is unrelated to Gemma pipeline roundtrip failure on \u2581." + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on \u2581 are distinct bugs." } ] }, "evaluator_result": { - "accept": false, - "feedback": "Rejecting because one soft-edge acceptance is not sufficiently grounded: issue:43994 is about nonsensical results with AutoModel/pipeline for SigLIP2, but the reason claims a missing-vision-backend crash similar to issue:46726. That connection is too strong from the packet alone. The rest of the summary is broadly conservative, but this overstates evidence." + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted duplicate cluster is the Qwen3.5-Moe regression, which is supported by matching titles and the explicit root-cause note in issue:45375. The rest of the rejections are appropriately cautious and do not overclaim similarity." }, "evaluator_used": true, - "retried": true, + "retried": false, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -14116,8 +14164,8 @@ "issue:44038|issue:45310", "issue:36010|issue:41628", "issue:44222|issue:45310", - "issue:46202|issue:46701", "issue:43278|issue:45137", + "issue:46202|issue:46701", "issue:42994|issue:43824", "issue:45081|issue:45103", "issue:42831|issue:44222", @@ -14137,20 +14185,20 @@ "issue:44623|issue:45310", "issue:42673|issue:46693" ], - "prepared_review_unit_hash": "0376a5a01a7b1d17fb34288152ab842e94d5440032ff5b1e58a3820e40fc2bcb", + "prepared_review_unit_hash": "3d51abb6f295b79d32931c50ccf270a1a752b10e7747dbafcfae8f7503a4491b", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 30, - "serialized_chars": 36540, + "serialized_chars": 36537, "estimated_input_tokens": 9135, "estimated_eval_tokens": 18526 }, @@ -14159,198 +14207,198 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated bug reports that only share broad surface area (imports, model loading, generation, tokenizer behavior). The only clear duplicate is the Qwen3ForCausalLM VRAM leak pair (42673/43704).", - "confidence": 0.84, - "canonical_issue_reason": "issue:42673 is the clearest duplicate anchor: it matches issue:43704 by title and bug description, and it has the stronger discussion/reference signal to serve as the kept issue.", + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues; the only clear duplicate is the repeated Qwen3ForCausalLM VRAM leak report. Most other soft matches are merely thematically similar (imports, model loading, tokenizer regressions, FP8, etc.) and should stay separate.", + "confidence": 0.63, + "canonical_issue_reason": "issue:41628 is the best anchor for the cluster\u2019s recurring import/export failures: it is a generic, high-signal missing-symbol bug in `transformers` and sits closest to several other import-related reports.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42673 is the best representative issue in this set because it is a concrete, actionable bug report and the only one with an exact duplicate in the cluster.", + "best_issue_reason": "issue:41628 is the most representative issue in the import-error subgroup and is broader/more central than the more model-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43873", "right": "issue:44945", "accept": false, - "reason": "Both are generation-related, but one is a quantization/offloading problem and the other is pipeline-parallelism output corruption; different code paths and symptoms." + "reason": "Different symptoms and code paths: quantization/offloading vs pipeline-parallel generation output correctness." }, { "left": "issue:44945", "right": "issue:45753", "accept": false, - "reason": "Incorrect output under pipeline parallelism is not the same bug as an infinite loop on a specific image in Qwen3_5." + "reason": "One is pipeline-parallel output corruption; the other is an image-specific infinite loop in Qwen3_5." }, { "left": "issue:43784", "right": "issue:46399", "accept": false, - "reason": "Both are runtime/import failures, but they involve different symbols and subsystems (`nn` import vs Qwen3.5 kernel registration)." + "reason": "Both are import/runtime errors, but they affect different packages and mechanisms (`nn` import vs rotary registration)." }, { "left": "issue:41628", "right": "issue:44351", "accept": false, - "reason": "These are separate missing-export issues for different classes (`AutoImageProcessor` vs `HybridCache`); same pattern, not the same bug." + "reason": "Both are import errors, but for different missing symbols and likely different fixes." }, { "left": "issue:44857", "right": "issue:46877", "accept": false, - "reason": "A CUDA/AMP loss crash is unrelated to auto_docstring printing diagnostics at import time." + "reason": "Unrelated: AMP/CUDA loss crash versus docstring logging on import." }, { "left": "issue:45593", "right": "issue:46877", "accept": false, - "reason": "D-FINE auxiliary-loss behavior and docstring import-time stdout are unrelated." + "reason": "Different subsystems entirely: model loss behavior vs import-time doc-lint output." }, { "left": "issue:42673", "right": "issue:43704", "accept": true, - "reason": "Exact same Qwen3ForCausalLM VRAM leak reported with the same title and threading context; clear duplicate." + "reason": "Same exact reported bug: Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads." }, { "left": "issue:44038", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen3 Moe models, but one is a general 5.0 bug and the other is a from_pretrained error on newer versions; not the same concrete failure." + "reason": "Related model family, but different model versions and different failure conditions; not the same concrete bug." }, { "left": "issue:36010", "right": "issue:41628", "accept": false, - "reason": "Different import errors for different APIs (`GenerationMixin` vs `AutoImageProcessor`)." + "reason": "Both are missing-import complaints, but they are different exported symbols and separate failures." }, { "left": "issue:44222", - "right": "issue:45310", - "accept": false, - "reason": "One is FP8 save_pretrained behavior for moe models; the other is a model loading error. Similar area, different bug." - }, - { - "left": "issue:46202", - "right": "issue:46701", + "right": "issue:45310", "accept": false, - "reason": "Both involve Gemma-related pipelines, but one is tokenizer roundtrip failure and the other is an image-text-to-text pipeline failure." + "reason": "Saving FP8 MoE models vs `from_pretrained` loading failure are distinct code paths." }, { "left": "issue:43278", "right": "issue:45137", "accept": false, - "reason": "Embedding dtype drift in eval and a DeepSpeed ZeRO3 deque IndexError are unrelated." + "reason": "Training/eval dtype mismatch is unrelated to the DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Both involve Gemma-family pipelines, but one is tokenizer roundtrip and the other is an image-text pipeline output issue." }, { "left": "issue:42994", "right": "issue:43824", "accept": false, - "reason": "Quantized model saving and a missing Qwen2_5_VLForConditionalGeneration export are separate problems." + "reason": "Quantized model saving and a missing `Qwen2_5_VLForConditionalGeneration` import are unrelated." }, { "left": "issue:45081", "right": "issue:45103", "accept": false, - "reason": "Tokenizer regex patching crash and auto_docstring annotation parsing crash are unrelated code paths." + "reason": "Both raise AttributeError, but in unrelated utilities and code paths." }, { "left": "issue:42831", "right": "issue:44222", "accept": false, - "reason": "Both mention FP8, but one is an accuracy regression and the other is a save_pretrained/moe issue; not the same bug." + "reason": "FP8 accuracy regression versus FP8 save/load failure are different bugs." }, { "left": "issue:43232", "right": "issue:44792", "accept": false, - "reason": "Generation kwargs handling after sync_gpus is unrelated to a janus image-generation test failure." + "reason": "Generation kwargs handling after GPU sync is unrelated to the Janus image-generation test failure." }, { "left": "issue:43504", "right": "issue:45561", "accept": false, - "reason": "Pretrained preset loading and pytest-xdist file races are unrelated." + "reason": "Model preset loading bug vs pytest-xdist race in testing utils; no shared underlying defect." }, { "left": "issue:42175", "right": "issue:44933", "accept": false, - "reason": "A missing TensorFlow backend dependency and a missing image_utils import are different issues." + "reason": "Packaging/backend installation issue vs a missing import from `image_utils`; different problems." }, { "left": "issue:45561", "right": "issue:45600", "accept": false, - "reason": "Test-worker file races and stale auto_mappings references are separate CI failures." + "reason": "Parallel test-file race vs stale auto-mapping references; unrelated." }, { "left": "issue:43976", "right": "issue:44933", "accept": false, - "reason": "Python-version incompatibility and an import from image_utils are not the same underlying bug." + "reason": "Python-version compatibility issue vs missing import from `image_utils`; not the same bug." }, { "left": "issue:44496", "right": "issue:45561", "accept": false, - "reason": "Model config loading failure and a captured_info.txt race are unrelated." + "reason": "Model config loading failure vs test utility race; unrelated." }, { "left": "issue:30333", "right": "issue:33290", "accept": false, - "reason": "Both are training/job failures, but one is MLflow status reporting and the other is an OOM during optimization." + "reason": "MLflow reporting on failed training vs OOM in Adafactor/DeepSpeed are different failures." }, { "left": "issue:45405", "right": "issue:46877", "accept": false, - "reason": "A bumped PEFT minimum version and docstring stderr/stdout behavior are unrelated." + "reason": "Dependency-version bump issue vs docstring import-time logging; unrelated." }, { "left": "issue:30990", "right": "issue:35141", "accept": false, - "reason": "Loading hang in sentence-transformers and post_init reinitializing embeddings are different failure modes." + "reason": "Loading hang vs embedding reinitialization on resize are separate model behaviors." }, { "left": "issue:42898", "right": "issue:44568", "accept": false, - "reason": "Both are tokenizer behavior changes in v5, but one is cleanup-space behavior and the other is add_special_tokens/BOS-EOS handling." + "reason": "Both are tokenizer regressions in v5, but they concern different APIs and symptoms." }, { "left": "issue:42175", "right": "issue:45712", "accept": false, - "reason": "Backend dependency installation and leftover dummy classes are unrelated." + "reason": "Extras/backend packaging issue vs dummy class leakage/check-repo failure; different root causes." }, { "left": "issue:43479", "right": "issue:46829", "accept": false, - "reason": "Different multimodal model bugs: default config initialization vs incorrect hidden_states assignment." + "reason": "Incorrect default multimodal config initialization vs output-field mismatch in video classification are unrelated." }, { "left": "issue:46077", "right": "issue:46123", "accept": false, - "reason": "Bidirectional attention masking and pipeline batch-finalization are unrelated." + "reason": "Attention-mask construction bug vs pipeline batching/drop-results bug; not the same change." }, { "left": "issue:44623", "right": "issue:45310", "accept": false, - "reason": "Missing files on processor.save_pretrained and a Qwen3.5 Moe from_pretrained error are distinct issues." + "reason": "Processor save_pretrained file omission vs Qwen3.5 from_pretrained error are different code paths." }, { "left": "issue:42673", "right": "issue:46693", "accept": false, - "reason": "VRAM leak in Qwen3ForCausalLM threads and FA2 long-context performance regression are different bugs." + "reason": "VRAM leak in multi-threaded dataloaders vs FA2 long-context performance regression; unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "Grounded and conservative overall. The cluster summary matches the packet: most links are only surface-similar, and the 42673/43704 pair is the only clearly identical duplicate by title with stronger activity on 42673. The soft-edge rejections stay appropriately cautious and do not overclaim shared root causes." + "feedback": "The assessment is grounded in the packet and stays conservative on the soft edges. The only accepted pair, issue:42673 vs issue:43704, is a clear duplicate by title. The other pair rejections are cautious and consistent with the issue titles; no overstatement of same-bug evidence stands out." }, "evaluator_used": true, "retried": false, @@ -14430,10 +14478,10 @@ "issue:43479|issue:44079", "issue:43874|issue:45092", "issue:43784|issue:43824", - "issue:44734|issue:45103", "issue:27957|issue:30333", - "issue:42222|issue:45200", + "issue:44734|issue:45103", "issue:30333|issue:34689", + "issue:42222|issue:45200", "issue:42898|issue:43295", "issue:42898|issue:45356", "issue:44295|issue:45656", @@ -14441,217 +14489,218 @@ "issue:30333|issue:35141", "issue:45405|issue:45712", "issue:44222|issue:44297", - "issue:44060|issue:45125", "issue:22355|issue:28282", + "issue:44060|issue:45125", "issue:44075|issue:46231", "issue:44933|issue:45712", "issue:44360|issue:46762", "issue:42831|issue:44623", "issue:44297|issue:45656", - "issue:42796|issue:43723" + "issue:42796|issue:43723", + "issue:44222|issue:44623" ], - "prepared_review_unit_hash": "67f2ee993beeede8a9568fd77e151104e9b0d88bce43f9b320cee5b6f79393f3", + "prepared_review_unit_hash": "52264eef50f917ce4267bcce38fdfe3d054f25fbef1fa658eb408c7a6262c1bc", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 29, - "serialized_chars": 35957, - "estimated_input_tokens": 8990, - "estimated_eval_tokens": 18236 + "soft_pair_count": 30, + "serialized_chars": 36257, + "estimated_input_tokens": 9065, + "estimated_eval_tokens": 18386 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers issues that only look superficially similar by embedding. The soft edges are largely false positives across tokenizer regressions, import errors, config bugs, docs/notebooks, and training/runtime issues.", - "confidence": 0.82, - "canonical_issue_reason": "issue:42898 is the broadest and most central tokenizer-regression report in the set (`clean_up_tokenization_spaces` behavior changes in v5), which makes it the best representative issue among the tokenizer-related items.", + "summary": "This cluster is highly heterogeneous: most items are unrelated model/import/docs/training bugs. Several pairs are only loosely similar because they mention tokenizers, v5 regressions, or documentation, but they do not describe the same concrete bug or change.", + "confidence": 0.71, + "canonical_issue_reason": "issue:42898 is the best representative of the cluster because it is a core tokenizer regression report in v5 and is one of the few items that recurs across multiple soft-similarity links, even though the cluster is not actually duplicate-heavy.", "canonical_pr_reason": null, - "best_issue_reason": "issue:42898 is the most suitable issue anchor because it captures a general v5 tokenizer behavior regression and sits closest to several other tokenizer-themed false positives, even though they are not true duplicates.", + "best_issue_reason": "issue:42898 is the most central issue to this noisy cluster\u2019s recurring tokenizer-v5 theme; it is the closest thing to a hub, but the surrounding issues are too diverse to treat it as a true duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43479", "right": "issue:44869", "accept": false, - "reason": "Different components and failures: multimodal config initialization vs Whisper timestamp decode crash." + "reason": "Different models and failures: multimodal config initialization vs Whisper word timestamp decoding." }, { "left": "issue:43723", "right": "issue:44568", "accept": false, - "reason": "Both involve tokenizers, but one is AutoTokenizer loading in v5 and the other is missing BOS/EOS insertion for a specific tokenizer path." + "reason": "Both involve tokenizers in v5, but one is tokenizer loading and the other is add_special_tokens behavior for a specific model." }, { "left": "issue:44016", "right": "issue:46231", "accept": false, - "reason": "Notebook syntax error is unrelated to a documentation parameter-name mismatch." + "reason": "Both are documentation-related, but one is a notebook syntax error and the other is a parameter-name mismatch in docs." }, { "left": "issue:42898", "right": "issue:45812", "accept": false, - "reason": "Both are tokenizer regressions, but one is `clean_up_tokenization_spaces` semantics and the other is wrong token IDs for Granite models." + "reason": "Both are tokenizer regressions, but they affect different APIs and different model families; not the same bug." }, { "left": "issue:44625", "right": "issue:46877", "accept": false, - "reason": "A config propagation bug is not the same as auto_docstring printing diagnostics at import time." + "reason": "Unrelated: Qwen config propagation vs auto_docstring import-time logging." }, { "left": "issue:42898", "right": "issue:44779", "accept": false, - "reason": "Related theme, but different concrete bugs: a global cleanup-behavior change vs Deepseek tokenization output regression." + "reason": "Both are tokenizer regressions, but one is a cleanup/decoding behavior change and the other is a Deepseek tokenization correctness bug." }, { "left": "issue:39692", "right": "issue:46909", "accept": false, - "reason": "Both are docs/examples, but they concern different example pages and different failure modes." + "reason": "Both are broken docs examples, but they are different examples with different failure modes." }, { "left": "issue:43479", "right": "issue:44079", "accept": false, - "reason": "Multimodal config defaults are unrelated to ModelOutput key assignment behavior." + "reason": "Different subsystems: multimodal config defaults vs ModelOutput key assignment logic." }, { "left": "issue:43874", "right": "issue:45092", "accept": false, - "reason": "Both touch multimodal loading, but one is a missing processor method and the other is remote-code checkpoint/meta-init incompatibility." + "reason": "Both are multimodal model issues, but one is a missing processor method and the other is remote-code/meta-init compatibility." }, { "left": "issue:43784", "right": "issue:43824", "accept": false, - "reason": "Import errors in different packages for different missing symbols; not the same import-path bug." - }, - { - "left": "issue:44734", - "right": "issue:45103", - "accept": false, - "reason": "Serve KV-cache indexing crash is unrelated to the auto_docstring annotations crash." + "reason": "Different import errors affecting different packages/models." }, { "left": "issue:27957", "right": "issue:30333", "accept": false, - "reason": "Flash Attention model support issue and MLFlow reporting bug are unrelated." + "reason": "Completely unrelated: Flash Attention on XLM-R vs MLflow reporting after failed training jobs." }, { - "left": "issue:42222", - "right": "issue:45200", + "left": "issue:44734", + "right": "issue:45103", "accept": false, - "reason": "Both are vision/multimodal model issues, but one is a broken vitpose family and the other is Gemma 4 token-type defaults." + "reason": "Different crash sites and causes: KV-cache tensor indexing vs annotations-related AttributeError in auto_docstring." }, { "left": "issue:30333", "right": "issue:34689", "accept": false, - "reason": "Training-job tracking in MLFlow is not the same as model-loading regression for Llama 3.2 Vision." + "reason": "Unrelated training-job reporting bug vs model-loading regression for Llama 3.2 Vision." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "Different models and fixes: vitpose breakage vs Gemma 4 mm_token_type_ids defaults." }, { "left": "issue:42898", "right": "issue:43295", "accept": false, - "reason": "Both involve v5 tokenizer/processor behavior, but one is a cleanup-semantics change and the other is processor.tokenizer/image-passing compatibility." + "reason": "Both are processor/tokenizer regressions, but the concrete breakage is different: cleanup behavior vs processor.tokenizer access/images passing." }, { "left": "issue:42898", "right": "issue:45356", "accept": false, - "reason": "Both are tokenizer regressions, but Kimi codec/regex handling is a different concrete defect from clean-up-space behavior." + "reason": "Both mention tokenizer regressions, but they are model-specific issues with different root causes and symptoms." }, { "left": "issue:44295", "right": "issue:45656", "accept": false, - "reason": "Registering/reading position_ids is unrelated to deepspeed causing optimizer.step to run twice." + "reason": "Unrelated: position_ids buffer handling vs DeepSpeed optimizer stepping twice." }, { "left": "issue:43116", "right": "issue:44190", "accept": false, - "reason": "Both are example-script issues, but one is empty multi-label outputs and the other is local dataset loading." + "reason": "Different example scripts and different failures." }, { "left": "issue:30333", "right": "issue:35141", "accept": false, - "reason": "MLFlow failed-reporting is unrelated to output-embedding reinitialization after resizing token embeddings." + "reason": "Different bugs: MLflow reporting vs embeddings being reinitialized in post_init." }, { "left": "issue:45405", "right": "issue:45712", "accept": false, - "reason": "PEFT version bump and leftover dummy classes are unrelated maintenance issues." + "reason": "Unrelated: dependency version bump vs leftover dummy classes in package exports." }, { "left": "issue:44222", "right": "issue:44297", "accept": false, - "reason": "Both involve saving artifacts, but one is FP8/moe save_pretrained behavior and the other is tokenizer class metadata mismatch." + "reason": "Different save_pretrained problems affecting different object types and outputs." }, { - "left": "issue:44060", - "right": "issue:45125", + "left": "issue:22355", + "right": "issue:28282", "accept": false, - "reason": "Wrong tied-weights warning is not the same bug as missing `_tp_plan` for tensor parallelism." + "reason": "Both are import errors, but one is a missing submodule and the other is a missing PyTorch dependency." }, { - "left": "issue:22355", - "right": "issue:28282", + "left": "issue:44060", + "right": "issue:45125", "accept": false, - "reason": "Both are import/dependency errors, but they stem from different missing modules/libraries." + "reason": "Different Qwen model configuration issues with different consequences." }, { "left": "issue:44075", "right": "issue:46231", "accept": false, - "reason": "Optimizer argument handling is unrelated to a documentation parameter-name mismatch." + "reason": "One is a code bug in SGD argument handling; the other is just a docs parameter mismatch." }, { "left": "issue:44933", "right": "issue:45712", "accept": false, - "reason": "Missing import from `image_utils` is not the same as leftover dummy classes leaking into `dir(transformers)`." + "reason": "Both concern package surface/imports, but they are different namespace problems." }, { "left": "issue:44360", "right": "issue:46762", "accept": false, - "reason": "A missing ReLU in the DSA indexer is a different code-path issue from Minimax attention numerical differences." + "reason": "Different technical areas: DSA indexer implementation vs Minimax attention numerical differences." }, { "left": "issue:42831", "right": "issue:44623", "accept": false, - "reason": "FP8 accuracy concerns and processor save_pretrained missing files are unrelated." + "reason": "Different issues: FP8 accuracy vs processor save_pretrained file emission." }, { - "left": "issue:44297", - "right": "issue:45656", + "left": "issue:42796", + "right": "issue:43723", "accept": false, - "reason": "Tokenizer metadata serialization and deepspeed double optimizer-step are unrelated." + "reason": "Both are tokenizer-related, but one is a BPE marker handling bug and the other is a tokenizer loading issue." }, { - "left": "issue:42796", - "right": "issue:43723", + "left": "issue:44222", + "right": "issue:44623", "accept": false, - "reason": "Tokenizer BPE-marker handling and AutoTokenizer loading failure are different bugs in different paths." + "reason": "Both mention save_pretrained, but they involve different objects and different missing outputs." } ] }, @@ -14725,7 +14774,6 @@ ], "soft_pairs": [ "issue:44448|issue:46132", - "issue:44222|issue:44623", "issue:45245|issue:46132", "issue:43479|issue:43644", "issue:41628|issue:42757", @@ -14742,9 +14790,9 @@ "issue:43232|issue:45834", "issue:44295|issue:44623", "issue:43475|issue:46305", - "issue:46305|issue:46655", - "issue:44263|issue:44360", "issue:46209|issue:46431", + "issue:44263|issue:44360", + "issue:46305|issue:46655", "issue:44263|issue:44373", "issue:30990|issue:38175", "issue:42831|issue:45310", @@ -14756,220 +14804,214 @@ "issue:44261|issue:45092", "issue:44483|issue:45406" ], - "prepared_review_unit_hash": "d6bb3858b604b4d0d40f1a92315e643ebbccf45262c80a2701137cb4c559f112", + "prepared_review_unit_hash": "50d08cc11e0530099f77b06e5367a47946e3c4fece0d481d5db56b2c64adc3b7", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, - "soft_pair_count": 31, - "serialized_chars": 35969, - "estimated_input_tokens": 8993, - "estimated_eval_tokens": 18242 + "soft_pair_count": 30, + "serialized_chars": 35671, + "estimated_input_tokens": 8918, + "estimated_eval_tokens": 18092 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is largely a mixed bag of unrelated Transformers issues, with a loose common theme of v5 regressions around config, loading, saving, and generation. There are no clear duplicate pairs here; most soft links only share a subsystem name, not the same bug.", - "confidence": 0.42, - "canonical_issue_reason": "Issue 45071 is the broadest core-regression report in the set (v5.4.0 type-checking breakage in PretrainedConfig). It is more representative of the cluster than the narrower model-specific save/load or architecture-specific bugs.", + "summary": "The cluster is broadly about Transformers v5 regressions and compatibility issues, but the soft-edge pairs are mostly only superficially similar. I would not merge any of the suggested pairs as duplicates.", + "confidence": 0.66, + "canonical_issue_reason": "issue:45092 is the clearest, most concrete representative of the cluster\u2019s recurring v5 compatibility/loading regressions: it has a specific failure mode, a defined affected class of checkpoints, and a likely actionable fix path.", "canonical_pr_reason": null, - "best_issue_reason": "45071 is the best anchor for triage because it targets a general library-wide regression rather than a single model family, import symbol, or documentation problem.", + "best_issue_reason": "issue:45092 is the strongest overall issue to keep as the representative triage item because it is specific, reproducible, and captures a high-impact regression pattern better than the more vague or narrowly scoped alternatives.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44448", "right": "issue:46132", "accept": false, - "reason": "Different bugs: Pegasus output mismatch vs AttentionInterface.register changing callback behavior." - }, - { - "left": "issue:44222", - "right": "issue:44623", - "accept": false, - "reason": "Both mention save_pretrained, but one is FP8 model saving and the other is processor file persistence." + "reason": "Different problems: Pegasus output regression vs AttentionInterface.register changing function behavior. Same broad version/regression theme, but not the same bug." }, { "left": "issue:45245", "right": "issue:46132", "accept": false, - "reason": "Unrelated: category-count runtime limit vs attention registration semantics." + "reason": "Unrelated failures: category-cardinality RuntimeError vs attention registration behavior. No shared code-path evidence." }, { "left": "issue:43479", "right": "issue:43644", "accept": false, - "reason": "Both involve initialization/configuration, but one is default multimodal config construction and the other is non-persistent buffer initialization corruption." + "reason": "Both involve config/init semantics, but one is default multimodal sub-config initialization and the other is non-persistent buffer serialization junk; too different to treat as duplicates." }, { "left": "issue:41628", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but for different symbols and likely different dependency breakages." + "reason": "Both are import errors, but they concern different symbols and different packages; likely separate dependency/version issues." }, { "left": "issue:44373", "right": "issue:44485", "accept": false, - "reason": "Docstring typo/question vs a real RoPE implementation bug; not the same change." + "reason": "A docstring issue is not the same as a RoPE implementation discussion; the titles suggest different scope and impact." }, { "left": "issue:44297", "right": "issue:45310", "accept": false, - "reason": "Tokenizer save metadata mismatch vs Qwen3.5 MoE from_pretrained failure; different code paths." + "reason": "Tokenizer save metadata mismatch vs Qwen3.5 MOE from_pretrained error. Overlap is only around model I/O, not the same concrete failure." }, { "left": "issue:46132", "right": "issue:46710", "accept": false, - "reason": "Registered-function behavior change is unrelated to the DeepSeek accuracy regression." + "reason": "Attention registration behavior change vs accuracy regression after upgrading Transformers. Different symptoms and likely different code paths." }, { "left": "issue:45071", "right": "issue:46082", "accept": false, - "reason": "Both are config-related, but one is a broad type-checking regression and the other is a specific LlamaConfig validation rule." + "reason": "Type-checking breakage in PretrainedConfig is not the same as LlamaConfig rejecting explicit head_dim. Related config area, but distinct bugs." }, { "left": "issue:44222", "right": "issue:44484", "accept": false, - "reason": "Shared save_pretrained wording only; one is quantized model serialization, the other is shard-size policy." + "reason": "FP8 save_pretrained bug vs max_shard_size question. One is a serialization bug, the other is a size-limit behavior question." }, { "left": "issue:43723", "right": "issue:44977", "accept": false, - "reason": "Tokenizer loading failure vs generation failure with flash-attention; different layers." + "reason": "Tokenizer loading issue and Qwen3.5 flash-attention generation issue are unrelated." }, { "left": "issue:43122", "right": "issue:46209", "accept": false, - "reason": "Tokenization regression vs FP8 inference failure on B200 are unrelated." + "reason": "Tokenizer output differences across versions vs FP8 inference failure on B200. Different subsystems and failure modes." }, { "left": "issue:44263", "right": "issue:45092", "accept": false, - "reason": "Same general model family area, but different bugs: split handling in a DSA indexer vs meta-init incompatibility." + "reason": "DSA indexer split semantics vs remote-code checkpoint meta-initialization incompatibility. No concrete shared bug." }, { "left": "issue:44190", "right": "issue:46082", "accept": false, - "reason": "Local dataset loading issue in an example script vs LlamaConfig head_dim validation." + "reason": "Local dataset loading in an example script is unrelated to LlamaConfig head_dim validation." }, { "left": "issue:43232", "right": "issue:46424", "accept": false, - "reason": "Both touch generation/cache code, but one is sync_gpus kwargs update and the other is cache length honoring." + "reason": "Generation state update after sync_gpus vs cache length handling for quantized cache. Similar cache/generation area, but different behavior." }, { "left": "issue:43232", "right": "issue:45834", "accept": false, - "reason": "Generation helper bug vs Kosmos2.5 OCR long-input index error." + "reason": "Generation kwargs bug and Kosmos2.5 OCR long-input index error are different code paths." }, { "left": "issue:44295", "right": "issue:44623", "accept": false, - "reason": "Position_ids buffer access bug vs processor.save_pretrained missing files." + "reason": "Reading a registered buffer vs processor.save_pretrained missing files. Both serialization-adjacent, but not the same bug." }, { "left": "issue:43475", "right": "issue:46305", "accept": false, - "reason": "SAM 3 video encoder attribute error vs MoonshineStreaming mask off-by-one." + "reason": "SAM 3 Vision encoder attribute error and MoonshineStreaming window-mask off-by-one are unrelated model-specific issues." }, { - "left": "issue:46305", - "right": "issue:46655", + "left": "issue:46209", + "right": "issue:46431", "accept": false, - "reason": "Both are masking/padding issues, but in different models and different failure modes." + "reason": "FP8 inference failure and missing chat template error are entirely different problems." }, { "left": "issue:44263", "right": "issue:44360", "accept": false, - "reason": "Same DSA indexer area, but split return handling and missing ReLU are separate defects." + "reason": "Both mention the DSA indexer, but one is about torch.split outputs and the other about a missing ReLU; not enough to merge." }, { - "left": "issue:46209", - "right": "issue:46431", + "left": "issue:46305", + "right": "issue:46655", "accept": false, - "reason": "FP8 inference on B200 vs missing chat template in a processor." + "reason": "Different media processors and different off-by-one/padding bugs; only superficial similarity." }, { "left": "issue:44263", "right": "issue:44373", "accept": false, - "reason": "Implementation bug vs documentation bug; not duplicates." + "reason": "Indexer implementation bug vs docstring issue. No duplicate relationship." }, { "left": "issue:30990", "right": "issue:38175", "accept": false, - "reason": "Sentence Transformers loading hang vs zero probabilities for SigLIP2 are different model issues." + "reason": "Sentence Transformers loading hang and zero probabilities in SigLIP2 are unrelated." }, { "left": "issue:42831", "right": "issue:45310", "accept": false, - "reason": "FP8 accuracy regression and Qwen3.5 from_pretrained error are not the same bug." + "reason": "FP8 accuracy issue and Qwen3.5 MOE from_pretrained error are separate model/runtime bugs." }, { "left": "issue:45092", "right": "issue:45357", "accept": false, - "reason": "Both are multimodal/v5 issues, but one is meta-init compatibility and the other is save_pretrained key naming." + "reason": "Different checkpoints and different failure modes: meta-init incompatibility vs incorrect visual encoder keys on save_pretrained." }, { "left": "issue:46369", "right": "issue:46909", "accept": false, - "reason": "Model memory overflow vs a documentation example problem." + "reason": "Model memory overflow and broken documentation example are unrelated." }, { "left": "issue:42907", "right": "issue:46050", "accept": false, - "reason": "Both concern quantized saving/loading, but different quantization schemes and different failure points." + "reason": "Both concern save/load of quantized models, but one is dequantized Ministrals/Devstrals serialization and the other is sinq model round-tripping; too different for duplicate triage." }, { "left": "issue:44279", "right": "issue:46328", "accept": false, - "reason": "General dependency complaint vs a performance bug in SequenceFeatureExtractor.pad." + "reason": "Generic dependency complaint vs a performance problem in SequenceFeatureExtractor.pad; no shared underlying bug." }, { "left": "issue:45381", "right": "issue:46305", "accept": false, - "reason": "Different sequence/video bugs in different models; no shared concrete code path." + "reason": "Qwen2.5-VL vision_position_ids bug and MoonshineStreaming mask off-by-one are unrelated sequence-positioning problems." }, { "left": "issue:44261", "right": "issue:45092", "accept": false, - "reason": "Missing rms_norm_eps precision issue vs meta-initialization incompatibility are unrelated." + "reason": "MLA q_a_layernorm epsilon precision issue is different from remote-code checkpoint meta initialization incompatibility." }, { "left": "issue:44483", "right": "issue:45406", "accept": false, - "reason": "Serving request handling failure vs Gemma4Processor missing _tokenizer; different code paths and fixes." + "reason": "Chat-completions request rejection in serve and Gemma4Processor missing _tokenizer are distinct serving/processor bugs." } ] }, @@ -15043,8 +15085,8 @@ "issue:46909" ], "soft_pairs": [ - "issue:44829|issue:46002", "issue:22355|issue:27957", + "issue:44829|issue:46002", "issue:43644|issue:46305", "issue:42831|issue:44038", "issue:41628|issue:43824", @@ -15070,243 +15112,225 @@ "issue:42371|issue:44263", "issue:39401|issue:44488", "issue:42907|issue:44964", - "issue:46172|issue:46762", "issue:43976|issue:44945", + "issue:46172|issue:46762", "issue:42907|issue:43937", "issue:41628|issue:45278" ], - "prepared_review_unit_hash": "db22f66c0df78f11fb0f6c704c88c0470c6868b0dd0cddad4da8742bd7ee462a", + "prepared_review_unit_hash": "c85f9d90a89df609d832dd25afc406684a2c0f42791cadee3596d9b5407a8c50", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36346, - "estimated_input_tokens": 9087, - "estimated_eval_tokens": 18430 + "serialized_chars": 36352, + "estimated_input_tokens": 9088, + "estimated_eval_tokens": 18432 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is highly heterogeneous; the soft-similarity links are mostly superficial and do not describe the same underlying bug or change. No PRs are present. I would not merge any of the proposed soft pairs.", - "confidence": 0.97, - "canonical_issue_reason": "issue:41628 is the broadest representative of the cluster\u2019s recurring import-regression theme, but it is only a loose thematic anchor rather than a true duplicate target.", + "summary": "This set is not a true duplicate cluster: it mixes unrelated tokenizer, import, model-loading, training, generation, and numerical-regression issues. None of the soft pairs looks like the same underlying bug/change, so all should be rejected.", + "confidence": 0.95, + "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report in the set ('many import errors after update from 4.57.0 to 5.5.0'), so it is the least-specific anchor if one must pick a canonical issue.", "canonical_pr_reason": null, - "best_issue_reason": "No issue here is a strong global duplicate target; if one must be chosen as a representative, issue:41628 is the least specific and most broadly descriptive.", + "best_issue_reason": "issue:45278 is the best triage anchor among these because it is the most general regression report; the rest are narrower and largely unrelated to one another.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:44829", - "right": "issue:46002", + "left": "issue:22355", + "right": "issue:27957", "accept": false, - "reason": "Both mention training/classification-head problems, but one is an attention-kernel training degradation and the other is NaNs after init/checkpointing; different failure modes." + "reason": "Import error for transformers.onnx vs XLMRoberta Flash Attention 2 behavior; different subsystem and symptom." }, { - "left": "issue:22355", - "right": "issue:27957", + "left": "issue:44829", + "right": "issue:46002", "accept": false, - "reason": "A missing transformers.onnx module and an XLM-RoBERTa Flash Attention 2 issue are unrelated code paths." + "reason": "Both involve training instability, but one is flash-attention-3 degeneracy and the other is NaNs after checkout/init; not the same bug path." }, { "left": "issue:43644", "right": "issue:46305", "accept": false, - "reason": "Non-persistent buffer junk and MoonshineStreaming window off-by-one are different bugs in different subsystems." + "reason": "Non-persistent buffer junk on init is unrelated to MoonshineStreaming sliding-window off-by-one masking." }, { "left": "issue:42831", "right": "issue:44038", "accept": false, - "reason": "FineGrainedFP8 accuracy loss and Qwen3-VL-Moe breakage are not the same concrete problem." + "reason": "FineGrainedFP8 accuracy regression and Qwen3-VL-Moe model bug are different problems, not one underlying issue." }, { "left": "issue:41628", "right": "issue:43824", "accept": false, - "reason": "Both are import errors, but they concern different missing symbols and likely different export paths." + "reason": "Both are import errors, but they miss different symbols from different model families; no shared root cause is evident." }, { "left": "issue:43408", "right": "issue:46402", "accept": false, - "reason": "A model-type warning and missing text-encoder weight loading are different SAM3 issues." + "reason": "SAM3 type warning during instantiation vs SAM3 text-encoder weights not loading; different failure modes." }, { "left": "issue:42831", "right": "issue:45753", "accept": false, - "reason": "FP8 accuracy drift and Qwen3_5 infinite-loop behavior are unrelated." + "reason": "FP8 accuracy issues are unrelated to Qwen3_5 infinite-loop behavior on a specific image." }, { "left": "issue:22355", "right": "issue:28218", "accept": false, - "reason": "ONNX import failure and tokenizer spacing after added tokens are not the same bug." + "reason": "Module import failure vs tokenizer adding an extra space after an added token; different code paths." }, { "left": "issue:39401", "right": "issue:44568", "accept": false, - "reason": "Wrong offset_mapping and missing BOS/EOS insertion are distinct tokenizer behaviors." + "reason": "Wrong offset_mapping and missing BOS/EOS on add_special_tokens are both tokenizer bugs but not the same concrete behavior." }, { "left": "issue:42886", "right": "issue:43066", "accept": false, - "reason": "Offline cache loading and wrong decoder type in v5 are different tokenizer issues." + "reason": "Offline cache load failure is unrelated to wrong tokenizer decoder type serialization in v5." }, { "left": "issue:42886", "right": "issue:42994", "accept": false, - "reason": "Tokenizer offline cache loading and quantized model saving are unrelated." + "reason": "Tokenizer cache-loading offline and quantized model saving are separate persistence issues." }, { "left": "issue:42757", "right": "issue:45278", "accept": false, - "reason": "A specific missing hub import is not the same as a broad set of import errors after upgrade." + "reason": "Missing is_offline_mode from huggingface_hub is a dependency/API mismatch; 45278 is a broader transformers import-regression report." }, { "left": "issue:27957", "right": "issue:30990", "accept": false, - "reason": "Flash Attention 2 on XLM-RoBERTa and Sentence Transformers loading hangs do not share the same root cause." + "reason": "Flash Attention 2 model-specific bug vs Sentence-Transformers loading hang; different symptoms and paths." }, { "left": "issue:44188", "right": "issue:44373", "accept": false, - "reason": "Torch.compile attention-kernel divergence and a docstring typo for position_ids are unrelated." + "reason": "torch.compile attention-kernel divergence is unrelated to a docstring issue for position_ids." }, { "left": "issue:43408", "right": "issue:46369", "accept": false, - "reason": "SAM3 model-type mismatch and memory overflow in sam3-litetext are different bugs." + "reason": "SAM3 model-type warning vs memory overflow in sam3-litetext; different failures." }, { "left": "issue:43232", "right": "issue:45593", "accept": false, - "reason": "Generation kwargs after sync_gpus and D-FINE auxiliary-loss handling are unrelated." + "reason": "Generation helper after sync_gpus and D-FINE auxiliary-loss behavior are unrelated subsystems." }, { "left": "issue:42796", "right": "issue:43122", "accept": false, - "reason": "Both are tokenizer regressions, but one is a specific BPE-marker handling bug and the other is a broader version-to-version tokenization change." + "reason": "Ministral tokenizer BPE-marker handling vs a version-wide tokenization change; similar area but not the same underlying bug." }, { "left": "issue:43646", "right": "issue:46002", "accept": false, - "reason": "Broken custom model initialization and NaNs in classification heads are different initialization-time symptoms." + "reason": "Custom model initialization breakage vs NaNs in classification heads after checkout/init; not the same code path." }, { "left": "issue:45381", "right": "issue:46909", "accept": false, - "reason": "Qwen2.5-VL video position_ids and a broken DIA docs example are not the same issue." + "reason": "Qwen2.5-VL video position_ids bug vs a broken documentation example for DIA; different artifact types." }, { "left": "issue:43540", "right": "issue:46909", "accept": false, - "reason": "Qwen3OmniMoe video ValueError and a documentation example bug are unrelated." + "reason": "Qwen3OmniMoe video-input ValueError is unrelated to a documentation example bug." }, { "left": "issue:30990", "right": "issue:33357", "accept": false, - "reason": "A loading hang in Sentence Transformers and a MacOS bus error are different failure modes." + "reason": "Loading hang for Sentence Transformers vs MacOS bus error on a CLIP model; different runtime failures." }, { "left": "issue:42907", "right": "issue:44945", "accept": false, - "reason": "Quantized-model save failures and pipeline-parallel output errors are unrelated." + "reason": "Saving dequantized Ministral/Devstral models is unrelated to incorrect outputs under pipeline parallelism." }, { "left": "issue:43576", "right": "issue:43784", "accept": false, - "reason": "A broken env CLI command and a sentence-transformers import NameError are different code paths." + "reason": "Broken transformers env CLI and sentence-transformers import NameError are different issues." }, { "left": "issue:44295", "right": "issue:44297", "accept": false, - "reason": "Position_ids buffer reading and tokenizer_class metadata mismatch are unrelated." + "reason": "Reading a registered buffer position_ids and tokenizer_class metadata mismatch are unrelated serialization bugs." }, { "left": "issue:42371", "right": "issue:44263", "accept": false, - "reason": "TF32 configuration guidance and torch.split return-value behavior are not the same bug." + "reason": "TF32 configuration API warning is unrelated to torch.split return-value handling in GlmMoeDsaIndexer." }, { "left": "issue:39401", "right": "issue:44488", "accept": false, - "reason": "Two tokenizer-loading/tokenization complaints, but with different models and symptoms; no concrete shared defect is evident." + "reason": "Qwen3 tokenizer offset mapping bug vs failure to load cjvt/sleng-bert; different tokenizer/model-loading problems." }, { "left": "issue:42907", "right": "issue:44964", "accept": false, - "reason": "Dequantized Ministral save failure and Phi-4 multimodal loading failure are different issues." - }, - { - "left": "issue:46172", - "right": "issue:46762", - "accept": false, - "reason": "Both involve attention math, but one is a specific code expression bug and the other is a numerical-difference comparison against reference." + "reason": "Dequantized saving failure and loading Phi-4-multimodal-instruct are separate model persistence/load issues." }, { "left": "issue:43976", "right": "issue:44945", "accept": false, - "reason": "Python-version incompatibility and pipeline-parallel output corruption are unrelated." - }, - { - "left": "issue:43576", - "right": "issue:43784", - "accept": false, - "reason": "CLI environment command breakage and an import NameError in sentence-transformers are different problems." - }, - { - "left": "issue:44295", - "right": "issue:44297", - "accept": false, - "reason": "Buffer access after registration and tokenizer save metadata mismatch do not share a root cause." + "reason": "Python-version compatibility regression and pipeline-parallelism output corruption are unrelated." }, { - "left": "issue:42371", - "right": "issue:44263", + "left": "issue:46172", + "right": "issue:46762", "accept": false, - "reason": "TF32 behavior settings and GlmMoeDsaIndexer torch.split semantics are unrelated." + "reason": "DeepSeek-V4 tensor transpose/dtype bug and Minimax MSA numerical differences are different attention/model-path issues." }, { - "left": "issue:39401", - "right": "issue:44488", + "left": "issue:42907", + "right": "issue:43937", "accept": false, - "reason": "Offset-mapping wrongness and a model-loading failure for cjvt/sleng-bert are distinct tokenizer issues." + "reason": "Saving dequantized models and invalid GenerationConfig are distinct failures." }, { - "left": "issue:42907", - "right": "issue:43937", + "left": "issue:41628", + "right": "issue:45278", "accept": false, - "reason": "Saving dequantized models and GLM-5 GenerationConfig validation are different bugs." + "reason": "Single missing AutoImageProcessor import vs broad import errors after upgrade; not the same concrete bug." } ] }, @@ -15402,215 +15426,215 @@ "issue:42757|issue:44351", "issue:36296|issue:43116", "issue:43278|issue:44945", - "issue:42175|issue:44938", "issue:22355|issue:29127", + "issue:42175|issue:44938", "issue:43493|issue:45753", "issue:46231|issue:46877", "issue:44188|issue:46328", "issue:43937|issue:44485" ], - "prepared_review_unit_hash": "e1b8a2372e123d0846d4d35771724c73ebfd8f6441e65b26ede4434ca9dee0e9", + "prepared_review_unit_hash": "ed3deab9cc0692ac1bcb15c9fe8142bb1823dd4ed537995335f5e4153cd0c274", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 29, - "serialized_chars": 35674, - "estimated_input_tokens": 8919, - "estimated_eval_tokens": 18094 + "serialized_chars": 35672, + "estimated_input_tokens": 8918, + "estimated_eval_tokens": 18092 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The set is mostly a grab-bag of unrelated bugs; none of the soft-similarity pairs are close enough to merge as true duplicates. I chose the broad tokenizer-regression report as the best representative issue, but there is no strong canonical duplicate for the whole cluster.", - "confidence": 0.9, - "canonical_issue_reason": "issue:43122 is the broadest and most representative report here: it describes a tokenizer regression across versions, which is the closest fit for the tokenizer-heavy subset of this cluster.", + "summary": "These artifacts are mostly unrelated, spanning different models, APIs, and failure modes. None of the soft pairs look like the same underlying bug/change, so all soft edges are rejected.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest representative issue here: it is open, actively discussed, and describes a concrete, reproducible API bug in a widely used path (`apply_chat_template`) with multimodal inputs.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43122 has the clearest general regression framing and the widest potential to subsume nearby tokenizer complaints, even though the cluster as a whole is not a true duplicate set.", + "best_issue_reason": "issue:44521 is the best overall issue candidate because it is well-scoped, user-facing, and likely to attract/anchor related reports about the same chat-template behavior.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43756", "right": "issue:44060", "accept": false, - "reason": "Different model families and different failure modes: RoPE-layer omission vs tied-weights warning in Qwen3-Next." + "reason": "Different bugs in different model families: RoPE layer selection vs a tied-weights warning in Qwen3-Next." }, { "left": "issue:43784", "right": "issue:45600", "accept": false, - "reason": "Both are import/config breakages, but they point to different code paths and symbols; not the same underlying bug." + "reason": "Import-time `nn` NameError in sentence-transformers is unrelated to stale `auto_mappings.py` config references and CI breakage." }, { "left": "issue:44117", "right": "issue:44704", "accept": false, - "reason": "One is a tokenizer mapping lookup issue, the other is kwargs not forwarded in AutoProcessor.from_pretrained." + "reason": "Tokenizer mapping `None` handling is a different code path from `AutoProcessor.from_pretrained` kwarg forwarding." }, { "left": "issue:39401", "right": "issue:44521", "accept": false, - "reason": "Both involve tokenization, but one is wrong offset mappings and the other is assistant masks for multimodal chat templates." + "reason": "Wrong `offset_mapping` in Qwen3 tokenizer is not the same as zeroed assistant masks in multimodal chat templating." }, { "left": "issue:44704", "right": "issue:45290", "accept": false, - "reason": "Unrelated code paths: cached_file argument forwarding vs chat-template crashes on tool-call messages." + "reason": "`cached_file` kwargs forwarding and chat-template crashes with tool-call messages are separate failures." }, { "left": "issue:44625", "right": "issue:44829", "accept": false, - "reason": "Different problems: config propagation for num_labels vs degenerate training under flash_attention_3." + "reason": "`num_labels` propagation and flash-attention-3 training degeneration are unrelated regression types." }, { "left": "issue:44360", "right": "issue:44485", "accept": false, - "reason": "DSA indexer activation bug and GLM-5 RoPE implementation issue are unrelated." + "reason": "A discussion about DSA indexer activation is unrelated to GLM-5 RoPE implementation details." }, { "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 output attribute." + "reason": "Wrong SAM3 model-type warning and missing `fpn_position_embeddings` attribute are distinct SAM3-video issues." }, { "left": "issue:43122", "right": "issue:44521", "accept": false, - "reason": "General tokenization regression vs multimodal assistant-mask bug; not the same defect." + "reason": "Tokenizer version drift is a different problem from multimodal assistant-mask generation in chat templates." }, { "left": "issue:44038", "right": "issue:44297", "accept": false, - "reason": "Qwen3-VL-Moe runtime bug and tokenizer save_pretrained metadata mismatch are different issues." + "reason": "Qwen3-VL-Moe/transformers 5.0 bug and tokenizer save metadata mismatch are separate paths." }, { "left": "issue:43122", "right": "issue:46326", "accept": false, - "reason": "Both touch tokenizers, but one is a version-regression report and the other is a specific continue_final_message bug." + "reason": "General tokenization differences across versions are not the same as `continue_final_message` handling for mistral_common tokenizers." }, { "left": "issue:42907", "right": "issue:43452", "accept": false, - "reason": "Saving dequantized models and gguf_file loading failures are distinct workflows and failure modes." + "reason": "Dequantized save failures and GGUF/from_pretrained loading breakage are different serialization/loading bugs." }, { "left": "issue:44038", "right": "issue:44295", "accept": false, - "reason": "Different bugs: Qwen3-VL-Moe regression vs position_ids buffer access error." + "reason": "Qwen3-VL-Moe model behavior and `position_ids` buffer access are unrelated." }, { "left": "issue:45216", "right": "issue:45478", "accept": false, - "reason": "Same model family, but save_pretrained corruption and from_pretrained import error are separate code paths." + "reason": "Both involve Qwen3.5, but one is a save-pretrained regression and the other is a from-pretrained error; not the same concrete bug." }, { "left": "issue:39401", "right": "issue:46614", "accept": false, - "reason": "Different tokenizer bugs: Qwen3 offset_mapping vs stale phonemizer language state." + "reason": "Tokenizer offset mapping for Qwen3 is unrelated to stale phonemizer backend state in Wav2Vec2 phoneme tokenization." }, { "left": "issue:22355", "right": "issue:29159", "accept": false, - "reason": "Importing transformers.onnx and slow/fast tokenizer inconsistency are unrelated." + "reason": "Missing `transformers.onnx` import and slow/fast tokenizer inconsistency are different issues." }, { "left": "issue:44038", "right": "issue:44222", "accept": false, - "reason": "Both mention MoE/FP8 contexts, but one is a model import/runtime bug and the other is save_pretrained behavior." + "reason": "A Qwen3-VL-Moe issue does not match FP8 `save_pretrained` behavior for MoE models." }, { "left": "issue:43824", "right": "issue:44351", "accept": false, - "reason": "Separate missing exports: Qwen2_5_VLForConditionalGeneration vs HybridCache." + "reason": "Two separate missing-symbol import errors (`Qwen2_5_VLForConditionalGeneration` vs `HybridCache`)." }, { "left": "issue:43116", "right": "issue:45200", "accept": false, - "reason": "Example script returns empty results and mm_token_type_ids defaults are different bugs." + "reason": "Empty results in the example classification script and Gemma 4 `mm_token_type_ids` defaults are different bugs." }, { "left": "issue:43122", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but there is no evidence they share the same tokenizer implementation or code path." + "reason": "Tokenizer behavior drift across versions is unrelated to Deepseek v5 incorrect tokenization." }, { "left": "issue:42757", "right": "issue:44351", "accept": false, - "reason": "Two unrelated import errors for different missing names from transformers/huggingface_hub." + "reason": "Both are import errors, but for different symbols from different packages and code paths." }, { "left": "issue:36296", "right": "issue:43116", "accept": false, - "reason": "Tensor-parallel training and multi-label classification script output are unrelated." + "reason": "Tensor-parallel training failure is not the same as multi-label example script returning empty outputs." }, { "left": "issue:43278", "right": "issue:44945", "accept": false, - "reason": "Dtype drift during evaluate and incorrect pipeline-parallel output are different training/runtime bugs." + "reason": "Training/eval dtype mismatch is unrelated to incorrect outputs under pipeline parallelism." }, { - "left": "issue:42175", - "right": "issue:44938", + "left": "issue:22355", + "right": "issue:29127", "accept": false, - "reason": "Backend extra-install packaging and Python 3.14 import failure are not the same issue." + "reason": "Missing module import and LayoutLMv3 error-message clarity are not the same bug." }, { - "left": "issue:22355", - "right": "issue:29127", + "left": "issue:42175", + "right": "issue:44938", "accept": false, - "reason": "Missing module import and LayoutLMv3 error-message clarity are entirely different." + "reason": "Missing TensorFlow in a torch extra install is unrelated to Python 3.14 load failures." }, { "left": "issue:43493", "right": "issue:45753", "accept": false, - "reason": "SigLIP2 implementation discrepancy and Qwen3_5 infinite-loop on an image are unrelated model bugs." + "reason": "SigLIP2 implementation discrepancy and Qwen3_5 image infinite loop are different model-specific failures." }, { "left": "issue:46231", "right": "issue:46877", "accept": false, - "reason": "Documentation parameter mismatch vs import-time doc-lint logging are separate documentation/tooling issues." + "reason": "Documentation parameter mismatch and noisy import-time doc-lint output are not the same underlying issue." }, { "left": "issue:44188", "right": "issue:46328", "accept": false, - "reason": "torch.compile attention-kernel divergence and SequenceFeatureExtractor.pad inefficiency are unrelated." + "reason": "`torch.compile`-dependent attention divergence and `SequenceFeatureExtractor.pad` performance are unrelated." }, { "left": "issue:43937", "right": "issue:44485", "accept": false, - "reason": "Both are GLM-5 related, but GenerationConfig validation and RoPE implementation are different defects." + "reason": "GLM-5 generation-config validation and GLM-5 RoPE implementation are separate problems, not one duplicate bug." } ] }, @@ -15688,10 +15712,10 @@ "issue:42898|issue:46614", "issue:43066|issue:44779", "issue:43611|issue:46002", - "issue:44038|issue:45478", "issue:43976|issue:45446", - "issue:43116|issue:43493", + "issue:44038|issue:45478", "issue:44355|issue:46231", + "issue:43116|issue:43493", "issue:42673|issue:45072", "issue:45357|issue:45478", "issue:44625|issue:46002", @@ -15700,8 +15724,8 @@ "issue:44610|issue:46909", "issue:43618|issue:46829", "issue:43992|issue:44704", - "issue:43116|issue:43825", "issue:44075|issue:44936", + "issue:43116|issue:43825", "issue:43066|issue:46710", "issue:28348|issue:46640", "issue:44263|issue:46762", @@ -15713,20 +15737,20 @@ "issue:33453|issue:39401", "issue:43064|issue:43673" ], - "prepared_review_unit_hash": "8567b16524e730fe52e729b0282cbae7b8b3483a1b2c4358f3ae758de1608846", + "prepared_review_unit_hash": "39fa5fc3b249e95e1787a212d04ca47483c3121350098f6819bac0fbb32c1be5", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 28, - "serialized_chars": 35742, + "serialized_chars": 35744, "estimated_input_tokens": 8936, "estimated_eval_tokens": 18128 }, @@ -15735,180 +15759,180 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These items are mostly unrelated Transformers regressions across different subsystems (tokenizers, loading, multimodal, training, docs). I found no soft-edge pair that clearly shares the same underlying bug/change, so all candidate duplicate links are rejected.", - "confidence": 0.81, - "canonical_issue_reason": "issue:43611 is the strongest representative issue: it describes a concrete, reproducible v5 loading regression with clear user impact and a specific code path.", + "summary": "This cluster is highly heterogeneous: most items are separate Transformers v5 regressions or model-specific bugs, not duplicates. No soft pair looks like the same concrete underlying issue/change, so all soft edges should be rejected.", + "confidence": 0.34, + "canonical_issue_reason": "If a representative issue is needed, issue 45278 is the broadest umbrella-style report in the set (\u201cmany import errors after update\u201d), but the cluster overall does not contain a true duplicate group.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43611 is the best issue to anchor the cluster because it is specific, actionable, and tied to a clear runtime failure rather than a vague compatibility complaint.", + "best_issue_reason": "Issue 45278 is the closest thing to a global representative because it is the most general regression report; however, it still only covers import errors and does not unify the rest of the cluster.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38617", "right": "issue:45278", "accept": false, - "reason": "Both mention import breakage, but one is a missing symbol and the other is a broad upgrade-wide import regression; not the same bug." + "reason": "One is a specific missing-symbol import error; the other is a broad report of many import errors after upgrade. Related theme, but not the same bug." }, { "left": "issue:42898", "right": "issue:46614", "accept": false, - "reason": "Tokenizer cleanup behavior changes and stale phonemizer backend state are different failure modes in different code paths." + "reason": "Tokenizer cleanup behavior change vs stale phonemizer backend state. Different code paths and symptoms." }, { "left": "issue:43066", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer-related, but one is a decoder-type regression and the other is incorrect DeepSeek tokenization results; no concrete shared fix." + "reason": "Both involve tokenizers, but one is about decoder type in v5 and the other is DeepSeek incorrect results. Too different to be duplicates." }, { "left": "issue:43611", "right": "issue:46002", "accept": false, - "reason": "Model loading via base_model_prefix and NaNs in classification heads are unrelated code-path problems." - }, - { - "left": "issue:44038", - "right": "issue:45478", - "accept": false, - "reason": "Both concern Qwen3.5/Qwen3-VL-Moe loading, but the reported failures are not the same concrete from_pretrained bug." + "reason": "Model loading via base_model_prefix versus NaNs in classification heads. Not the same underlying problem." }, { "left": "issue:43976", "right": "issue:45446", "accept": false, - "reason": "A Python-version compatibility report is not the same as an incorrect PyTorch version gate in flex_attention." + "reason": "Python-version compatibility issue vs a PyTorch import/version check in flex_attention. Different failure modes." }, { - "left": "issue:43116", - "right": "issue:43493", + "left": "issue:44038", + "right": "issue:45478", "accept": false, - "reason": "Example-script multi-label output failure and SigLIP2 implementation discrepancy are different bugs." + "reason": "Both are Qwen3/Moe-related, but one is Qwen3-VL-Moe and the other is Qwen3.5 Moe from_pretrained. Not the same concrete bug." }, { "left": "issue:44355", "right": "issue:46231", "accept": false, - "reason": "Compiled-file execution errors and a documentation parameter-name mismatch are not the same issue." + "reason": "Compiled-Python import/runtime errors versus a documentation parameter-name mismatch. Not duplicates." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script multi-label classification failure versus a SigLIP2 implementation discrepancy. Different components and symptoms." }, { "left": "issue:42673", "right": "issue:45072", "accept": false, - "reason": "VRAM leakage in multithreaded inference and bf16 dtype mismatches are separate runtime bugs." + "reason": "VRAM leak in Qwen3ForCausalLM dataloader threads versus dtype mismatch in inference. Different issues." }, { "left": "issue:45357", "right": "issue:45478", "accept": false, - "reason": "Save-pretrained key mapping and from-pretrained loading failures are related only loosely; they are distinct bugs." + "reason": "Both are Qwen3.5-related, but one is save_pretrained visual-key regression and the other is from_pretrained load failure. Not one bug." }, { "left": "issue:44625", "right": "issue:46002", "accept": false, - "reason": "Num-label propagation and NaNs in classification heads do not share the same underlying code path." + "reason": "num_labels propagation bug versus NaNs in classification heads. Different root causes." }, { "left": "issue:42371", "right": "issue:43874", "accept": false, - "reason": "TF32 API guidance and a missing image-patch helper method are unrelated." + "reason": "TF32 API behavior guidance versus a missing image-processor method. No overlap in bug." }, { "left": "issue:45083", "right": "issue:45446", "accept": false, - "reason": "A helper length-calculation regression is not the same as a PyTorch version-check bug." + "reason": "Qwen3_omni_moe feature-length helper bug versus flex_attention version check. Unrelated." }, { "left": "issue:44610", "right": "issue:46909", "accept": false, - "reason": "An image-size mismatch in OmDet-Turbo and a broken DIA documentation example are unrelated." + "reason": "OmDet-Turbo processor size mismatch versus broken DIA documentation example. Different problems." }, { "left": "issue:43618", "right": "issue:46829", "accept": false, - "reason": "Both are output-dataclass issues, but they affect different models and fields with different fixes." + "reason": "Both are output-field assignment bugs, but they affect different model outputs and code paths; too broad to merge." }, { "left": "issue:43992", "right": "issue:44704", "accept": false, - "reason": "Missing encoder weights and missing kwargs passthrough to cached_file are separate from_pretrained/processor issues." + "reason": "Missing weight in UMT5Encoder loading versus AutoProcessor kwargs not forwarded. Different APIs." }, { - "left": "issue:43116", - "right": "issue:43825", + "left": "issue:44075", + "right": "issue:44936", "accept": false, - "reason": "Empty classification results and a misleading pipeline error message are different example/pipeline problems." + "reason": "SGD optimizer args ignored versus trainer.evaluate failing after train. Not the same bug." }, { - "left": "issue:44075", - "right": "issue:44936", + "left": "issue:43116", + "right": "issue:43825", "accept": false, - "reason": "Optimizer-argument handling and trainer.evaluate() failing after train are not the same bug." + "reason": "Example script behavior versus a pipeline() error message/doc mismatch. Different symptoms and fixes." }, { "left": "issue:43066", "right": "issue:46710", "accept": false, - "reason": "Tokenizer decoder regression and DeepSeek accuracy drift are not the same underlying failure." + "reason": "Tokenizer decoder-type regression versus an accuracy regression in a DeepSeek model. Too different to be duplicates." }, { "left": "issue:28348", "right": "issue:46640", "accept": false, - "reason": "Both reference FlashAttention 2, but one is a feature request for GPT2 and the other is an unspecified issue; not the same concrete change." + "reason": "FlashAttention 2 feature support request for GPT2LMHeadModel versus a separate Flash Attention 2 issue. Not the same concrete code-path problem." }, { "left": "issue:44263", "right": "issue:46762", "accept": false, - "reason": "A torch.split return-value issue in one model and numerical attention differences in another are distinct." + "reason": "torch.split indexing bug versus numerical differences in Minimax attention. Different issues." }, { "left": "issue:46614", "right": "issue:46752", "accept": false, - "reason": "Stale phonemizer language state and empty-list chat template crashes are unrelated." + "reason": "Stale phonemizer language state versus apply_chat_template crash on empty input. No shared bug." }, { "left": "issue:44995", "right": "issue:46697", "accept": false, - "reason": "A stale GLM indexer cache and Swin2SR over-padding are different model-specific bugs." + "reason": "GlmMoeDsa stale cache crash versus Swin2SR over-padding. Unrelated." }, { "left": "issue:45092", "right": "issue:46762", "accept": false, - "reason": "Meta-initialization incompatibility and attention numerical drift are unrelated regressions." + "reason": "Meta-init incompatibility for InternVL2 checkpoints versus Minimax MSA numerical mismatch. Different model paths." }, { "left": "issue:42222", "right": "issue:43441", "accept": false, - "reason": "Vitpose import breakage and Ministral FlashAttention failure are different subsystems and fixes." + "reason": "VitPose import/breakage issue versus Ministral-3 FlashAttention failure. Different subsystems." }, { "left": "issue:43827", "right": "issue:45446", "accept": false, - "reason": "Documentation still referencing pipeline() and a flex_attention version-check bug are unrelated." + "reason": "Documentation still references removed pipeline() versus a PyTorch version check bug. Not duplicates." }, { "left": "issue:33453", "right": "issue:39401", "accept": false, - "reason": "Generic tokenizer-loading regression and wrong offset_mapping for Qwen3 are different tokenizer bugs." + "reason": "Generic tokenizer loading regression versus a specific offset_mapping bug in Qwen3 tokenizer. Related area only." }, { "left": "issue:43064", "right": "issue:43673", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption and missing generation cache during chunked_prefill are separate training vs generation issues." + "reason": "Trainer/FSDP2/PEFT optimizer-state corruption versus missing GenerationMixin cache during chunked prefill. Different code paths." } ] }, @@ -15982,17 +16006,17 @@ "issue:46752" ], "soft_pairs": [ - "issue:43937|issue:45563", "issue:42673|issue:42796", + "issue:43937|issue:45563", "issue:44380|issue:46077", - "issue:43673|issue:46328", "issue:45356|issue:46614", + "issue:43673|issue:46328", "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:44466|issue:46002", "issue:44410|issue:45310", "issue:39401|issue:44779", "issue:43937|issue:44977", @@ -16012,20 +16036,20 @@ "issue:44877|issue:46231", "issue:42371|issue:44261" ], - "prepared_review_unit_hash": "95e2681dc25b32d7941f2d48b8d97d45df9e95ac344d22fd50fe6cf92103f299", + "prepared_review_unit_hash": "12cc3301578552ca3613cf0268b3dcf450a672d977b77904b025790bf4f638e8", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 29, - "serialized_chars": 36390, + "serialized_chars": 36391, "estimated_input_tokens": 9098, "estimated_eval_tokens": 18452 }, @@ -16034,186 +16058,192 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No soft edge looks like the same underlying bug/change. The items span distinct regressions across tokenizer behavior, model loading, generation, attention backends, video processing, and test utilities, so all candidate merges are rejected.", - "confidence": 0.79, - "canonical_issue_reason": "issue:43122 is the broadest regression report in the set: it compares behavior across versions, has clear user impact, and is framed as a general tokenizer regression rather than a model-specific edge case.", + "summary": "This is a heterogeneous set of unrelated transformer bugs and regressions. The soft pairs are mostly same-subsystem but different concrete failures, so none are strong duplicate candidates.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 43122 is the broadest regression-style report in the set and the safest representative issue, but it is not a true duplicate hub for the rest.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43122 is the strongest anchor among these issues because it is a versioned, reproducible regression with broad applicability; the others are narrower, model- or subsystem-specific bugs.", + "best_issue_reason": "Issue 43122 has the most generic, version-regression framing and is the closest thing to a cluster representative among otherwise unrelated reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { - "left": "issue:43937", - "right": "issue:45563", + "left": "issue:42673", + "right": "issue:42796", "accept": false, - "reason": "Both involve generation, but one is an invalid GenerationConfig error and the other is a stale warning about num_return_sequences; different bugs and code paths." + "reason": "One is a Qwen3ForCausalLM VRAM leak under multithreaded dataloading; the other is a tokenizer BPE-marker handling bug for Ministral. Different subsystem and failure mode." }, { - "left": "issue:42673", - "right": "issue:42796", + "left": "issue:43937", + "right": "issue:45563", "accept": false, - "reason": "VRAM leakage in dataloader threads is unrelated to a tokenizer BPE-marker parsing bug." + "reason": "Both touch generation behavior, but one is an invalid GenerationConfig error and the other is a stale warning about num_return_sequences in paged generate(). Not the same bug." }, { "left": "issue:44380", "right": "issue:46077", "accept": false, - "reason": "Different attention problems: GPT2 scaling ignored in backend implementations versus Gemma4 building the wrong attention mask." + "reason": "Different model families and different code paths: GPT2 attention scaling under SDPA/FlashAttention versus Gemma4 bidirectional mask construction." }, { - "left": "issue:43673", - "right": "issue:46328", + "left": "issue:45356", + "right": "issue:46614", "accept": false, - "reason": "Generation cache chunked_prefill bug is unrelated to a padding performance issue in SequenceFeatureExtractor.pad." + "reason": "Both involve tokenizer/processor state, but one is Kimi-K2.5 codec/fix_mistral_regex regression and the other is Wav2Vec2 phonemizer language state retention. Distinct bugs." }, { - "left": "issue:45356", - "right": "issue:46614", + "left": "issue:43673", + "right": "issue:46328", "accept": false, - "reason": "Both are tokenizer-related regressions, but they affect different tokenizers and different state-handling failures." + "reason": "Chunked prefill cache handling and SequenceFeatureExtractor.pad list conversion are unrelated; one is a generation correctness issue, the other a performance inefficiency." }, { "left": "issue:42673", "right": "issue:46655", "accept": false, - "reason": "VRAM leakage in Qwen3ForCausalLM threads is not the same as a video processor padding miscalculation." + "reason": "A VRAM leak in Qwen3ForCausalLM is unrelated to SmolVLM video padding calculation." }, { "left": "issue:42222", "right": "issue:46752", "accept": false, - "reason": "VitPose model breakage and an empty-list apply_chat_template crash are unrelated failures." + "reason": "VitPose model breakage and empty-list apply_chat_template IndexError are unrelated failures in different parts of the stack." }, { "left": "issue:44291", "right": "issue:45071", "accept": false, - "reason": "One is an init_empty_weights argument error; the other is PretrainedConfig type-checking regression. Same release window, different concrete bug." + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a model-loading bug; PretrainedConfig type checking in v5.4.0 is a separate validation regression." }, { "left": "issue:45405", "right": "issue:45446", "accept": false, - "reason": "A dependency-version bump issue is unrelated to a PyTorch version check in flex_attention." + "reason": "A dependency version bump to an unreleased PEFT version is unrelated to a PyTorch version check bug in flex_attention." }, { - "left": "issue:44466", - "right": "issue:46002", + "left": "issue:44117", + "right": "issue:45588", "accept": false, - "reason": "Serialization of lm_head.weight is a different issue from NaNs appearing in classification heads." + "reason": "TOKENIZER_MAPPING_NAMES returning None during from_pretrained is a mapping/registry issue; flash_attention.py s_aux=None crash is an attention integration bug." }, { - "left": "issue:44117", - "right": "issue:45588", + "left": "issue:44466", + "right": "issue:46002", "accept": false, - "reason": "TOKENIZER_MAPPING_NAMES None handling and flash_attention s_aux=None crashes are different subsystems and failures." + "reason": "Serialization of tied lm_head weights differs by device; NaNs in classification heads after checkout/init is a separate training/initialization problem." }, { "left": "issue:44410", "right": "issue:45310", "accept": false, - "reason": "Both mention Qwen, but one is missing projections in qwen3next and the other is a from_pretrained error in Qwen3.5 MoE; not the same bug." + "reason": "Both mention Qwen3.5, but one is missing projections in qwen3next layer 0 and the other is a from_pretrained load error. Not the same concrete defect." }, { "left": "issue:39401", "right": "issue:44779", "accept": false, - "reason": "Both are tokenizer regressions, but they concern different models and different symptom sets (offset mapping vs overall incorrect results)." + "reason": "Qwen3 offset_mapping errors and Deepseek tokenizer regressions are both tokenization issues but clearly distinct models and behaviors." }, { "left": "issue:43937", "right": "issue:44977", "accept": false, - "reason": "GenerationConfig validation and flash-attention generation failure are different issues." + "reason": "GenerationConfig validation failure and Qwen3.5 flash-attention generation problems are different code paths and symptoms." }, { "left": "issue:45341", "right": "issue:46328", "accept": false, - "reason": "A testing_utils bug and a feature-extractor padding inefficiency are unrelated." + "reason": "A testing_utils bug and a SequenceFeatureExtractor.pad inefficiency are unrelated; one is correctness, the other performance." }, { "left": "issue:43450", "right": "issue:43986", "accept": false, - "reason": "Video processor output-shape bug and AutoProcessor dependency crash are different code paths." + "reason": "Incorrect batched video processor shapes and AutoProcessor crashing without torchvision are different video-processing issues." }, { "left": "issue:45341", "right": "issue:45561", "accept": false, - "reason": "Testing-utils cleanup bug is unrelated to a pytest-xdist file-race issue, even though both touch test infrastructure." + "reason": "Both involve testing utilities, but one is a generic bug in testing_utils.py and the other is an xdist race on captured_info.txt. Different concrete problems." }, { "left": "issue:44829", "right": "issue:45216", "accept": false, - "reason": "FlashAttention-3 training degeneration and Qwen3.5 checkpoint serialization regression are different problems." + "reason": "FlashAttention-3 training degeneration and Qwen3.5 save_pretrained regression are unrelated model/backends and do not share the same fix." }, { "left": "issue:45071", "right": "issue:46512", "accept": false, - "reason": "Type checking on PretrainedConfig and stale config.dtype after from_config are related only at a high level; not the same underlying defect." + "reason": "PretrainedConfig type checking regression and stale model.config.dtype after from_config are separate configuration issues." }, { "left": "issue:44117", "right": "issue:45081", "accept": false, - "reason": "A mapping-lookup None bug is not the same as the Mistral regex patch crashing on backend_tokenizer absence." + "reason": "Registry None handling during from_pretrained and a Mistral regex patch crash are different failure points in tokenizer loading." }, { "left": "issue:38617", "right": "issue:41628", "accept": false, - "reason": "Both are import errors, but they involve different missing symbols from different modules and do not indicate one root cause." + "reason": "Missing layer_type_validation import and missing AutoImageProcessor import are both import errors but in unrelated modules." }, { "left": "issue:43408", "right": "issue:43742", "accept": false, - "reason": "A model-type warning and a key error while loading a specific model are unrelated." + "reason": "A sam3_video/sam3_tracker type warning is not the same as the facebook/MobileLLM-125M key error when loading." }, { "left": "issue:43122", "right": "issue:46298", "accept": false, - "reason": "A tokenizer regression across versions is not the same as a noisy logging message." + "reason": "Different tokenization regression versus a noisy logging message; no shared bug or code path." }, { "left": "issue:42617", "right": "issue:45341", "accept": false, - "reason": "Running a 3d_parallel script and a small testing_utils bug are unrelated." + "reason": "Failure to run 3d_parallel.py and a testing_utils bug are unrelated." }, { "left": "issue:45362", "right": "issue:46328", "accept": false, - "reason": "Qwen3.5 chat crashes and SequenceFeatureExtractor padding inefficiency are different bugs." + "reason": "Qwen3.5-35B chat crash and SequenceFeatureExtractor.pad inefficiency are unrelated." }, { "left": "issue:42222", "right": "issue:46178", "accept": false, - "reason": "VitPose model breakage and a backend-mapping validation error are unrelated." + "reason": "VitPose model breakage and a backend mapping ValueError for tensorflow_text are different problems." }, { "left": "issue:44484", "right": "issue:45310", "accept": false, - "reason": "A save_pretrained shard-size question is not the same as a Qwen3.5 MoE loading regression." + "reason": "max_shard_size documentation/behavior question is unrelated to the Qwen3.5 Moe from_pretrained error." }, { "left": "issue:44877", "right": "issue:46231", "accept": false, - "reason": "Strict config loading and a documentation parameter-name mismatch are different severities and different fixes." + "reason": "granite_speech strict config loading and a set_training doc parameter-name mismatch are unrelated." }, { "left": "issue:42371", "right": "issue:44261", "accept": false, - "reason": "TF32 API guidance and MLA rms_norm_eps precision mismatch are unrelated." + "reason": "TF32 behavior API settings and MLA q_a_layernorm rms_norm_eps precision error are unrelated." + }, + { + "left": "issue:45561", + "right": "issue:46752", + "accept": false, + "reason": "xdist captured_info.txt race and empty-list apply_chat_template IndexError are different bugs in different subsystems." } ] }, @@ -16317,214 +16347,214 @@ "issue:43986|issue:45341", "issue:44261|issue:44485" ], - "prepared_review_unit_hash": "efbd2aea51aab5ad90b31ecd3bfd17d8599ad867cc967c43684ab1f49fec8259", + "prepared_review_unit_hash": "a3f61105b700e4cdeb16d2469178d3b99f64a15ddbff08ac91f3d677443a2182", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 30, - "serialized_chars": 35568, - "estimated_input_tokens": 8892, - "estimated_eval_tokens": 18040 + "serialized_chars": 35564, + "estimated_input_tokens": 8891, + "estimated_eval_tokens": 18038 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is loosely related but heterogeneous: most items are distinct Transformers regressions around imports, tokenizer/model loading, multimodal model wiring, and a few optimizer/runtime issues. No soft edge looks strong enough to merge as the same underlying bug.", - "confidence": 0.79, - "canonical_issue_reason": "issue:45278 is the broadest and most representative user-facing regression report in the set, covering multiple import failures after an upgrade and fitting the cluster\u2019s general theme better than the narrower model-specific bugs.", + "summary": "This cluster is a loose mix of Transformers regression reports, mostly around unrelated model-loading, tokenizer, and API-breaking issues. None of the soft-paired artifacts look like the same concrete bug, so all soft edges are rejected. The best representative issue is the broad v5 import/regression umbrella.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report in the set: generic import breakages after upgrading versions, which best matches the cluster\u2019s overall regression-heavy theme.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best single issue to stand in for the cluster because it captures the widest shared symptom category (post-upgrade import breakage) and is more general than the other, more specific reports.", + "best_issue_reason": "issue:45278 is the most representative global issue because it is broad, version-regression oriented, and closer to the cluster\u2019s common failure mode than the many highly specific model bugs.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43937", "right": "issue:44496", "accept": false, - "reason": "Different models and different failure modes: invalid GenerationConfig vs unrecognized model/config loading." + "reason": "Different failures and code paths: GLM-5 generation config validation vs model config recognition for Olmo-Hybrid." }, { "left": "issue:39401", "right": "issue:46209", "accept": false, - "reason": "Tokenizer offset mapping bug vs FP8 inference failure on B200; no shared code-path evidence." + "reason": "Tokenizer offset_mapping bug is unrelated to FP8 inference failure on B200 GPUs." }, { "left": "issue:44410", "right": "issue:45478", "accept": false, - "reason": "Missing projections in qwen3next layer 0 is unrelated to Qwen3.5-Moe from_pretrained loading error." + "reason": "Both mention Qwen3 variants, but one is missing projections in a layer and the other is a from_pretrained loading error." }, { "left": "issue:42371", "right": "issue:46617", "accept": false, - "reason": "TF32 API behavior change and cache FileNotFoundError are unrelated issues." + "reason": "TF32 configuration guidance and trust_remote_code cache FileNotFoundError are unrelated issues." }, { "left": "issue:44261", "right": "issue:46762", "accept": false, - "reason": "Different numerical bugs in different model families; no indication they are the same defect." + "reason": "Both are numerical-difference reports, but they involve different models/modules and distinct root causes." }, { "left": "issue:43208", "right": "issue:46894", "accept": false, - "reason": "Separate training bugs in different models; only a vague similarity in symptom type." + "reason": "Different model families and different training bugs; the shared \u201ctraining loss\u201d wording is not enough." }, { "left": "issue:44933", "right": "issue:45278", "accept": false, - "reason": "Both are import-related, but one is a specific nonexistent image_utils import and the other is a broad upgrade regression; not the same underlying bug." + "reason": "One is a specific missing image_utils import, the other is a broad upgrade-wide import regression report." }, { "left": "issue:38617", "right": "issue:44220", "accept": false, - "reason": "Missing configuration_utils symbol import vs an fbank feature extraction issue; unrelated." + "reason": "ImportError for configuration_utils has nothing to do with the _torch_extract_fbank_features issue." }, { "left": "issue:44075", "right": "issue:45656", "accept": false, - "reason": "SGD args being ignored is not the same as optimizer.step() being called twice under DeepSpeed." + "reason": "Both are optimizer-related, but one is ignored SGD args and the other is double optimizer steps under deepspeed." }, { "left": "issue:43066", "right": "issue:43122", "accept": false, - "reason": "Both involve tokenizer behavior changes, but one is a decoder-type mismatch and the other is a broader tokenization difference; too weak to treat as the same bug." + "reason": "Both involve tokenizer behavior changes, but the symptoms and likely root causes are different." }, { "left": "issue:44263", "right": "issue:44485", "accept": false, - "reason": "torch.split return values in a GlmMoeDsaIndexer is unrelated to GLM-5 RoPE implementation." + "reason": "GlmMoeDsaIndexer split-return bug and GLM-5 RoPE implementation are separate components." }, { "left": "issue:43097", "right": "issue:44493", "accept": false, - "reason": "Embedding/encoder-decoder tying removal and unexpected position-id keys are different compatibility problems." + "reason": "Removed tie_embeddings API and unexpected position_id key errors are different regressions." }, { "left": "issue:43931", "right": "issue:44410", "accept": false, - "reason": "Weight-shape mismatch for a VL model and missing projections in qwen3next are different model-loading defects." + "reason": "Both are Qwen3-family, but one is a weight-shape mismatch and the other is missing attention projections." }, { "left": "issue:43673", "right": "issue:45341", "accept": false, - "reason": "GenerationMixin cache regression during chunked prefill is unrelated to a testing_utils bug." + "reason": "Generation cache handling and a testing_utils bug are unrelated." }, { "left": "issue:41628", "right": "issue:45071", "accept": false, - "reason": "AutoImageProcessor import failure and PreTrainedConfig type-checking regression are different API breakages." + "reason": "Importing AutoImageProcessor and PretrainedConfig type-checking are different breakages." }, { "left": "issue:42222", "right": "issue:45542", "accept": false, - "reason": "VitPose model breakage and TensorFlow-backend confusion from tensorboard-only installs are unrelated." + "reason": "VitPose model breakage and TensorBoard-without-TensorFlow backend errors are unrelated." }, { "left": "issue:45071", "right": "issue:45542", "accept": false, - "reason": "Different subsystems and symptoms; no shared bug." + "reason": "PretrainedConfig typing regression does not match the TensorFlow backend import issue." }, { "left": "issue:44483", "right": "issue:45464", "accept": false, - "reason": "Both touch chat/completions, but one is a general request acceptance failure and the other is a streaming inference failure for a specific model." + "reason": "Chat/completions request rejection and Qwen3.5 streaming failure are distinct API/model-path problems." }, { "left": "issue:44484", "right": "issue:45478", "accept": false, - "reason": "save_pretrained shard-size question is unrelated to Qwen3.5-Moe from_pretrained loading failure." + "reason": "save_pretrained shard-size behavior is unrelated to Qwen3.5 MoE loading failure." }, { "left": "issue:43874", "right": "issue:44261", "accept": false, - "reason": "Missing image patch count helper and missing rms_norm_eps in MLA are separate multimodal/model math bugs." + "reason": "Missing image patch count helper and MLA layernorm epsilon are different multimodal model bugs." }, { "left": "issue:44263", "right": "issue:46172", "accept": false, - "reason": "GlmMoeDsaIndexer torch.split behavior and deepseek-v4 comb transpose issue are different code paths." + "reason": "GlmMoeDsaIndexer torch.split behavior and deepseek-v4 tensor transpose are separate code paths." }, { "left": "issue:44360", "right": "issue:46172", "accept": false, - "reason": "ReLU missing in DSA indexer is unrelated to deepseek-v4 tensor transpose behavior." + "reason": "Missing ReLU in the DSA indexer is unrelated to the deepseek-v4 transpose issue." }, { "left": "issue:27957", "right": "issue:43352", "accept": false, - "reason": "Both mention Flash Attention 2 support, but they concern different models and likely different implementation gaps." + "reason": "Both concern Flash Attention 2, but they are different model-specific support limitations." }, { "left": "issue:43931", "right": "issue:44038", "accept": false, - "reason": "Both are Qwen3-VL-related, but one is a shape-mismatch load error and the other is a separate from_pretrained failure on a moe variant." + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-VL-Moe from_pretrained failure are not the same bug." }, { "left": "issue:43012", "right": "issue:43577", "accept": false, - "reason": "Compile-time warning on bf16 models and Blip2 dtype remaining float32 are not the same bug." + "reason": "Both mention dtype behavior, but one is a warning during compilation and the other is incorrect dtype after loading Blip2." }, { "left": "issue:43441", "right": "issue:44829", "accept": false, - "reason": "FlashAttention problems in different models and scenarios; not enough evidence they are the same concrete defect." + "reason": "Ministral FlashAttention failure and degenerate training with flash_attention_3 are different issues and code paths." }, { "left": "issue:43097", "right": "issue:43452", "accept": false, - "reason": "Embedding-tie removal and gguf_file/pretrained loading breakage are unrelated." + "reason": "An API removal warning and gguf_file loading failures are unrelated." }, { "left": "issue:45003", "right": "issue:45341", "accept": false, - "reason": "Unsafe sys.modules access and a testing_utils bug are unrelated." + "reason": "Unsafe sys.modules access and a testing_utils bug do not describe the same underlying defect." }, { "left": "issue:43986", "right": "issue:45341", "accept": false, - "reason": "AutoProcessor crash without torchvision and testing_utils bug are unrelated." + "reason": "AutoProcessor crashing without torchvision and a testing_utils bug are unrelated." }, { "left": "issue:44261", "right": "issue:44485", "accept": false, - "reason": "Different GLM-5-related model bugs: one is a layernorm epsilon issue, the other is a RoPE implementation question." + "reason": "MLA q_a_layernorm epsilon handling and GLM-5 RoPE implementation are distinct model internals." } ] }, @@ -16598,15 +16628,15 @@ ], "soft_pairs": [ "issue:45753|issue:46369", - "issue:43874|issue:44263", "issue:44117|issue:44987", + "issue:43874|issue:44263", "issue:45440|issue:46172", "issue:43531|issue:45341", "issue:43673|issue:46123", "issue:43097|issue:45588", "issue:38617|issue:42757", - "issue:44410|issue:45375", "issue:44038|issue:44410", + "issue:44410|issue:45375", "issue:43208|issue:46897", "issue:43064|issue:44315", "issue:43986|issue:45372", @@ -16615,8 +16645,8 @@ "issue:46655|issue:46664", "issue:42890|issue:46298", "issue:46172|issue:46710", - "issue:46298|issue:46752", "issue:38617|issue:44351", + "issue:46298|issue:46752", "issue:44062|issue:44117", "issue:43097|issue:43873", "issue:43012|issue:46821", @@ -16626,202 +16656,202 @@ "issue:44704|issue:45103", "issue:44380|issue:46421" ], - "prepared_review_unit_hash": "262a131e24d0770c525f7af03840f9e39e2f7a33a72aeb9a93c31a8d363cc75d", + "prepared_review_unit_hash": "2f918ec9ce39aa1e34912192571527e1fcde5eae6e26f5d7fcd9129202b3eb5d", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 47, "item_count": 47, "soft_pair_count": 28, - "serialized_chars": 35336, - "estimated_input_tokens": 8834, - "estimated_eval_tokens": 17924 + "serialized_chars": 35332, + "estimated_input_tokens": 8833, + "estimated_eval_tokens": 17922 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is a grab-bag of mostly unrelated Transformers issues; the soft links are only superficial similarities by subsystem or model family, not the same underlying bug. No PRs are present, and none of the issue pairs look safe to merge as duplicates.", - "confidence": 0.84, - "canonical_issue_reason": "issue:43502 is the cleanest self-contained bug report in the set: it names a single concrete failure mode, has an obvious repro condition, and maps to a specific code path. That makes it a good representative duplicate target compared with the more model-specific or multi-bug reports.", + "summary": "This cluster is broadly heterogeneous: most items are distinct Transformers bugs or regressions, and the soft pairs are mostly same-subsystem lookalikes rather than true duplicates. I would not merge any of the proposed soft edges.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 44117 is the cleanest representative: it describes a concrete upstream loading failure in from_pretrained/tokenizer mapping logic, which is more canonical than the model-specific or one-off reports in this set.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43502 is the best dedupe anchor because the symptom is precise and broadly understandable: local_files_only still allowing network access. It is concise, actionable, and less ambiguous than the other issues here.", + "best_issue_reason": "44117 is the most actionable and broadly representative issue in the cluster; it has a clear root-cause shape and is the closest thing here to a reusable duplicate anchor.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45753", "right": "issue:46369", "accept": false, - "reason": "Different bugs: one is a Qwen3_5 generation loop on a specific image, the other is a Trainer import conflict involving datasets." + "reason": "Different models and symptoms: an image-triggered Qwen3_5 loop vs a Trainer import/uninstall-datasets issue." }, { - "left": "issue:43874", - "right": "issue:44263", + "left": "issue:44117", + "right": "issue:44987", "accept": false, - "reason": "Both mention GLM-family code, but they describe different failures in different paths: image patch counting vs torch.split handling." + "reason": "Both involve loading, but one is a tokenizer mapping None bug and the other is a specific model-loading failure; no clear same root cause." }, { - "left": "issue:44117", - "right": "issue:44987", + "left": "issue:43874", + "right": "issue:44263", "accept": false, - "reason": "The tokenizer-mapping None bug is broader, but the loading failure for physical-intelligence/fast is a separate model-specific symptom." + "reason": "GLM46V image processor missing method vs a torch.split indexing bug; different code paths." }, { "left": "issue:45440", "right": "issue:46172", "accept": false, - "reason": "Different model implementations and different failure modes: DeepseekV3MoE divergence versus a deepseek-v4 tensor transpose bug." + "reason": "Different DeepSeek model families and different failures; not the same bug." }, { "left": "issue:43531", "right": "issue:45341", "accept": false, - "reason": "Unrelated: a Qwen3-MoE sliding_window problem versus a small testing_utils bug." + "reason": "Qwen3-MoE sliding-window behavior vs a testing_utils bug; unrelated." }, { "left": "issue:43673", "right": "issue:46123", "accept": false, - "reason": "Both involve generation/pipeline behavior, but they are distinct code paths and symptoms: chunked_prefill cache handling vs final partial-batch masking." + "reason": "Chunked-prefill cache bug vs MaskGenerationPipeline final-batch handling; different subsystems and failure modes." }, { "left": "issue:43097", "right": "issue:45588", "accept": false, - "reason": "A removed config option and a flash-attention crash on s_aux=None are not the same underlying bug." + "reason": "API removal in tie_embeddings_and_encoder_decoder vs flash_attention crash on s_aux=None; unrelated." }, { "left": "issue:38617", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they involve different modules and symbols from different dependencies." + "reason": "Both are ImportErrors, but for different symbols and different modules; not the same underlying issue." }, { - "left": "issue:44410", - "right": "issue:45375", + "left": "issue:44038", + "right": "issue:44410", "accept": false, - "reason": "Same broad Qwen space, but one is missing attention projections while the other drops a vision-config field under @strict." + "reason": "Same broad Qwen family, but one is a transformers 5.0 bug and the other is missing layer projections; distinct concrete problems." }, { - "left": "issue:44038", - "right": "issue:44410", + "left": "issue:44410", + "right": "issue:45375", "accept": false, - "reason": "Both mention Qwen variants, but the reported failures are different model/config issues rather than one shared bug." + "reason": "Both Qwen3-family config/model issues, but one is missing projections and the other is a missing config field; not one bug." }, { "left": "issue:43208", "right": "issue:46897", "accept": false, - "reason": "Multi-bug xLSTM training issues are unrelated to the Florence2 training-loss double-shift regression." + "reason": "xLSTM training blockers vs a Florence2 training-loss shift bug; unrelated." }, { "left": "issue:43064", "right": "issue:44315", "accept": false, - "reason": "One is an optimizer-state corruption bug under Trainer/FSDP2/PEFT; the other is Liger Kernel not being applied in model_init." + "reason": "FSDP2/PEFT optimizer-state corruption vs Liger Kernel not applied with model_init; different fixes." }, { "left": "issue:43986", "right": "issue:45372", "accept": false, - "reason": "Different missing-dependency failures: torchvision absence for video models versus a mistral_common import break in Gemma 4 loading." + "reason": "Video AutoProcessor without torchvision vs a mistral_common import crash; unrelated." }, { "left": "issue:43097", "right": "issue:46050", "accept": false, - "reason": "A removed API flag and broken quantized save/load behavior are separate issues with different triggers and fixes." + "reason": "Removed tie_embeddings API vs quantized save/load corruption; different code paths." }, { "left": "issue:43986", "right": "issue:44484", "accept": false, - "reason": "No overlap beyond being transformer-related; one is a video-model import crash, the other is a save_pretrained shard-size question." + "reason": "Video processor import crash vs max_shard_size default question; unrelated." }, { "left": "issue:46655", "right": "issue:46664", "accept": false, - "reason": "Same SmolVLM video processor area, but different bugs: padding computation versus num_frames semantics." + "reason": "Both are SmolVLM video-processor reports, but padding math and num_frames semantics are distinct bugs." }, { "left": "issue:42890", "right": "issue:46298", "accept": false, - "reason": "A flaky test due to missing set_seed is unrelated to a noisy logging message." + "reason": "Missing set_seed in tests vs a noisy logging message; not duplicates." }, { "left": "issue:46172", "right": "issue:46710", "accept": false, - "reason": "Different DeepSeek-related reports, but one is a concrete tensor operation bug and the other is a broad accuracy regression after upgrade." + "reason": "DeepSeek-V4 tensor transform bug vs a downstream accuracy regression report; no direct same-root-cause evidence." }, { - "left": "issue:46298", - "right": "issue:46752", + "left": "issue:38617", + "right": "issue:44351", "accept": false, - "reason": "A logging nuisance and an empty-list apply_chat_template IndexError are not the same bug." + "reason": "Both import-name errors, but for different missing exports and different failure contexts." }, { - "left": "issue:38617", - "right": "issue:44351", + "left": "issue:46298", + "right": "issue:46752", "accept": false, - "reason": "Both are import errors in transformers, but they concern different missing symbols and different failure sites." + "reason": "Logging noise vs apply_chat_template empty-list crash; unrelated." }, { "left": "issue:44062", "right": "issue:44117", "accept": false, - "reason": "Tokenizer construction error with AddedToken is a different bug from TOKENIZER_MAPPING_NAMES returning None." + "reason": "Tokenizer AddedToken keyword collision vs tokenizer mapping None; both tokenizer-related but different failures." }, { "left": "issue:43097", "right": "issue:43873", "accept": false, - "reason": "No shared code path: one is an API removal complaint, the other is an offloading/quantization behavior bug." + "reason": "API removal in model config vs quantization offloading behavior; different issues." }, { "left": "issue:43012", "right": "issue:46821", "accept": false, - "reason": "Unrelated issues: a PyTorch precision warning during bfloat16 compilation versus a TimesFM AttributeError." + "reason": "PyTorch precision warning during bf16 compilation vs TimesFM window_size AttributeError; unrelated." }, { "left": "issue:44117", "right": "issue:46701", "accept": false, - "reason": "A tokenizer-mapping assumption bug and a DiffusionGemma pipeline failure are separate problems." + "reason": "Tokenizer mapping None bug vs DiffusionGemma pipeline failure; different code paths." }, { "left": "issue:43502", "right": "issue:43992", "accept": false, - "reason": "One is about unintended network access with local_files_only; the other is a missing embed_tokens.weight during UMT5Encoder loading." + "reason": "local_files_only still making requests vs missing embed_tokens.weight on load; unrelated loading bugs." }, { "left": "issue:44936", "right": "issue:46365", "accept": false, - "reason": "Trainer.evaluate() failing after train is unrelated to the inability to import Trainer while datasets is installed." + "reason": "Trainer evaluate-after-train regression vs import failure until datasets is removed; not the same bug." }, { "left": "issue:44704", "right": "issue:45103", "accept": false, - "reason": "Different layers entirely: cached_file kwargs forwarding versus an auto_docstring crash under future annotations." + "reason": "AutoProcessor kwargs propagation bug vs auto_docstring failure under future annotations; unrelated." }, { "left": "issue:44380", "right": "issue:46421", "accept": false, - "reason": "Both are performance/behavior regressions, but one is attention-scaling being ignored and the other is spurious recompilation from cache initialization." + "reason": "GPT2 attention scaling ignored in backends vs cache lazy-init recompilation; distinct problems." } ] }, @@ -16902,8 +16932,8 @@ "issue:43825|issue:44655", "issue:43012|issue:43550", "issue:45712|issue:46298", - "issue:43012|issue:44368", "issue:43208|issue:43550", + "issue:43012|issue:44368", "issue:43992|issue:44484", "issue:43352|issue:43976", "issue:43352|issue:44380", @@ -16927,220 +16957,220 @@ "issue:44038|issue:44230", "issue:43756|issue:44485" ], - "prepared_review_unit_hash": "a3a8ea5ca3f2673c1b8f4bcf4da6e4fa6698d84bd1ec680bb00f93459de2537e", + "prepared_review_unit_hash": "c35b10bfd5205f636e10f508fcafb7860444d310c9295990c277a712a7c236f4", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 31, - "serialized_chars": 36746, - "estimated_input_tokens": 9187, - "estimated_eval_tokens": 18630 + "serialized_chars": 36743, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "No duplicate cluster emerged here: the soft pairs mostly link unrelated issues across different components, error classes, and model families. I would not merge any of these pairs as the same underlying bug/change.", - "confidence": 0.96, - "canonical_issue_reason": "No single issue clearly serves as a canonical duplicate target because the candidates span many unrelated problems rather than one shared defect.", - "canonical_pr_reason": "No pull requests were provided, and none can be selected as a canonical fix for this mixed set of issues.", - "best_issue_reason": "No best issue stands out as a global representative; the set is too heterogeneous to justify one canonical issue.", - "best_pr_reason": "No PR candidates are present.", + "summary": "This cluster is a grab bag of unrelated Transformers issues; none of the soft-edge pairs look like true duplicates. Most are different models, different code paths, or warning-vs-crash variants rather than the same underlying bug.", + "confidence": 0.74, + "canonical_issue_reason": "No strong duplicate center emerges; if one issue must anchor the cluster, issue 36683 is the highest-signal concrete runtime failure and has the most discussion.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 36683 is the best stand-in because it is a clear, user-facing crash with broad relevance and substantial conversation, unlike the many narrowly scoped reports here.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44938", "right": "issue:46178", "accept": false, - "reason": "Python 3.14 import/load failure vs a backend-mapping ValueError for tensorflow_text; different failure modes and likely different code paths." + "reason": "Different failure modes: Python 3.14 import/load failure vs an invalid backend mapping for tensorflow_text." }, { "left": "issue:43986", "right": "issue:44117", "accept": false, - "reason": "Video-model/AutoProcessor crash without torchvision is unrelated to TOKENIZER_MAPPING_NAMES returning None during from_pretrained." + "reason": "Both are import/loading-related, but one is missing torchvision handling and the other is a tokenizer mapping None bug; not the same code path." }, { "left": "issue:42907", "right": "issue:44655", "accept": false, - "reason": "Dequantized model saving failure is a model serialization bug, while Pipeline.save_pretrained concerns pipeline object persistence." + "reason": "Saving dequantized models and saving Pipeline objects are different serialization paths." }, { "left": "issue:42907", "right": "issue:45753", "accept": false, - "reason": "Saving dequantized Ministral/Devstral models and Qwen3.5 infinite-loop generation are different problems in different code paths." + "reason": "Model saving failure vs an infinite generation loop on a specific image are unrelated." }, { "left": "issue:43825", "right": "issue:44655", "accept": false, - "reason": "pipeline() task-message mismatch is an error-message/behavior issue, not the same as saving Pipeline objects." + "reason": "Error-message behavior in pipeline() is unrelated to save_pretrained() pipeline object persistence." }, { "left": "issue:43012", "right": "issue:43550", "accept": false, - "reason": "A PyTorch precision warning during bfloat16 compilation is unrelated to Bamba-9B-v2 torch.compile SDPA failure." + "reason": "A precision warning during compilation is not the same as a torch.compile SDPA failure on a specific model." }, { "left": "issue:45712", "right": "issue:46298", "accept": false, - "reason": "Leftover dummy classes/check_repo leak and a noisy logging message are both cleanup concerns but not the same bug." + "reason": "Left is about dummy class cleanup/export leaks; right is about a noisy log message." }, { - "left": "issue:43012", - "right": "issue:44368", + "left": "issue:43208", + "right": "issue:43550", "accept": false, - "reason": "Compile-time float32 warning for bf16 models is unrelated to a tie_word_embeddings warning in Qwen3.5 LoRA fine-tuning." + "reason": "xLSTM training bugs are unrelated to Bamba torch.compile/SDPA behavior." }, { - "left": "issue:43208", - "right": "issue:43550", + "left": "issue:43012", + "right": "issue:44368", "accept": false, - "reason": "xLSTM training bugs and Bamba-9B-v2 SDPA/torch.compile failure are distinct model-specific issues." + "reason": "Both mention warnings, but one is a bfloat16 compile warning and the other is a tie_word_embeddings config warning." }, { "left": "issue:43992", "right": "issue:44484", "accept": false, - "reason": "Missing embed_tokens.weight in UMT5Encoder.from_pretrained is unrelated to the default max_shard_size in save_pretrained." + "reason": "Missing embed_tokens.weight on load is unrelated to the max_shard_size default in save_pretrained()." }, { "left": "issue:43352", "right": "issue:43976", "accept": false, - "reason": "Nemotron Flash Attention 2 support and Python 3.9/3.10 compatibility are unrelated." + "reason": "Flash Attention support for a model is unrelated to a Python version compatibility regression." }, { "left": "issue:43352", "right": "issue:44380", "accept": false, - "reason": "Flash Attention 2 support gating for Nemotron is not the same as GPT2 attention scaling being ignored by backend implementations." + "reason": "Different attention/backend issues: model not supporting FA2 vs GPT2 scaling ignored under SDPA/FlashAttention." }, { "left": "issue:43066", "right": "issue:43906", "accept": false, - "reason": "Wrong tokenizer decoder type in v5 and an isolated reproduction of another issue are not the same defect." + "reason": "Tokenizer decoder type regression is not the same as an isolated reproduction of another issue." }, { "left": "issue:44162", "right": "issue:46172", "accept": false, - "reason": "ESM2 breakage is a model-specific regression; deepseek-v4 comb.transpose behavior is a different model/math bug." + "reason": "ESM2 model breakage and a deepseek matmul/transpose bug are different model-specific problems." }, { "left": "issue:43097", "right": "issue:46298", "accept": false, - "reason": "Removed tie_embeddings_and_encoder_decoder API and noisy logging are unrelated changes." + "reason": "Removed config flag behavior is unrelated to a noisy logging message." }, { "left": "issue:44380", "right": "issue:46132", "accept": false, - "reason": "GPT2 attention scaling ignored by SDPA/FlashAttention is unrelated to AttentionInterface.register changing registered-function behavior." + "reason": "Attention scaling being ignored is distinct from AttentionInterface.register changing function behavior." }, { "left": "issue:45584", "right": "issue:46365", "accept": false, - "reason": "Whisper empty-transcription generation failure is unrelated to a Trainer import conflict with datasets." + "reason": "Whisper generation edge case is unrelated to importing Trainer conflicting with datasets." }, { "left": "issue:44077", "right": "issue:46365", "accept": false, - "reason": "patchtsmixer post_init validation and Trainer import conflict are different problems in different subsystems." + "reason": "PatchTSMixer post_init validation is unrelated to a Trainer import conflict." }, { "left": "issue:44222", "right": "issue:45478", "accept": false, - "reason": "FP8 save_pretrained for moe models and Qwen3.5 Moe from_pretrained errors are related only broadly to MoE, but not the same concrete bug." + "reason": "FP8 save_pretrained behavior and Qwen3.5 MoE from_pretrained errors are different directions and likely different bugs." }, { "left": "issue:36683", "right": "issue:43582", "accept": false, - "reason": "Gemma3Config missing vocab_size and Apple Silicon TypeError in caching_allocator_warmup are entirely different issues." + "reason": "Gemma3Config vocab_size crash is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." }, { "left": "issue:44380", "right": "issue:45479", "accept": false, - "reason": "Attention backend scaling bug is unrelated to degenerate zero loss in single-label classification." + "reason": "GPT2 attention scaling under backend kernels is unrelated to zero-loss behavior for single-label classification." }, { "left": "issue:42175", "right": "issue:46365", "accept": false, - "reason": "TensorFlow not included in a torch-only install and Trainer import failing until datasets is uninstalled are distinct dependency-resolution problems." + "reason": "TensorFlow extra/backend packaging is unrelated to the datasets/Trainer import conflict." }, { "left": "issue:42757", "right": "issue:43824", "accept": false, - "reason": "Missing is_offline_mode import from huggingface_hub and missing Qwen2_5_VLForConditionalGeneration export are separate import-regression issues." + "reason": "Both are import errors, but they involve different symbols and different packages." }, { "left": "issue:43064", "right": "issue:44380", "accept": false, - "reason": "FSDP2/PEFT optimizer-state corruption on nonzero ranks is unrelated to GPT2 attention scaling being ignored." + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to GPT2 attention scaling behavior." }, { "left": "issue:42490", "right": "issue:43012", "accept": false, - "reason": "Trailing-slash local model load warning and bf16 compile precision warning are different user-facing issues." + "reason": "Local path trailing-slash warning is unrelated to a bfloat16 compile warning." }, { "left": "issue:43208", "right": "issue:43526", "accept": false, - "reason": "xLSTM training bugs are unrelated to BeitImageProcessorFast reduce_labels returning only one label." + "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated subsystems." }, { "left": "issue:44483", "right": "issue:44514", "accept": false, - "reason": "chat/completions request rejection in v5 is unrelated to Qwen2_5_VLProcessor.apply_chat_template crashing on batched padding=False input." + "reason": "Chat completions request handling is different from a batched apply_chat_template crash." }, { "left": "issue:43618", "right": "issue:44380", "accept": false, - "reason": "CLIPOutput attentions no longer assigned is a model-output regression, not the GPT2 attention-scaling backend bug." + "reason": "CLIPOutput attentions assignment regression is unrelated to GPT2 attention scaling under backend kernels." }, { "left": "issue:38617", "right": "issue:43502", "accept": false, - "reason": "Missing layer_type_validation import and API requests despite local_files_only=True are unrelated import vs networking bugs." + "reason": "Missing configuration_utils symbol import and local_files_only network leakage are different bugs." }, { "left": "issue:44038", "right": "issue:44230", "accept": false, - "reason": "Qwen3-VL-Moe v5 bug and fp8 Qwen3-VL-Moe inference support are not the same concrete failure." + "reason": "Qwen3-VL-Moe support issues here are about different model variants and different failure points." }, { "left": "issue:43756", "right": "issue:44485", "accept": false, - "reason": "Smollm3 RoPE-layer mismatch and GLM-5 RoPE implementation are both about RoPE but describe different models and different implementation problems." + "reason": "RoPE-layer dropping in Smollm3 and GLM-5 RoPE implementation questions are model-specific and not the same bug." } ] }, @@ -17219,9 +17249,9 @@ "issue:43299|issue:43867", "issue:36010|issue:42757", "issue:44625|issue:46365", + "issue:45276|issue:46077", "issue:44355|issue:46877", "issue:43352|issue:43937", - "issue:45276|issue:46077", "issue:44483|issue:46369", "issue:36683|issue:43475", "issue:43611|issue:44655", @@ -17249,238 +17279,238 @@ "issue:43867|issue:45375", "issue:42890|issue:45561" ], - "prepared_review_unit_hash": "457c4e50866584364086c0a2c86f05e766beddfdf40b0c606f983ca8e7b93b0f", + "prepared_review_unit_hash": "73b75afd55ce0ea8c3a48f22f0fe545d2a4790eace9eb3b74c92d742e486fab4", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, "soft_pair_count": 34, - "serialized_chars": 37581, - "estimated_input_tokens": 9396, - "estimated_eval_tokens": 19048 + "serialized_chars": 37588, + "estimated_input_tokens": 9397, + "estimated_eval_tokens": 19050 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions and model-specific bugs. There are some surface-level similarities (imports, loading, configs, saving), but none of the soft pairs look like the same underlying defect with enough certainty to merge.", - "confidence": 0.82, - "canonical_issue_reason": "Issue 43611 is the clearest broad regression in the set: a concrete v5 loading failure affecting model classes via `base_model_prefix`, and it is more representative than the narrower model-specific items.", + "summary": "This is a heterogeneous set of unrelated bugs and regressions. The soft pairs mostly share only broad themes like import/load failures or model-specific regressions, not the same concrete underlying defect, so no pair should be merged.", + "confidence": 0.39, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here (general import failures after an upgrade), so it is the least narrow representative if one issue must anchor the cluster.", "canonical_pr_reason": null, - "best_issue_reason": "43611 is the best single issue to anchor the cluster because it is concrete, user-facing, and broadly phrased enough to represent the general v5 loading-breakage theme without being as vague as the more generic complaints.", + "best_issue_reason": "issue:45278 is the closest thing to a cluster anchor because it describes a wide regression class rather than a model-specific symptom, but the cluster overall is not cohesive.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:42994", "right": "issue:44655", "accept": false, - "reason": "Both are about saving, but one is quantized model saving and the other is pipeline `save_pretrained`; different failure modes." + "reason": "Both concern saving, but one is quantized model saving and the other is Pipeline.save_pretrained; different failure modes and code paths." }, { "left": "issue:43618", "right": "issue:43761", "accept": false, - "reason": "Both touch CLIP v5 outputs, but one is missing attentions and the other missing hidden states; not the same bug." + "reason": "Both touch CLIP v5 regression behavior, but one is missing attentions and the other is missing hidden_states; same subsystem, not clearly the same bug." }, { "left": "issue:43299", "right": "issue:43867", "accept": false, - "reason": "Both are model-loading errors, but Qwen3VL MoE loading and sorted state_dict loading are different code paths and symptoms." + "reason": "Different problems: Qwen3VL MoE loading versus a generic sorted-state_dict load error." }, { "left": "issue:36010", "right": "issue:42757", "accept": false, - "reason": "Both are import errors, but they concern different symbols from different packages." + "reason": "Both are import errors, but they involve different symbols and different dependency breakages." }, { "left": "issue:44625", "right": "issue:46365", "accept": false, - "reason": "One is config propagation for `num_labels`; the other is an import-time Trainer failure tied to `datasets`." + "reason": "Unrelated: Qwen3.5 num_labels propagation versus importing Trainer being blocked by datasets." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Same model family, but one is embedding resize propagation and the other is attention-mask behavior; not the same underlying bug." }, { "left": "issue:44355", "right": "issue:46877", "accept": false, - "reason": "Both are import/runtime hygiene issues, but compiled Python file errors and auto-docstring stdout diagnostics are unrelated." + "reason": "One is about running compiled Python files, the other about import-time docstring lint output." }, { "left": "issue:43352", "right": "issue:43937", "accept": false, - "reason": "Flash Attention support rejection and invalid GenerationConfig are distinct failures." - }, - { - "left": "issue:45276", - "right": "issue:46077", - "accept": false, - "reason": "Same model family, but resizing embeddings and causal-vs-bidirectional attention masks are different bugs." + "reason": "Different model-specific errors: Flash Attention support versus invalid GenerationConfig." }, { "left": "issue:44483", "right": "issue:46369", "accept": false, - "reason": "Chat-completions request validation and SAM3 memory overflow are unrelated." + "reason": "Completely different areas: chat/completions request validation versus SAM3 memory overflow." }, { "left": "issue:36683", "right": "issue:43475", "accept": false, - "reason": "Different model families and different missing attributes; not the same defect." + "reason": "Different model configs and different missing attributes; only superficial AttributeError similarity." }, { "left": "issue:43611", "right": "issue:44655", "accept": false, - "reason": "Model-loading via `base_model_prefix` and pipeline serialization are separate issues." + "reason": "Model loading with base_model_prefix is unrelated to saving Pipeline objects." }, { "left": "issue:43901", "right": "issue:44655", "accept": false, - "reason": "Docs mismatch for `return_all_scores` is unrelated to pipeline save failures." + "reason": "Docs mismatch for TextClassificationPipeline is unrelated to pipeline save_pretrained failure." }, { "left": "issue:43867", "right": "issue:45216", "accept": false, - "reason": "A sorted-state_dict load error and a bad Qwen3.5 checkpoint save are not the same root cause." + "reason": "Both involve load/save regressions, but one is sorted state_dict loading and the other is checkpoint saving correctness for Qwen3.5." }, { "left": "issue:43976", "right": "issue:44938", "accept": false, - "reason": "Both are platform-version regressions, but Python 3.9/3.10 compatibility and Python 3.14 load failure are different problems." + "reason": "Both are Python-version compatibility issues, but they are different version bounds and different failures." }, { "left": "issue:36683", "right": "issue:44077", "accept": false, - "reason": "Missing config fields for Gemma3 and unused SGD args are unrelated." + "reason": "Unrelated: Gemma3 vocab_size missing versus SGD optimizer args not being used." }, { "left": "issue:44625", "right": "issue:46077", "accept": false, - "reason": "`num_labels` propagation and Gemma4 attention-mask behavior do not share a code path." + "reason": "Different Gemma/Qwen model bugs; no shared concrete code-path." }, { "left": "issue:46077", "right": "issue:46082", "accept": false, - "reason": "Both involve model configuration, but attention-mask construction and `head_dim` validation are different changes." + "reason": "Both are Llama/Gemma-style config issues, but one is attention masking and the other is head_dim validation." }, { "left": "issue:43618", "right": "issue:45216", "accept": false, - "reason": "CLIP output fields and Qwen3.5 checkpoint saving are unrelated regressions." + "reason": "Different regressions in different models: CLIP output fields versus Qwen3.5 checkpoint saving." }, { "left": "issue:43208", "right": "issue:46829", "accept": false, - "reason": "xLSTM training bugs and VideoPrism output-field mixup are unrelated." + "reason": "xLSTM training bugs are unrelated to VideoPrism output wiring." }, { "left": "issue:36683", "right": "issue:43525", "accept": false, - "reason": "Both are missing config attributes, but on different model families with different semantics." + "reason": "Different missing attributes in different configs; same error class only superficially." }, { "left": "issue:36683", "right": "issue:44625", "accept": false, - "reason": "Gemma3 missing `vocab_size` and Qwen3.5 `num_labels` propagation are different config bugs." + "reason": "Both are config-related, but the missing fields and affected models are different." }, { "left": "issue:29870", "right": "issue:36296", "accept": false, - "reason": "Bart generation crash and tensor-parallel training bug are unrelated." + "reason": "Bart generate decoder_attention_mask error is unrelated to tensor-parallel training bugs." }, { "left": "issue:43992", "right": "issue:45310", "accept": false, - "reason": "UMT5 missing `embed_tokens.weight` and Qwen3.5 MoE `from_pretrained` failure are different loading issues." + "reason": "Both involve from_pretrained, but they affect different models and different missing/invalid state handling." }, { "left": "issue:43064", "right": "issue:44492", "accept": false, - "reason": "A distributed-training optimizer-state bug is unrelated to a cache-strategy typo." + "reason": "Trainer/FSDP optimizer-state corruption is unrelated to a cache-strategy typo." }, { "left": "issue:43010", "right": "issue:45278", "accept": false, - "reason": "`@torch.no_grad` decoration for cache updates and broad import errors after upgrade are different concerns." + "reason": "Cache update no_grad decoration is unrelated to broad import errors after upgrade." }, { "left": "issue:44075", "right": "issue:46365", "accept": false, - "reason": "Unused optimizer args and import failure when `datasets` is installed are not the same defect." + "reason": "Optimizer argument handling is unrelated to importing Trainer." }, { "left": "issue:44908", "right": "issue:46851", "accept": false, - "reason": "A scheduler keyword bug and Qwen3.5 variable-length position_ids failure are unrelated." + "reason": "Scheduler kwargs are unrelated to Qwen3.5 variable-length position_ids behavior." }, { "left": "issue:29870", "right": "issue:33453", "accept": false, - "reason": "Bart generation masking crash and tokenizer loading regression are different issues." + "reason": "Bart generation error and tokenizer-loading regression are different failures." }, { "left": "issue:44938", "right": "issue:45071", "accept": false, - "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are distinct version-breaking problems." + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different compatibility issues." }, { "left": "issue:44355", "right": "issue:44855", "accept": false, - "reason": "Compiled-file import errors and a Python 3.13 `@torch.jit.script` parsing issue have different causes." + "reason": "Both are import-time problems, but one is compiled-file execution and the other is a torch.jit/script parsing issue." }, { "left": "issue:43867", "right": "issue:46202", "accept": false, - "reason": "Load failure with sorted state_dict and Gemma pipeline roundtrip failure are not the same bug." + "reason": "Generic load error and Gemma pipeline roundtrip failure are not the same bug." }, { "left": "issue:36683", "right": "issue:43064", "accept": false, - "reason": "A model config attribute bug and a distributed optimizer-state bug are unrelated." + "reason": "Gemma3 config attribute missing is unrelated to FSDP2/PEFT optimizer-state corruption." }, { "left": "issue:43867", "right": "issue:45375", "accept": false, - "reason": "Sorted state_dict load errors and missing Qwen3.5 vision config fields do not share a root cause." + "reason": "Both are config/load regressions, but they concern different models and different missing fields." }, { "left": "issue:42890", "right": "issue:45561", "accept": false, - "reason": "One is missing `set_seed()` in integration tests; the other is an xdist race on a captured-info file." + "reason": "One is test flakiness from missing set_seed in a model integration test; the other is an xdist race in patched testing utils." } ] }, @@ -17585,228 +17615,241 @@ "issue:45125|issue:45375", "issue:46664|issue:46726", "issue:43010|issue:43232", + "issue:43502|issue:44222", "issue:43502|issue:44961" ], - "prepared_review_unit_hash": "3ac81f82dca4dce5370566360dcc45a434dcc9d1440aa4f576370f2c70aa85b7", + "prepared_review_unit_hash": "d20a5b5d62049028d8d73c38649b249ce1599735b65559dca9f63885923681f1", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 32, - "serialized_chars": 36879, - "estimated_input_tokens": 9220, - "estimated_eval_tokens": 18696 + "soft_pair_count": 33, + "serialized_chars": 37183, + "estimated_input_tokens": 9296, + "estimated_eval_tokens": 18848 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, - "split": true, - "analyst_result": { - "summary": "This cluster is mostly a loose bundle of v5-era regressions, but the strongest duplicate-like subtheme is top-level import/export breakage and, secondarily, import-time performance. I only accepted pairs where the titles clearly point to the same concrete issue.", - "confidence": 0.63, - "canonical_issue_reason": "Issue 42548 is the clearest central representative of the cluster\u2019s import/export breakage theme: a concrete, user-facing top-level import failure in transformers v5.", - "canonical_pr_reason": null, - "best_issue_reason": "42548 is the best single issue anchor because it is specific, broadly relevant, and closest to the recurring class of import regressions present in this set.", - "best_pr_reason": null, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most pairs are false-positive similarity links across unrelated import errors, tokenizer/model regressions, training/test flakiness, and model-specific issues. Only the Qwen3.5 MoE tensor-parallel pair looks plausibly like the same underlying bug.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue is safe here; the items span unrelated code paths and models, with only a narrow Qwen3.5 MoE subset showing real overlap.", + "canonical_pr_reason": "No PRs are present in this cluster.", + "best_issue_reason": "No global best issue for the whole cluster. If anything, issue 45125 is the most concrete root-cause report, but it only fits a small subset of the items.", + "best_pr_reason": "No PRs are present in this cluster.", "soft_edge_verdicts": [ { "left": "issue:43010", "right": "issue:45850", "accept": false, - "reason": "Both are broad complaints about v5, but they describe different problems: cache update no_grad semantics vs general model breakage." + "reason": "Different scopes: a cache/no_grad code fix vs a broad complaint that many models break in v5." }, { "left": "issue:41669", "right": "issue:44246", - "accept": true, - "reason": "Both point to the same underlying problem: transformers import is slow, likely from wildcard imports in models." + "accept": false, + "reason": "Both mention import performance, but one is a specific import-star cleanup and the other is a generic slow-import report." }, { "left": "issue:36296", "right": "issue:43575", "accept": false, - "reason": "Both mention tensor parallelism, but one is a training bug and the other is an OOM on loading a specific model." + "reason": "Both touch tensor parallelism, but one is a generic training bug and the other is an OOM on loading a specific model." }, { "left": "issue:43010", "right": "issue:43976", "accept": false, - "reason": "Unrelated issues: torch.no_grad on cache updates vs Python version compatibility." + "reason": "Unrelated problems: cache update gradients vs Python version compatibility." }, { "left": "issue:43618", "right": "issue:44077", "accept": false, - "reason": "Different models and different bugs; no shared concrete code path." + "reason": "Different model/code paths; one is CLIP output fields, the other is a patchtsmixer init restriction." }, { "left": "issue:43992", "right": "issue:45478", "accept": false, - "reason": "Both are model loading errors, but they involve different models and different failure modes." + "reason": "Both are from_pretrained failures, but they affect different models and appear to have different causes." }, { "left": "issue:43867", "right": "issue:45357", "accept": false, - "reason": "One is a generic load failure, the other is incorrect save_pretrained keys for Qwen3.5; not the same bug." + "reason": "Serialization/loading bug vs save_pretrained key mismatch; not the same concrete defect." }, { "left": "issue:28282", "right": "issue:33453", "accept": false, - "reason": "ImportError for missing PyTorch is unrelated to tokenizer loading regression." + "reason": "Completely unrelated: missing PyTorch import vs tokenizer-loading regression." }, { "left": "issue:43618", "right": "issue:46285", "accept": false, - "reason": "Both are CLIP-related, but one is an attentions regression and the other is a structure inconsistency." + "reason": "Same broad CLIP area, but one is an output attribute issue and the other is a structural model API change." }, { "left": "issue:44360", "right": "issue:45440", "accept": false, - "reason": "Completely different topics: DSA indexer activation vs DeepSeekV3MoE implementation divergence." + "reason": "Different subsystems and symptoms: DSA indexer issue vs DeepSeek MoE implementation divergence." }, { "left": "issue:42890", "right": "issue:44964", "accept": false, - "reason": "Test flakiness from missing set_seed is unrelated to a model-loading failure." + "reason": "Test flakiness in SAM HQ is unrelated to a Phi-4 multimodal load failure." }, { "left": "issue:42890", "right": "issue:44112", "accept": false, - "reason": "Both are test-related, but they target different test failures and different root causes." + "reason": "Both are CI/test failures, but they are different tests with different failure mechanisms." }, { "left": "issue:42890", "right": "issue:45600", "accept": false, - "reason": "Unrelated: a flaky SAM HQ integration test vs broken auto_mappings references." + "reason": "Test seeding issue vs stale auto-mapping references; not the same bug." }, { "left": "issue:42898", "right": "issue:46664", "accept": false, - "reason": "Tokenizer cleanup behavior and video num_frames semantics are unrelated." + "reason": "Tokenizer behavior change vs video processor frame-count semantics; unrelated." }, { "left": "issue:43502", "right": "issue:44062", "accept": false, - "reason": "Different bugs: offline API requests vs AddedToken keyword argument conflict." + "reason": "Network access despite local_files_only is unrelated to AddedToken argument duplication." }, { "left": "issue:44734", "right": "issue:46617", "accept": false, - "reason": "One is a server KV-cache indexing crash, the other a cache FileNotFoundError; not the same issue." + "reason": "KV-cache continuation crash in serve vs cache FileNotFoundError for trust_remote_code; different paths." }, { "left": "issue:43723", "right": "issue:44556", "accept": false, - "reason": "Both concern loading after v5, but one is tokenizer loading and the other is checkpoint reload compatibility; too broad to merge." + "reason": "Tokenizer-loading regression vs broad checkpoint reload regression after upgrade; too broad to merge." }, { "left": "issue:44222", "right": "issue:45125", "accept": false, - "reason": "FP8 save_pretrained behavior and missing tp plan are different code paths." + "reason": "FP8 save_pretrained issue and missing tensor-parallel plan are different bugs." }, { "left": "issue:44387", "right": "issue:44556", "accept": false, - "reason": "Quantization memory growth and checkpoint reload failure are unrelated." + "reason": "Quantization memory regression vs checkpoint reload incompatibility; not the same defect." }, { "left": "issue:43824", "right": "issue:46617", "accept": false, - "reason": "Different import/load failures with different causes and targets." + "reason": "Different import/load failures affecting different symbols and code paths." }, { "left": "issue:43502", "right": "issue:44485", "accept": false, - "reason": "Offline-loading behavior and GLM-5 RoPE implementation are unrelated." + "reason": "Network-request bug vs GLM-5 RoPE implementation discussion; unrelated." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "Image patch-count helper missing vs trust_remote_code cache loading failure; distinct issues." }, { "left": "issue:43576", "right": "issue:46132", "accept": false, - "reason": "CLI/env breakage and attention registration behavior are distinct changes." + "reason": "Broken env CLI vs AttentionInterface behavior change; different features and failure modes." }, { "left": "issue:43010", "right": "issue:43065", "accept": false, - "reason": "No meaningful duplication: cache update semantics vs a dummy Conv2d implementation." + "reason": "Cache update gradient handling vs dummy Conv2d in Sam3PixelDecoder; unrelated." }, { "left": "issue:45125", "right": "issue:45478", "accept": true, - "reason": "Both describe the same Qwen3.5 MoE tensor-parallel loading problem, with 45478 sounding like the user-facing manifestation of the missing tp plan." + "reason": "Both point to the same Qwen3.5 MoE regression around tensor-parallel/from_pretrained behavior, and a single fix could plausibly address both." }, { "left": "issue:30333", "right": "issue:33453", "accept": false, - "reason": "MLFlow reporting after failed training and tokenizer loading regression are unrelated." + "reason": "MLFlow reporting on failed jobs vs tokenizer loading regression; no shared code path." }, { "left": "issue:44355", "right": "issue:44749", "accept": false, - "reason": "Compiled Python file errors and post-upgrade filtering slowdown are different problems." + "reason": "Compiled Python file errors vs model-filter slowdown after upgrade; unrelated." }, { "left": "issue:42548", "right": "issue:44351", "accept": false, - "reason": "Both are import errors, but for different exported symbols and likely different fixes." + "reason": "Both are import errors, but they refer to different missing exports and likely different root causes." }, { "left": "issue:45125", "right": "issue:45375", "accept": false, - "reason": "Same model family, but one is tensor-parallel setup and the other is a missing config field during save/load." + "reason": "Both involve Qwen3.5-related config, but one is tensor-parallel planning and the other is a missing vision config field." }, { "left": "issue:46664", "right": "issue:46726", "accept": false, - "reason": "Different processor bugs: video frame semantics vs missing image processor backend handling." + "reason": "Video processor frame semantics vs missing vision backend image-processor handling; different bugs." }, { "left": "issue:43010", "right": "issue:43232", "accept": false, - "reason": "Different generation/cache bugs with no shared concrete failure mode." + "reason": "Both touch generation/cache internals, but one is about no_grad decoration and the other about post-sync_gpus kwargs updates." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Local-files-only networking bug vs FP8 save_pretrained issue; unrelated." }, { "left": "issue:43502", "right": "issue:44961", "accept": false, - "reason": "No relationship between offline file loading and the unrelated 'racoon' issue." + "reason": "No meaningful overlap; one is an API/networking bug, the other is unrelated." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays appropriately cautious. The two accepted soft pairs are plausible from the titles alone, and the rejected pairs avoid over-merging issues with different concrete failure modes. The choice of 42548 as the anchor is defensible for the import/export breakage subtheme." + "feedback": "The cluster summary is well-grounded in the packet: the items are broadly heterogeneous, and the only clearly plausible same-bug link is the Qwen3.5 MoE tensor-parallel/from_pretrained pair (45125, 45478). The other soft-edge rejections are conservative and align with the issue titles without overstating evidence." }, "evaluator_used": true, "retried": false, @@ -17840,7 +17883,6 @@ "issue:42757", "issue:43299", "issue:43475", - "issue:43502", "issue:43576", "issue:43673", "issue:43824", @@ -17869,6 +17911,7 @@ "issue:45362", "issue:45468", "issue:45542", + "issue:45753", "issue:45850", "issue:46143", "issue:46144", @@ -17878,7 +17921,6 @@ ], "soft_pairs": [ "issue:30990|issue:31795", - "issue:43502|issue:44222", "issue:43827|issue:44961", "issue:30333|issue:36296", "issue:43299|issue:44556", @@ -17899,8 +17941,8 @@ "issue:46143|issue:46144", "issue:43475|issue:46617", "issue:44222|issue:44230", - "issue:44315|issue:46209", "issue:41628|issue:42548", + "issue:44315|issue:46209", "issue:44485|issue:46664", "issue:43673|issue:44492", "issue:45362|issue:45850", @@ -17909,234 +17951,242 @@ "issue:44360|issue:46664", "issue:44038|issue:45125", "issue:44273|issue:45003", - "issue:44393|issue:45071" + "issue:44393|issue:45071", + "issue:44393|issue:45753", + "issue:43576|issue:44556" ], - "prepared_review_unit_hash": "46519470211342ea4ac63d6c8dc15d95cb4669e4e4fd1f54d359ab283dbc1466", + "prepared_review_unit_hash": "ef33d5193e67a47f668e9091a43dc7e4ec81f7dafbc43b3419b4ad8e4057a3ac", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 48, "item_count": 48, - "soft_pair_count": 33, - "serialized_chars": 36630, - "estimated_input_tokens": 9158, - "estimated_eval_tokens": 18572 + "soft_pair_count": 34, + "serialized_chars": 36921, + "estimated_input_tokens": 9231, + "estimated_eval_tokens": 18718 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of unrelated Transformers issues that only share broad themes like loading regressions, imports, or model-specific bugs. I rejected all soft edges because none look like the same concrete defect or change.", - "confidence": 0.59, - "canonical_issue_reason": "Most broadly scoped, concrete loading regression with a clear version boundary and user impact.", + "summary": "The set is a grab bag of mostly unrelated Transformers issues: loading/import regressions, model-specific bugs, API/export breaks, and documentation problems. None of the soft pairs look safe to collapse into a single duplicate, especially for PR-style strictness.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 33453 is the cleanest concrete regression report in the set and is easier to triage than the very broad or vague titles; if one issue must anchor the group, this is the most representative user-facing bug report.", "canonical_pr_reason": null, - "best_issue_reason": "Strongest representative issue here because it is specific, reproducible, and captures the common version-upgrade loading failure theme.", + "best_issue_reason": "Issue 33453 is the best single issue to keep as a canonical example because it is specific, clearly describes a regression, and is more actionable than the broad/vague alternatives.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:30990", "right": "issue:31795", "accept": false, - "reason": "SentenceTransformers loading hang vs documentation confusion in model.forward; unrelated symptom and subsystem." - }, - { - "left": "issue:43502", - "right": "issue:44222", - "accept": false, - "reason": "Local-files-only network leakage vs FP8 save_pretrained behavior; different code paths." + "reason": "Stuck-loading behavior and documentation confusion are different problems; no shared failing code path." }, { "left": "issue:43827", "right": "issue:44961", "accept": false, - "reason": "Pipeline-removal docs issue vs unrelated placeholder issue." + "reason": "Docs about removed pipeline() usage vs an unrelated placeholder title; no duplicate signal." }, { "left": "issue:30333", "right": "issue:36296", "accept": false, - "reason": "MLflow reporting after failed training vs tensor-parallel training bug; no shared concrete defect." + "reason": "MLflow reporting on failed jobs and tensor-parallel training bugs are unrelated." }, { "left": "issue:43299", "right": "issue:44556", "accept": false, - "reason": "Both are loading regressions, but one is Qwen3VL-Moe-specific while the other is a broad checkpoint reload regression; not the same bug." + "reason": "Both mention version upgrades, but one is Qwen3VL-Moe loading and the other is checkpoint reload compatibility; too broad to merge." }, { "left": "issue:29870", "right": "issue:31795", "accept": false, - "reason": "Generation UnboundLocalError vs documentation ambiguity; unrelated." + "reason": "Runtime UnboundLocalError in generate() vs documentation confusion; different issue types and code paths." }, { "left": "issue:44355", "right": "issue:45071", "accept": false, - "reason": "Compiled Python file errors vs PreTrainedConfig typing regression; different failure modes." + "reason": "Compiled Python file errors and PretrainedConfig type-checking are unrelated failures." }, { "left": "issue:33453", "right": "issue:34689", "accept": false, - "reason": "Both are loading regressions, but tokenizer-loading regression and Llama 3.2 vision model-loading breakage are distinct." + "reason": "Both are loading regressions, but one is tokenizer loading and the other is a model-loading break for a specific Llama VLM; not the same bug." }, { "left": "issue:42757", "right": "issue:43576", "accept": false, - "reason": "Hub import error vs broken env CLI; different APIs and root causes." + "reason": "Missing import from huggingface_hub vs broken transformers env command; no evidence they share the same fix." }, { "left": "issue:43576", "right": "issue:45071", "accept": false, - "reason": "CLI/env command breakage vs type-checking regression; not the same change." + "reason": "CLI env command regression and config type-checking break are separate code paths." }, { "left": "issue:44315", "right": "issue:44360", "accept": false, - "reason": "Liger Kernel/model_init application vs DSA indexer ReLU discussion; unrelated." + "reason": "Liger Kernel application in model_init and DSA indexer/ReLU discussion are unrelated." }, { "left": "issue:31795", "right": "issue:38617", "accept": false, - "reason": "Docs confusion in forward signature vs missing import symbol; no shared code path." + "reason": "Docs clarity issue vs ImportError for a missing symbol; different underlying problems." }, { "left": "issue:43576", "right": "issue:43824", "accept": false, - "reason": "Broken env command vs missing Qwen2_5_VL import; different package export issues." + "reason": "Broken env command and Qwen2_5_VL import failure are separate API/CLI failures." }, { "left": "issue:44961", "right": "issue:45468", "accept": false, - "reason": "Placeholder issue vs Gemma4 audio positional encoding bug; unrelated." + "reason": "Placeholder title and Gemma4 audio positional encoding bug are not duplicates." }, { "left": "issue:44485", "right": "issue:44961", "accept": false, - "reason": "RoPE implementation discussion vs unrelated placeholder issue." + "reason": "GLM-5 RoPE implementation discussion is unrelated to the placeholder issue." }, { "left": "issue:44351", "right": "issue:46617", "accept": false, - "reason": "Missing HybridCache import vs FileNotFoundError for trust_remote_code cache loading; both load-related but different defects." + "reason": "Top-level HybridCache import failure and trust_remote_code cache FileNotFoundError are different loading failures." }, { "left": "issue:44908", "right": "issue:44961", "accept": false, - "reason": "Scheduler kwargs bug vs unrelated placeholder issue." + "reason": "Scheduler kwargs handling and placeholder title have no overlap." }, { "left": "issue:43576", "right": "issue:44938", "accept": false, - "reason": "CLI/env issue vs Python 3.14 import failure; not the same bug." + "reason": "Broken env subcommand and Python 3.14 load failure are unrelated." }, { "left": "issue:36296", "right": "issue:43901", "accept": false, - "reason": "Tensor-parallel training bug vs docs mentioning return_all_scores; different domains." + "reason": "Tensor-parallel training bug and docs about return_all_scores are not the same issue." }, { "left": "issue:46143", "right": "issue:46144", "accept": false, - "reason": "Both involve RoFormer, but one is kwargs propagation and the other attention interface architecture; not a single concrete fix." + "reason": "Both RoFormer-related, but one is kwargs propagation and the other is attention interface usage; not one concrete bug." }, { "left": "issue:43475", "right": "issue:46617", "accept": false, - "reason": "SAM3 video encoder attribute error vs trust_remote_code cache FileNotFoundError; unrelated." + "reason": "SAM 3 Video missing attribute and cache FileNotFoundError are unrelated." }, { "left": "issue:44222", "right": "issue:44230", "accept": false, - "reason": "FP8 save_pretrained bug vs FP8 inference support request; different operations." + "reason": "Saving FP8 MoE models and inference support for FP8 MoE models are different code paths." }, { - "left": "issue:44315", - "right": "issue:46209", + "left": "issue:41628", + "right": "issue:42548", "accept": false, - "reason": "model_init/Liger Kernel issue vs FP8 inference on B200; unrelated." + "reason": "Different missing top-level imports; likely separate export regressions, not one duplicate." }, { - "left": "issue:41628", - "right": "issue:42548", + "left": "issue:44315", + "right": "issue:46209", "accept": false, - "reason": "AutoImageProcessor import error vs PreTrainedModel import error; separate export regressions." + "reason": "Liger Kernel application bug and FP8 inference on B200 are unrelated." }, { "left": "issue:44485", "right": "issue:46664", "accept": false, - "reason": "GLM-5 RoPE implementation vs SmolVLM num_frames semantics; unrelated model areas." + "reason": "RoPE implementation discussion and SmolVLM num_frames semantics are different model areas." }, { "left": "issue:43673", "right": "issue:44492", "accept": false, - "reason": "chunked_prefill cache bug vs cache-strategy typo; not same underlying issue." + "reason": "Generation cache missing in chunked_prefill and a cache-strategy typo do not describe the same defect." }, { "left": "issue:45362", "right": "issue:45850", "accept": false, - "reason": "Qwen3.5 chat crash vs broad 'v5 breaks too many models' complaint; too vague to merge." + "reason": "Specific Qwen3.5 chat crash vs a vague 'breaks too many models' complaint; not the same concrete bug." }, { "left": "issue:44273", "right": "issue:45542", "accept": false, - "reason": "lazy loading bug vs TensorBoard/TensorFlow backend error; unrelated." + "reason": "Lazy loading regression and TensorBoard-without-TensorFlow backend error are unrelated." }, { "left": "issue:44623", "right": "issue:45230", "accept": false, - "reason": "missing files on processor.save_pretrained vs empty bug report; no evidence of same defect." + "reason": "Missing save_pretrained files is specific, while the other is a generic bug report with no concrete overlap." }, { "left": "issue:44360", "right": "issue:46664", "accept": false, - "reason": "DSA ReLU discussion vs video num_frames semantics; unrelated." + "reason": "DSA indexer/ReLU discussion and SmolVLM video-frame semantics are unrelated." }, { "left": "issue:44038", "right": "issue:45125", "accept": false, - "reason": "Qwen3-VL-Moe loading bug vs missing _tp_plan for tensor parallelism; distinct issues." + "reason": "Qwen3-VL-Moe loading breakage and missing _tp_plan for tensor parallelism are separate model-loading vs parallelism bugs." }, { "left": "issue:44273", "right": "issue:45003", "accept": false, - "reason": "lazy loading bug vs unsafe sys.modules access; different code paths." + "reason": "Lazy loading problems and unsafe sys.modules access are different implementation defects." }, { "left": "issue:44393", "right": "issue:45071", "accept": false, - "reason": "Qwen3-VL bounding box output bug vs PreTrainedConfig type checking regression; unrelated." + "reason": "Qwen3-VL bounding-box output issue and PretrainedConfig type checking are unrelated." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Bounding-box output errors and infinite-loop behavior on a specific image are different failure modes." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Broken env command and checkpoint reload incompatibility are both version-related but not the same bug." } ] }, @@ -18201,7 +18251,6 @@ "issue:45341", "issue:45468", "issue:45478", - "issue:45753", "issue:45850", "issue:46095", "issue:46218", @@ -18210,8 +18259,6 @@ "issue:46894" ], "soft_pairs": [ - "issue:44393|issue:45753", - "issue:43576|issue:44556", "issue:41669|issue:43673", "issue:43525|issue:45850", "issue:44556|issue:45092", @@ -18253,272 +18300,257 @@ "issue:41669|issue:43704", "issue:31795|issue:34689" ], - "prepared_review_unit_hash": "2773e47550ef21c190e4535990d8f61d327d596f2228c969d3afcf4f7893ac79", + "prepared_review_unit_hash": "379b810dcfa87142a013e703a625fcfe2d3c8ae1e70344abb271b13e76ba8a70", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { - "node_count": 48, - "item_count": 48, - "soft_pair_count": 42, - "serialized_chars": 39690, - "estimated_input_tokens": 9923, - "estimated_eval_tokens": 20102 + "node_count": 47, + "item_count": 47, + "soft_pair_count": 40, + "serialized_chars": 38545, + "estimated_input_tokens": 9637, + "estimated_eval_tokens": 19530 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly loose v5-era breakage and import/load regressions, with a few genuinely related loading/import-lazy-loading pairs; most soft links are too broad to merge.", - "confidence": 0.58, - "canonical_issue_reason": "issue:45278 is the clearest broad umbrella report here: concrete upgrade breakage from 4.57.0 to 5.5.0 with many import failures, making it a better cluster anchor than the vaguer or highly model-specific reports.", + "summary": "This cluster is dominated by v5-era regressions and unrelated model/subsystem-specific bugs that only look similar at a high level; none of the soft edges appear to be the same concrete bug/change.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45850 is the broadest umbrella report in the set and best matches the general v5 regression theme, while the other issues are narrower, model-specific symptoms.", "canonical_pr_reason": null, - "best_issue_reason": "issue:45278 is the best single issue to surface because it is concrete, version-specific, and representative of the larger v5 compatibility theme without being as vague as issue:45850.", + "best_issue_reason": "issue:45850 is the most representative issue for this cluster because it summarizes the broad cross-model breakage pattern better than the more specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ - { - "left": "issue:44393", - "right": "issue:45753", - "accept": false, - "reason": "Same model family, but different concrete failures: bbox output issues vs an infinite loop on one image." - }, - { - "left": "issue:43576", - "right": "issue:44556", - "accept": false, - "reason": "Unrelated surfaces: the v5 env CLI vs checkpoint reload compatibility after upgrade." - }, { "left": "issue:41669", "right": "issue:43673", "accept": false, - "reason": "Import-time cleanup/performance regression is not the same bug as missing GenerationMixin cache in chunked prefill." + "reason": "One is import-star/lazy-loading performance, the other is a generation cache regression; not the same code-path bug." }, { "left": "issue:43525", "right": "issue:45850", "accept": false, - "reason": "A missing config attribute for Llama4 is a specific model bug, not the same underlying problem as a broad 'v5 breaks too many models' report." + "reason": "Specific missing pad_token_id in Llama4Config versus a broad v5 breakage complaint; too generic to be a duplicate." }, { "left": "issue:44556", "right": "issue:45092", - "accept": true, - "reason": "Both describe v5-era checkpoint loading incompatibility tied to initialization/loading behavior, and could plausibly be fixed by the same code path." + "accept": false, + "reason": "Both involve loading failures after v5 upgrades, but they affect different checkpoint types and likely different causes." }, { "left": "issue:43572", "right": "issue:45850", "accept": false, - "reason": "StableLm pad_token_idx regression is a distinct model config issue, not the same as the generic v5 breakage complaint." + "reason": "StableLmConfig pad_token_idx regression is a concrete config bug, not the same underlying issue as the broad v5-breakage report." }, { "left": "issue:44492", "right": "issue:45850", "accept": false, - "reason": "A cache-strategy typo is not the same underlying bug as broad version-wide breakage." + "reason": "A typo in cache strategies is unrelated to a general claim that v5 breaks many models." }, { "left": "issue:45278", "right": "issue:45850", "accept": false, - "reason": "Both are broad upgrade complaints, but they do not identify the same concrete failure or fixable code path." + "reason": "General import errors after upgrade are broader, but this is still a separate report from the umbrella complaint." }, { "left": "issue:44230", "right": "issue:46095", "accept": false, - "reason": "One is fp8 inference support for Qwen3-VL/Qwen3.5, the other asks about a weight conversion script for DeepSeekV4; different changes." + "reason": "FP8 inference support for Qwen models is a different feature request from a DeepSeek weight-conversion question." }, { "left": "issue:42371", "right": "issue:44492", "accept": false, - "reason": "TF32 API guidance and a cache-strategy typo are unrelated." + "reason": "TF32 API guidance and a cache-strategy typo are unrelated issues." }, { "left": "issue:41669", "right": "issue:46218", "accept": false, - "reason": "Import surface/performance changes are not the same as failing to import Qwen3_5ForCausalLM." + "reason": "Import-star cleanup/performance and a missing Qwen3_5 module import error are not the same bug." }, { "left": "issue:43408", "right": "issue:44315", "accept": false, - "reason": "Model-type mismatch warning vs Liger kernel application under model_init are different problems." + "reason": "Wrong model-type warning versus Liger Kernel not being applied in model_init; different code paths." }, { "left": "issue:42890", "right": "issue:44908", "accept": false, - "reason": "Test seeding flakiness and scheduler kwargs handling are unrelated." + "reason": "Test flakiness from missing set_seed and a scheduler kwarg bug are unrelated." }, { "left": "issue:44246", "right": "issue:44273", - "accept": true, - "reason": "Both point to lazy-loading/import-path behavior being broken or ineffective, which plausibly explains the import slowdown and functional lazy-loading issue." + "accept": false, + "reason": "Both touch import/lazy-loading performance, but one is about slow import and the other claims lazy loading is broken; not clearly the same defect." }, { "left": "issue:43576", "right": "issue:44908", "accept": false, - "reason": "The env command and inverse_sqrt scheduler kwargs are unrelated." + "reason": "Broken `transformers env` command and inverse_sqrt scheduler kwargs are unrelated." }, { "left": "issue:43352", "right": "issue:44393", "accept": false, - "reason": "Flash Attention 2 support gating is unrelated to Qwen3-VL bounding-box hallucination." + "reason": "Flash Attention 2.0 support gating and Qwen3-VL bounding-box output errors are different problems." }, { "left": "issue:43976", "right": "issue:44393", "accept": false, - "reason": "Python version compatibility is unrelated to Qwen3-VL output errors." + "reason": "Python version compatibility and Qwen3-VL output errors are unrelated." }, { "left": "issue:44908", "right": "issue:45071", "accept": false, - "reason": "Scheduler kwargs handling is unrelated to PretrainedConfig type-checking regressions." + "reason": "Scheduler kwarg handling and PretrainedConfig type-checking regressions are different code paths." }, { "left": "issue:44295", "right": "issue:45230", "accept": false, - "reason": "A position_ids buffer read error is not the same as a generic bug report." + "reason": "Position_ids buffer access issue versus a generic bug report; no concrete match." }, { "left": "issue:44297", "right": "issue:45230", "accept": false, - "reason": "Tokenizer save_pretrained mismatch is a specific serialization bug, not the generic bug report." + "reason": "Tokenizer save_pretrained mismatch is a specific serialization bug, not the same as a generic bug report." }, { "left": "issue:44704", "right": "issue:44908", "accept": false, - "reason": "AutoProcessor cached_file kwargs forwarding and scheduler kwargs handling are different code paths." + "reason": "AutoProcessor kwargs forwarding and scheduler kwargs are unrelated subsystems." }, { "left": "issue:44492", "right": "issue:46328", "accept": false, - "reason": "A typo in cache-strategy text is unrelated to SequenceFeatureExtractor.pad performance." + "reason": "Cache strategy typo and SequenceFeatureExtractor.pad performance are unrelated." }, { "left": "issue:41669", "right": "issue:44351", "accept": false, - "reason": "Broad import-time cleanup and a HybridCache import failure are different import regressions." + "reason": "Import-star cleanup/performance does not match a missing HybridCache import error." }, { "left": "issue:41669", "right": "issue:45341", "accept": false, - "reason": "Import-surface refactoring is unrelated to a testing_utils bug." + "reason": "Model import cleanup and a testing_utils bug are unrelated." }, { "left": "issue:43976", "right": "issue:45850", "accept": false, - "reason": "Python support/version constraints are not the same as generic model breakage." + "reason": "Python compatibility issues are not the same as the broad 'many models break' complaint." }, { "left": "issue:44393", "right": "issue:45478", "accept": false, - "reason": "Different Qwen3.5-related failures: hallucination/error on a specific output vs from_pretrained error." + "reason": "Both mention Qwen3.5/Qwen3-VL, but one is bbox hallucination and the other is from_pretrained failure; different defects." }, { "left": "issue:43976", "right": "issue:46218", "accept": false, - "reason": "Python version compatibility is unrelated to a specific missing module import." + "reason": "Python version compatibility and missing Qwen3_5 module import are unrelated." }, { "left": "issue:44077", "right": "issue:44861", "accept": false, - "reason": "A post_init policy issue is not the same bug as tied-weight key handling." + "reason": "Optional post_init policy and an AttributeError in tied-weight key handling are different bugs." }, { "left": "issue:31795", "right": "issue:33453", "accept": false, - "reason": "Documentation confusion and tokenizer loading regression are unrelated." + "reason": "Documentation confusion in model.forward versus tokenizer loading regression are unrelated." }, { "left": "issue:43576", "right": "issue:44273", "accept": false, - "reason": "The env CLI bug does not match the lazy-loading bug." + "reason": "Broken env CLI and lazy-loading issues are too different to merge as duplicates." }, { "left": "issue:44315", "right": "issue:44829", "accept": false, - "reason": "Liger-kernel application under model_init and flash_attention_3 training degeneration are different problems." + "reason": "Liger Kernel not being applied in model_init and flash_attention_3 training degeneration are different failures." }, { "left": "issue:44360", "right": "issue:45468", "accept": false, - "reason": "DSA indexer/ReLU discussion and Gemma4 audio positional encoding are unrelated." + "reason": "DSA indexer activation discussion and Gemma-4 audio positional encoding are unrelated." }, { - "left": "issue:45230", - "right": "issue:45310", + "left": "issue:43519", + "right": "issue:46655", "accept": false, - "reason": "Generic bug-report framing is not enough to treat this as the same issue as the Qwen3.5 MoE from_pretrained error." + "reason": "Incorrect timestamp calculation and incorrect padding calculation are both processor bugs but on different modalities and code paths." }, { - "left": "issue:44295", - "right": "issue:45230", + "left": "issue:44273", + "right": "issue:44908", "accept": false, - "reason": "Position_ids buffer access is not the same as the generic bug report." + "reason": "Lazy-loading failure and scheduler kwarg handling are unrelated." }, { - "left": "issue:44297", - "right": "issue:45230", + "left": "issue:44655", + "right": "issue:44861", "accept": false, - "reason": "Tokenizer save_pretrained mismatch is not the same as the generic bug report." + "reason": "Pipeline save_pretrained failure and tied-weight key AttributeError are different bugs." }, { - "left": "issue:45468", - "right": "issue:46894", + "left": "issue:45230", + "right": "issue:45310", "accept": false, - "reason": "Gemma-4 audio positional encoding and CohereASR training loss shifting are unrelated model bugs." + "reason": "Generic bug report versus a specific Qwen3.5 MoE from_pretrained regression." }, { "left": "issue:43519", - "right": "issue:46655", + "right": "issue:44485", "accept": false, - "reason": "Both are video-processor related, but timestamp calculation and padding calculation are different concrete bugs." + "reason": "Qwen3VL timestamp handling and GLM-5 RoPE implementation are not the same bug." }, { - "left": "issue:43519", - "right": "issue:44908", + "left": "issue:45468", + "right": "issue:46894", "accept": false, - "reason": "Qwen3VL timestamp math and scheduler kwargs handling are unrelated." + "reason": "Gemma-4 audio positional encoding and CohereASR double-shift training loss are different model-specific problems." } ] }, - "evaluator_result": { - "accept": true, - "feedback": "Grounded overall. issue:45278 is a reasonable cluster anchor because it is the clearest concrete v5 upgrade regression in the packet. The rejected links stay conservative, and the two accepted links are appropriately hedged as plausible same-path loading/lazy-loading regressions rather than asserted as certain duplicates. No major overstatement of 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 }, @@ -18586,8 +18618,8 @@ "issue:46018|issue:46172", "issue:43704|issue:44485", "issue:44485|issue:45468", - "issue:42371|issue:43704", "issue:43502|issue:43519", + "issue:42371|issue:43704", "issue:45850|issue:46218", "issue:43519|issue:44623", "issue:44263|issue:45468", @@ -18628,14 +18660,14 @@ "issue:37934|issue:43668", "issue:42936|issue:43668" ], - "prepared_review_unit_hash": "8037d579e2ef110d7c11d82c927e1dc8e2c1375f023d687fc6cc91a82282b607", + "prepared_review_unit_hash": "936027c0d61717483ac8f2016545a69b310a858626fc055eb658f579c2d94da9", "component_budget": { "node_count": 500, "item_count": 500, "soft_pair_count": 1839, - "serialized_chars": 953865, - "estimated_input_tokens": 238467, - "estimated_eval_tokens": 477190 + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 }, "budget": { "node_count": 46, @@ -18650,276 +18682,276 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "The cluster is heterogeneous and mostly contains unrelated model-specific regressions, import/load failures, and tokenizer/v5 complaints. None of the soft-edge pairs look like true duplicates; they are mostly same-subsystem or same-version similarity rather than the same concrete bug.", - "confidence": 0.66, - "canonical_issue_reason": "Issue 45850 is the closest umbrella representative because it captures the shared \"5.0 breaks many models\" theme that recurs across several nearby reports, even though the set is not a clean duplicate cluster.", + "summary": "The cluster is largely heterogeneous. A recurring theme is Flash Attention 2.0 / version-compatibility regressions, but the items still point to different models or different failure modes, so I would not merge any soft pair as a duplicate.", + "confidence": 0.61, + "canonical_issue_reason": "issue:45250 is the broadest and most representative artifact in the set, since several nearby items relate to Flash Attention 2.0 support or related upgrade breakage. Even so, the cluster is not cohesive enough to treat as a duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45850 is the best single representative of the cluster because it is broad enough to summarize the common v5-regression theme without pretending the narrower reports are the same bug.", + "best_issue_reason": "issue:45250 is the best single issue to keep as the cluster representative because it captures the only clear recurring theme and is broader than the model-specific or symptom-specific reports.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46018", "right": "issue:46172", "accept": false, - "reason": "Same model family, but different concrete failures: shared-expert gating vs a dtype/transpose bug in DeepSeek-V4." + "reason": "Same model family, but different bugs: expert gating logic vs a tensor transpose/cast issue." }, { "left": "issue:43704", "right": "issue:44485", "accept": false, - "reason": "Both are model implementation bugs, but VRAM leakage in Qwen3 is unrelated to GLM-5 RoPE logic." + "reason": "Different models and different code paths; VRAM leakage is unrelated to GLM-5 RoPE behavior." }, { "left": "issue:44485", "right": "issue:45468", "accept": false, - "reason": "Different models and different code paths: GLM-5 RoPE implementation versus Gemma-4 audio positional encoding." + "reason": "Both are model-specific architecture issues, but they concern different components and symptoms." }, { - "left": "issue:42371", - "right": "issue:43704", + "left": "issue:43502", + "right": "issue:43519", "accept": false, - "reason": "TF32 API settings and a Qwen3 VRAM leak are unrelated issues." + "reason": "One is a local_files_only network request bug; the other is a Qwen3VL timestamp calculation issue." }, { - "left": "issue:43502", - "right": "issue:43519", + "left": "issue:42371", + "right": "issue:43704", "accept": false, - "reason": "Both are runtime bugs, but local_files_only network leakage and Qwen3VL timestamp calculation are separate defects." + "reason": "TF32 API/deprecation guidance is unrelated to a multi-thread VRAM leak." }, { "left": "issue:45850", "right": "issue:46218", "accept": false, - "reason": "One is a broad version-5 breakage complaint; the other is a specific import failure for Qwen3_5ForCausalLM." + "reason": "Both are version-related regressions, but one is broad model breakage and the other is a specific import failure." }, { "left": "issue:43519", "right": "issue:44623", "accept": false, - "reason": "Processor timestamp math and save_pretrained missing files are different processor bugs." + "reason": "Both involve processors, but the failures are in different methods and unrelated behaviors." }, { "left": "issue:44263", "right": "issue:45468", "accept": false, - "reason": "torch.split return handling in GlmMoeDsaIndexer is unrelated to Gemma-4 audio positional encoding." + "reason": "Different model internals and different bug classes; no shared code-path evidence." }, { "left": "issue:44261", "right": "issue:45468", "accept": false, - "reason": "MLA q_a_layernorm epsilon precision issue is not the same bug as Gemma-4 audio positional encoding." + "reason": "Different model-specific numerical/configuration issue versus Gemma-4 audio encoding behavior." }, { "left": "issue:43298", "right": "issue:46095", "accept": false, - "reason": "Audio continuous batching support and DeepSeek-V4 weight conversion are unrelated feature requests." + "reason": "Audio continuous batching support and DeepSeek-V4 weight conversion are unrelated problems." }, { "left": "issue:43519", "right": "issue:43704", "accept": false, - "reason": "Qwen3VL timestamp calculation and Qwen3 VRAM leakage are different bugs in different paths." + "reason": "Timestamp computation in a processor is unrelated to VRAM leakage across dataloader threads." }, { "left": "issue:44351", "right": "issue:46218", "accept": false, - "reason": "Both are import errors, but they affect different modules/models and are not the same missing-symbol problem." + "reason": "Both are import errors, but they reference different missing symbols/modules and likely different fixes." }, { "left": "issue:43352", "right": "issue:46218", "accept": false, - "reason": "Flash Attention 2.0 support for NemotronH and a Qwen3_5 import failure are unrelated." + "reason": "Flash Attention support and an import failure for Qwen3_5ForCausalLM are not the same underlying issue." }, { "left": "issue:42913", "right": "issue:43066", "accept": false, - "reason": "Both mention tokenizer v5 regressions, but one is a broad behavior change and the other is a decoder-type mismatch; not enough to call them the same bug." + "reason": "Both are tokenizer v5-related, but one is a behavior regression and the other is a wrong decoder type bug." }, { "left": "issue:28348", "right": "issue:45250", "accept": false, - "reason": "GPT2-specific Flash Attention 2.0 support is narrower than the generic Flash Attention 2.0 request and does not clearly describe the same defect." + "reason": "Related on Flash Attention 2.0, but 28348 is GPT2LMHeadModel-specific while 45250 is only a generic request." }, { "left": "issue:44908", "right": "issue:44998", "accept": false, - "reason": "Scheduler kwargs handling and an unrelated placeholder issue title are not duplicates." + "reason": "A scheduler kwargs bug is unrelated to an unrelated miscellaneous issue." }, { "left": "issue:44492", "right": "issue:44998", "accept": false, - "reason": "A cache-strategy typo is unrelated to the placeholder/unemployment title." + "reason": "Cache-strategy typo and the unrelated issue title do not match any underlying bug." }, { "left": "issue:43618", "right": "issue:44998", "accept": false, - "reason": "CLIPOutput attentions assignment regression is unrelated to the placeholder title." + "reason": "CLIPOutput attentions regression is unrelated to the other issue." }, { "left": "issue:44998", "right": "issue:45850", "accept": false, - "reason": "The placeholder title is not the same as the broad v5 regression report." + "reason": "No shared bug or code path; one is a placeholder/misc title and the other is a broad compatibility complaint." }, { "left": "issue:44938", "right": "issue:44998", "accept": false, - "reason": "Python 3.14 loading failure is unrelated to the placeholder title." + "reason": "Python 3.14 import failure is unrelated to the other issue." }, { "left": "issue:44861", "right": "issue:45290", "accept": false, - "reason": "A tied-weights helper crash and a chat-template tool-call crash are different code paths." + "reason": "Different crash sites: tied-weights handling versus chat template/tool-call serialization." }, { "left": "issue:31515", "right": "issue:38617", "accept": false, - "reason": "Checkpoint load slowness and a missing import symbol are unrelated." + "reason": "Checkpoint loading performance and a missing import are unrelated." }, { "left": "issue:43572", "right": "issue:44861", "accept": false, - "reason": "Missing pad_token_idx and tied-weights helper crashes are different issues." + "reason": "Missing pad_token_idx after v5 update is unrelated to list handling in tied weights." }, { "left": "issue:43525", "right": "issue:44861", "accept": false, - "reason": "Llama4Config pad_token_id absence is not the same as the tied-weights list crash." + "reason": "Missing pad_token_id on Llama4Config is a different config-attribute bug." }, { "left": "issue:28348", "right": "issue:43352", "accept": false, - "reason": "Both concern Flash Attention 2.0, but they target different models and different support gaps." + "reason": "Both mention Flash Attention 2.0, but they target different models and likely different support paths." }, { "left": "issue:44230", "right": "issue:44623", "accept": false, - "reason": "FP8 inference support and missing processor save files are unrelated." + "reason": "FP8 inference support and missing processor.save_pretrained files are unrelated." }, { "left": "issue:29870", "right": "issue:31515", "accept": false, - "reason": "BART generate() decoder_attention_mask bug and checkpoint loading slowness are unrelated." + "reason": "BART generate() UnboundLocalError is unrelated to slow checkpoint loading." }, { "left": "issue:30333", "right": "issue:31515", "accept": false, - "reason": "MLFlow finalization on failure and checkpoint load slowness are separate problems." + "reason": "MLflow reporting on failed jobs is unrelated to from_pretrained performance." }, { "left": "issue:31515", "right": "issue:36296", "accept": false, - "reason": "from_pretrained slowness and tensor-parallel training bugs do not share the same concrete failure." + "reason": "Checkpoint load slowness and tensor parallel training bugs are different problems." }, { "left": "issue:31515", "right": "issue:31795", "accept": false, - "reason": "Loading checkpoint slowness and documentation confusion are different kinds of issues." + "reason": "Performance on loading checkpoints is unrelated to documentation confusion in forward()." }, { "left": "issue:44230", "right": "issue:45310", "accept": false, - "reason": "FP8 support request and Qwen3.5 MoE from_pretrained error are unrelated." + "reason": "Both touch Qwen3.5 MoE, but one is fp8 support and the other is a from_pretrained error; not the same fix." }, { "left": "issue:43298", "right": "issue:44230", "accept": false, - "reason": "Audio continuous batching support and FP8 inference support are different feature requests." + "reason": "Audio continuous batching and Qwen3-VL/Qwen3.5 FP8 inference support are different features." }, { "left": "issue:43717", "right": "issue:46144", "accept": false, - "reason": "Mamba init_weights usage and RoFormer attention interface usage are unrelated architecture issues." + "reason": "Mamba init_weights behavior and RoFormer attention interface usage are unrelated." }, { "left": "issue:42936", "right": "issue:46018", "accept": false, - "reason": "Mask2Former ignore_value handling and DeepSeek-V4 gating are unrelated." + "reason": "Mask2Former ignore_value handling is unrelated to DeepSeek-V4 gating." }, { "left": "issue:43668", "right": "issue:44031", "accept": false, - "reason": "A config type-hint bug and a tokenizer regex warning are different defects." + "reason": "A config type hint issue and a tokenizer regex warning are not the same bug." }, { "left": "issue:42936", "right": "issue:43240", "accept": false, - "reason": "Ignore_value not used and kwargs not passed to loss calculation are different bugs." + "reason": "Different subsystems and different symptoms; no shared code path." }, { "left": "issue:43240", "right": "issue:46144", "accept": false, - "reason": "Loss kwargs propagation and RoFormer attention interface usage are unrelated." + "reason": "Loss kwargs propagation is unrelated to RoFormer attention implementation." }, { "left": "issue:42936", "right": "issue:44031", "accept": false, - "reason": "Mask2Former ignore_value handling is unrelated to the tokenizer regex warning." + "reason": "Mask2Former ignore_value behavior and tokenizer regex warnings are unrelated." }, { "left": "issue:42936", "right": "issue:46144", "accept": false, - "reason": "Mask2Former ignore_value handling and RoFormer attention interface usage are different bugs." + "reason": "Different model families and different bug types." }, { "left": "issue:42936", "right": "issue:43813", "accept": false, - "reason": "Mask2Former ignore_value handling is unrelated to the quantization-operation typo." + "reason": "A Mask2Former bug and a typo in quantization conversion naming are unrelated." }, { "left": "issue:43240", "right": "issue:43668", "accept": false, - "reason": "Passing kwargs to loss calculation and a ModernBERT type-hint mistake are unrelated." + "reason": "Loss-function kwargs and a ModernBERT type-hint problem are unrelated." }, { "left": "issue:43668", "right": "issue:43813", "accept": false, - "reason": "ModernBERT norm_eps typing and the quantization-operation typo are separate issues." + "reason": "Type-hint correctness and quantization-operation typo are different issues." }, { "left": "issue:37934", "right": "issue:43668", "accept": false, - "reason": "Llama4TextL2Norm normalization semantics and a ModernBERT type-hint error are not the same bug." + "reason": "Llama4 normalization semantics and ModernBERT type hints do not describe the same change." }, { "left": "issue:42936", "right": "issue:43668", "accept": false, - "reason": "Mask2Former ignore_value and ModernBERT norm_eps typing are unrelated." + "reason": "Mask2Former ignore_value handling is unrelated to ModernBERT config typing." } ] }, @@ -18952,52 +18984,52 @@ "issue:10519|issue:11005", "issue:11005|issue:11307" ], - "prepared_review_unit_hash": "cda7107b460078b8f23b0f480d4528f6659ad7065e46b168bccf65b1cecb0562", + "prepared_review_unit_hash": "eb7c71b70f2b1831364dfe97eb6da3ef168ca8e810465b610ee3df3f24f280cc", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2877, - "estimated_input_tokens": 720, - "estimated_eval_tokens": 1696 + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 2877, - "estimated_input_tokens": 720, - "estimated_eval_tokens": 1696 + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is a false positive: the three issues cover unrelated feature requests in different parts of the project, so none should be merged as duplicates.", - "confidence": 0.98, - "canonical_issue_reason": "No canonical issue: issue #10519 (truncation direction), #11005 (ReduceLROnPlateau-like functionality), and #11307 (Wav2Vec2 word time offsets) are unrelated topics.", + "summary": "The cluster is not a duplicate set: one issue is about tokenizer truncation direction, one about ReduceLROnPlateau-style scheduling, and one about Wav2Vec2 word time offsets. They only share vague ML/NLP similarity, not the same underlying problem.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue: the three issues describe unrelated feature requests in different parts of the library.", "canonical_pr_reason": null, - "best_issue_reason": "No single issue is a good canonical representative because the cluster is not a duplicate set; the overlap is only superficial wording similarity.", + "best_issue_reason": "If one issue must anchor the cluster, #11307 is the strongest representative by discussion activity, but it is not a duplicate of the others.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:10519", "right": "issue:11307", "accept": false, - "reason": "Different feature areas and code paths: token truncation behavior vs word boundary timing in Wav2Vec2. No evidence they describe the same bug/change." + "reason": "Tokenizer truncation from the beginning and Wav2Vec2 word offset extraction are unrelated feature requests affecting different code paths." }, { "left": "issue:10519", "right": "issue:11005", "accept": false, - "reason": "Unrelated requests: truncation option vs learning-rate scheduler functionality. Similarity is lexical only." + "reason": "One request is about truncation behavior in tokenization; the other is about learning-rate scheduler behavior. No shared underlying bug." }, { "left": "issue:11005", "right": "issue:11307", "accept": false, - "reason": "Different domains and expected fixes: scheduler behavior vs audio model timestamp extraction. Not the same underlying issue." + "reason": "ReduceLROnPlateau-like functionality and Wav2Vec2 token timing offsets are separate feature requests with no concrete overlap." } ] }, @@ -19029,12 +19061,12 @@ "issue:13244|issue:16618", "issue:15980|issue:16618" ], - "prepared_review_unit_hash": "5e9226a192edcf741df924ac88eb0e35d80da688ecd3818247962478f2f56bf0", + "prepared_review_unit_hash": "362a78a1970b535576e8eb50b88b506d0932ca74d5a1c3a88c4d9256231f0264", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2478, + "serialized_chars": 2477, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -19042,7 +19074,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 2478, + "serialized_chars": 2477, "estimated_input_tokens": 620, "estimated_eval_tokens": 1496 }, @@ -19051,24 +19083,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate issue reports: one about Tapas tokenization mismatch, one about a bad auth-related download error message, and one about a generic model-loading failure. They do not look like the same underlying bug.", - "confidence": 0.82, - "canonical_issue_reason": "Issue #16618 is the broadest and most central of the three because it concerns a core model-loading failure, whereas #15980 is specifically about an unauthenticated private-model download error message and #13244 is about Tapas tokenization behavior.", + "summary": "These three closed issues are not duplicates of one another: one is about TAPAS tokenization mismatching TensorFlow, one is about a misleading auth/download error for private models, and one is about failing to load bert-base-uncased. The soft similarities are superficial and do not indicate the same underlying bug.", + "confidence": 0.89, + "canonical_issue_reason": "Issue #16618 is the most representative user-facing failure in the set, but it is still distinct from the other two issues.", "canonical_pr_reason": null, - "best_issue_reason": "#16618 is the best umbrella issue for the cluster since it captures the general 'can't load model' symptom. #15980 is narrower and more about error handling/message quality, and #13244 is unrelated to model loading.", + "best_issue_reason": "#16618 is the clearest, most concrete model-loading problem and the best single representative if one issue must be chosen.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:13244", "right": "issue:16618", "accept": false, - "reason": "Different subsystems and failure modes: Tapas tokenization behavior versus generic model loading. No indication they describe the same bug." + "reason": "Different problem areas: TAPAS tokenization parity vs generic model loading for bert-base-uncased. No shared code-path or same bug." }, { "left": "issue:15980", "right": "issue:16618", "accept": false, - "reason": "Both involve model access/loading, but #15980 is specifically about the error shown when downloading a private model while logged out, not the same concrete failure as the generic inability to load bert-base-uncased." + "reason": "Both mention loading models, but #15980 is specifically an authentication/error-message issue for private models, while #16618 is a separate load failure for a public model." } ] }, @@ -19098,7 +19130,7 @@ "soft_pairs": [ "issue:26350|issue:33260" ], - "prepared_review_unit_hash": "59e647f1ca3432f99d82f752c795a9fb249837582471165213199f1d12293f43", + "prepared_review_unit_hash": "87ddbecc582afcac93f2ffe8cedb1f50ab30f59170e1cb4bd01ccb54c1cc1122", "component_budget": { "node_count": 2, "item_count": 2, @@ -19120,18 +19152,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two separate community contribution feature requests: one for Flash Attention 2 support across more architectures, and one for GGUF support across more architectures. They overlap only in being broad architecture-support proposals, not in the underlying change.", - "confidence": 0.98, - "canonical_issue_reason": "No canonical issue: the requests target different features and different code paths.", + "summary": "These are both community-contribution feature requests about expanding architecture support, but they target different technologies: Flash Attention 2 vs GGUF. They are related at a high level only and should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "Issue #26350 is the stronger representative because it is older and has substantially higher discussion activity, indicating broader engagement around that request.", "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a duplicate candidate for the other; both should remain separate tracking items.", + "best_issue_reason": "Issue #26350 is the best single issue to represent the cluster because it is the more prominent and more established feature request, even though the two issues are not duplicates.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:26350", "right": "issue:33260", "accept": false, - "reason": "Different requested capabilities (Flash Attention 2 vs GGUF) and no shared concrete bug or implementation path." + "reason": "Both are feature requests for broader architecture support, but they concern different concrete additions and code paths: Flash Attention 2 support versus GGUF support. They are not the same underlying change." } ] }, @@ -19190,184 +19222,184 @@ "issue:44370|issue:46226", "issue:46222|issue:46226" ], - "prepared_review_unit_hash": "08185f0eea802173391c3b7133dafab89fa4cf186e12536ea6d8aac1974f727a", + "prepared_review_unit_hash": "4525b0bb5a4776a6532f9c825c997c3ca58e7d1d76dfa846d54f8201badd0dce", "component_budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13527, - "estimated_input_tokens": 3382, - "estimated_eval_tokens": 7020 + "serialized_chars": 13530, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 }, "budget": { "node_count": 8, "item_count": 8, "soft_pair_count": 24, - "serialized_chars": 13527, - "estimated_input_tokens": 3382, - "estimated_eval_tokens": 7020 + "serialized_chars": 13530, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The cluster is mostly a mix of separate i18n translation-request issues for different languages, but there are two exact-duplicate subgroups: the placeholder-title issues (43834/44365/44370) and the Thai-title pair (46222/46226).", - "confidence": 0.84, - "canonical_issue_reason": "issue:43834 is the best canonical representative of the exact duplicate placeholder-title subgroup: it is the earliest of the three and has two near-identical matching peers.", + "summary": "This cluster is mostly i18n docs-translation requests, but they split into separate language-specific tasks. Only the exact duplicate placeholder-title issues and the identical Thai-titled pair look like true duplicates; the Turkish/Spanish/Romanian requests are distinct issues, not the same bug/change.", + "confidence": 0.77, + "canonical_issue_reason": "Use issue 43834 as the canonical representative of the exact-duplicate placeholder-template subgroup: it matches the other placeholder-title issues closely and is the cleanest anchor among the true duplicates.", "canonical_pr_reason": null, - "best_issue_reason": "issue:43834 is the strongest global issue anchor because it belongs to the largest exact-duplicate subgroup in this cluster; the other language-specific translation requests are distinct topics.", + "best_issue_reason": "43834 is the best issue to anchor deduplication for the placeholder-template copies, but it should not be treated as canonical for the language-specific translation requests.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:27088", "right": "issue:44370", "accept": false, - "reason": "Different language-specific translation requests; same broad i18n theme is not enough to treat them as duplicates." + "reason": "Different target languages; these are separate translation requests, not the same underlying issue." }, { "left": "issue:27088", "right": "issue:44365", "accept": false, - "reason": "Different language-specific translation requests; only the boilerplate/template looks similar." + "reason": "Turkish vs. generic placeholder issue; not the same concrete task." }, { "left": "issue:27088", "right": "issue:43834", "accept": false, - "reason": "Turkish translation request vs generic placeholder request; not the same concrete issue." + "reason": "Specific Turkish translation request does not duplicate the generic placeholder-template issue." }, { "left": "issue:27088", "right": "issue:38435", "accept": false, - "reason": "Different target languages (Turkish vs Romanian), so these are separate requests." + "reason": "Different language translation work; distinct issues." }, { "left": "issue:27088", "right": "issue:28936", "accept": false, - "reason": "Turkish vs Spanish translation request; same subsystem, different issue." + "reason": "Turkish and Spanish translation requests are separate issues." }, { "left": "issue:28936", "right": "issue:43834", "accept": false, - "reason": "Spanish translation request vs generic placeholder request; not the same underlying task." + "reason": "Spanish translation request is not the same as the generic placeholder issue." }, { "left": "issue:28936", "right": "issue:44365", "accept": false, - "reason": "Spanish request and placeholder-title issue are not the same concrete translation task." + "reason": "Different language target and different issue intent; not duplicates." }, { "left": "issue:28936", "right": "issue:38435", "accept": false, - "reason": "Different languages (Spanish vs Romanian); these are distinct requests." + "reason": "Spanish vs. Romanian translation requests are distinct tasks." }, { "left": "issue:28936", "right": "issue:44370", "accept": false, - "reason": "Spanish request vs placeholder-title issue; similarity is only at the template level." + "reason": "Spanish translation request does not match the placeholder-template issue." }, { "left": "issue:38435", "right": "issue:44370", "accept": false, - "reason": "Romanian translation request is distinct from the generic placeholder issue." + "reason": "Romanian translation request and placeholder-template copy are not the same underlying issue." }, { "left": "issue:38435", "right": "issue:44365", "accept": false, - "reason": "Romanian request vs placeholder-title issue; not the same specific change." + "reason": "Romanian request is distinct from the generic placeholder issue." }, { "left": "issue:38435", "right": "issue:43834", "accept": false, - "reason": "Romanian translation request and generic placeholder request are different issue texts/tasks." + "reason": "Concrete Romanian translation issue is not a duplicate of the placeholder-template issue." }, { "left": "issue:38435", "right": "issue:46222", "accept": false, - "reason": "Romanian translation request vs Thai-titled request; different languages, different issues." + "reason": "Different translation targets and different issue content; not duplicates." }, { "left": "issue:38435", "right": "issue:46226", "accept": false, - "reason": "Romanian translation request vs Thai-titled request; not duplicates." + "reason": "Romanian translation request is unrelated to the Thai-titled issue pair." }, { "left": "issue:43834", "right": "issue:44370", "accept": true, - "reason": "Same generic placeholder title and near-identical issue metadata/body shape; likely the same underlying request." + "reason": "Same placeholder-title translation issue; appears to be an exact duplicate." }, { "left": "issue:43834", "right": "issue:44365", "accept": true, - "reason": "Same placeholder-title translation request; appears to be an exact duplicate of the same issue." + "reason": "Identical placeholder-template translation issue with no substantive difference." }, { "left": "issue:43834", "right": "issue:46222", "accept": false, - "reason": "Placeholder-title issue vs Thai-titled issue; these are not the same request." + "reason": "Only superficially similar template text; different title/content, so not the same issue." }, { "left": "issue:43834", "right": "issue:46226", "accept": false, - "reason": "Placeholder-title issue vs Thai-titled issue; only superficial similarity." + "reason": "Different title/content from the placeholder-template issue; reject as duplicate." }, { "left": "issue:44365", "right": "issue:44370", "accept": true, - "reason": "Same exact placeholder title and very similar metadata; this looks like a duplicate submission." + "reason": "Same exact placeholder-title issue; these look like duplicate copies." }, { "left": "issue:44365", "right": "issue:46222", "accept": false, - "reason": "Different titles and likely different language requests; not the same issue." + "reason": "Different issue text and title; not the same underlying request." }, { "left": "issue:44365", "right": "issue:46226", "accept": false, - "reason": "Placeholder-title issue vs Thai-titled issue; not a duplicate pair." + "reason": "Not the same issue; only the placeholder-template subgroup matches." }, { "left": "issue:44370", "right": "issue:46222", "accept": false, - "reason": "Different language/request text; no evidence they describe the same underlying issue." + "reason": "Different title/content and different underlying translation request." }, { "left": "issue:44370", "right": "issue:46226", "accept": false, - "reason": "Placeholder-title issue vs Thai-titled issue; different requests." + "reason": "Not a duplicate; the Thai-titled issue is a separate request." }, { "left": "issue:46222", "right": "issue:46226", "accept": true, - "reason": "Same Thai title and matching metadata/body shape; these appear to be duplicate issues for the same request." + "reason": "Exact same Thai title and likely same body; clear duplicate pair." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet: the duplicate claims are supported by identical titles, near-identical body lengths, and very high soft-similarity scores, and the rejected cross-language pairs are handled conservatively. No clear overstatement beyond what the metadata supports." + "feedback": "The assessment is grounded in the packet and stays conservative. The exact-duplicate conclusions are limited to the identical placeholder-title subgroup and the identical Thai-titled pair, while the Turkish/Spanish/Romanian issues are appropriately treated as separate translation requests. No soft-edge verdict appears to overstate the evidence." }, "evaluator_used": true, "retried": false, @@ -19386,8 +19418,9 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-36817-18", + "cluster_id": "cluster-36718-19", "nodes": [ + "issue:36718", "issue:36817", "issue:36879", "issue:37015", @@ -19408,6 +19441,7 @@ "issue:46653" ], "soft_pairs": [ + "issue:36718|issue:41115", "issue:36817|issue:41211", "issue:36817|issue:43671", "issue:36817|issue:36879", @@ -19448,268 +19482,274 @@ "issue:42503|issue:46653", "issue:45027|issue:46653" ], - "prepared_review_unit_hash": "988ac2365026ccf38a8ec47682c6c7af167d7d9445924a67fa81a70f9e511543", + "prepared_review_unit_hash": "745a0fae795c8824cbe0959fc0c0b0fa62d3f68d5db81025b9a9a4960476a025", "component_budget": { - "node_count": 18, - "item_count": 18, - "soft_pair_count": 39, - "serialized_chars": 23786, - "estimated_input_tokens": 5947, - "estimated_eval_tokens": 12150 + "node_count": 19, + "item_count": 19, + "soft_pair_count": 40, + "serialized_chars": 24673, + "estimated_input_tokens": 6169, + "estimated_eval_tokens": 12594 }, "budget": { - "node_count": 18, - "item_count": 18, - "soft_pair_count": 39, - "serialized_chars": 23786, - "estimated_input_tokens": 5947, - "estimated_eval_tokens": 12150 + "node_count": 19, + "item_count": 19, + "soft_pair_count": 40, + "serialized_chars": 24673, + "estimated_input_tokens": 6169, + "estimated_eval_tokens": 12594 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are mostly separate model-addition/support requests that share a broad template, but they do not describe the same underlying bug or change. No soft pairs look mergeable as duplicates.", - "confidence": 0.91, - "canonical_issue_reason": "issue:46133 is the strongest representative: it is detailed, discussion-heavy, and has a clearly scoped model request.", + "summary": "This cluster is a loose topical grouping of model-addition/support requests, not a set of true duplicates. The soft-similarity links are too broad to merge: the issues span different model families and tasks (BERT, DETR, audio ASR/TTS, OCR, vision-language, sparse transformers).", + "confidence": 0.88, + "canonical_issue_reason": "issue:43837 is the most central representative of the cluster: a concrete model-support request with the broadest topical overlap to the other model-addition issues, especially the audio/model-support requests.", "canonical_pr_reason": null, - "best_issue_reason": "issue:46133 is the best overall anchor for this cluster because it is the most active and well-defined request among otherwise unrelated model additions.", + "best_issue_reason": "issue:43837 is the best issue-level representative for this cluster because it is specific, current, and sits near the densest area of soft similarity without collapsing distinct requests into one duplicate.", "best_pr_reason": null, "soft_edge_verdicts": [ + { + "left": "issue:36718", + "right": "issue:41115", + "accept": false, + "reason": "Different concrete requests in different domains: keypoint matching vs a model architecture addition." + }, { "left": "issue:36817", "right": "issue:41211", "accept": false, - "reason": "Different model requests (EuroBert vs DEIMv2); same broad add-model pattern only." + "reason": "Both are model requests, but for different architectures and code paths; not the same change." }, { "left": "issue:36817", "right": "issue:43671", "accept": false, - "reason": "EuroBert config support and Qwen3-TTS are unrelated model features." + "reason": "EuroBert config support and Qwen3-TTS support are unrelated model additions." }, { "left": "issue:36817", "right": "issue:36879", "accept": false, - "reason": "Different models entirely (EuroBert vs RF-DETR), not the same change." + "reason": "Different model families (EuroBert vs RF-DETR); no shared underlying fix." }, { "left": "issue:36879", "right": "issue:41211", "accept": false, - "reason": "RF-DETR and DEIMv2 are separate model additions with no shared code-path shown." + "reason": "RF-DETR and DEIMv2 are distinct model integrations, not one duplicate request." }, { "left": "issue:36879", "right": "issue:43909", "accept": false, - "reason": "Object detection model request vs audio model request; not duplicates." + "reason": "Different models and modalities; separate implementation work." }, { "left": "issue:36879", "right": "issue:44121", "accept": false, - "reason": "RF-DETR and a sparse-transformer request are distinct model targets." + "reason": "RF-DETR model support is unrelated to a sparse-transformer model request." }, { "left": "issue:36879", "right": "issue:43837", "accept": false, - "reason": "RF-DETR and Qwen3-ASR are different model families and tasks." + "reason": "Different models and tasks (detection vs ASR); not mergeable as one PR/issue." }, { "left": "issue:36879", "right": "issue:43671", "accept": false, - "reason": "RF-DETR and Qwen3-TTS are unrelated additions." + "reason": "RF-DETR and Qwen3-TTS are unrelated model requests." }, { "left": "issue:37015", "right": "issue:43837", "accept": false, - "reason": "NeoBERT and Qwen3-ASR are unrelated model requests." + "reason": "NeoBERT and Qwen3-ASR are distinct model-support requests." }, { "left": "issue:37015", "right": "issue:44121", "accept": false, - "reason": "NeoBERT support and sparse transformer support are not the same concrete change." + "reason": "Different model families with no shared concrete code-path problem." }, { "left": "issue:37015", "right": "issue:43909", "accept": false, - "reason": "NeoBERT and LFM2.5 Audio are different model types and code paths." + "reason": "NeoBERT and LFM2.5 Audio are unrelated model additions." }, { "left": "issue:36879", "right": "issue:37015", "accept": false, - "reason": "RF-DETR and NeoBERT are unrelated model additions." + "reason": "RF-DETR and NeoBERT are separate model integrations." }, { "left": "issue:37015", "right": "issue:42503", "accept": false, - "reason": "NeoBERT and ModernVBERT are separate model requests, not one duplicate bug." + "reason": "NeoBERT and ModernVBERT are different model requests, even if both are BERT-like." }, { "left": "issue:41115", "right": "issue:43837", "accept": false, - "reason": "MiniCPM3 and Qwen3-ASR are different model support requests." + "reason": "MiniCPM3 architecture work and Qwen3-ASR support are different models and tasks." }, { "left": "issue:41115", "right": "issue:44121", "accept": false, - "reason": "MiniCPM3 and sparse-transformer support are distinct changes." + "reason": "MiniCPM3 architecture and sparse-transformer support are unrelated changes." }, { "left": "issue:41115", "right": "issue:42503", "accept": false, - "reason": "MiniCPM3 and ModernVBERT are separate model-add requests." + "reason": "Different model families; not the same underlying addition." }, { "left": "issue:41115", "right": "issue:45027", "accept": false, - "reason": "MiniCPM3 and Voxtral TTS support are unrelated model targets." + "reason": "Model architecture work and Voxtral TTS support are unrelated." }, { "left": "issue:41211", "right": "issue:43671", "accept": false, - "reason": "DEIMv2 and Qwen3-TTS are different requests with no shared underlying fix." + "reason": "DEIMv2 and Qwen3-TTS are different model requests, not one duplicate." }, { "left": "issue:42503", "right": "issue:43837", "accept": false, - "reason": "ModernVBERT and Qwen3-ASR are unrelated model support requests." + "reason": "ModernVBERT and Qwen3-ASR are distinct model-support requests." }, { "left": "issue:42503", "right": "issue:43671", "accept": false, - "reason": "ModernVBERT and Qwen3-TTS are different model additions." + "reason": "ModernVBERT and Qwen3-TTS are unrelated model additions." }, { "left": "issue:42503", "right": "issue:42738", "accept": false, - "reason": "ModernVBERT and generic BERT-like RoPE support are related only at a broad conceptual level, not the same change." + "reason": "ModernVBERT models and BERT-like models with RoPE are related only at a broad theme level, not the same concrete change." }, { "left": "issue:42738", "right": "issue:45027", "accept": false, - "reason": "BERT-like RoPE support and Voxtral TTS are unrelated requests." + "reason": "BERT-like RoPE support and Voxtral TTS support are different tasks and code paths." }, { "left": "issue:42738", "right": "issue:46653", "accept": false, - "reason": "BERT-like RoPE support and a person-named issue are not the same artifact type or change." + "reason": "The named request and the person-only issue title are clearly unrelated." }, { "left": "issue:43671", "right": "issue:43837", "accept": false, - "reason": "Qwen3-TTS and Qwen3-ASR are sibling requests, but they are different modalities and not one mergeable PR/issue." + "reason": "Related product line, but TTS vs ASR are distinct model-support requests with different implementations." }, { "left": "issue:43671", "right": "issue:45027", "accept": false, - "reason": "Qwen3-TTS and Voxtral TTS are different model support requests." + "reason": "Qwen3-TTS and Voxtral TTS are different models; same modality is not enough to merge." }, { "left": "issue:43671", "right": "issue:45177", "accept": false, - "reason": "TTS support and DeepSeek-OCR2 are unrelated model requests." + "reason": "Qwen3-TTS support and DeepSeek-OCR2 are unrelated model additions." }, { "left": "issue:43671", "right": "issue:44121", "accept": false, - "reason": "Qwen3-TTS and sparse-transformer support are not the same code-path problem." + "reason": "Audio TTS support and sparse-transformer support are unrelated changes." }, { "left": "issue:43837", "right": "issue:43909", "accept": false, - "reason": "Qwen3-ASR and LFM2.5 Audio are both audio-related but target different models and changes." + "reason": "Both audio-related, but ASR and audio model support are different concrete requests." }, { "left": "issue:43837", "right": "issue:45322", "accept": false, - "reason": "Qwen3-ASR and EUPE are unrelated model requests." + "reason": "Qwen3-ASR and EUPE are different models and different code paths." }, { "left": "issue:43837", "right": "issue:44121", "accept": false, - "reason": "ASR model support and sparse-transformer support are distinct changes." + "reason": "ASR support and sparse-transformer support are unrelated." }, { "left": "issue:43909", "right": "issue:44205", "accept": false, - "reason": "LFM2.5 Audio and SAM3-LiteText are separate model additions." + "reason": "LFM2.5 Audio and SAM3-LiteText are different model families/tasks." }, { "left": "issue:43909", "right": "issue:46645", "accept": false, - "reason": "Audio model support and a Granite attention option are different concrete changes." + "reason": "Audio model support and a Granite SWA+sink option are not the same change." }, { "left": "issue:44121", "right": "issue:45322", "accept": false, - "reason": "Sparse-transformer support and EUPE are different model requests." + "reason": "Sparse-transformer support and EUPE are distinct model requests." }, { "left": "issue:43837", "right": "issue:45027", "accept": false, - "reason": "Qwen3-ASR and Voxtral TTS are different speech-model support requests." + "reason": "Different TTS models; similar modality does not make them duplicates." }, { "left": "issue:37015", "right": "issue:45027", "accept": false, - "reason": "NeoBERT and Voxtral TTS are unrelated model requests." + "reason": "NeoBERT and Voxtral TTS are unrelated model additions." }, { "left": "issue:43837", "right": "issue:45177", "accept": false, - "reason": "Qwen3-ASR and DeepSeek-OCR2 are different model tasks and code paths." + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different models and tasks." }, { "left": "issue:45322", "right": "issue:46133", "accept": false, - "reason": "EUPE and TIPSv2 are separate vision-language model requests, not duplicates." + "reason": "EUPE and TIPSv2 are both vision-language related, but they are distinct model requests." }, { "left": "issue:42503", "right": "issue:46653", "accept": false, - "reason": "ModernVBERT and the unrelated closed issue 'Mirko Privitera' are not the same change." + "reason": "ModernVBERT and the person-only issue title are clearly unrelated." }, { "left": "issue:45027", "right": "issue:46653", "accept": false, - "reason": "Voxtral TTS support and the unrelated closed issue 'Mirko Privitera' do not match." + "reason": "Voxtral TTS support and the person-only issue title are unrelated." } ] }, @@ -19739,7 +19779,7 @@ "soft_pairs": [ "issue:38740|issue:46443" ], - "prepared_review_unit_hash": "24af7d7f8f916129276ac332b0be86ac3b3a6e4ff083afbdf813833ec6595451", + "prepared_review_unit_hash": "cf2c65dadb356864ff08fc2d3e4998b4f0fd5aa75e99e07dfee6b32ed7a6f393", "component_budget": { "node_count": 2, "item_count": 2, @@ -19761,18 +19801,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are unrelated: one is a docs request for adding `pruna` as an optimization framework, while the other is a technical refactor for Vision RoPE. They do not describe the same underlying problem or change.", - "confidence": 0.96, + "summary": "These issues are unrelated: one is a docs request about adding `pruna` as an optimization framework, while the other is a Vision RoPE refactor. No duplicate relationship is supported.", + "confidence": 0.98, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #46443 is the more substantive technical issue and is the better candidate to keep as the representative issue if one must be chosen.", + "best_issue_reason": "Issue 46443 is the more substantive technical item and is the better representative issue; issue 38740 is a standalone documentation request and not part of the same bug/change.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:38740", "right": "issue:46443", "accept": false, - "reason": "No meaningful overlap in problem statement, code path, or requested change; a docs addition request is not the same as a Vision RoPE refactor." + "reason": "Different subjects and scopes: documentation for `pruna` vs a Vision RoPE refactor. No shared files, targets, or concrete code-path overlap." } ] }, @@ -19802,7 +19842,7 @@ "soft_pairs": [ "issue:41084|issue:43037" ], - "prepared_review_unit_hash": "4082abf6af0732d849ecf600a2f1a8c48bb1018506ab50ca8c94ae5d142093ae", + "prepared_review_unit_hash": "c67c095a5a7911926c7488eb42e352a34c375b0dbe1d35bce9675fdf420324c4", "component_budget": { "node_count": 2, "item_count": 2, @@ -19824,7 +19864,7 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one asks about Set Block Decoding, while the other requests DeepSeek v3.2 support. They describe different feature areas and no shared issue target or code path is evident.", + "summary": "The two issues are unrelated: one is about block decoding configuration, the other is about adding DeepSeek v3.2 model support. They do not describe the same underlying bug or feature request.", "confidence": 0.97, "canonical_issue_reason": null, "canonical_pr_reason": null, @@ -19835,7 +19875,7 @@ "left": "issue:41084", "right": "issue:43037", "accept": false, - "reason": "The titles and likely problem domains are unrelated: block decoding behavior versus model support. No shared target, filename, or concrete code-path overlap is shown, so this is not a duplicate." + "reason": "Different problems and likely different code areas; no shared target, filename, or concrete symptom indicating a duplicate." } ] }, @@ -19865,40 +19905,40 @@ "soft_pairs": [ "issue:41306|issue:41826" ], - "prepared_review_unit_hash": "99625f2433eb3eb776a8e565605c7f51f2155ce2995f76f375bb2f01bb8444cd", + "prepared_review_unit_hash": "a49ccbd33ff88f5b7b1704092a6a7922040fc475d59099f5ca363fdfbbfe55e8", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, - "estimated_input_tokens": 396, - "estimated_eval_tokens": 1048 + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1584, - "estimated_input_tokens": 396, - "estimated_eval_tokens": 1048 + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one requests a `from_pretrained()` flag for TiledFusedLogitsLoss / LigerFusedLinearCrossEntropy, while the other proposes integrating TiledMLP for lower memory use. They target different model components and different changes.", - "confidence": 0.98, - "canonical_issue_reason": "Issue 41306 is the more concrete, narrowly scoped request, but it is not a duplicate of 41826.", + "summary": "These two issues are about different model components and different requested changes: one is about exposing a TiledFusedLogitsLoss/FLCE flag in `from_pretrained()`, the other is about integrating TiledMLP for lower memory use. They do not look like the same underlying bug or feature request.", + "confidence": 0.97, + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "If one must be chosen as the cluster representative, 41306 is more actionable and specific; however, the two issues should remain separate.", + "best_issue_reason": "No canonical issue in this pair; if forced to pick one, #41306 is more concrete and actionable, but it is not a duplicate of #41826.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:41306", "right": "issue:41826", "accept": false, - "reason": "Different features and code paths: one is about a logits-loss integration flag in `from_pretrained()`, the other is about adding TiledMLP for memory reduction. They do not look like the same underlying bug or change." + "reason": "Different requested functionality and code paths: FLCE/from_pretrained configuration versus TiledMLP memory-footprint integration. No strong evidence they describe the same change." } ] }, @@ -19928,7 +19968,7 @@ "soft_pairs": [ "issue:43598|issue:43600" ], - "prepared_review_unit_hash": "9e68edd57401fadca934f0f7952bf58891d20a884a1558375a3dbf6c76f33796", + "prepared_review_unit_hash": "facbe6c96bc6c51317ca93f3842260167c1293019191aefd0f3fb11e14e1ae4f", "component_budget": { "node_count": 2, "item_count": 2, @@ -19950,18 +19990,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues both mention custom trainers, but they describe different problems: one is about `remove_unused_column` dropping needed inputs, the other is about FLOPS estimation logic for custom trainers. They are related in theme but not duplicates.", - "confidence": 0.94, + "summary": "These are both Trainer-related, but they describe different problems: one is about `remove_unused_column` hurting customizability, the other is about FLOPS estimation for custom trainers. They do not look like the same underlying bug.", + "confidence": 0.93, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue 43598 is the broader and more concrete custom-trainer compatibility concern; issue 43600 is a separate FLOPS-accounting problem.", + "best_issue_reason": "No single canonical issue stands out; if forced to pick a representative, #43598 is the broader Trainer customizability topic, but it does not cover #43600.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43598", "right": "issue:43600", "accept": false, - "reason": "Shared mention of custom trainers is too broad; one concerns input-column pruning, the other concerns FLOPS estimation, so they are not the same underlying bug." + "reason": "Different concrete issues: `remove_unused_column` behavior vs FLOPS accounting for custom trainers. Shared Trainer context is too broad to treat as a duplicate." } ] }, @@ -19991,12 +20031,12 @@ "soft_pairs": [ "issue:43684|issue:44405" ], - "prepared_review_unit_hash": "b354738058770ba10cdb397c0a3ea7b912028c6988b7bcb777bcd4acec5c5e9b", + "prepared_review_unit_hash": "d29f07799638e464074169bebbdc292b22c5f2ff81ceb202126ca096b8e7a83b", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -20004,7 +20044,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1611, + "serialized_chars": 1612, "estimated_input_tokens": 403, "estimated_eval_tokens": 1062 }, @@ -20013,18 +20053,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are related only at a very broad Qwen/auto-model level, but they request different features for different model families and different AutoModel heads, so they are not duplicates.", + "summary": "These are two different feature requests for different Qwen model families and different auto-class behaviors, so they should not be deduplicated.", "confidence": 0.98, - "canonical_issue_reason": "Issue 43684 is the broader registration request (Qwen3-Omni plus AutoModel/AutoModelForConditionalGeneration), so it is the better representative of this small cluster, but it is still distinct from the Qwen3.5 sequence-classification request.", + "canonical_issue_reason": "Issue 43684 is the broader and more central request: it asks for model registration in AutoModel and AutoModelForConditionalGeneration for Qwen3-Omni.", "canonical_pr_reason": null, - "best_issue_reason": "43684 is the most general issue in scope and best represents the auto-registration theme, though it does not subsume the separate Qwen3.5 AutoModelForSequenceClassification request.", + "best_issue_reason": "Issue 43684 is the better representative issue because it targets core model registration rather than a narrower task-specific auto-class addition.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43684", "right": "issue:44405", "accept": false, - "reason": "Different model families (Qwen3-Omni vs Qwen3.5) and different target heads (ConditionalGeneration vs SequenceClassification); not the same underlying bug/change." + "reason": "Different underlying changes: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 sequence classification support for a different config and auto class." } ] }, @@ -20061,7 +20101,7 @@ "issue:43845|issue:43859", "issue:43846|issue:43859" ], - "prepared_review_unit_hash": "7d39f703fb5f42fb7c87be55d436f5b10f5e5929e7d08a08651b211a1e2dd936", + "prepared_review_unit_hash": "f342c57353ea723913f76f5a1080b9fb68fce35083e93bb526fb5b3f2c6239c2", "component_budget": { "node_count": 4, "item_count": 4, @@ -20083,48 +20123,48 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is made up of very low-information issues with generic/duplicated titles and no concrete bug details. They do not provide enough evidence to confirm true duplicates, but they look like the same spammy or placeholder topic group.", - "confidence": 0.94, - "canonical_issue_reason": "Issue 43859 is the best representative only because it has the only discussion activity; otherwise the issues are too sparse to distinguish meaningfully.", + "summary": "The cluster is dominated by three very sparse issues titled \"huggingface \" and one unrelated issue titled \"Fett\". There is not enough concrete description, linkage, or target information to confirm any true duplicate bug, so all soft edges should be rejected.", + "confidence": 0.31, + "canonical_issue_reason": "If one issue must represent the cluster, 43845 is the best placeholder because it is one of the identical \"huggingface \" reports and the earliest of that subgroup, but the evidence is too thin to treat it as a proven duplicate set.", "canonical_pr_reason": null, - "best_issue_reason": "43859 is the strongest issue to keep as canonical because it appears slightly more active and the cluster contains no technical detail to justify a different choice.", + "best_issue_reason": "43845 is the best overall issue representative since it matches the repeated \"huggingface \" pattern and is earliest among the near-identical entries; 43835 appears unrelated by title.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:43835", "right": "issue:43845", "accept": false, - "reason": "Both are too vague and lack enough body detail to prove the same underlying issue." + "reason": "Different titles and no shared concrete problem details; only weak text similarity." }, { "left": "issue:43835", "right": "issue:43846", "accept": false, - "reason": "Same generic title pattern, but no concrete bug description or target is available to establish duplication." + "reason": "Different titles and no substantive evidence they describe the same bug." }, { "left": "issue:43835", "right": "issue:43859", "accept": false, - "reason": "Insufficient evidence beyond superficial similarity; cannot confirm the same underlying issue." + "reason": "Different titles with no shared code-path or issue target evidence." }, { "left": "issue:43845", "right": "issue:43846", "accept": false, - "reason": "The titles are identical, but the bodies are too sparse to verify that they describe the same bug or request." + "reason": "Identical generic title, but the bodies are too sparse to confirm the same underlying issue." }, { "left": "issue:43845", "right": "issue:43859", "accept": false, - "reason": "Too little detail to confirm these are duplicates rather than separate placeholder reports." + "reason": "Same vague title, but no concrete details, targets, or links proving duplication." }, { "left": "issue:43846", "right": "issue:43859", "accept": false, - "reason": "No concrete problem statement is available, so the match cannot be accepted as the same issue." + "reason": "Same vague title, but insufficient evidence to conclude they are the same issue." } ] }, @@ -20165,88 +20205,88 @@ "issue:44572|issue:44937", "issue:44573|issue:44937" ], - "prepared_review_unit_hash": "bc401ca08100242925dd5d871976caac2c5afab781dc5925dd14678c896eb8e5", + "prepared_review_unit_hash": "537206ca7e6a754c64abbe20f7e9046cb60cefb461e93a3c1285ededec7aa0d8", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5788, - "estimated_input_tokens": 1447, - "estimated_eval_tokens": 3150 + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 }, "budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 9, - "serialized_chars": 5788, - "estimated_input_tokens": 1447, - "estimated_eval_tokens": 3150 + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster looks like mostly low-signal spam/placeholder issues with no concrete bug report or shared technical detail. The items do not provide enough evidence to treat them as duplicates of the same underlying problem.", - "confidence": 0.19, - "canonical_issue_reason": "Issue 44418 is the only item with a somewhat descriptive title, but it is still extremely vague and lacks any technical details; it is the least-bad anchor if one must be chosen.", + "summary": "This cluster looks like unrelated low-content/spam issues rather than a real duplicate set. The only semi-descriptive item is a documentation suggestion, but there is no evidence that any pair refers to the same underlying bug or request.", + "confidence": 0.16, + "canonical_issue_reason": "Issue 44418 is the least noisy and most descriptive item in the set, so it is the best available representative if one must be chosen.", "canonical_pr_reason": null, - "best_issue_reason": "44418 is marginally the most suitable issue to represent the cluster because it is the only non-placeholder title, but the overall cluster is too noisy to be a meaningful canonical issue set.", + "best_issue_reason": "No issue here is a strong canonical candidate because the cluster lacks substantive bodies, targets, filenames, or shared problem details; 44418 is only the least-bad representative.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:44371", "right": "issue:44572", "accept": false, - "reason": "Both are placeholder spam titles with no body content or technical context; no evidence they describe the same bug." + "reason": "Both are just '' placeholders with no substantive content; identical titles alone are not enough to prove the same underlying issue." }, { "left": "issue:44371", "right": "issue:44573", "accept": false, - "reason": "Both are placeholder spam titles with no body content or technical context; no evidence they describe the same bug." + "reason": "Both are spam-like placeholders and provide no concrete bug or request details to establish duplication." }, { "left": "issue:44371", "right": "issue:44418", "accept": false, - "reason": "A spam placeholder and a vague documentation suggestion are not enough to establish the same underlying issue." + "reason": "One is spam and the other is a generic documentation suggestion; they do not describe the same problem." }, { "left": "issue:44371", "right": "issue:44937", "accept": false, - "reason": "The titles are unrelated and there is no shared technical description or target." + "reason": "A spam placeholder and a random named title do not indicate the same underlying issue." }, { "left": "issue:44418", "right": "issue:44572", "accept": false, - "reason": "A vague documentation suggestion and a spam placeholder do not indicate the same bug or change." + "reason": "Generic documentation feedback is unrelated to a '' placeholder." }, { "left": "issue:44418", "right": "issue:44573", "accept": false, - "reason": "A vague documentation suggestion and a spam placeholder do not indicate the same bug or change." + "reason": "Generic documentation feedback is unrelated to a '' placeholder." }, { "left": "issue:44572", "right": "issue:44573", "accept": false, - "reason": "They are both placeholder spam issues, but there is no substantive content showing they refer to the same underlying problem." + "reason": "Although the titles match, there is no body or evidence showing the same concrete bug/request; this looks like duplicate spam, not a confirmed duplicate issue." }, { "left": "issue:44572", "right": "issue:44937", "accept": false, - "reason": "No overlap in topic or technical details; one is spam-like and the other is a generic brand-name title." + "reason": "A spam placeholder and a named title about 'Google LLC' are not the same underlying issue." }, { "left": "issue:44573", "right": "issue:44937", "accept": false, - "reason": "No overlap in topic or technical details; one is spam-like and the other is a generic brand-name title." + "reason": "A spam placeholder and a named title about 'Google LLC' are not the same underlying issue." } ] }, @@ -20276,7 +20316,7 @@ "soft_pairs": [ "issue:45145|issue:45151" ], - "prepared_review_unit_hash": "d800bb2ea87f294900f526fd58c0244757ebfb2b6229fe6292c9d827f1cf0fcb", + "prepared_review_unit_hash": "a6b3bc795eca430edfda6785191100a046569d7af085578aeef65362bf5afe4b", "component_budget": { "node_count": 2, "item_count": 2, @@ -20298,24 +20338,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two closed issues are highly similar and likely related, but the evidence is not strong enough to confidently mark them as duplicates based on the packet alone. Identical titles and similar body sizes suggest a duplicate cluster, but there is no body-content or link evidence establishing the same underlying proposal.", - "confidence": 0.78, - "canonical_issue_reason": "Issue #45145 is the better canonical representative because it was created earlier and has an inbound reference, but the duplicate relationship is only tentative from the available signals.", + "summary": "The pair looks like a strong near-duplicate match: identical titles, very similar body sizes, and no distinguishing links, files, or explicit targets. But the available evidence still doesn\u2019t prove they are the same underlying issue, so this should remain a non-deterministic/low-confidence duplicate candidate rather than a definite merge.", + "confidence": 0.68, + "canonical_issue_reason": "Issue #45145 is the better canonical placeholder because it is slightly earlier and has one inbound reference, making it the more established thread without claiming definitive duplication.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #45145 is the best cluster representative due to earliest creation and slightly better linkage; #45151 does not add a clearly distinct problem statement, though the duplicate status is not fully proven.", + "best_issue_reason": "Issue #45145 is the best representative for the cluster because it was created first and has marginally more reference activity; the evidence is otherwise symmetric.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45145", "right": "issue:45151", "accept": false, - "reason": "Rejecting as a confirmed duplicate: the packet only shows identical titles, similar body lengths, and soft similarity. That is not enough to establish the same underlying energy-reduction proposal without body-content evidence or explicit linkage." + "reason": "Identical titles and similar body length are suggestive, but there is not enough concrete evidence to conclude these are the same underlying issue. With no links, files, or explicit targets, this should be treated as a possible duplicate rather than accepted as one." } ] }, "evaluator_result": { "accept": false, - "feedback": "The clustering is plausible, but the soft-edge verdict overstates the evidence. The packet only shows identical titles, similar body lengths, and a soft-similarity signal; it does not establish the \u201csame underlying energy-reduction proposal\u201d or that the pair should be confidently accepted as duplicates. Please downgrade the verdict to tentative/conservative unless body-content evidence is available." + "feedback": "The conclusion is directionally reasonable, but it overstates the evidence. The packet supports identical titles, close body lengths, and no links/files/targets, but not \"identical subject matter\" or that they are definitely duplicates. The soft pair is explicitly non-deterministic, so the accept verdict is too strong for the available evidence." }, "evaluator_used": true, "retried": true, @@ -20396,40 +20436,40 @@ "soft_pairs": [ "issue:45762|issue:45763" ], - "prepared_review_unit_hash": "08798f53e1eded28245cef46b73d3569c4eb89a3ba5862744a7bfb3f45dab0a4", + "prepared_review_unit_hash": "66d668db9c22dce970cac0f46e4f173f1ba34c4f7eda38e8cd3d64c9e27c0037", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1476, - "estimated_input_tokens": 369, - "estimated_eval_tokens": 994 + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1476, - "estimated_input_tokens": 369, - "estimated_eval_tokens": 994 + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues do not appear to be duplicates or part of the same bug report; they are unrelated name/date entries with no shared problem description.", + "summary": "The two issues are not duplicates: they refer to different people and different dates, with no shared bug report or code change context.", "confidence": 0.98, - "canonical_issue_reason": "No canonical issue: the titles reference different people and dates, with no overlap in subject matter or any supporting discussion/content.", + "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Neither issue is a good representative duplicate target because they do not describe the same underlying problem.", + "best_issue_reason": "No single canonical issue here; these are unrelated one-off issues with different titles and no overlapping content.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45762", "right": "issue:45763", "accept": false, - "reason": "The issues are unrelated; the titles identify different individuals and dates, and there is no evidence of a shared bug or request." + "reason": "Different named subjects and dates; no evidence they describe the same underlying problem or request." } ] }, @@ -20459,7 +20499,7 @@ "soft_pairs": [ "issue:45764|issue:45765" ], - "prepared_review_unit_hash": "34e9bbf2578bb85540b84b334dcd62d9c12dedc2b5774f693d5ed7da31df62db", + "prepared_review_unit_hash": "69c7cc4c5cde221d5be92fd140435795f924b1e326a30abb302226778fc90d27", "component_budget": { "node_count": 2, "item_count": 2, @@ -20481,18 +20521,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues do not describe the same bug or change; they look like unrelated low-signal placeholder/spam reports containing only names and dates.", + "summary": "These two closed issues are unrelated personal-name/date entries and do not describe the same bug or request. The soft similarity is superficial only, so the pair should not be merged as duplicates.", "confidence": 0.98, - "canonical_issue_reason": null, + "canonical_issue_reason": "No canonical issue: neither item contains actionable problem details, and the titles are different personal-name/date strings rather than duplicate bug reports.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 45764 is only the earliest item, but neither issue is a meaningful canonical duplicate target because they are unrelated and lack bug-report content.", + "best_issue_reason": "Neither issue is a good canonical representative for a duplicate cluster because both appear non-technical and unrelated.", "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 bug, code path, or actionable problem; this is not a duplicate pair." + "reason": "Different names and dates with no shared technical subject, target, or problem statement; not the same underlying issue." } ] }, @@ -20522,12 +20562,12 @@ "soft_pairs": [ "issue:45810|issue:46295" ], - "prepared_review_unit_hash": "b6fa603c0a7e73a13301dd69aae3fed0a5884006321f940a75646a5ef97a192c", + "prepared_review_unit_hash": "2a19dd98194cfbc824c3fdaf1f95865712819ace28f3822e08a31961d0f50a11", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1541, + "serialized_chars": 1542, "estimated_input_tokens": 386, "estimated_eval_tokens": 1028 }, @@ -20535,7 +20575,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1541, + "serialized_chars": 1542, "estimated_input_tokens": 386, "estimated_eval_tokens": 1028 }, @@ -20544,18 +20584,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are not duplicates: one requests a new Qwen3_5 token-classification/value-model head, while the other asks for FP8 training support. They concern different features and code paths.", - "confidence": 0.98, + "summary": "The two issues are not duplicates: one requests adding a Qwen3_5 token-classification model class for value-model use, while the other asks for FP8 training support. They concern different code paths and different kinds of functionality.", + "confidence": 0.95, "canonical_issue_reason": null, "canonical_pr_reason": null, - "best_issue_reason": "Issue #45810 is the more concrete feature request and would be the better representative if one had to be chosen, but it is not a duplicate of #46295.", + "best_issue_reason": "issue 45810 is the more concrete and self-contained request, with a specific model-class addition target, so it is the better anchor for this cluster even though it is unrelated to the FP8 training request.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45810", "right": "issue:46295", "accept": false, - "reason": "Different underlying requests: model/task head support vs FP8 training support. They do not describe the same bug or change." + "reason": "No shared bug or change: model-class support for token classification/value models is unrelated to FP8 training support, and the titles/bodies do not indicate the same underlying work." } ] }, @@ -20595,12 +20635,12 @@ "issue:45871|issue:45873", "issue:45872|issue:45873" ], - "prepared_review_unit_hash": "df96a3b7e7edeb72f4123f2ef988f1e93a8a4f75989444eb1c29d83650a556c8", + "prepared_review_unit_hash": "ba3470b9831d79898ba249f4cbe80f78f6b8c5a87e2289c4fed58efdd1761dd7", "component_budget": { "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5727, + "serialized_chars": 5728, "estimated_input_tokens": 1432, "estimated_eval_tokens": 3120 }, @@ -20608,7 +20648,7 @@ "node_count": 5, "item_count": 5, "soft_pair_count": 8, - "serialized_chars": 5727, + "serialized_chars": 5728, "estimated_input_tokens": 1432, "estimated_eval_tokens": 3120 }, @@ -20617,66 +20657,66 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This set is really two separate exact-duplicate pairs plus one related-but-distinct issue: 45870/45871 are duplicates, 45872/45873 are duplicates, and 45869 is a different dataset-promo issue that only shares the financial/DataForge theme.", - "confidence": 0.93, - "canonical_issue_reason": "45870 is the cleanest representative of the 45870/45871 duplicate pair: same title/body signal as 45871, slightly more activity, and it is the earlier of the exact matches.", + "summary": "There are two clear duplicate issue pairs: 45870/45871 and 45872/45873. Issue 45869 is a separate dataset-promo request and does not match the DataForge example/service issues.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45870 is a clean canonical representative for the exact-duplicate 'Example: Financial time series prediction with DataForge' pair; it is fully redundant with 45871.", "canonical_pr_reason": null, - "best_issue_reason": "45870 is the best overall issue to keep as the canonical thread for the exact-duplicate 'Example: Financial time series prediction with DataForge' pair.", + "best_issue_reason": "Issue 45872 is the strongest standalone issue to anchor because its title is the most specific and it also has an exact duplicate (45873), making it the clearest representative of a distinct topic.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:45869", "right": "issue:45870", "accept": false, - "reason": "Same broad financial theme, but the titles describe different artifacts: a dataset bundle vs. an example workflow." + "reason": "Different topics: dataset bundle promo vs. a DataForge prediction example." }, { "left": "issue:45869", "right": "issue:45871", "accept": false, - "reason": "Different underlying request; this is a dataset listing, not the same example issue." + "reason": "Different topics: dataset bundle promo vs. a DataForge prediction example." }, { "left": "issue:45870", "right": "issue:45871", "accept": true, - "reason": "Exact duplicate pair: identical title/body length and same example request." + "reason": "Exact duplicate titles and matching metadata/body size indicate the same issue." }, { "left": "issue:45870", "right": "issue:45872", "accept": false, - "reason": "Both mention Financial/DataForge, but one is an example and the other is a scraping-service topic; not the same issue." + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." }, { "left": "issue:45870", "right": "issue:45873", "accept": false, - "reason": "Related subsystem, but different concrete request: example prediction vs. scraping service." + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." }, { "left": "issue:45871", "right": "issue:45872", "accept": false, - "reason": "Only topical overlap; they do not appear to describe the same underlying bug/change." + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." }, { "left": "issue:45871", "right": "issue:45873", "accept": false, - "reason": "Different issue topic and different requested change." + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." }, { "left": "issue:45872", "right": "issue:45873", "accept": true, - "reason": "Exact duplicate pair: identical title/body length and same DataForge scraping-service request." + "reason": "Exact duplicate titles and matching metadata/body size indicate the same issue." } ] }, "evaluator_result": { "accept": true, - "feedback": "The clustering is grounded in the packet and the conservative non-duplicate calls are reasonable. The two accepted duplicate pairs are well supported by identical titles, identical body lengths, and jaccard 1.0, and 45870 as the canonical representative is justified by earlier creation time and slightly more activity." + "feedback": "The summary is grounded in the packet and stays conservative. The exact-duplicate calls for 45870/45871 and 45872/45873 are supported by identical titles, identical body lengths, and jaccard 1.0. The rejection of 45869 against the DataForge issues is also appropriately cautious based on the title mismatch and topic difference. No soft-edge verdict appears overstated." }, "evaluator_used": true, "retried": false, @@ -20703,49 +20743,46 @@ "soft_pairs": [ "issue:46439|issue:46441" ], - "prepared_review_unit_hash": "35763dc147bcfa9fe451f518618cd892ef301daeeb3b566a3ff330bf33906bf7", + "prepared_review_unit_hash": "4c7033d0843bee84cc965d327fd77442d97a28880dd39a51545637a244f0a5a0", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1636, - "estimated_input_tokens": 409, - "estimated_eval_tokens": 1074 + "serialized_chars": 1637, + "estimated_input_tokens": 410, + "estimated_eval_tokens": 1076 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1636, - "estimated_input_tokens": 409, - "estimated_eval_tokens": 1074 + "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 plausibly related through the static-cache/linear-attention path, but the packet only shows title-level similarity and a soft-similarity signal. With no shared targets, filenames, or explicit links, this is not enough to confidently merge them as the same bug.", - "confidence": 0.42, - "canonical_issue_reason": "Issue 46439 is the better canonical report because it describes the broader mechanism (`early_initialization` corrupting linear-attention layers) rather than a single model-specific crash surface.", + "summary": "Both issues involve static cache behavior around linear-attention/Qwen3.5, but they report different failure modes: one is shape corruption during early initialization, the other is a KeyError for a missing `linear_attention` cache entry. They look related in subsystem, not the same underlying bug.", + "confidence": 0.8, + "canonical_issue_reason": "Issue 46439 is the better representative because it describes the concrete mechanism (`early_initialization`) and the resulting shape corruption, which is a clearer root-cause style report than the model-specific KeyError in 46441.", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46439 is the best global issue anchor for the cluster since it is more mechanism-oriented and can reasonably subsume the narrower Qwen3.5 symptom if later evidence confirms they match.", + "best_issue_reason": "46439 is the strongest standalone issue for triage since it is more specific about the failure path and likely applies beyond one model variant.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46439", "right": "issue:46441", "accept": false, - "reason": "The evidence is too thin to treat these as the same underlying bug. They share a static-cache/linear-attention theme, but one is a RuntimeError from shape corruption and the other is a Qwen3.5 KeyError on `linear_attention`; without shared targets, filenames, or explicit links, this should remain only possibly related." + "reason": "Related static-cache/linear-attention failures, but the symptoms differ materially: wrong-shape RuntimeError vs `KeyError: 'linear_attention'`. Not enough evidence they are the same bug or that one PR would fix both." } ] }, - "evaluator_result": { - "accept": false, - "feedback": "The cluster is plausible, but the evidence in the packet is too thin to support a firm same-bug merge. We only have titles plus a soft-similarity signal, with no shared targets, filenames, or explicit links. Issue 46439 describes an early_initialization shape-corruption RuntimeError, while 46441 reports a Qwen3.5 KeyError on `linear_attention`; these may be related, but the analyst overstates the certainty by treating them as the same underlying cache-initialization bug and accepting the edge outright. A more conservative verdict would be 'possibly related' rather than accepted." - }, - "evaluator_used": true, - "retried": true, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, "accepted_nontrivial_soft_edge": false, "error_kind": null, "error_message": null @@ -20769,12 +20806,12 @@ "soft_pairs": [ "issue:46551|issue:46552" ], - "prepared_review_unit_hash": "ddfaf6c70a4d02d25b74097caacecf1698c64e29af37a4e23ae5b4480ae7c8f3", + "prepared_review_unit_hash": "9cc0d2e6562aefd3d7d6c86a7d0bebe03224fad474d621606e18b03e5548852c", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1707, + "serialized_chars": 1706, "estimated_input_tokens": 427, "estimated_eval_tokens": 1110 }, @@ -20782,7 +20819,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 1707, + "serialized_chars": 1706, "estimated_input_tokens": 427, "estimated_eval_tokens": 1110 }, @@ -20791,27 +20828,27 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are near-perfect duplicate issue reports for the same Wav2Vec2CTCTokenizer v5 special-token leak bug. The open issue is the better canonical target.", - "confidence": 0.99, - "canonical_issue_reason": "Issue #46552 is open, has inbound references, and has the same exact title/body as #46551, making it the better canonical tracker for this bug.", + "summary": "The two issues are very likely duplicates: they have identical titles and identical body length, and both point to the same Wav2Vec2CTCTokenizer v5 special-token regression. The packet does not include the body text, so the duplicate call should be phrased based on metadata rather than claiming the bodies are identical in content. #46552 is the better canonical issue because it is open and already has inbound references.", + "confidence": 0.93, + "canonical_issue_reason": "#46552 is open and has inbound references, so it is the most actionable representative of this duplicate pair.", "canonical_pr_reason": null, - "best_issue_reason": "#46552 is the best issue to keep because it is still open and already attracting references, while #46551 appears to be the earlier closed duplicate.", + "best_issue_reason": "#46552 best represents the currently active report for this tokenizer regression.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46551", "right": "issue:46552", "accept": true, - "reason": "Identical titles and body content describe the same concrete tokenizer bug, so this is a true duplicate pair." + "reason": "Same title and identical body length, with matching metadata pointing to the same tokenizer bug; the body text is not available, but these look like duplicate reports of the same issue." } ] }, "evaluator_result": { - "accept": true, - "feedback": "The summary is well grounded in the packet: the two issues have identical titles and bodies, and #46552 is open with inbound references while #46551 is closed. The duplicate verdict is conservative and supported by the evidence." + "accept": false, + "feedback": "The pair is well-supported as a duplicate by the identical titles and identical body length, but the summary overstates the evidence by claiming the bodies have the same content. The packet does not include body text, so that part should be stated more conservatively. The canonical choice of #46552 as the open issue with inbound references is grounded." }, "evaluator_used": true, - "retried": false, + "retried": true, "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null @@ -20835,7 +20872,7 @@ "soft_pairs": [ "issue:46657|issue:46692" ], - "prepared_review_unit_hash": "d72ef512e28207c15d736872a5a8f11b382204110263d266231bde963437c98b", + "prepared_review_unit_hash": "6795f94fe4c483d467eaa2c0e629d8e5c21168695fbe4c95b37d8b65c0d87529", "component_budget": { "node_count": 2, "item_count": 2, @@ -20857,18 +20894,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two issues are related to training object detection models, but they describe different failures: a tensor-size mismatch versus an out-of-memory condition after several epochs. They do not look like the same underlying bug.", + "summary": "These two issues are not the same underlying bug: one reports a tensor shape mismatch during object-detection training, while the other reports out-of-memory after several epochs in RT-DETR v2. They overlap only at a broad training/model level.", "confidence": 0.95, - "canonical_issue_reason": null, + "canonical_issue_reason": "Issue #46657 is the better canonical issue because it describes a concrete, specific failure mode (tensor size mismatch) rather than a generic training resource problem.", "canonical_pr_reason": null, - "best_issue_reason": "Issue #46657 is the broader, more generic report, but it is still not a good canonical for this cluster because the failure mode differs from #46692.", + "best_issue_reason": "Issue #46657 is the strongest representative of the cluster since it is more specific and actionable; #46692 appears to be a different symptom class (OOM) and should not be grouped as the same bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46657", "right": "issue:46692", "accept": false, - "reason": "Different symptom and likely different root cause: shape mismatch during training vs OOM after some epochs in RT-DETR v2." + "reason": "Different failure modes: shape mismatch vs. memory exhaustion. They share only the general context of training an object detection model, which is not enough to treat them as the same issue." } ] }, @@ -20898,7 +20935,7 @@ "soft_pairs": [ "issue:46822|issue:46823" ], - "prepared_review_unit_hash": "a13d06ae1cfc12f6b7f2e17d829964a41c4a7dde8d31b10ed10e47f4f830e62b", + "prepared_review_unit_hash": "eae3edc28a2a7247f14bae452c047970cc69e164be13ca7a83a7a2054399e755", "component_budget": { "node_count": 2, "item_count": 2, @@ -20920,18 +20957,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two issues are both about model loading/dispatched placement failures, but they describe different code paths and different failure modes: ZeRO-3 zero.Init partitioning for a composite submodule versus 4-bit device_map/max_memory dispatch incorrectly sending modules to CPU. They should not be merged as duplicates.", + "summary": "These two issues are only superficially similar: both involve model loading and memory placement, but one is a ZeRO-3/zero.Init partitioning failure on a composite multimodal submodule, while the other is a 4-bit device_map/max_memory dispatch bug sending a model to CPU unnecessarily. They do not look like the same underlying defect.", "confidence": 0.97, - "canonical_issue_reason": "Issue 46822 is the more specific bug report with a concrete subsystem (ZeRO-3 zero.Init) and a clearly identified failing composite submodule path, making it a better canonical issue for that cluster.", + "canonical_issue_reason": "No canonical issue for the pair: the reports describe different code paths and different failure modes (ZeRO-3 partitioning vs. 4-bit auto device mapping).", "canonical_pr_reason": null, - "best_issue_reason": "Issue 46822 is the best standalone issue because it names the exact loading mechanism and the observed OOM symptom, which is more actionable and narrower than the broader quantization/device_map error in 46823.", + "best_issue_reason": "Issue 46822 is the more concrete, higher-signal report with clearer repro/impact and some discussion activity; 46823 appears to be a separate, unrelated loading bug.", "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "issue:46822", "right": "issue:46823", "accept": false, - "reason": "Both concern model loading, but one is a ZeRO-3 partitioning/OOM bug and the other is a 4-bit device_map/max_memory routing bug. They affect different mechanisms and would not plausibly be fixed by one change." + "reason": "Different underlying bugs: 46822 is about ZeRO-3 zero.Init not partitioning a composite submodule, while 46823 is about 4-bit loading with device_map='auto' incorrectly dispatching to CPU despite available memory." } ] }, @@ -20945,7 +20982,7 @@ { "label": "pull_request", "component_index": 1, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -20961,12 +20998,12 @@ "soft_pairs": [ "pull_request:40546|pull_request:42781" ], - "prepared_review_unit_hash": "3e13e51d8cbbe9ef860350395afb6382d76bd4c26ccf84a7f9de61fbbee93394", + "prepared_review_unit_hash": "10000ca40582744b7e4c57678d139fc0250aef0dc0650a5efba4cce6547e359a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7222, + "serialized_chars": 7223, "estimated_input_tokens": 1806, "estimated_eval_tokens": 3868 }, @@ -20974,7 +21011,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 7222, + "serialized_chars": 7223, "estimated_input_tokens": 1806, "estimated_eval_tokens": 3868 }, @@ -20983,18 +21020,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related VibeVoice PRs, but not duplicates: #40546 is the base VibeVoice integration, while #42781 is a broader follow-up adding VibeVoice Realtime and acoustic-tokenizer support.", - "confidence": 0.9, + "summary": "Both pull requests are about VibeVoice, but they are not the same change: #40546 implements the core VibeVoice integration, while #42781 adds a separate VibeVoice Realtime path plus acoustic tokenizer work. The overlap is largely shared model-family scaffolding, not a single mergeable fix.", + "confidence": 0.86, "canonical_issue_reason": null, - "canonical_pr_reason": "#42781 is the broader, newer VibeVoice PR and appears to subsume the base integration with extra realtime/tokenizer work, so it is the better representative PR if one must be chosen.", + "canonical_pr_reason": "#40546 is the better canonical PR because it is the foundational VibeVoice implementation and appears to be the broader, more central change in this cluster.", "best_issue_reason": null, - "best_pr_reason": "#42781 is the most complete VibeVoice-related PR in this cluster, covering the additional realtime and tokenizer pieces beyond the core model port.", + "best_pr_reason": "#40546 best represents the cluster as the base VibeVoice implementation; #42781 extends into a distinct realtime/tokenizer feature set rather than duplicating the same PR.", "soft_edge_verdicts": [ { "left": "pull_request:40546", "right": "pull_request:42781", "accept": false, - "reason": "Shared VibeVoice files indicate overlap, but the scopes differ: #40546 is the core model implementation, while #42781 adds realtime and acoustic-tokenizer functionality. They are related follow-up work, not the same concrete change." + "reason": "Reject: the PRs overlap on VibeVoice infrastructure, but #42781 adds a separate realtime feature and acoustic tokenizer work, so they are not the same concrete code-path change and would not plausibly collapse into one PR." } ] }, @@ -21008,7 +21045,7 @@ { "label": "pull_request", "component_index": 2, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21024,7 +21061,7 @@ "soft_pairs": [ "pull_request:40755|pull_request:41224" ], - "prepared_review_unit_hash": "aa87e434f8c062096a5a0b3e2866fff06f8ffb0a363763b7c308d7a15d5dfe00", + "prepared_review_unit_hash": "f987aca1022b2655246f86c7b323cf4a805e2617d7f2ad630db3f8b8a9f0e2ed", "component_budget": { "node_count": 2, "item_count": 2, @@ -21046,18 +21083,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests are unrelated enhancements in different model families: TimesFM covariate forecasting vs DINOv3 image classification support. They do not look like the same underlying change, so the soft duplicate edge should be rejected.", - "confidence": 0.99, + "summary": "These two PRs are unrelated: one adds covariate-based forecasting support for TimesFM, while the other adds DINOv3 image-classification support. They touch different models, different code paths, and solve different product changes, so they are not duplicates.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 41224 is the better representative of the cluster because it is a focused, well-scoped feature addition with substantial review activity; however, it is not a duplicate of PR 40755.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 41224 is the strongest PR in the set as a concrete, mature feature implementation with clearer review evidence and a tighter scope than the TimesFM covariates PR.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:40755", "right": "pull_request:41224", "accept": false, - "reason": "Different models and different changes: TimesFM forecasting with covariates vs DINOv3 image classification support. They do not share the same underlying bug or code path." + "reason": "Different features for different models: TimesFM covariate forecasting vs DINOv3 image classification. No shared bug, no shared code path, and they would not plausibly be merged into one PR." } ] }, @@ -21071,7 +21108,7 @@ { "label": "pull_request", "component_index": 3, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21134,7 +21171,7 @@ { "label": "pull_request", "component_index": 4, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21188,7 +21225,7 @@ "pull_request:43823|pull_request:46911", "pull_request:43251|pull_request:43254" ], - "prepared_review_unit_hash": "84e9f57d6d3feb76052d4c3102e5fc305fbcf9a71817df769ebbbf34644045c3", + "prepared_review_unit_hash": "0142246c909ff26b86df4b14c309735e6a7457400a18d1f8034450a82c45ce0a", "component_budget": { "node_count": 15, "item_count": 15, @@ -21205,179 +21242,29 @@ "estimated_input_tokens": 10563, "estimated_eval_tokens": 21382 }, - "overflow_policy": "truncate_then_skip", - "trimmed": false, - "aggressively_trimmed": false, - "split": false, - "analyst_result": { - "summary": "This cluster is overwhelmingly heterogeneous: tokenizer methods, processor registration, model additions, docs, FSDP migration, RoPE changes, and loss/kernel work are all unrelated. The only true duplicate pair is the two `fixed_cross_entropy` PRs, which both implement the same `cross_entropy` kwargs passthrough fix. Everything else should stay separate.", - "confidence": 0.96, - "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43251 is the clearest representative of the only real duplicate pair: it directly targets issue 43240 and implements the `cross_entropy` kwargs passthrough in `fixed_cross_entropy` with the broader, cleaner signature update.", - "best_issue_reason": null, - "best_pr_reason": "PR 43251 is the strongest single PR in the cluster to anchor deduping because it is issue-linked, changes the exact loss code path, and is materially the same fix as 43254.", - "soft_edge_verdicts": [ - { - "left": "pull_request:41855", - "right": "pull_request:43823", - "accept": false, - "reason": "Different model areas and changes: Mistral tokenizer methods vs adding MobileLLM model files. No shared bug or change." - }, - { - "left": "pull_request:41855", - "right": "pull_request:43424", - "accept": false, - "reason": "Tokenizer API addition and an ExecuTorch dynamic-shape test are unrelated." - }, - { - "left": "pull_request:42668", - "right": "pull_request:43823", - "accept": false, - "reason": "Both are processor/model-adjacent, but one is auto/processor robustness work while the other is a new model addition; not the same change." - }, - { - "left": "pull_request:43251", - "right": "pull_request:43823", - "accept": false, - "reason": "Loss utility kwargs passthrough and MobileLLM model addition are unrelated." - }, - { - "left": "pull_request:43251", - "right": "pull_request:44827", - "accept": false, - "reason": "Same broad library area, but one is cross-entropy kwargs support and the other is Mistral4 test/model work; not the same concrete fix." - }, - { - "left": "pull_request:43424", - "right": "pull_request:46911", - "accept": false, - "reason": "ExecuTorch dynamic-shape testing and Olmo3 RoPE-per-layer support are different features." - }, - { - "left": "pull_request:43424", - "right": "pull_request:46707", - "accept": false, - "reason": "Dynamic-shape export testing vs native FSDP2 migration are unrelated." - }, - { - "left": "pull_request:43424", - "right": "pull_request:44827", - "accept": false, - "reason": "Different subsystems and objectives: export test coverage vs Mistral4 model/test fixes." - }, - { - "left": "pull_request:43424", - "right": "pull_request:43823", - "accept": false, - "reason": "ExecuTorch export testing and a new model implementation do not address the same underlying issue." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45655", - "accept": false, - "reason": "New model addition vs config-resolution order fix; no shared code-path problem." - }, - { - "left": "pull_request:43823", - "right": "pull_request:43995", - "accept": false, - "reason": "MobileLLM addition and Falcon output-interface refactor are different changes." - }, - { - "left": "pull_request:43995", - "right": "pull_request:45213", - "accept": false, - "reason": "Falcon refactor and model-creation skill/docs PR are unrelated." - }, - { - "left": "pull_request:43995", - "right": "pull_request:44827", - "accept": false, - "reason": "Falcon output-capture refactor and Mistral4 test fixes are not the same bug/change." - }, - { - "left": "pull_request:44827", - "right": "pull_request:46911", - "accept": false, - "reason": "Mistral4 test fixes and Olmo3 RoPE-per-layer logic are unrelated." - }, - { - "left": "pull_request:44827", - "right": "pull_request:46707", - "accept": false, - "reason": "Mistral4 fixes and FSDP2 migration address different code paths." - }, - { - "left": "pull_request:44827", - "right": "pull_request:45170", - "accept": false, - "reason": "Mistral4 test/model work and CLIP-like checkpoint renaming are unrelated." - }, - { - "left": "pull_request:44827", - "right": "pull_request:46267", - "accept": false, - "reason": "Mistral4 fixes and chat-response parsing updates are different features." - }, - { - "left": "pull_request:45170", - "right": "pull_request:46707", - "accept": false, - "reason": "Weight-name renaming for CLIP-like models is unrelated to FSDP2 migration." - }, - { - "left": "pull_request:46048", - "right": "pull_request:46707", - "accept": false, - "reason": "TDT loss kernel integration and FSDP2 migration are different subsystems and fixes." - }, - { - "left": "pull_request:44827", - "right": "pull_request:46048", - "accept": false, - "reason": "Mistral4 test fixes and TDT loss kernel work are unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:46707", - "accept": false, - "reason": "New MobileLLM model code and distributed FSDP2 migration are not the same change." - }, - { - "left": "pull_request:43823", - "right": "pull_request:46911", - "accept": false, - "reason": "MobileLLM model addition and Olmo3 RoPE support are unrelated." - }, - { - "left": "pull_request:44827", - "right": "pull_request:46048", - "accept": false, - "reason": "Mistral4 fixes and TDT kernel integration do not share the same concrete bug." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45213", - "accept": false, - "reason": "Model addition and a repo skill/docs PR are unrelated." - }, - { - "left": "pull_request:43823", - "right": "pull_request:45213", - "accept": false, - "reason": "Model addition and a repo skill/docs PR are unrelated." - }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated PRs across tokenizers, processors, model additions, docs, and infra. The only clear duplicate is the pair of `fixed_cross_entropy` kwargs fixes in `src/transformers/loss/loss_utils.py`; they address the same concrete bug and should be treated as one change, with PR #43251 as the stronger canonical candidate.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43251 is the better canonical PR for the loss-utils fix: it targets the issue explicitly, has the fuller implementation, and shows more review activity than #43254 while fixing the same `fixed_cross_entropy` code path.", + "best_issue_reason": null, + "best_pr_reason": "PR #43251 is the best PR to represent the duplicate pair because it is the more complete and better-reviewed version of the same kwargs-support fix in `fixed_cross_entropy`.", + "soft_edge_verdicts": [ { "left": "pull_request:43251", "right": "pull_request:43254", "accept": true, - "reason": "Both PRs fix the same `fixed_cross_entropy` bug by forwarding supported kwargs to `nn.functional.cross_entropy`; they are the same concrete code-path change and could be merged into one." + "reason": "Same file, same function, same issue target, and the same concrete fix: pass supported kwargs through to `nn.functional.cross_entropy`. They are plausible duplicates of the exact same change." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is well grounded in the packet: the cluster is mostly heterogeneous by filenames and diff scope, and the only clearly duplicate pair is 43251/43254, which both target `fixed_cross_entropy`/`cross_entropy` kwargs passthrough for the same issue. The soft-edge verdicts are conservative and do not overstate similarity." + "feedback": "The summary is grounded in the packet and stays conservative overall. The only minor nuance is that PR #43251 does not literally pass arbitrary kwargs through; both PRs add the same supported `weight`/`label_smoothing` cross-entropy arguments, so describing them as the same bugfix is fine. The duplicate verdict on #43251/#43254 is supported and mergeable into one PR for the same fix." }, "evaluator_used": true, "retried": false, @@ -21388,7 +21275,7 @@ { "label": "pull_request", "component_index": 5, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21404,12 +21291,12 @@ "soft_pairs": [ "pull_request:42975|pull_request:42976" ], - "prepared_review_unit_hash": "57329bf97c0aec97569361afe420621eeb9b6724dd6341ebc192605377d08eb8", + "prepared_review_unit_hash": "2cbea396c86d80ccced39b5ca94e56fe5470d3731b7bbadcc176c28d420de98e", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5935, + "serialized_chars": 5934, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -21417,7 +21304,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5935, + "serialized_chars": 5934, "estimated_input_tokens": 1484, "estimated_eval_tokens": 3224 }, @@ -21426,18 +21313,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests are workflow dependency upgrade batches, but they are not the same concrete change: PR 42975 is specifically for Node 24 compatibility across many GitHub Actions, while PR 42976 is a broader \u201clatest versions\u201d bump centered on Docker build actions. They overlap in a few workflow files, but the underlying fix goals differ.", - "confidence": 0.89, + "summary": "Two open PRs both modernize GitHub Actions in workflow files, but they update different action families and address different maintenance goals: one is for Node 24 compatibility of core GitHub actions, the other is for newer Docker build/publish action versions. They are related but not the same change.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 42975 is the better canonical PR because it has the clearest intent (\u201cNode 24 compatibility\u201d) and covers the shared workflow-upgrade theme more explicitly than the vaguer \u201clatest versions\u201d bump in PR 42976.", + "canonical_pr_reason": "PR 42975 is the stronger representative of the cluster because it is broader and more clearly framed around a specific platform-driven compatibility upgrade across many workflows.", "best_issue_reason": null, - "best_pr_reason": "PR 42975 is the best representative of the cluster since it targets a specific platform compatibility issue and appears to be the more comprehensive workflow-action upgrade set.", + "best_pr_reason": "PR 42975 is the best standalone PR to keep: it covers a wider set of workflow action upgrades and has the clearest single-purpose maintenance objective, whereas PR 42976 is narrower and focused on Docker action version bumps.", "soft_edge_verdicts": [ { "left": "pull_request:42975", "right": "pull_request:42976", "accept": false, - "reason": "Reject: both are GitHub Actions upgrade PRs, but they address different upgrade goals and different concrete actions; they would not plausibly be merged as one duplicate fix." + "reason": "Reject: they both touch GitHub Actions workflows, but 42975 upgrades core GitHub actions for Node 24 compatibility while 42976 upgrades docker/build-push-action and related workflow steps. Different action families and different concrete fixes, so they are not the same underlying change." } ] }, @@ -21451,7 +21338,7 @@ { "label": "pull_request", "component_index": 6, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21478,82 +21365,82 @@ "pull_request:45490|pull_request:46562", "pull_request:46562|pull_request:46658" ], - "prepared_review_unit_hash": "175d190ad8e9888e8402a542e85a75b38b1e0911cb334828f2db9276f8af6647", + "prepared_review_unit_hash": "30158ac5da763df8f35cb0036a8cd36b18d46fd65377a6149205e427932c80aa", "component_budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 21197, - "estimated_input_tokens": 5300, - "estimated_eval_tokens": 10856 + "serialized_chars": 21192, + "estimated_input_tokens": 5298, + "estimated_eval_tokens": 10852 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 8, - "serialized_chars": 21197, - "estimated_input_tokens": 5300, - "estimated_eval_tokens": 10856 + "serialized_chars": 21192, + "estimated_input_tokens": 5298, + "estimated_eval_tokens": 10852 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is heterogeneous: five independent model-addition PRs plus one Step3p7-related script/model-harness PR. The shared docs/auto-mapping edits are boilerplate, not evidence of duplicate bugs or mergeable overlapping changes.", - "confidence": 0.98, + "summary": "This cluster is not a duplicate set: it contains several unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM, Step 3.7 Flash) plus one WIP Step 3.7 Flash scripting PR. They share some autogenerated framework files and docs plumbing, but not the same underlying bug or change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: these PRs target different model families (ViT NEPA, Molmo, Molmo2, CTSM, Step3p7) and are not duplicates of one another.", + "canonical_pr_reason": "No canonical PR: the items are separate model implementations or support scripts for different models, not alternate fixes for the same code-path.", "best_issue_reason": null, - "best_pr_reason": "If forced to pick a representative, #46562 is the strongest standalone candidate because it is a substantive model addition and the only PR with a plausible near-neighbor (#46658), but it still is not a duplicate of the rest of the cluster.", + "best_pr_reason": "46562 is the most substantive standalone PR in the set, but it still represents a distinct model addition rather than a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:42978", "right": "pull_request:43448", "accept": false, - "reason": "Different model additions (ViT NEPA vs Molmo); overlap is only in docs/auto-registration boilerplate." + "reason": "Different model additions (ViT NEPA vs Molmo); overlap is only in shared framework/docs files." }, { "left": "pull_request:42978", "right": "pull_request:43451", "accept": false, - "reason": "Different model families (ViT NEPA vs Molmo2); shared auto/docs files do not indicate the same change." + "reason": "Different model additions (ViT NEPA vs Molmo2); no shared concrete bug or implementation target." }, { "left": "pull_request:42978", "right": "pull_request:45490", "accept": false, - "reason": "ViT NEPA and CTSM are unrelated model additions; same broad subsystem only." + "reason": "Different model additions (ViT NEPA vs CTSM); similar repo plumbing only." }, { "left": "pull_request:43448", "right": "pull_request:43451", "accept": false, - "reason": "Molmo and Molmo2 are separate model releases with different codebases and filenames; not the same concrete change." + "reason": "Molmo and Molmo2 are separate model families, not the same change, despite shared auto-registration/docs edits." }, { "left": "pull_request:43448", "right": "pull_request:45490", "accept": false, - "reason": "Molmo vs CTSM are unrelated additions; shared framework plumbing is incidental." + "reason": "Molmo vs CTSM are unrelated model additions; shared auto/model registry files are not enough." }, { "left": "pull_request:43451", "right": "pull_request:45490", "accept": false, - "reason": "Molmo2 and CTSM are different model integrations; overlap is limited to registry/docs updates." + "reason": "Molmo2 and CTSM are unrelated additions with only generic framework overlap." }, { "left": "pull_request:45490", "right": "pull_request:46562", "accept": false, - "reason": "CTSM and Step3p7 Flash are different model additions; shared auto-mapping/docs files are too generic to merge as duplicates." + "reason": "CTSM and Step 3.7 Flash are different models; overlap is limited to shared transformer registry files." }, { "left": "pull_request:46562", "right": "pull_request:46658", "accept": false, - "reason": "Related theme, but not the same change: #46562 adds the Step3p7 model, while #46658 adds a mini-model/script harness and debug artifacts, not the same code-path fix." + "reason": "Both are Step 3.7 Flash-related, but 46562 is the core model implementation while 46658 is a WIP script/demo bundle for building or converting a mini model; not the same concrete code-path change." } ] }, @@ -21567,7 +21454,7 @@ { "label": "pull_request", "component_index": 7, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21575,129 +21462,176 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-43291-6", + "cluster_id": "cluster-43291-8", "nodes": [ "pull_request:43291", "pull_request:43665", "pull_request:44815", "pull_request:46380", "pull_request:46539", - "pull_request:46870" + "pull_request:46870", + "pull_request:46886", + "pull_request:46922" ], "soft_pairs": [ "pull_request:43291|pull_request:43665", "pull_request:43291|pull_request:44815", "pull_request:43291|pull_request:46380", + "pull_request:43291|pull_request:46922", "pull_request:43665|pull_request:46539", "pull_request:43665|pull_request:46870", "pull_request:43665|pull_request:44815", "pull_request:43665|pull_request:46380", + "pull_request:43665|pull_request:46922", "pull_request:44815|pull_request:46539", "pull_request:44815|pull_request:46380", - "pull_request:46380|pull_request:46539" + "pull_request:44815|pull_request:46922", + "pull_request:46380|pull_request:46539", + "pull_request:46380|pull_request:46922", + "pull_request:46539|pull_request:46922", + "pull_request:46886|pull_request:46922" ], - "prepared_review_unit_hash": "57fa0316a5cf34748e2a04a879540cf7dfdae344c087427e27610a186b669700", + "prepared_review_unit_hash": "27f1b3e790dca36b6a6f0522f5b7b5afe1b14c50fa6ed5fbde5de3b04c34b19f", "component_budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 10, - "serialized_chars": 16287, - "estimated_input_tokens": 4072, - "estimated_eval_tokens": 8400 + "node_count": 8, + "item_count": 8, + "soft_pair_count": 16, + "serialized_chars": 21719, + "estimated_input_tokens": 5430, + "estimated_eval_tokens": 11116 }, "budget": { - "node_count": 6, - "item_count": 6, - "soft_pair_count": 10, - "serialized_chars": 16287, - "estimated_input_tokens": 4072, - "estimated_eval_tokens": 8400 + "node_count": 8, + "item_count": 8, + "soft_pair_count": 16, + "serialized_chars": 21719, + "estimated_input_tokens": 5430, + "estimated_eval_tokens": 11116 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These pull requests are not duplicates of one another: they span unrelated areas such as Whisper decoding tests, FP8 dequantization, generation-config validation, CI memory limits, and ASR pipeline dispatch. The only overlapping signal is broad similarity in test or model code, not the same underlying bug or change.", - "confidence": 0.88, + "summary": "The cluster is mostly heterogeneous, with one clear duplicate pair around YOLOS device-mismatch fixes. Most other PRs are unrelated and should not be merged as duplicates.", + "confidence": 0.87, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46380 is the strongest canonical candidate because it contains a real functional bug fix in a core runtime path (generation config validation/watermarking conversion) with broad downstream impact, whereas several others are test-only, CI-only, or narrower subsystem fixes.", + "canonical_pr_reason": "PR 46922 is the best canonical PR: it addresses the same YOLOS device_map bug as 46886, is tied to the explicit issue target, and contains the more complete fix (the device move plus the related no-split-module adjustment).", "best_issue_reason": null, - "best_pr_reason": "PR 46380 is the best standalone PR in the set: it fixes an actual runtime correctness issue rather than adding placeholder tests or CI plumbing, and the change is concrete and user-visible.", + "best_pr_reason": "PR 46922 is the strongest representative of the cluster\u2019s only clear code-fix duplicate because it directly fixes the YOLOS device mismatch and subsumes the smaller patch in 46886.", "soft_edge_verdicts": [ { "left": "pull_request:43291", "right": "pull_request:43665", "accept": false, - "reason": "Different artifacts: 43291 changes Whisper tokenization/decoding behavior, while 43665 is just test-file assertions and does not fix the same bug." + "reason": "Whisper tokenizer/test cleanup vs. unrelated CLIP/ViT test edits; no shared bug or concrete change." }, { "left": "pull_request:43291", "right": "pull_request:44815", "accept": false, - "reason": "Whisper decoding vs FP8 dequantization are unrelated code paths and failure modes." + "reason": "Whisper decoding changes are unrelated to the FP8 dequant/loading fix." }, { "left": "pull_request:43291", "right": "pull_request:46380", "accept": false, - "reason": "Both touch model/runtime code, but one is Whisper decoding and the other is generation-config watermarking validation; not the same underlying issue." + "reason": "Different areas and problems: Whisper decode logic vs. generation watermarking validation." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46922", + "accept": false, + "reason": "Whisper tokenizer fix and YOLOS device-mismatch fix are unrelated." }, { "left": "pull_request:43665", "right": "pull_request:46539", "accept": false, - "reason": "43665 is placeholder/failing test edits, while 46539 is a CI memory-limit/integration-test infrastructure change." + "reason": "Bogus model tests vs. CI memory/setup changes for Phimoe; not the same underlying issue." }, { "left": "pull_request:43665", "right": "pull_request:46870", "accept": false, - "reason": "43665 is not a real product fix, and 46870 is an ASR pipeline model-type dispatch change." + "reason": "Model test placeholders are unrelated to ASR pipeline model-type routing." }, { "left": "pull_request:43665", "right": "pull_request:44815", "accept": false, - "reason": "43665 only touches tests; 44815 fixes FP8 dequant/padding logic in runtime code." + "reason": "Test-only edits for CLIP/ViT do not match the FP8 dequant/loading bug fix." }, { "left": "pull_request:43665", "right": "pull_request:46380", "accept": false, - "reason": "43665 is test-only noise, not the same runtime bug as 46380's watermarking/generation validation fix." + "reason": "No overlap beyond being PRs; one is test noise, the other is generation/watermarking behavior." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46922", + "accept": false, + "reason": "CLIP/ViT test changes are not the same change as the YOLOS fix." }, { "left": "pull_request:44815", "right": "pull_request:46539", "accept": false, - "reason": "FP8 dequantization and CI memory patching are unrelated." + "reason": "FP8 dequant/loading changes and Phimoe CI infrastructure are different bugs." }, { "left": "pull_request:44815", "right": "pull_request:46380", "accept": false, - "reason": "Different concrete problems: FP8 kernel/dequant behavior versus generation-config validation." + "reason": "Different code paths: FP8 quantization/load conversion vs. generation config validation." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46922", + "accept": false, + "reason": "FP8 dequant fix is unrelated to YOLOS device placement." }, { "left": "pull_request:46380", "right": "pull_request:46539", "accept": false, - "reason": "46380 fixes generation config validation; 46539 fixes CI runner memory reporting for integration tests." + "reason": "Generation watermarking validation and CI memory patching are unrelated changes." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46922", + "accept": false, + "reason": "Watermarking/generation config work does not match the YOLOS device bug." + }, + { + "left": "pull_request:46539", + "right": "pull_request:46922", + "accept": false, + "reason": "Phimoe integration-test CI fixes are unrelated to YOLOS model runtime behavior." + }, + { + "left": "pull_request:46886", + "right": "pull_request:46922", + "accept": true, + "reason": "Both fix the same YOLOS device-mismatch problem in the same file; 46922 is the broader version of the same change." } ] }, - "evaluator_result": null, - "evaluator_used": false, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: most PRs are clearly unrelated, and the only well-supported duplicate pair is 46886/46922 on the same YOLOS device-mismatch bug. The canonical/representative choice is conservative, and the accepted pair looks mergeable as a single bugfix PR." + }, + "evaluator_used": true, "retried": false, - "accepted_nontrivial_soft_edge": false, + "accepted_nontrivial_soft_edge": true, "error_kind": null, "error_message": null }, { "label": "pull_request", "component_index": 8, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21716,7 +21650,7 @@ "pull_request:43488|pull_request:43532", "pull_request:43532|pull_request:46747" ], - "prepared_review_unit_hash": "7977f1395bbc0f982e16653929b24fdb45e35b61445292389154784d8f4aea27", + "prepared_review_unit_hash": "a940ac07c7d8af4269b9fc5fb4a5f1deb60f267f785bf15ee15e1ad51f9fa444", "component_budget": { "node_count": 3, "item_count": 3, @@ -21738,30 +21672,30 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three unrelated '[don't merge]' test/demo PRs touching different CI/setup areas; they do not look like the same underlying bug or change, so none should be clustered as duplicates.", - "confidence": 0.93, + "summary": "All three pull requests are distinct '[don't merge]' experiment/verification changes. They overlap superficially around CI/testing/workflow files, but they do not fix the same concrete code-path problem.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No PR is a true duplicate of the others. If one must serve as a representative, PR #46747 is the cleanest single-file workflow change, but it is still unrelated to the others.", + "canonical_pr_reason": "No true canonical duplicate exists in this cluster; the PRs change different mechanisms (setup/dependency/test tweaks vs. workflow trigger rewrites), so any single pick would be arbitrary.", "best_issue_reason": null, - "best_pr_reason": "PR #46747 is the most self-contained and easiest to use as a cluster exemplar, but the cluster itself is heterogeneous and not a duplicate set.", + "best_pr_reason": "If one PR must represent the cluster, 43532 is the broadest and most feature-complete CI/test experiment, but it is still not a duplicate of the others.", "soft_edge_verdicts": [ { "left": "pull_request:43488", "right": "pull_request:46747", "accept": false, - "reason": "43488 changes packaging/dependency metadata and a vit test, while 46747 only switches a security-gate workflow trigger. Different files, intent, and code paths." + "reason": "43488 changes packaging/dependency/test files, while 46747 only switches a workflow trigger for a security gate. Different subsystems and different fixes." }, { "left": "pull_request:43488", "right": "pull_request:43532", "accept": false, - "reason": "43488 is a setup/dependency/version-bump style PR; 43532 edits multiple workflows plus testing/notification helpers. They do not fix the same concrete problem." + "reason": "43488 is a setup/version/test tweak; 43532 rewrites several workflows and testing utilities. They share a test file, but not the same underlying change or bug." }, { "left": "pull_request:43532", "right": "pull_request:46747", "accept": false, - "reason": "43532 is a broad CI/workflow demonstration PR, but 46747 is a narrow workflow trigger change. Same subsystem broadly, but not the same underlying change." + "reason": "Both touch CI workflows, but 43532 is about self-comment CI and related test/util changes, whereas 46747 changes the PR security gate trigger. Not the same concrete fix." } ] }, @@ -21775,7 +21709,7 @@ { "label": "pull_request", "component_index": 9, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21791,12 +21725,12 @@ "soft_pairs": [ "pull_request:43612|pull_request:43613" ], - "prepared_review_unit_hash": "fb0c834082b5d2d99a2a4b72545641c9bca67d860158125df105d37383c496c8", + "prepared_review_unit_hash": "c766b58cbf56e15efe0b0513d6577c768d6090f4217a4da356c5421f1a129e63", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6083, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -21804,7 +21738,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6082, + "serialized_chars": 6083, "estimated_input_tokens": 1521, "estimated_eval_tokens": 3298 }, @@ -21813,18 +21747,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related only at the infrastructure level. The two PRs add different pipeline features for different segmentation tasks, so they should not be treated as duplicates or merged into one change.", - "confidence": 0.96, + "summary": "These are related but distinct PRs: both add new promptable segmentation pipelines, but they target different tasks, different model integrations, and different concrete code paths.", + "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 43613 is the better representative of the pair because it is the broader and more complete pipeline addition, but it is still a distinct feature from 43612 rather than a duplicate.", + "canonical_pr_reason": "No canonical PR: 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation with different pipeline implementations, model/test coverage, and task docs.", "best_issue_reason": null, - "best_pr_reason": "43613 has the more expansive scope and supporting changes (more model integrations and tests), so it is the stronger standalone PR to represent the cluster.", + "best_pr_reason": "43613 is the broader/more substantial PR, covering more model integrations and tests, so it is the better representative of the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:43612", "right": "pull_request:43613", "accept": false, - "reason": "Different underlying features: 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation with different model support and code paths. Shared docs/registry updates are incidental and not enough to treat them as the same PR." + "reason": "Different features and code paths: concept segmentation vs visual segmentation. The overlap is mainly shared docs/registration scaffolding, not the same underlying change." } ] }, @@ -21838,7 +21772,7 @@ { "label": "pull_request", "component_index": 10, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21854,7 +21788,7 @@ "soft_pairs": [ "pull_request:43651|pull_request:43663" ], - "prepared_review_unit_hash": "04a55dba0b2d97534985f30938e85908a65c1905a6a75d0947bb78caca948a51", + "prepared_review_unit_hash": "cf52ad3d2a73780fac1fd7c4f137303b2ae4242766d67d7888b0aa6bfbcf82bd", "component_budget": { "node_count": 2, "item_count": 2, @@ -21876,18 +21810,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs both add override hooks in Trainer, but they target different code paths: gradient-accumulation loss scaling vs. signature-based column filtering. They are not duplicates and should not be merged together.", - "confidence": 0.94, + "summary": "Two open PRs touch the same trainer file but address different extension points: gradient-accumulation loss scaling vs. signature-column filtering. They are related in theme but not duplicates.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs are about different Trainer behaviors, not the same underlying fix.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "PR 43651 is the more concrete standalone change, since it directly alters training_step loss-scaling behavior and adds an explicit hook for custom trainers.", + "best_pr_reason": "pull_request:43651 is the better representative of the cluster because it is a focused trainer-behavior change with a clear user-facing effect; 43663 is a separate internal refactor for column filtering.", "soft_edge_verdicts": [ { "left": "pull_request:43651", "right": "pull_request:43663", "accept": false, - "reason": "Different underlying changes in the same file: 43651 changes gradient-accumulation loss scaling, while 43663 extracts signature-column filtering logic. They do not fix the same bug or form one merged PR." + "reason": "Both modify src/transformers/trainer.py, but they target different code paths and problems: 43651 changes training_step loss scaling for gradient accumulation, while 43663 extracts signature-column filtering for custom trainer overrideability. Shared file alone is not enough to treat them as the same change." } ] }, @@ -21901,7 +21835,7 @@ { "label": "pull_request", "component_index": 11, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -21917,7 +21851,7 @@ "soft_pairs": [ "pull_request:43656|pull_request:43842" ], - "prepared_review_unit_hash": "c6594dc0c18f385c09e485472245cf6677d866c3e959f099127cbce43b7bc62a", + "prepared_review_unit_hash": "ab996854e7ea5d75d2ef167ec8b1a3da156abcffa80b8f9ade2e586d818bdcf0", "component_budget": { "node_count": 2, "item_count": 2, @@ -21939,18 +21873,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs touch `src/transformers/cli/serve.py` and are about annotation-related `TypeAdapter`/typing runtime issues, but they appear to fix different concrete failures. The later PR is a narrow pydantic-absence fix; the earlier PR is a broader annotations/future-import cleanup. I would not merge them as duplicates.", - "confidence": 0.77, + "summary": "Two open PRs target related TypeAdapter NameError problems in the same CLI file, but they appear to fix different runtime causes rather than the same exact bug.", + "confidence": 0.84, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #43842 is the most specific representation of the cluster: it directly targets the `TypeAdapter` NameError path when pydantic is unavailable, with a minimal, focused change in `serve.py`.", + "canonical_pr_reason": "PR #43656 is the broader and more direct fix for the CLI TypeAdapter NameError, with a larger runtime-annotation cleanup in serve.py and a title that matches the reported bug closely.", "best_issue_reason": null, - "best_pr_reason": "PR #43842 is the best standalone representative because it is narrowly scoped, explicit about the failure mode, and easier to reason about than the broader typing/future-annotations change in #43656.", + "best_pr_reason": "PR #43656 is the best representative of the cluster because it addresses the underlying annotation-evaluation problem in the CLI more generally, rather than only a narrow optional-dependency import path.", "soft_edge_verdicts": [ { "left": "pull_request:43656", "right": "pull_request:43842", "accept": false, - "reason": "They overlap in file and symptom class, but the fixes are not the same concrete change: #43656 is a broader annotations/future-import cleanup, while #43842 specifically guards `TypeAdapter` for missing pydantic. They could coexist, but they are not a duplicate PR." + "reason": "Both touch serve.py and mention TypeAdapter NameError, but #43656 fixes postponed-annotation/runtime typing issues broadly, while #43842 specifically guards the TypeAdapter import when pydantic is absent. They do not look like the same concrete bug or a single mergeable change." } ] }, @@ -21964,7 +21898,7 @@ { "label": "pull_request", "component_index": 12, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -22030,7 +21964,7 @@ { "label": "pull_request", "component_index": 13, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 2, "status": "reviewed", @@ -22088,8 +22022,8 @@ "pull_request:44018|pull_request:44019", "pull_request:44019|pull_request:44071", "pull_request:44068|pull_request:44722", - "pull_request:44018|pull_request:44116", "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", @@ -22118,8 +22052,8 @@ "pull_request:43997|pull_request:44056", "pull_request:44071|pull_request:44076", "pull_request:44074|pull_request:44101", - "pull_request:44056|pull_request:44161", "pull_request:43997|pull_request:44018", + "pull_request:44056|pull_request:44161", "pull_request:43997|pull_request:44161", "pull_request:44074|pull_request:44161", "pull_request:44018|pull_request:44161", @@ -22143,238 +22077,232 @@ "pull_request:44044|pull_request:44154", "pull_request:44024|pull_request:44076" ], - "prepared_review_unit_hash": "5c0e0c6520cee04d955829c716986bcfdce02a2963fa45dcb41c800f6a4e4fbc", + "prepared_review_unit_hash": "dc9ff42294aa7ca9265154f7dd5214676279b5d8faa7979756b4b9e44ef2c126", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84502, - "estimated_input_tokens": 21126, - "estimated_eval_tokens": 42508 + "serialized_chars": 84506, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 }, "budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 72, - "serialized_chars": 76934, - "estimated_input_tokens": 19234, - "estimated_eval_tokens": 38724 + "serialized_chars": 76939, + "estimated_input_tokens": 19235, + "estimated_eval_tokens": 38726 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "This cluster is mostly a set of model-specific output-tracing refactors. Only same-model pairs look like true duplicates; cross-model pairs share a theme but not the same concrete code path.", - "confidence": 0.84, - "canonical_issue_reason": null, - "canonical_pr_reason": "PR #44722 is the clearest representative: it targets the same GPT-J output-tracing refactor, is later and more complete, and explicitly captures the standardized decorator-based change.", + "summary": "This cluster is mostly a broad multi-model migration to standardized output tracing/decorators. Most pairs are only thematically similar; the clear duplicate-like edges are the GPT-Neo pair and the GPT-J/CodeGen pair(s).", + "confidence": 0.91, + "canonical_issue_reason": "No issue artifact is present in the cluster; #43979 is the shared umbrella tracking issue referenced by many of these PRs.", + "canonical_pr_reason": "44722 is the best canonical PR: it directly targets the shared refactor and captures the GPT-J/CodeGen output-tracing change in one coherent implementation.", "best_issue_reason": null, - "best_pr_reason": "Best fit for the cluster because it is the most complete and explicit implementation of the shared GPT-J/CodeGen output-tracing refactor.", + "best_pr_reason": "44722 is the most representative PR for the family, with an explicit issue link and the broadest concrete implementation of the same refactor.", "soft_edge_verdicts": [ { "left": "pull_request:43996", "right": "pull_request:44085", "accept": false, - "reason": "Different model/code paths; same output-tracing theme is too broad to treat as a duplicate." + "reason": "Different model families (CVT/FNet vs GPT-J); same tracking issue is not enough." }, { "left": "pull_request:43996", "right": "pull_request:44044", "accept": false, - "reason": "Different models and implementations; only the generic refactor theme overlaps." + "reason": "Unrelated model implementations; both are output-tracing refactors, but not the same code path." }, { "left": "pull_request:44066", "right": "pull_request:44085", "accept": true, - "reason": "Both change GPT-J output tracing in the same code path and could plausibly be one PR." + "reason": "Both hit the same GPT-J output-tracing code path; 44066 is a broader variant that also includes CodeGen." }, { "left": "pull_request:44007", "right": "pull_request:44072", "accept": false, - "reason": "RegNet and EfficientNet are separate model implementations; not the same bug/change." + "reason": "Different backbones and code paths (RegNet/ResNet/RT-DETR vs EfficientNet)." }, { "left": "pull_request:44072", "right": "pull_request:44722", "accept": false, - "reason": "EfficientNet vs GPT-J; shared tracking issue alone is not enough." + "reason": "Different model families and independent refactors; only the umbrella issue is shared." }, { "left": "pull_request:44066", "right": "pull_request:44072", "accept": false, - "reason": "GPT-J and EfficientNet are unrelated code paths." + "reason": "GPT-J/CodeGen vs EfficientNet are unrelated concrete changes." }, { "left": "pull_request:44013", "right": "pull_request:44044", "accept": false, - "reason": "MobileNetV2 and DeBERTa-v2 are unrelated implementations." + "reason": "MobileNetV2 and DeBERTa are separate model-specific refactors, not one duplicate bug." }, { "left": "pull_request:44066", "right": "pull_request:44086", "accept": false, - "reason": "GPT-J and MGP-STR are different model families with different code paths." + "reason": "GPT-J/CodeGen refactor versus MGP-STR output tracing; same theme, different code path." }, { "left": "pull_request:44018", "right": "pull_request:44068", "accept": true, - "reason": "Both are GPT-Neo output-tracing refactors in the same model code path." + "reason": "Both are GPT-Neo output-tracing refactors on the same model/file family." }, { "left": "pull_request:44066", "right": "pull_request:44071", "accept": false, - "reason": "GPT-J and MPT are separate model implementations; not a duplicate." + "reason": "GPT-J/CodeGen and MPT are unrelated model implementations." }, { "left": "pull_request:44066", "right": "pull_request:44068", "accept": false, - "reason": "GPT-J vs GPT-Neo: same theme, but different concrete model code." + "reason": "GPT-J/CodeGen versus GPT-Neo; shared pattern only, not the same fix." }, { "left": "pull_request:44018", "right": "pull_request:44066", "accept": false, - "reason": "GPT-Neo and GPT-J are different model code paths." + "reason": "GPT-Neo and GPT-J are different concrete implementations, despite similar decorator migration." }, { "left": "pull_request:44019", "right": "pull_request:44722", "accept": false, - "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen." + "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen output tracing." }, { "left": "pull_request:44086", "right": "pull_request:44722", "accept": false, - "reason": "MGP-STR and GPT-J are different model implementations." + "reason": "MGP-STR and GPT-J/CodeGen touch different model code paths." }, { "left": "pull_request:44019", "right": "pull_request:44085", "accept": false, - "reason": "ResNet and GPT-J are separate code paths." + "reason": "ResNet and GPT-J are separate refactors in different files." }, { "left": "pull_request:44018", "right": "pull_request:44019", "accept": false, - "reason": "GPT-Neo and ResNet are unrelated models." + "reason": "GPT-Neo and ResNet are different model families; same refactor style only." }, { "left": "pull_request:44019", "right": "pull_request:44071", "accept": false, - "reason": "ResNet and MPT are unrelated model implementations." + "reason": "ResNet and MPT are unrelated concrete code paths." }, { "left": "pull_request:44068", "right": "pull_request:44722", "accept": false, - "reason": "GPT-Neo and GPT-J are different model code paths." + "reason": "GPT-Neo is a different model family from GPT-J/CodeGen." }, { - "left": "pull_request:44018", - "right": "pull_request:44116", + "left": "pull_request:44068", + "right": "pull_request:44076", "accept": false, - "reason": "GPT-Neo and Flaubert do not share the same concrete change." + "reason": "GPT-Neo and ImageGPT are separate model-specific changes." }, { - "left": "pull_request:44068", + "left": "pull_request:44018", "right": "pull_request:44116", "accept": false, - "reason": "GPT-Neo and Flaubert are different implementations." + "reason": "GPT-Neo and Flaubert are unrelated implementations." }, { "left": "pull_request:44071", "right": "pull_request:44722", "accept": false, - "reason": "MPT and GPT-J are different model code paths." + "reason": "MPT and GPT-J/CodeGen do not share the same concrete change." }, { - "left": "pull_request:44085", + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated model implementations." + }, + { + "left": "pull_request:44068", "right": "pull_request:44086", "accept": false, - "reason": "Different model families; same broad refactor pattern is not enough." + "reason": "GPT-Neo and MGP-STR are different concrete refactors." }, { "left": "pull_request:44059", "right": "pull_request:44086", "accept": false, - "reason": "GPT-2 and MGP-STR are unrelated model implementations." + "reason": "GPT-2 and MGP-STR are separate model-specific changes." }, { "left": "pull_request:44098", "right": "pull_request:44101", "accept": false, - "reason": "ViLT and Flaubert are different model code paths." + "reason": "ViLT and Flaubert/XLM are unrelated model families." }, { "left": "pull_request:44059", "right": "pull_request:44722", "accept": false, - "reason": "GPT-2 and GPT-J are different concrete implementations." + "reason": "GPT-2 and GPT-J/CodeGen are different concrete code paths." }, { "left": "pull_request:44017", "right": "pull_request:44161", "accept": false, - "reason": "SegFormer and LongT5 are unrelated code paths." - }, - { - "left": "pull_request:44073", - "right": "pull_request:44086", - "accept": false, - "reason": "VisualBERT and MGP-STR are different model families." - }, - { - "left": "pull_request:44073", - "right": "pull_request:44101", - "accept": false, - "reason": "VisualBERT and Flaubert are unrelated implementations." + "reason": "SegFormer and LongT5 are unrelated model implementations." }, { "left": "pull_request:44056", "right": "pull_request:44101", "accept": false, - "reason": "MPNet and Flaubert are separate model code paths." + "reason": "MPNet and Flaubert/XLM are different model families." }, { "left": "pull_request:43997", "right": "pull_request:44017", "accept": false, - "reason": "RegNet and SegFormer are different vision backbones with different fixes." + "reason": "RegNet and SegFormer are separate model-specific output-tracing changes." }, { "left": "pull_request:43997", "right": "pull_request:44056", "accept": false, - "reason": "RegNet and MPNet are unrelated model implementations." + "reason": "RegNet and MPNet are unrelated code paths." }, { "left": "pull_request:44071", "right": "pull_request:44076", "accept": false, - "reason": "MPT and ImageGPT are different code paths." + "reason": "MPT and ImageGPT are distinct model implementations." }, { "left": "pull_request:44074", "right": "pull_request:44101", "accept": false, - "reason": "TextNet and Flaubert are separate model implementations." - }, - { - "left": "pull_request:44056", - "right": "pull_request:44161", - "accept": false, - "reason": "MPNet and LongT5 are unrelated code paths." + "reason": "TextNet and Flaubert/XLM are unrelated changes." }, { "left": "pull_request:43997", @@ -22382,161 +22310,149 @@ "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 separate concrete refactors." + }, { "left": "pull_request:43997", "right": "pull_request:44161", "accept": false, - "reason": "RegNet and LongT5 are unrelated implementations." + "reason": "RegNet and LongT5 do not fix the same underlying change." }, { "left": "pull_request:44074", "right": "pull_request:44161", "accept": false, - "reason": "TextNet and LongT5 do not share the same concrete change." + "reason": "TextNet and LongT5 are unrelated model-specific refactors." }, { "left": "pull_request:44018", "right": "pull_request:44161", "accept": false, - "reason": "GPT-Neo and LongT5 are unrelated model code paths." + "reason": "GPT-Neo and LongT5 are different model code paths." }, { "left": "pull_request:44010", "right": "pull_request:44071", "accept": false, - "reason": "SqueezeBert and MPT are different implementations." + "reason": "SqueezeBert and MPT are separate implementations." }, { "left": "pull_request:44010", "right": "pull_request:44073", "accept": false, - "reason": "SqueezeBert and VisualBERT are separate model code paths." + "reason": "SqueezeBert and VisualBert are different model-specific changes." }, { "left": "pull_request:44018", "right": "pull_request:44129", "accept": false, - "reason": "GPT-Neo and SpeechT5 are unrelated model changes." + "reason": "GPT-Neo and SpeechT5 are unrelated model families." }, { "left": "pull_request:43997", "right": "pull_request:44074", "accept": false, - "reason": "RegNet and TextNet are different model families." + "reason": "RegNet and TextNet are different concrete output-tracing changes." }, { "left": "pull_request:44010", "right": "pull_request:44018", "accept": false, - "reason": "SqueezeBert and GPT-Neo are unrelated implementations." + "reason": "SqueezeBert and GPT-Neo are unrelated code paths." }, { "left": "pull_request:43997", "right": "pull_request:44129", "accept": false, - "reason": "RegNet and SpeechT5 are different concrete code paths." + "reason": "RegNet and SpeechT5 are separate model refactors." }, { "left": "pull_request:44071", "right": "pull_request:44129", "accept": false, - "reason": "MPT and SpeechT5 are unrelated model implementations." + "reason": "MPT and SpeechT5 are different model implementations." }, { "left": "pull_request:43996", "right": "pull_request:44161", "accept": false, - "reason": "CVT/FNet and LongT5 are different code paths." + "reason": "CVT/FNet and LongT5 are not the same underlying change." }, { "left": "pull_request:43996", "right": "pull_request:44017", "accept": false, - "reason": "CVT/FNet and SegFormer are separate model implementations." + "reason": "CVT/FNet and SegFormer are distinct model-specific refactors." }, { "left": "pull_request:44010", "right": "pull_request:44044", "accept": false, - "reason": "SqueezeBert and DeBERTa-v2 are different model code paths." + "reason": "SqueezeBert and DeBERTa are separate model implementations." }, { "left": "pull_request:44044", "right": "pull_request:44074", "accept": false, - "reason": "DeBERTa-v2 and TextNet are unrelated models." - }, - { - "left": "pull_request:44044", - "right": "pull_request:44129", - "accept": false, - "reason": "DeBERTa-v2 and SpeechT5 are different implementations." - }, - { - "left": "pull_request:44044", - "right": "pull_request:44154", - "accept": false, - "reason": "DeBERTa-v2 and Vits are unrelated code paths." - }, - { - "left": "pull_request:44024", - "right": "pull_request:44076", - "accept": false, - "reason": "FocalNet and ImageGPT are different model implementations." + "reason": "DeBERTa and TextNet are unrelated concrete changes." }, { "left": "pull_request:44066", "right": "pull_request:44722", "accept": true, - "reason": "Same GPT-J output-tracing refactor; the later PR is a direct superset-style duplicate candidate." + "reason": "Same GPT-J/CodeGen output-tracing refactor family; 44722 is a later, broader version." }, { "left": "pull_request:44085", "right": "pull_request:44722", "accept": true, - "reason": "Both target the same GPT-J output-tracing change in the same file family." + "reason": "Both center on the same GPT-J output-tracing change; 44722 also includes CodeGen." }, { "left": "pull_request:44085", "right": "pull_request:44129", "accept": false, - "reason": "GPT-J and SpeechT5 are unrelated model code paths." + "reason": "GPT-J output tracing and SpeechT5 output tracing are different code paths." }, { "left": "pull_request:43996", "right": "pull_request:44007", "accept": false, - "reason": "CVT/FNet and RegNet are different model implementations." + "reason": "They share the same umbrella theme but cover different model backbones and implementations." }, { "left": "pull_request:43996", "right": "pull_request:44129", "accept": false, - "reason": "CVT/FNet and SpeechT5 are unrelated changes." + "reason": "CVT/FNet and SpeechT5 are unrelated model-specific refactors." }, { "left": "pull_request:44044", "right": "pull_request:44129", "accept": false, - "reason": "DeBERTa-v2 and SpeechT5 are different code paths." + "reason": "DeBERTa and SpeechT5 are different concrete changes." }, { "left": "pull_request:44044", "right": "pull_request:44154", "accept": false, - "reason": "DeBERTa-v2 and Vits are unrelated model implementations." + "reason": "DeBERTa and ViTS are unrelated model implementations." }, { "left": "pull_request:44024", "right": "pull_request:44076", "accept": false, - "reason": "FocalNet and ImageGPT are different concrete model changes." + "reason": "FocalNet and ImageGPT are separate model-specific output-tracing changes." } ] }, "evaluator_result": { "accept": true, - "feedback": "The summary is grounded in the packet and stays conservative: it treats cross-model overlaps as thematic rather than duplicate, and the accepted PR pairs are all plausibly mergeable into single bugfix PRs within the same model/code path. The canonical PR choice is also well supported by the later, broader GPT-J/CodeGen refactor in #44722." + "feedback": "The summary is grounded in the packet: it correctly identifies the cluster as a broad standardized-output-tracing migration, and the accepted soft edges stay within the same concrete model/file families (GPT-Neo and GPT-J/CodeGen). The rejected edges are conservative and mostly rely on shared theme or umbrella issue alone, which is not enough to merge them into a single bugfix PR." }, "evaluator_used": true, "retried": false, @@ -22547,7 +22463,7 @@ { "label": "pull_request", "component_index": 13, - "component_count": 38, + "component_count": 39, "review_unit_index": 2, "review_unit_count": 2, "status": "reviewed", @@ -22571,20 +22487,20 @@ "pull_request:43997|pull_request:44010", "pull_request:44010|pull_request:44129" ], - "prepared_review_unit_hash": "d91eadb432a87d8a793b2f156f0fa07d782d2d45a72d94ccd881f08aba24fd6c", + "prepared_review_unit_hash": "9d1c34d1741458e842c2335bfae1c682cb1f1c8f08dba694d36fc66d7cf59d3d", "component_budget": { "node_count": 28, "item_count": 28, "soft_pair_count": 77, - "serialized_chars": 84502, - "estimated_input_tokens": 21126, - "estimated_eval_tokens": 42508 + "serialized_chars": 84506, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 }, "budget": { "node_count": 6, "item_count": 6, "soft_pair_count": 5, - "serialized_chars": 13068, + "serialized_chars": 13067, "estimated_input_tokens": 3267, "estimated_eval_tokens": 6790 }, @@ -22593,42 +22509,42 @@ "aggressively_trimmed": false, "split": true, "analyst_result": { - "summary": "These PRs are all model-specific refactors to standardized output tracing/capture_outputs, but they affect different code paths and models. They are related thematically, not duplicates, so no soft edge should be merged.", - "confidence": 0.88, + "summary": "These PRs share a common theme\u2014migrating model forward passes to standardized output-capture utilities\u2014but they touch different model code paths, so they are related refactors rather than duplicates.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44074 is the most representative cluster member: it cleanly refactors a model to the new output-tracing pattern and also updates tests, making it the best exemplar of this refactor family.", + "canonical_pr_reason": "No single canonical PR: each change is model-specific (RegNet, SqueezeBert, MPNet, VisualBert, TextNet, SpeechT5). If a representative is needed, PR 44129 is the strongest candidate because it has an explicit issue target and the most complete output-tracing rewrite.", "best_issue_reason": null, - "best_pr_reason": "PR 44074 is the strongest standalone example because it combines the core model refactor with test coverage, while the others are narrower model-specific variants of the same migration pattern.", + "best_pr_reason": "PR 44129 is the best representative of the cluster\u2019s theme: it has an explicit issue link, a substantive refactor, and clear standardized output-capture changes, while the others are similar but isolated to different models.", "soft_edge_verdicts": [ { "left": "pull_request:44073", "right": "pull_request:44074", "accept": false, - "reason": "Same refactor theme, but different models (VisualBert vs TextNet) and different concrete code paths; not mergeable as one PR." + "reason": "Same refactor pattern, but different model implementations and code paths. This is not a single concrete change that could be merged as one PR." }, { "left": "pull_request:44056", "right": "pull_request:44073", "accept": false, - "reason": "Both migrate output tracing, but for different models (MPNet vs VisualBert); this is parallel cleanup, not the same change." + "reason": "Both migrate output tracing, but MPNet and VisualBert are distinct model backends with different implementation details. Not the same underlying bug/change." }, { "left": "pull_request:44056", "right": "pull_request:44074", "accept": false, - "reason": "Different model implementations and layer wiring; they share a pattern, not an underlying bug or single fix." + "reason": "Different models and different refactor shapes; they share a maintenance pattern only, not a duplicate fix." }, { "left": "pull_request:43997", "right": "pull_request:44010", "accept": false, - "reason": "RegNet and SqueezeBert are unrelated model code paths; these are separate standardized-output refactors, not duplicates." + "reason": "RegNet and SqueezeBert are unrelated code paths. Similar standardized-output migration, but not the same concrete change." }, { "left": "pull_request:44010", "right": "pull_request:44129", "accept": false, - "reason": "Both involve output tracing changes, but SqueezeBert and SpeechT5 are distinct model stacks with different behaviors and targets." + "reason": "Different model families and different output-tracing implementations; 44129 is tied to SpeechT5-specific layer behavior and an explicit issue, so it is not a duplicate of SqueezeBert." } ] }, @@ -22642,7 +22558,7 @@ { "label": "pull_request", "component_index": 14, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -22708,7 +22624,7 @@ "pull_request:44028|pull_request:44030", "pull_request:44029|pull_request:44030" ], - "prepared_review_unit_hash": "71f3b74478a02c8314d3c6bfd870cf9a393ce6eafae22782c51aa50b2934f8b0", + "prepared_review_unit_hash": "058e2b674f763390a24a12bcf78aedbed6e0a3ce76adcc0024179cd08046fb70", "component_budget": { "node_count": 13, "item_count": 13, @@ -22730,252 +22646,252 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are highly similar refactors of output tracing, but they target different model implementations and code paths, so they are not true duplicates.", - "confidence": 0.97, + "summary": "These PRs share a broad theme of output-tracing refactors, but they target different model implementations and forward paths, so they are not duplicates of one concrete change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 44003 is the strongest representative if one must anchor the cluster: it is a broader output-tracing refactor and touches more than one model file, making it more central than the narrower single-model PRs.", + "canonical_pr_reason": "PR 44003 is the strongest representative of the cluster: it is a broader output-tracing refactor spanning two related Mamba model files and introduces the shared capture/output plumbing seen across the group.", "best_issue_reason": null, - "best_pr_reason": "PR 44003 is the best single representative of this cluster because it shows the shared pattern most clearly while remaining substantive; the others are narrower, model-specific variants.", + "best_pr_reason": "PR 44003 is the best single anchor for this cluster because it has the broadest scope and the clearest shared tracing plumbing, making it the most representative refactor among the PRs.", "soft_edge_verdicts": [ { "left": "pull_request:43998", "right": "pull_request:44002", "accept": false, - "reason": "Both are output-tracing refactors, but `timm_backbone` and `upernet` are unrelated model code paths; not the same change." + "reason": "Both are output-tracing refactors, but timm_backbone and upernet are different model code paths and not the same underlying fix." }, { "left": "pull_request:43998", "right": "pull_request:44000", "accept": false, - "reason": "Different models (`timm_backbone` vs `vision_text_dual_encoder`) and different forward paths; parallel refactors, not duplicates." + "reason": "Same theme, but timm_backbone and vision_text_dual_encoder are unrelated implementations with different forward behavior." }, { "left": "pull_request:43998", "right": "pull_request:44001", "accept": false, - "reason": "`timm_backbone` and `univnet` are separate architectures with unrelated implementations; same theme, not same underlying bug/change." + "reason": "Both touch output handling, but timm_backbone and univnet are separate model paths and cannot be merged as one fix." }, { "left": "pull_request:43998", "right": "pull_request:44003", "accept": false, - "reason": "One is `timm_backbone`, the other is `mamba/falcon_mamba`; the shared theme is too broad to treat as one duplicate change." + "reason": "These are different model families; the similarity is only in refactoring style, not a shared bug or change." }, { "left": "pull_request:43998", "right": "pull_request:44029", "accept": false, - "reason": "`timm_backbone` and `rwkv` are distinct model paths; similar refactor pattern, but not mergeable as one PR." + "reason": "timm_backbone and rwkv have distinct forward paths; this is a thematic similarity, not a duplicate change." }, { "left": "pull_request:43999", "right": "pull_request:44001", "accept": false, - "reason": "`mobilenet_v1` and `univnet` are unrelated models; same output-tracing cleanup pattern, not the same concrete change." + "reason": "MobileNetV1 and UnivNet are unrelated model code paths, so this is not the same concrete fix." }, { "left": "pull_request:43999", "right": "pull_request:44002", "accept": false, - "reason": "Different targets (`mobilenet_v1` vs `upernet`), so this is similarity by template rather than duplicate functionality." + "reason": "Both are refactors, but MobileNetV1 and UperNet have different output behavior and cannot be treated as one PR." }, { "left": "pull_request:43999", "right": "pull_request:44025", "accept": false, - "reason": "`mobilenet_v1` and `depth_anything` are separate model implementations; not the same underlying bug or fix." + "reason": "MobileNetV1 and DepthAnything are different model implementations; only the refactor pattern matches." }, { "left": "pull_request:43999", "right": "pull_request:44026", "accept": false, - "reason": "These touch different wrapper/model code paths (`mobilenet_v1` vs `vision_encoder_decoder`); too broad to merge as duplicates." + "reason": "These modify different model wrappers and forward signatures, so they are not duplicates." }, { "left": "pull_request:43999", "right": "pull_request:44027", "accept": false, - "reason": "`mobilenet_v1` and `speech_encoder_decoder` are unrelated code paths; same refactor style, not same concrete change." + "reason": "MobileNetV1 and SpeechEncoderDecoder are unrelated code paths; same broad subsystem is not enough." }, { "left": "pull_request:44000", "right": "pull_request:44001", "accept": false, - "reason": "`vision_text_dual_encoder` and `univnet` are different models; they share a refactor motif but not the same bug/change." + "reason": "VisionTextDualEncoder and UnivNet are separate architectures; the overlap is only the tracing refactor theme." }, { "left": "pull_request:44000", "right": "pull_request:44002", "accept": false, - "reason": "`vision_text_dual_encoder` vs `upernet` is only broad output-tracing similarity; the code paths are distinct." + "reason": "These are different model paths with different output contracts, so they are not the same underlying change." }, { "left": "pull_request:44000", "right": "pull_request:44003", "accept": false, - "reason": "`vision_text_dual_encoder` and `mamba/falcon_mamba` are different model families; not a single duplicate PR." + "reason": "VisionTextDualEncoder and the Mamba PR target unrelated models; not mergeable as one concrete fix." }, { "left": "pull_request:44000", "right": "pull_request:44025", "accept": false, - "reason": "Different model targets (`vision_text_dual_encoder` vs `depth_anything`), so this is not one concrete change." + "reason": "Same style of cleanup, but different model code paths and forward APIs." }, { "left": "pull_request:44001", "right": "pull_request:44002", "accept": false, - "reason": "`univnet` and `upernet` are unrelated implementations; the shared pattern is too generic to count as a duplicate." + "reason": "UnivNet and UperNet are unrelated implementations; the shared label is only output-tracing refactor." }, { "left": "pull_request:44001", "right": "pull_request:44003", "accept": false, - "reason": "These affect different model families (`univnet` vs `mamba/falcon_mamba`); not the same underlying fix." + "reason": "These touch different model families and do not fix the same concrete bug." }, { "left": "pull_request:44001", "right": "pull_request:44025", "accept": false, - "reason": "`univnet` and `depth_anything` are different models; same refactor style, different code paths." + "reason": "UnivNet and DepthAnything are distinct model paths; not a single duplicate PR." }, { "left": "pull_request:44002", "right": "pull_request:44029", "accept": false, - "reason": "`upernet` and `rwkv` are distinct models with independent forward logic; not duplicates." + "reason": "UperNet and RWKV are unrelated model code paths, so this is only thematic similarity." }, { "left": "pull_request:44002", "right": "pull_request:44003", "accept": false, - "reason": "`upernet` and `mamba/falcon_mamba` are separate model code paths; too broad to merge as one PR." + "reason": "These are different model implementations with different forward/output behavior." }, { "left": "pull_request:44003", "right": "pull_request:44028", "accept": false, - "reason": "`mamba/falcon_mamba` and `superpoint` are unrelated architectures; same refactor family, not the same change." + "reason": "Falcon/Mamba and SuperPoint are unrelated code paths; same refactor style, not same change." }, { "left": "pull_request:44003", "right": "pull_request:44004", "accept": false, - "reason": "`mamba/falcon_mamba` and `codegen` touch different subsystems and forward contracts; not a duplicate." + "reason": "Mamba/Falcon-Mamba and CodeGen are different model implementations, so they are not duplicates." }, { "left": "pull_request:43998", "right": "pull_request:44004", "accept": false, - "reason": "`timm_backbone` and `codegen` are unrelated model paths; only the refactor pattern is shared." + "reason": "timm_backbone and CodeGen are unrelated model code paths; similar tracing refactor only." }, { "left": "pull_request:44001", "right": "pull_request:44004", "accept": false, - "reason": "`univnet` and `codegen` are separate implementations; similar output-tracing cleanup, not one bug." + "reason": "UnivNet and CodeGen are different architectures and not the same concrete fix." }, { "left": "pull_request:44002", "right": "pull_request:44004", "accept": false, - "reason": "`upernet` and `codegen` are different models with different forward semantics; not the same underlying change." + "reason": "UperNet and CodeGen are unrelated outputs/refactors, not one mergeable PR." }, { "left": "pull_request:44000", "right": "pull_request:44004", "accept": false, - "reason": "`vision_text_dual_encoder` and `codegen` are unrelated code paths; the overlap is only in refactoring style." + "reason": "VisionTextDualEncoder and CodeGen modify different forward paths and cannot be merged as one fix." }, { "left": "pull_request:44025", "right": "pull_request:44026", "accept": false, - "reason": "`depth_anything` and `vision_encoder_decoder` are different model families; not the same concrete fix." + "reason": "DepthAnything and VisionEncoderDecoder are different model wrappers; the overlap is only the refactoring pattern." }, { "left": "pull_request:44025", "right": "pull_request:44027", "accept": false, - "reason": "`depth_anything` and `speech_encoder_decoder` are separate implementations; this is similarity, not duplication." + "reason": "DepthAnything and SpeechEncoderDecoder are unrelated model paths, not a duplicate bug or change." }, { "left": "pull_request:44025", "right": "pull_request:44028", "accept": false, - "reason": "`depth_anything` and `superpoint` are unrelated vision models; same refactor motif but different code paths." + "reason": "DepthAnything and SuperPoint target different architectures and output contracts." }, { "left": "pull_request:44025", "right": "pull_request:44029", "accept": false, - "reason": "`depth_anything` and `rwkv` are entirely different architectures; not mergeable as one PR." + "reason": "DepthAnything and RWKV are separate model implementations; no shared concrete fix." }, { "left": "pull_request:44025", "right": "pull_request:44030", "accept": false, - "reason": "`depth_anything` and `dpr` are distinct model implementations; same output-tracing theme, not one change." + "reason": "DepthAnything and DPR are different code paths, so this is not the same underlying change." }, { "left": "pull_request:44026", "right": "pull_request:44029", "accept": false, - "reason": "`vision_encoder_decoder` and `rwkv` are unrelated code paths; shared theme is too broad." + "reason": "VisionEncoderDecoder and RWKV are unrelated models; similar output-tracing work is not enough." }, { "left": "pull_request:44026", "right": "pull_request:44027", "accept": false, - "reason": "Both are encoder-decoder wrappers, but vision and speech variants are different model implementations; not the same underlying fix." + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are different wrapper implementations and not a single fix." }, { "left": "pull_request:44026", "right": "pull_request:44028", "accept": false, - "reason": "`vision_encoder_decoder` and `superpoint` are separate models; no single concrete duplicate change." + "reason": "These are separate model paths with different output handling, so they are not duplicates." }, { "left": "pull_request:44026", "right": "pull_request:44030", "accept": false, - "reason": "`vision_encoder_decoder` and `dpr` are different code paths; only the refactor pattern matches." + "reason": "VisionEncoderDecoder and DPR are unrelated code paths despite both touching forward/output behavior." }, { "left": "pull_request:44027", "right": "pull_request:44028", "accept": false, - "reason": "`speech_encoder_decoder` and `superpoint` are unrelated implementations; not duplicates." + "reason": "SpeechEncoderDecoder and SuperPoint are different architectures; only the refactor theme matches." }, { "left": "pull_request:44027", "right": "pull_request:44029", "accept": false, - "reason": "`speech_encoder_decoder` and `rwkv` are different model families; too broad to merge." + "reason": "SpeechEncoderDecoder and RWKV are not the same underlying bug or code change." }, { "left": "pull_request:44027", "right": "pull_request:44030", "accept": false, - "reason": "`speech_encoder_decoder` and `dpr` are separate model code paths; not the same underlying change." + "reason": "SpeechEncoderDecoder and DPR modify different model-specific output paths." }, { "left": "pull_request:44028", "right": "pull_request:44029", "accept": false, - "reason": "`superpoint` and `rwkv` are unrelated architectures; same output-tracing refactor style, not duplicate code." + "reason": "SuperPoint and RWKV are unrelated model implementations, so this pair is not a duplicate." }, { "left": "pull_request:44028", "right": "pull_request:44030", "accept": false, - "reason": "`superpoint` and `dpr` are different models and different forward contracts; not the same fix." + "reason": "SuperPoint and DPR are different code paths; the shared theme is too broad." }, { "left": "pull_request:44029", "right": "pull_request:44030", "accept": false, - "reason": "`rwkv` and `dpr` are distinct model implementations; this is parallel cleanup, not one mergeable PR." + "reason": "RWKV and DPR are distinct model families with different output contracts, not one mergeable change." } ] }, @@ -22989,7 +22905,7 @@ { "label": "pull_request", "component_index": 15, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23005,12 +22921,12 @@ "soft_pairs": [ "pull_request:44662|pull_request:45401" ], - "prepared_review_unit_hash": "2edb9f6541088257dfaec09b034310447de7e37f30dea0c27393bed82488f9aa", + "prepared_review_unit_hash": "aa0a71ceba6b29f1eaab3bc7db2491e97acb16b52f6023f3b5d30eba9ee5df91", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6551, + "serialized_chars": 6552, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -23018,7 +22934,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 6551, + "serialized_chars": 6552, "estimated_input_tokens": 1638, "estimated_eval_tokens": 3532 }, @@ -23027,18 +22943,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two unrelated model-integration PRs: one adds PenguinVL, the other adds Voxtral TTS. Shared transformer registry/docs files are expected boilerplate and do not indicate the same underlying change.", - "confidence": 0.97, + "summary": "The two PRs are unrelated model-addition efforts: one adds PenguinVL, the other adds Voxtral TTS. The shared auto-registration/docs files are expected boilerplate and do not indicate the same underlying fix.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "44662 is the more general, non-draft integration PR and the better representative of the cluster, even though it is not a duplicate of 45401.", + "canonical_pr_reason": "45401 is the better representative of its topic because it is explicitly tied to issue #45027 and has a narrower, well-scoped Voxtral TTS implementation. 44662 is a separate PenguinVL model addition.", "best_issue_reason": null, - "best_pr_reason": "44662 is the only plausible PR anchor here; 45401 targets a different model and should not be used as a duplicate destination.", + "best_pr_reason": "45401 is the stronger overall PR in this cluster: it is issue-linked, more focused, and easier to reason about as a standalone change. 44662 is broader and belongs to a different model family.", "soft_edge_verdicts": [ { "left": "pull_request:44662", "right": "pull_request:45401", "accept": false, - "reason": "Reject: PenguinVL and Voxtral TTS are different model additions with different codepaths; overlapping auto/docs files are shared integration scaffolding, not evidence of the same change." + "reason": "Reject: these implement different models (PenguinVL vs Voxtral TTS) and do not appear to fix the same concrete code-path problem. Shared framework/docs files are just model-integration boilerplate." } ] }, @@ -23052,7 +22968,7 @@ { "label": "pull_request", "component_index": 16, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23068,12 +22984,12 @@ "soft_pairs": [ "pull_request:44965|pull_request:46845" ], - "prepared_review_unit_hash": "349f93a1dac71f4891e6e86ee8b749fcdb78202ed4696fe8f5da73db5f48dd30", + "prepared_review_unit_hash": "8b211751f9d6a68b9440d8533a199f3821fd06aef9e7594640de4424fe5e1219", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4219, + "serialized_chars": 4220, "estimated_input_tokens": 1055, "estimated_eval_tokens": 2366 }, @@ -23081,7 +22997,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4219, + "serialized_chars": 4220, "estimated_input_tokens": 1055, "estimated_eval_tokens": 2366 }, @@ -23090,18 +23006,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two pull requests are not duplicates: one changes a documentation/PR workflow to convert PRs to draft, while the other adjusts CI routing and includes an unrelated code tweak. They do not target the same bug or change.", - "confidence": 0.97, + "summary": "These two open PRs are not duplicates. PR 44965 rewrites a docs workflow to convert PRs to draft, while PR 46845 changes CI branch routing and includes an unrelated modeling_utils edit. The soft similarity is superficial only.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46845 is the less suitable canonical PR because it bundles an unrelated CI workflow branch change with an apparent placeholder code edit, so it is not a clean representative of a single fix.", + "canonical_pr_reason": "No true duplicate canonical exists here; if one PR must represent the cluster, 46845 is the more substantive change, but it is not the same underlying change as 44965.", "best_issue_reason": null, - "best_pr_reason": "PR 44965 is the better representative of its own change because it has a focused purpose: modifying the PR documentation workflow to convert PRs to draft.", + "best_pr_reason": "46845 is the better representative of the set because it contains the more meaningful CI-related change, whereas 44965 is a separate workflow rewrite with a different purpose.", "soft_edge_verdicts": [ { "left": "pull_request:44965", "right": "pull_request:46845", "accept": false, - "reason": "Different underlying changes: 44965 alters a workflow to change PRs to draft, while 46845 changes CI branch selection and contains an unrelated code edit. They are not the same concrete fix and would not plausibly merge into one PR." + "reason": "Reject: the PRs address different problems and code paths. 44965 changes a workflow to convert PRs to draft, while 46845 updates CI branch handling and touches modeling_utils; they could not plausibly be merged as one fix." } ] }, @@ -23115,7 +23031,7 @@ { "label": "pull_request", "component_index": 17, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23131,12 +23047,12 @@ "soft_pairs": [ "pull_request:44979|pull_request:45363" ], - "prepared_review_unit_hash": "8846b28b9fa5b68fc3a036e987f871bd0687149cfe8137d480a76d1e35cdb8d1", + "prepared_review_unit_hash": "b9c3586792d8230f19978f6d41d7fe39b61b2a763912b8f4f7be720643e0a3c4", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4408, + "serialized_chars": 4407, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -23144,7 +23060,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4408, + "serialized_chars": 4407, "estimated_input_tokens": 1102, "estimated_eval_tokens": 2460 }, @@ -23153,18 +23069,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related by the broad theme of module/kernel fusion, but they target different abstractions and code paths: one introduces a generic module-fusion API, the other wires up n-to-1 kernel fusion via KernelConfig and integration changes.", - "confidence": 0.87, + "summary": "The two PRs are related to module/kernel fusion, but they implement different layers of the feature: one adds a general Module Fusion API, while the other wires n-to-1 kernel fusion through KernelConfig and model loading. They are not the same underlying change and should not be deduplicated.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45363 is the more concrete end-user feature: it adds KernelConfig-driven fusion across modeling and integration code, with an example and broader runtime impact.", + "canonical_pr_reason": "PR #44979 is the better canonical representative for the cluster because it introduces the core Module Fusion API and test coverage; PR #45363 is a separate integration of KernelConfig-based kernel fusion built on top of that broader idea.", "best_issue_reason": null, - "best_pr_reason": "PR #45363 is the strongest representative of the cluster because it implements a specific fusion mechanism in the core loading/execution path rather than just adding a standalone abstraction layer.", + "best_pr_reason": "PR #44979 is the best single representative because it is the more foundational, self-contained feature PR. PR #45363 is narrower and more implementation-specific, so it is less representative of the overall fusion topic.", "soft_edge_verdicts": [ { "left": "pull_request:44979", "right": "pull_request:45363", "accept": false, - "reason": "They are not the same underlying change: #44979 adds a generic Module Fusion API and tests, while #45363 implements n-to-1 kernel fusion via KernelConfig across model loading/integration files. Similar theme, but different concrete functionality and code paths." + "reason": "Different concrete changes: #44979 adds a new module fusion abstraction/API, while #45363 adds KernelConfig-driven n-to-1 fusion plumbing and an example. Related area, but not the same bug or mergeable PR." } ] }, @@ -23178,7 +23094,7 @@ { "label": "pull_request", "component_index": 18, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23194,7 +23110,7 @@ "soft_pairs": [ "pull_request:45086|pull_request:45317" ], - "prepared_review_unit_hash": "69556d6afde928d8ade89af162653fb5104d5a442caaa70b6219e93f56506db5", + "prepared_review_unit_hash": "f23bce768d1eecdc60303c90ba6042ff7a649c61ec10e87393b098a7ee1b1923", "component_budget": { "node_count": 2, "item_count": 2, @@ -23216,18 +23132,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both pull requests fix the same AttributeError in `_patch_mistral_regex` by switching from `tokenizer.backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer` and adjusting the same code path in `tokenization_utils_tokenizers.py`. PR 45317 is the stronger representative because it includes the same fix plus test coverage.", - "confidence": 0.98, + "summary": "Both PRs fix the same Mistral regex patch bug in `tokenization_utils_tokenizers.py` by switching from `tokenizer.backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer`; PR 45317 is the more complete version because it also adds tests.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45317 is the best canonical PR: it implements the same fix as 45086 but is more complete, adding a test file alongside the code change.", + "canonical_pr_reason": "PR 45317 is the better canonical PR because it implements the same fix as 45086 but also includes test coverage and a clearer title for the AttributeError path.", "best_issue_reason": null, - "best_pr_reason": "PR 45317 is the best fit overall because it directly fixes the bug and includes validation, while 45086 is the narrower version of the same change.", + "best_pr_reason": "PR 45317 is the strongest representative of this cluster: same concrete bug fix, plus tests, making it the most merge-ready and informative.", "soft_edge_verdicts": [ { "left": "pull_request:45086", "right": "pull_request:45317", "accept": true, - "reason": "Same underlying bug and same fix in the same function/file; 45317 is effectively a superset with tests, so these are duplicates for triage." + "reason": "Same underlying code-path bug in `_patch_mistral_regex` with the same one-line fix; 45317 is essentially the tested, slightly expanded version of 45086, so they look like duplicate PRs for the same change." } ] }, @@ -23241,7 +23157,7 @@ { "label": "pull_request", "component_index": 19, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23257,7 +23173,7 @@ "soft_pairs": [ "pull_request:45115|pull_request:45569" ], - "prepared_review_unit_hash": "8d8b7769da8cd9c2a98731a46a678bdadf5fbd2b83b2bf24f4dcb3658d7f844e", + "prepared_review_unit_hash": "677e3b5d98aa1c7786cc7c1b3b65013868eb55a9aa1b7941994e8b2e798120a4", "component_budget": { "node_count": 2, "item_count": 2, @@ -23279,18 +23195,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related NemotronH architecture refactors, but they target different changes: one is a focused inheritance/conversion mapping refactor, the other is a broader split into Dense/Sparse model variants plus docs and auto-mapping updates.", - "confidence": 0.92, + "summary": "Two open NemotronH-related PRs, but they are not duplicates: one is a refactor/conversion-mapping cleanup, the other is a broader feature split adding NemotronH dense/sparse model variants, docs, and tests.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "45569 is the broader, more complete architectural PR: it introduces the new nemotron_h_dense and nemotron_h_sparse variants, updates docs, and adjusts auto-mapping/model registration. 45115 is narrower and mostly about refactoring inheritance plus conversion-key remapping.", + "canonical_pr_reason": "PR #45569 is the better canonical PR because it is the more complete, user-visible implementation: it introduces the dense/sparse model split, updates auto-mapping/docs, and includes tests.", "best_issue_reason": null, - "best_pr_reason": "45569 is the best representative PR for the cluster because it captures the major NemotronH model-structure change rather than just a supporting conversion refactor.", + "best_pr_reason": "PR #45569 is the strongest representative of the cluster since it covers the main architecture change and its surrounding API/docs support, whereas #45115 is a narrower refactor/conversion-mapping update.", "soft_edge_verdicts": [ { "left": "pull_request:45115", "right": "pull_request:45569", "accept": false, - "reason": "They overlap on NemotronH files, but the concrete changes differ: 45115 focuses on refactoring NemotronH to inherit GraniteMoeHybrid and adding checkpoint conversion mappings, while 45569 splits the model into dense/sparse variants and updates docs/auto-mapping. Not the same underlying fix and not something that would plausibly be merged as one PR." + "reason": "Both concern NemotronH, but they do not appear to fix the same concrete code-path problem. #45115 focuses on conversion-mapping and inheritance/refactor work, while #45569 introduces a broader dense/sparse model split with docs and tests. They would not plausibly be merged as one PR." } ] }, @@ -23304,7 +23220,7 @@ { "label": "pull_request", "component_index": 20, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23320,40 +23236,40 @@ "soft_pairs": [ "pull_request:45332|pull_request:45333" ], - "prepared_review_unit_hash": "0b225d4d6843d690402bb15bb9445eca1dbc69acd67bf47a9f61260b14f5ae92", + "prepared_review_unit_hash": "f0e3aef2b3a89697a70686e666fcee532485f165f510bd12597ead200d2ea9f6", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5952, - "estimated_input_tokens": 1488, - "estimated_eval_tokens": 3232 + "serialized_chars": 5953, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 }, "budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5952, - "estimated_input_tokens": 1488, - "estimated_eval_tokens": 3232 + "serialized_chars": 5953, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Two open PRs on heterogeneous support: one is a broad implementation of per-layer configuration plus modeling, while the other is a narrower config-only subset. They overlap heavily in docs/config code, but they do not look like the same concrete change.", - "confidence": 0.83, + "summary": "These are closely related but not duplicates: #45333 adds per-layer heterogeneous configuration support, while #45332 extends that work into full heterogeneous model/runtime support. They overlap on the config/doc/test scaffolding, but they target different scopes and code paths.", + "confidence": 0.9, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:45332 is the better canonical PR because it includes the full heterogeneous feature set, extending beyond configuration into modeling/runtime support; the other PR is a narrower config-only slice.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "pull_request:45332 is the best representative of the cluster because it is the more complete implementation and subsumes the config-only work.", + "best_pr_reason": "#45332 is the better cluster representative because it is the broader, end-to-end heterogeneous modeling PR and subsumes the config-layer groundwork rather than only the configuration piece.", "soft_edge_verdicts": [ { "left": "pull_request:45332", "right": "pull_request:45333", "accept": false, - "reason": "Reject: 45333 appears to be a config-only precursor/subset, while 45332 adds additional modeling and integration changes. They overlap, but they are not the same underlying PR-level change and would not naturally merge as one duplicate fix." + "reason": "Related scope, but not the same change: #45332 covers heterogeneous modeling/runtime integration across cache, masking, model utils, and model support, while #45333 is limited to per-layer config plumbing. They could be sequential PRs, not one duplicate PR." } ] }, @@ -23367,7 +23283,7 @@ { "label": "pull_request", "component_index": 21, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23383,12 +23299,12 @@ "soft_pairs": [ "pull_request:45415|pull_request:45425" ], - "prepared_review_unit_hash": "3f574df17165e3c50956b62af7a6d257450b36401ddcbb268f0222f00f327ba2", + "prepared_review_unit_hash": "a91af0794df3a69dd19919667b2a0e36349d231dfe89488b0940c4494557c823", "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 }, @@ -23396,7 +23312,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5514, + "serialized_chars": 5513, "estimated_input_tokens": 1379, "estimated_eval_tokens": 3014 }, @@ -23405,18 +23321,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both items are typing-related PRs, but they target different changes: one is a broad type-checking sweep across many transformer files, while the other adds/adjusts typing helpers and protocols for specific modeling code. The filename overlap is not enough to make them duplicates.", - "confidence": 0.92, + "summary": "These are both typing-focused PRs, but they are not the same change: one is a broad repository-wide type-checking sweep, while the other is a narrower typing support update centered on `_typing.py` and `modeling_utils`-related protocols.", + "confidence": 0.89, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45425 is the more focused, concrete change: it adds typing infrastructure and specific protocol aliases needed by a small set of modules. It is a better representative of the narrower underlying work than the broad type-checking sweep in 45415.", + "canonical_pr_reason": "PR 45425 is the more coherent concrete typing change: it introduces specific helper protocols and typing aliases used by `modeling_utils` and related code. PR 45415 is a much broader type-checking sweep across many files.", "best_issue_reason": null, - "best_pr_reason": "PR 45425 is the best standalone PR in this cluster because it has a single coherent purpose and a clear code-path impact, whereas 45415 is a wide, repo-spanning typing cleanup rather than one concrete fix.", + "best_pr_reason": "45425 is the best representative PR for the cluster because it is focused on one typing problem domain and appears more self-contained; 45415 is too broad and mostly a repo-wide annotation pass.", "soft_edge_verdicts": [ { "left": "pull_request:45415", "right": "pull_request:45425", "accept": false, - "reason": "Do not merge: both are typing-related, but 45415 is a broad type-checking pass over many files while 45425 adds specific typing protocols/helpers for modeling and integration code. They are not the same underlying bug/change and would not plausibly be merged into one PR." + "reason": "Both are typing-related and share some touched files, but they address different scopes: 45415 is a broad annotation/type-checking sweep, while 45425 adds specific typing aliases/protocols for modeling utilities. They do not look like the same underlying change." } ] }, @@ -23430,7 +23346,7 @@ { "label": "pull_request", "component_index": 22, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23446,12 +23362,12 @@ "soft_pairs": [ "pull_request:45548|pull_request:45662" ], - "prepared_review_unit_hash": "dc5214ed52f16ea95356c6ec63331ae813d6858691bda6cf7642bc8a84cb1862", + "prepared_review_unit_hash": "fcc16111e712199093ec244257c70540cc898355d82270f159fca746a4b8dfa4", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -23459,7 +23375,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4507, + "serialized_chars": 4508, "estimated_input_tokens": 1127, "estimated_eval_tokens": 2510 }, @@ -23468,18 +23384,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related by expert parallelism, but they target different bugs and loading paths: one fixes EP + DeepSpeed ZeRO-3/accelerate config handling, the other fixes EP + FSDP2 expert weights being overwritten by rank-0 broadcast. They should not be treated as duplicates.", - "confidence": 0.94, + "summary": "Both pull requests concern expert-parallelism loading behavior, but they fix different concrete failure modes: 45548 targets EP + DeepSpeed ZeRO-3 loading via accelerate launch and config serialization, while 45662 targets EP + FSDP2 where experts can be overwritten by rank-0 broadcast/post-shard handling. They are related by subsystem, not the same underlying change.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #45662 is the better canonical PR for this cluster because it addresses a concrete, well-scoped corruption bug in the model loading path and has a more specific failure mode than #45548.", + "canonical_pr_reason": "45662 is the better representative PR because it addresses a concrete EP loading overwrite bug with a broader, more explicit fix across the loading path and EP integrations.", "best_issue_reason": null, - "best_pr_reason": "PR #45662 is the strongest representative change set: it fixes a distinct code-path problem with expert weights being overwritten during distributed loading, and its edits align tightly with that bug.", + "best_pr_reason": "45662 is the strongest standalone PR in this cluster: the title and diff point to a specific bug (experts being silently overwritten) and the code changes align with that failure mode.", "soft_edge_verdicts": [ { "left": "pull_request:45548", "right": "pull_request:45662", "accept": false, - "reason": "Do not merge: both concern EP, but #45548 fixes distributed_config serialization and accelerate/DeepSpeed ZeRO-3 loading, while #45662 fixes rank-0 broadcast overwriting experts under FSDP2. Different concrete bugs and different fix paths." + "reason": "Related EP-loading area, but different bugs and code paths: 45548 is about serialize/config handling for accelerate+ZeRO-3, while 45662 is about post-shard wrapping and rank-0 overwrite in FSDP2." } ] }, @@ -23493,7 +23409,7 @@ { "label": "pull_request", "component_index": 23, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23509,7 +23425,7 @@ "soft_pairs": [ "pull_request:45639|pull_request:45645" ], - "prepared_review_unit_hash": "1911561d479c46b18cd04b03a47e179a7029b325f25f58eaef55d619b8e88740", + "prepared_review_unit_hash": "e5006ba79292d7e33f6f3a01301321f894b2378d0379bd78e7c2e592efc7ef69", "component_budget": { "node_count": 2, "item_count": 2, @@ -23531,18 +23447,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs target the same xdist-related captured_info collision bug. PR 45645 is the broader, more complete fix because it includes the worker-safe log handling plus the CI workflow adjustment to collect multiple captured_info files.", - "confidence": 0.91, + "summary": "Both PRs address the same underlying xdist/captured_info logging collision in testing_utils, with PR 45645 expanding the fix to CI artifact collection as well. PR 45645 is the broader, more complete change set.", + "confidence": 0.88, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45645 is the better canonical PR because it covers the full underlying problem: fixing xdist collisions in the patched testing logs and preserving those logs in CI.", + "canonical_pr_reason": "PR 45645 is the better canonical PR because it covers the core xdist-safe file handling plus the workflow-side log aggregation needed to preserve multiple captured_info artifacts.", "best_issue_reason": null, - "best_pr_reason": "PR 45645 is the best representative PR for the cluster since it contains the core code-path fix and the follow-up workflow change needed to keep debug artifacts visible.", + "best_pr_reason": "PR 45645 is the strongest representative of the cluster: it fixes the collision in the code path and also updates CI handling, making it the more complete end-to-end solution.", "soft_edge_verdicts": [ { "left": "pull_request:45639", "right": "pull_request:45645", "accept": true, - "reason": "Both PRs address the same captured_info xdist collision in testing_utils; 45645 is a superset that extends the same fix to CI log collection, so they look like the same underlying change." + "reason": "Both PRs target the same captured_info/xdist collision problem in the testing debug-log path, and PR 45645 looks like a superset that could plausibly absorb PR 45639's core change." } ] }, @@ -23556,7 +23472,7 @@ { "label": "pull_request", "component_index": 24, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23572,7 +23488,7 @@ "soft_pairs": [ "pull_request:45720|pull_request:46259" ], - "prepared_review_unit_hash": "57f5bc271aff5b5b0b6935ddfc15d023967900bdb8395f4a26d976a8cac1848a", + "prepared_review_unit_hash": "0e2717cffb963ceaa8b8606f2ac72780343e88e0f9e771f96d3881e86274dace", "component_budget": { "node_count": 2, "item_count": 2, @@ -23594,18 +23510,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These two pull requests are both linter-driven model refactors in transformers, but they target different rules and different underlying changes. PR 45720 adds TRF003 coverage for return tuple handling in forward methods, while PR 46259 adds TRF018 modeling-rule changes around weight initialization. They should not be merged as duplicates.", - "confidence": 0.96, + "summary": "These are two unrelated linter/chore PRs: one implements TRF003 across model forwards, the other adds TRF018 initialization-rule changes. They touch overlapping model files, but the underlying code paths and fixes are different, so they should not be merged as duplicates.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: the PRs implement different lint rules and fix different code-path patterns, so they are not the same change.", + "canonical_pr_reason": "PR #45720 is the more focused representative: it addresses TRF003 by adding `can_return_tuple` and adjusting `forward` signatures/return handling across multiple model implementations.", "best_issue_reason": null, - "best_pr_reason": "Neither PR is a good global representative of the other; 45720 is about forward/output API decoration, while 46259 is about initialization logic cleanup. The overlap in filenames is only because both sweep across many model implementations.", + "best_pr_reason": "PR #45720 is the best representative for the cluster because its scope is narrower and more directly tied to a single concrete rule change, making it easier to summarize as the canonical PR.", "soft_edge_verdicts": [ { "left": "pull_request:45720", "right": "pull_request:46259", "accept": false, - "reason": "Different concrete changes: TRF003 adds can_return_tuple/return_dict handling in model forwards, while TRF018 rewrites model weight initialization logic. Shared subsystem and shared files do not make these the same PR." + "reason": "Different lint rules and different fix patterns: TRF003 changes model `forward` return behavior, while TRF018 refactors weight initialization. Shared model files are incidental and not enough to indicate the same underlying change." } ] }, @@ -23619,7 +23535,7 @@ { "label": "pull_request", "component_index": 25, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23685,7 +23601,7 @@ { "label": "pull_request", "component_index": 26, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23704,52 +23620,52 @@ "pull_request:45932|pull_request:46891", "pull_request:46100|pull_request:46891" ], - "prepared_review_unit_hash": "ae1ae610cd0bb628e66e2f702708421fbdcd3976dd3fdfe7357dafb242a7f331", + "prepared_review_unit_hash": "5dcacce9dccc81055f460d5437752d45829d1572675b789b932580830d8cf682", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7879, - "estimated_input_tokens": 1970, - "estimated_eval_tokens": 4196 + "serialized_chars": 7881, + "estimated_input_tokens": 1971, + "estimated_eval_tokens": 4198 }, "budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 3, - "serialized_chars": 7879, - "estimated_input_tokens": 1970, - "estimated_eval_tokens": 4196 + "serialized_chars": 7881, + "estimated_input_tokens": 1971, + "estimated_eval_tokens": 4198 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are three separate open PRs that all happen to touch testing/CI infrastructure, but they target different changes: a grafana/CI secret and model tweak PR, a broad test migration PR adding kernel/model tests, and a PEFT integration CI PR updating workflow branching and peft integration code. They are not duplicates.", + "summary": "These three PRs are only superficially similar because they all touch tests/CI, but they address different changes: grafana/telemetry plumbing, test migration/kernel coverage, and PEFT integration CI. I would not deduplicate any pair.", "confidence": 0.93, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 45932 is the most representative only in a loose cluster sense: it has the broadest diff footprint and the most discussion/review activity, but it still does not match the others as the same underlying change.", + "canonical_pr_reason": "No issue artifact is present; among the PRs, #46891 is the most coherent representative because it has the clearest concrete change set centered on PEFT integration and related CI/test plumbing.", "best_issue_reason": null, - "best_pr_reason": "PR 45932 is the best single representative of the set because it is the most active and cross-cutting PR, but it should not be treated as a duplicate of the others.", + "best_pr_reason": "#46891 is the best single representative: it has the most specific, self-contained purpose and a clearer code-path focus than the mostly infrastructure/test-migration PRs.", "soft_edge_verdicts": [ { "left": "pull_request:45932", "right": "pull_request:46100", "accept": false, - "reason": "Different underlying work: 45932 changes CI secrets plus model code in aimv2/auto-mappings/other model files, while 46100 is a test migration adding kernel tests and DDP YAML. Same general testing theme, but not the same fix/change." + "reason": "Both are test-related, but they change different things: #45932 is grafana/telemetry CI plus model config tweaks, while #46100 adds AltCLIP/ViT kernel and DDP tests. No shared concrete bug or mergeable single change." }, { "left": "pull_request:45932", "right": "pull_request:46891", "accept": false, - "reason": "Both touch .github/workflows/pr-ci-caller.yml, but 45932 adds OTEL secrets and model-related edits, whereas 46891 changes the push branch/workflow target and peft integration code. They are not the same concrete change." + "reason": "They only overlap on a workflow file. #45932 is testing grafana/otel plumbing with unrelated model edits, while #46891 is PEFT integration CI and peft.py/test-fetcher changes. Different underlying changes." }, { "left": "pull_request:46100", "right": "pull_request:46891", "accept": false, - "reason": "46100 is a large test migration adding model/kernel tests, while 46891 is a PEFT integration CI change with workflow edits and small peft-related code. Shared testing/CI context is too broad to merge as duplicates." + "reason": "Both involve tests, but one is a kernel-test migration for AltCLIP/ViT and the other is PEFT integration CI with peft integration code. They do not fix the same code-path problem and would not plausibly merge as one PR." } ] }, @@ -23763,7 +23679,7 @@ { "label": "pull_request", "component_index": 27, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23779,12 +23695,12 @@ "soft_pairs": [ "pull_request:46126|pull_request:46269" ], - "prepared_review_unit_hash": "d11d51fcb7873039c207311da705feb80f5e4e9c0f5963266998d315814774c4", + "prepared_review_unit_hash": "6a0d4528e46620f371a11239f3aef067211851a9e8b39031b7a5fed07e83f67a", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5526, + "serialized_chars": 5525, "estimated_input_tokens": 1382, "estimated_eval_tokens": 3020 }, @@ -23792,7 +23708,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 5526, + "serialized_chars": 5525, "estimated_input_tokens": 1382, "estimated_eval_tokens": 3020 }, @@ -23801,18 +23717,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs overlap in distributed-module filenames, but they are not the same underlying change: one is a focused EP/trainer integration, while the other is a broad distributed branch-base PR with many unrelated edits.", - "confidence": 0.94, + "summary": "These are not duplicates: PR 46126 is a focused EP/Trainer integration feature, while PR 46269 is a broad distributed branch base touching many unrelated areas. The overlap in a few distributed files is not enough to merge them as the same change.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46126 is the more focused and coherent change set, centered on EP router parallel and trainer integration; PR 46269 is a broad umbrella branch-base PR spanning many distributed areas, so it is not a good canonical duplicate anchor.", + "canonical_pr_reason": "PR 46126 is the more canonical change because it targets one concrete feature area (EP + Trainer integration) with a narrower, coherent scope. PR 46269 is a broad base branch update spanning many subsystems, so it is less suitable as the representative PR.", "best_issue_reason": null, - "best_pr_reason": "PR 46126 best represents the specific concrete code-path change in this cluster. PR 46269 is too wide in scope and mixes multiple distributed subsystems, making it a poor representative for duplicate triage.", + "best_pr_reason": "PR 46126 is the best representative of a specific code-path change. PR 46269 is too wide and mixed to serve as a clean duplicate-triage anchor for this cluster.", "soft_edge_verdicts": [ { "left": "pull_request:46126", "right": "pull_request:46269", "accept": false, - "reason": "Shared distributed filenames are not enough here: 46126 is a narrow EP/trainer feature, while 46269 is a large branch-base PR touching many unrelated components. They do not appear to fix the same concrete bug or change." + "reason": "Reject: despite shared distributed filenames, the PRs do not appear to fix the same concrete bug or implement the same change. 46126 is a targeted EP/Trainer integration PR, while 46269 is a broad distributed branch base that bundles many unrelated updates." } ] }, @@ -23826,7 +23742,7 @@ { "label": "pull_request", "component_index": 28, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -23878,208 +23794,208 @@ "pull_request:46504|pull_request:46563", "pull_request:46504|pull_request:46716" ], - "prepared_review_unit_hash": "54f3aa9a77d17ace25d8638d8427ee831119765a6829cf98d87d654d70d0b717", + "prepared_review_unit_hash": "e6a31b6b72a229c62c8b6c75abe8e43081cc6f631adf05fd77549696f2cd2f70", "component_budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 29, - "serialized_chars": 25917, - "estimated_input_tokens": 6480, - "estimated_eval_tokens": 13216 + "serialized_chars": 25924, + "estimated_input_tokens": 6481, + "estimated_eval_tokens": 13218 }, "budget": { "node_count": 10, "item_count": 10, "soft_pair_count": 29, - "serialized_chars": 25917, - "estimated_input_tokens": 6480, - "estimated_eval_tokens": 13216 + "serialized_chars": 25924, + "estimated_input_tokens": 6481, + "estimated_eval_tokens": 13218 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is mostly a grab bag of small workflow-maintenance PRs. Despite the generic similarity signals, they target different workflows and different changes; I do not see any pair that clearly fixes the same underlying bug or could be safely merged into one PR.", - "confidence": 0.92, + "summary": "This cluster is mostly unrelated workflow-maintenance PRs. The pr-ci-security-gate items are thematically similar, but they change different security controls/steps, so they are not true duplicates under a strict mergeability test.", + "confidence": 0.91, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46554 is the strongest representative of the cluster because it makes the most substantive workflow/security hardening change in the shared pr-ci-security-gate area, and the later PRs around that file are closer variants of the same workflow than the others.", + "canonical_pr_reason": "PR 46554 is the most concrete and self-contained workflow-security hardening change in the set: it pins the downloaded monitor script, verifies its checksum, and clears HF_TOKEN before SSH.", "best_issue_reason": null, - "best_pr_reason": "PR 46554 is the best standalone representative: it is concrete, security-relevant, and more central than the other small one-off workflow edits.", + "best_pr_reason": "PR 46554 is the best representative PR for this cluster because it has the clearest behavioral impact and strongest security rationale, unlike the other mostly generic workflow tweak PRs.", "soft_edge_verdicts": [ { "left": "pull_request:46138", "right": "pull_request:46554", "accept": false, - "reason": "Different workflow files and different behaviors: stdin handling for self-comment CI vs. PR security-gate logic. Similar maintenance style, but not the same bug or change." + "reason": "Different workflow files and different behavior; only the generic chore style is shared." }, { "left": "pull_request:46138", "right": "pull_request:46558", "accept": false, - "reason": "Unrelated workflows and different code paths. One updates self-comment CI command piping; the other changes PR security-gate permissions/approval flow." + "reason": "Unrelated workflows: one changes self-comment CI input plumbing, the other hardens the PR CI security gate." }, { "left": "pull_request:46138", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow responsibilities: self-comment CI vs. a caller note in pr-ci-security-gate. No shared concrete fix." + "reason": "Different workflow purposes and code paths; the edits are not the same change." }, { "left": "pull_request:46138", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow files and unrelated changes. The first edits a CI helper invocation; the second changes reusable workflow secrets handling." + "reason": "Different workflow files and different semantics; no shared concrete bug fix." }, { "left": "pull_request:46138", "right": "pull_request:46504", "accept": false, - "reason": "Different workflows with unrelated fixes: comment handling in one workflow vs. commit-message parsing in build-ci-docker-images." + "reason": "One updates self-comment CI stdin handling, the other tweaks commit-message matching in build image CI." }, { "left": "pull_request:46432", "right": "pull_request:46503", "accept": false, - "reason": "Both are workflow edits, but one is SSH-runner hardening and the other enables benchmark_v2 on workflow_dispatch. Not the same underlying change." + "reason": "SSH runner hardening and benchmark trigger logic are unrelated changes." }, { "left": "pull_request:46432", "right": "pull_request:46501", "accept": false, - "reason": "Different workflow files and distinct fixes: SSH-runner secret handling versus anti-slop permissions placement." + "reason": "Different workflows and different fixes; no common underlying bug." }, { "left": "pull_request:46432", "right": "pull_request:46502", "accept": false, - "reason": "Different workflows and unrelated behavior changes. The checkout ref tweak in assign-reviewers is not the same as SSH-runner security updates." + "reason": "SSH runner changes do not match reviewer-assignment checkout behavior." }, { "left": "pull_request:46432", "right": "pull_request:46554", "accept": false, - "reason": "Both concern CI security, but they modify different workflows and different mechanisms. Not one concrete bug/change." + "reason": "Different workflows with unrelated security/automation changes." }, { "left": "pull_request:46432", "right": "pull_request:46558", "accept": false, - "reason": "SSH-runner hardening and PR CI security-gate permission changes are separate workflow concerns, not a single mergeable fix." + "reason": "Not the same problem: SSH runner token handling vs PR CI gate permissions/result flow." }, { "left": "pull_request:46501", "right": "pull_request:46503", "accept": false, - "reason": "Different workflow files and different triggers/permissions. Similar broad area only." + "reason": "Both touch CI workflows, but one is anti-slop permissions and the other is benchmark dispatch behavior." }, { "left": "pull_request:46501", "right": "pull_request:46502", "accept": false, - "reason": "Anti-slop permissions and assign-reviewers checkout ref are unrelated workflow changes." + "reason": "Different workflow files and different automation concerns." }, { "left": "pull_request:46501", "right": "pull_request:46504", "accept": false, - "reason": "Different files and distinct fixes. No shared concrete code-path problem." + "reason": "Anti-slop permissions and build-ci image triggering are unrelated." }, { "left": "pull_request:46501", "right": "pull_request:46554", "accept": false, - "reason": "Security-related in a broad sense, but these are different workflows with different logic and different failure modes." + "reason": "Different workflows and different concrete changes; not mergeable as one fix." }, { "left": "pull_request:46502", "right": "pull_request:46503", "accept": false, - "reason": "Assign-reviewers checkout ref and benchmark dispatch support are unrelated workflow adjustments." + "reason": "Reviewer checkout ref fix and benchmark trigger expansion are unrelated." }, { "left": "pull_request:46502", "right": "pull_request:46504", "accept": false, - "reason": "Different workflows and changes: reviewer assignment checkout ref vs. commit-message parsing in docker-image build CI." + "reason": "Assign-reviewers checkout semantics do not match build-ci commit parsing." }, { "left": "pull_request:46502", "right": "pull_request:46554", "accept": false, - "reason": "Not the same bug or code path; one affects reviewer assignment, the other PR security-gate behavior." + "reason": "Different workflows and different code paths; no shared bug/change." }, { "left": "pull_request:46503", "right": "pull_request:46504", "accept": false, - "reason": "Different workflows and unrelated operational fixes." + "reason": "Benchmark dispatch logic and build-ci message parsing are distinct automation changes." }, { "left": "pull_request:46503", "right": "pull_request:46554", "accept": false, - "reason": "Benchmark trigger support and PR security gating are separate workflow changes." + "reason": "Different workflows with no common underlying fix." }, { "left": "pull_request:46504", "right": "pull_request:46554", "accept": false, - "reason": "Same general CI ecosystem, but different workflows and distinct fixes." + "reason": "Build-ci commit-message parsing and SSH runner security hardening are unrelated." }, { "left": "pull_request:46504", "right": "pull_request:46558", "accept": false, - "reason": "Build-ci-docker-images commit parsing and PR CI security-gate permissions are unrelated." + "reason": "Different workflows and different fixes; only the broad CI theme overlaps." }, { "left": "pull_request:46554", "right": "pull_request:46558", "accept": false, - "reason": "Both touch pr-ci-security-gate.yml, but one edits the reusable security gate implementation while the other changes the caller workflow's permissions/approval wiring. Different concrete changes." + "reason": "Same workflow file, but different hardening edits affecting different steps and semantics." }, { "left": "pull_request:46554", "right": "pull_request:46563", "accept": false, - "reason": "The reusable workflow implementation change and the caller-side documentation comment are not the same underlying bug or fix." + "reason": "Both touch pr-ci-security-gate.yml, but one adds checks/checkout while the other changes reusable-workflow commentary; not the same fix." }, { "left": "pull_request:46554", "right": "pull_request:46716", "accept": false, - "reason": "One changes the reusable workflow implementation; the other changes the caller's secrets passing. Related file area, but not one mergeable fix." + "reason": "Security-gate hardening vs secrets-passing change in a reusable workflow call; different concrete changes." }, { "left": "pull_request:46558", "right": "pull_request:46563", "accept": false, - "reason": "Caller workflow permission/approval logic and a comment-only safety note are not the same change." + "reason": "Same file, but one changes permissions and gating, the other adds safety notes to a reusable workflow call." }, { "left": "pull_request:46558", "right": "pull_request:46716", "accept": false, - "reason": "Both are caller-side changes, but one adjusts permissions/approval flow and the other changes secrets inheritance. Different fixes and could not be treated as one bug." + "reason": "Permissions/approval-flow tweak is not the same change as replacing inherited secrets with an empty secrets map." }, { "left": "pull_request:46563", "right": "pull_request:46716", "accept": false, - "reason": "A documentation comment and a functional secrets-handling change are not the same underlying issue." + "reason": "Both relate to the same reusable workflow, but one only documents safety and the other changes secrets handling; not duplicate." }, { "left": "pull_request:46504", "right": "pull_request:46563", "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Different workflow files and different automation logic." }, { "left": "pull_request:46504", "right": "pull_request:46716", "accept": false, - "reason": "Different workflow files and unrelated changes." + "reason": "Build-ci docker image tagging and PR CI reusable workflow secrets are unrelated." } ] }, @@ -24093,7 +24009,7 @@ { "label": "pull_request", "component_index": 29, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24109,7 +24025,7 @@ "soft_pairs": [ "pull_request:46152|pull_request:46207" ], - "prepared_review_unit_hash": "e787edfe9aaaa687b7f8664f1bd1260da40edefb6837081ba7e27b254de38718", + "prepared_review_unit_hash": "d9fcb4695f21a615f1d0ede6ef7b643885a033f71411d7c337bbe9171a737670", "component_budget": { "node_count": 2, "item_count": 2, @@ -24131,18 +24047,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are related in theme (replacing file-open source introspection), but they target different code paths and are not the same fix. They should not be deduplicated as one PR.", - "confidence": 0.89, + "summary": "These PRs share a broad theme (replacing file-open source introspection), but they fix different code paths in different modules, so they are not duplicates.", + "confidence": 0.95, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:46207 is the clearest representative because it directly replaces source-file opening in a specific runtime check with inspect.getsource and includes test coverage.", + "canonical_pr_reason": "No canonical PR: PR 46152 and PR 46207 address separate source-reading implementations in different files with different behavior changes.", "best_issue_reason": null, - "best_pr_reason": "pull_request:46207 is the best standalone PR in this cluster: the change is concrete, narrowly scoped, and the bug/fix relationship is explicit.", + "best_pr_reason": "PR 46207 is the strongest standalone PR here because it has a clear, localized fix in modeling_utils.py and includes tests, making it the better representative of the cluster.", "soft_edge_verdicts": [ { "left": "pull_request:46152", "right": "pull_request:46207", "accept": false, - "reason": "Both avoid reading source via open(), but they modify different modules and different mechanisms (_resolve_traversable in import_utils vs _can_set_attn_implementation in modeling_utils). They do not appear to be the same underlying bug or mergeable into one PR." + "reason": "Both replace file-open source access, but one fixes import_utils.py traversal logic and the other fixes _can_set_attn_implementation in modeling_utils.py; different underlying bugs and not mergeable into one PR." } ] }, @@ -24156,7 +24072,7 @@ { "label": "pull_request", "component_index": 30, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24219,7 +24135,7 @@ { "label": "pull_request", "component_index": 31, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24235,12 +24151,12 @@ "soft_pairs": [ "pull_request:46281|pull_request:46890" ], - "prepared_review_unit_hash": "65d03bf996a4490edf35e9d6f0f57b04aa43f805d1fa8343f6793e47f629cda6", + "prepared_review_unit_hash": "858b17423563424d455f7837bf2a659640e4a9762ea9e447ace26284964c4e57", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4269, + "serialized_chars": 4270, "estimated_input_tokens": 1068, "estimated_eval_tokens": 2392 }, @@ -24248,7 +24164,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4269, + "serialized_chars": 4270, "estimated_input_tokens": 1068, "estimated_eval_tokens": 2392 }, @@ -24257,18 +24173,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs harden local checkpoint loading against path traversal, but they protect different untrusted inputs in different code paths (`weight_map` shard filenames vs `transformers_weights` explicit filename), so they are not duplicates.", - "confidence": 0.95, + "summary": "Both pull requests harden local-loading code against path traversal, but they fix different untrusted inputs in different code paths: sharded checkpoint filenames from `weight_map` vs. `transformers_weights` from config. Similar security theme, not the same underlying bug or mergeable change.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "No single canonical PR for a duplicate cluster: PR #46281 fixes shard-file path traversal from the checkpoint index, while PR #46890 fixes a separate explicit-filename path traversal in `transformers_weights`.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "If forced to pick one representative, #46281 is the cleaner canonical example of the cluster\u2019s security theme, but it should not subsume #46890 because the underlying bugs differ.", + "best_pr_reason": "PR 46890 is the cleaner standalone representative of the security-hardening theme: it adds explicit containment checks for one untrusted local config field with a narrowly scoped fix and test coverage.", "soft_edge_verdicts": [ { "left": "pull_request:46281", "right": "pull_request:46890", "accept": false, - "reason": "Different vulnerable inputs and call paths: #46281 validates shard filenames from the index `weight_map` in `hub.py`, while #46890 validates `transformers_explicit_filename` from config in `modeling_utils.py`. Same broad security theme, not the same bug." + "reason": "Different attack surfaces and loaders: 46281 validates shard paths from a checkpoint index, while 46890 validates `transformers_weights` from config. Same broad CWE-22 theme, but not the same concrete bug or a single PR candidate." } ] }, @@ -24282,7 +24198,7 @@ { "label": "pull_request", "component_index": 32, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24298,7 +24214,7 @@ "soft_pairs": [ "pull_request:46310|pull_request:46546" ], - "prepared_review_unit_hash": "c1d0e855200c90e14d4067786409bb1f4fcc92bda1cc0849463721999c96a1fc", + "prepared_review_unit_hash": "802991a032ccebc67464dcddfd5b4c3e0687506ac95b88b994fc7d49f874f1d8", "component_budget": { "node_count": 2, "item_count": 2, @@ -24320,18 +24236,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "Both PRs fix the same off-by-one sliding-window boundary in Moonshine streaming\u2019s mask function; PR 46546 is the broader, more complete version with tests and an additional Gemma4 fix.", - "confidence": 0.91, + "summary": "Both PRs address the same sliding-window right-boundary off-by-one fix in Moonshine streaming; the later PR expands the same correction to Gemma4 and adds tests.", + "confidence": 0.87, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46546 is the better canonical PR because it addresses the same Moonshine boundary bug and includes a test plus an additional related fix, making it the more complete change set.", + "canonical_pr_reason": "PR #46546 is the better canonical PR because it contains the same Moonshine streaming boundary fix plus a test update and a related Gemma4 application, making it the more complete change set.", "best_issue_reason": null, - "best_pr_reason": "PR 46546 is the best representative of the cluster since it contains the same core bug fix as 46310 and adds validation and a second implementation update.", + "best_pr_reason": "PR #46546 is the strongest representative of the bug fix because it covers the core Moonshine streaming code path, includes validation, and appears to subsume the smaller Moonshine-only patch in #46310.", "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 from `<` to `<=`, fixing the same underlying off-by-one bug in the same code path; 46546 is a superset, so they are duplicate-compatible." + "reason": "Same off-by-one right-window boundary bug in the Moonshine streaming mask; the second PR applies the same fix in the same code path and could plausibly subsume the first." } ] }, @@ -24345,7 +24261,7 @@ { "label": "pull_request", "component_index": 33, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24353,72 +24269,87 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-46376-5", + "cluster_id": "cluster-46376-6", "nodes": [ "pull_request:46376", "pull_request:46417", "pull_request:46425", "pull_request:46509", - "pull_request:46850" + "pull_request:46850", + "pull_request:46917" ], "soft_pairs": [ "pull_request:46376|pull_request:46417", "pull_request:46376|pull_request:46425", "pull_request:46425|pull_request:46509", - "pull_request:46425|pull_request:46850" + "pull_request:46425|pull_request:46917", + "pull_request:46425|pull_request:46850", + "pull_request:46850|pull_request:46917" ], - "prepared_review_unit_hash": "6aaf5a1e829f3c7e018f3cc7e35a9b88a6312b3bf1ebdf083f151c74a55794e8", + "prepared_review_unit_hash": "aa876c0dfcac8c902d32330dffae7955c0572bdedef44c3a1462fc1e2dbc536c", "component_budget": { - "node_count": 5, - "item_count": 5, - "soft_pair_count": 4, - "serialized_chars": 14158, - "estimated_input_tokens": 3540, - "estimated_eval_tokens": 7336 + "node_count": 6, + "item_count": 6, + "soft_pair_count": 6, + "serialized_chars": 16835, + "estimated_input_tokens": 4209, + "estimated_eval_tokens": 8674 }, "budget": { - "node_count": 5, - "item_count": 5, - "soft_pair_count": 4, - "serialized_chars": 14158, - "estimated_input_tokens": 3540, - "estimated_eval_tokens": 7336 + "node_count": 6, + "item_count": 6, + "soft_pair_count": 6, + "serialized_chars": 16835, + "estimated_input_tokens": 4209, + "estimated_eval_tokens": 8674 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All candidates are distinct PRs: three are separate model-support additions, one is a continuous-batching multimodal feature, and one is a Gemma masking bugfix. The soft-similarity signals are broad and do not indicate true duplicates.", - "confidence": 0.93, + "summary": "This cluster is heterogeneous: multimodal continuous batching support, new model additions (HunYuan VL, RADIO, Nemotron Omni), a Gemma masking bug fix, and a deprecation cleanup. They are not duplicates of one another.", + "confidence": 0.96, "canonical_issue_reason": null, - "canonical_pr_reason": "pull_request:46425 is the most self-contained standalone model-support PR, but this cluster does not contain a true duplicate canonical pair.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "pull_request:46425 is the most representative standalone PR in this mixed set; the others are either a much larger feature or an unrelated bugfix.", + "best_pr_reason": null, "soft_edge_verdicts": [ { "left": "pull_request:46376", "right": "pull_request:46417", "accept": false, - "reason": "Different subsystems and code paths: continuous batching/multimodal runtime vs adding a specific model implementation and docs." + "reason": "Different changes: continuous batching multimodal infrastructure vs adding a new HunYuan VL model and docs/tests." }, { "left": "pull_request:46376", "right": "pull_request:46425", "accept": false, - "reason": "One is a continuous-batching feature; the other is a new RADIO model support PR. Same broad ecosystem, not the same change." + "reason": "Different targets and code paths: continuous batching cache/scheduler work vs RADIO model support and auto-mapping/docs." }, { "left": "pull_request:46425", "right": "pull_request:46509", "accept": false, - "reason": "Both add model support, but for different architectures and files; they are separate model integrations, not one fix." + "reason": "Both are model additions, but for different architectures and code paths (RADIO vs Nemotron Omni), so not the same change." + }, + { + "left": "pull_request:46425", + "right": "pull_request:46917", + "accept": false, + "reason": "RADIO model support is unrelated to removing deprecated backend/training-args behavior." }, { "left": "pull_request:46425", "right": "pull_request:46850", "accept": false, - "reason": "RADIO model support and Gemma sliding-window attention masking are unrelated changes with no shared concrete bug or patch target." + "reason": "RADIO model support and Gemma bidirectional/sliding-window masking fixes are unrelated changes." + }, + { + "left": "pull_request:46850", + "right": "pull_request:46917", + "accept": false, + "reason": "Gemma masking logic fix is unrelated to the deprecation cleanup in image processing/training args." } ] }, @@ -24432,7 +24363,7 @@ { "label": "pull_request", "component_index": 34, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24448,7 +24379,7 @@ "soft_pairs": [ "pull_request:46574|pull_request:46580" ], - "prepared_review_unit_hash": "9307db0811f971f8a5317e83dbb28e9ff519c969b59353296ccdcdfa4c6f50cf", + "prepared_review_unit_hash": "4c89d7176224392603d7e05e08077f0129adc27dc78e1c0784e2ece1d15ec673", "component_budget": { "node_count": 2, "item_count": 2, @@ -24470,18 +24401,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These PRs are not duplicates: one fixes CodeLlama decode behavior in the tokenizer implementation, the other fixes DeepSeek-Coder/DeepSeek-LLM loading to the correct tokenizer class in auto-tokenization. They affect different code paths and should remain separate.", - "confidence": 0.99, + "summary": "These two PRs are unrelated: one fixes CodeLlama decode behavior around leading spaces, the other fixes DeepSeek checkpoint routing to the correct tokenizer class. They touch different code paths, different models, and would not plausibly merge into one change.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": "No canonical PR: 46574 is a CodeLlama decode-space fix, while 46580 is a model-class selection fix for DeepSeek checkpoints.", + "canonical_pr_reason": null, "best_issue_reason": null, - "best_pr_reason": "No single best PR for the cluster; the two PRs solve unrelated bugs in different modules and do not subsume one another.", + "best_pr_reason": "No single PR is a good canonical representative because the changes address different bugs in different tokenizer code paths; neither should be treated as the cluster\u2019s sole fix.", "soft_edge_verdicts": [ { "left": "pull_request:46574", "right": "pull_request:46580", "accept": false, - "reason": "Different underlying problems: CodeLlama decoder stripping leading spaces vs. DeepSeek checkpoints loading the wrong tokenizer class. They do not share the same concrete code path or merge into one PR cleanly." + "reason": "Different underlying problems: CodeLlama decode stripping vs. DeepSeek tokenizer-class misloading. Separate models, separate files, separate fixes." } ] }, @@ -24495,7 +24426,7 @@ { "label": "pull_request", "component_index": 35, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24511,7 +24442,7 @@ "soft_pairs": [ "pull_request:46603|pull_request:46604" ], - "prepared_review_unit_hash": "bd0b6e72a1ee96d9b252868a9e1b5b3dfaaec2077b61a2b201a2e70ca5bf0eff", + "prepared_review_unit_hash": "299fc604e791f9bc3f4e1e380c871cf9e04185c22ae6a3cbc05e8c1ab85c19cd", "component_budget": { "node_count": 2, "item_count": 2, @@ -24533,18 +24464,18 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "These are two distinct Mistral PRs that overlap in touched files but not in underlying change: one moves `MistralConverter` into the `integrations/mistral` package, while the other adds tekken tokenizer detection/conversion/save utilities. Shared filenames are incidental, not a duplicate signal.", - "confidence": 0.95, + "summary": "These are related to Mistral internals but they are different PRs: one refactors/moves the existing MistralConverter, the other adds tekken tokenizer detection/conversion/save support. They should not be deduplicated.", + "confidence": 0.97, "canonical_issue_reason": null, - "canonical_pr_reason": "PR 46604 is the more substantial and representative Mistral tokenizer change; PR 46603 is mainly a refactor/move of existing converter code.", + "canonical_pr_reason": "No canonical PR: 46603 is a package move/refactor for MistralConverter, while 46604 introduces new tekken tokenizer functionality. Different underlying changes.", "best_issue_reason": null, - "best_pr_reason": "PR 46604 is the strongest standalone PR here because it introduces new tokenizer behavior and utilities, whereas 46603 is a narrower packaging refactor.", + "best_pr_reason": "Neither PR is a good global duplicate representative because they solve different problems in the same subsystem.", "soft_edge_verdicts": [ { "left": "pull_request:46603", "right": "pull_request:46604", "accept": false, - "reason": "Do not merge: 46603 relocates `MistralConverter` into `integrations/mistral`, while 46604 adds tekken tokenizer detection, conversion, and save support. They touch the same Mistral area but fix different problems and would not plausibly be one PR." + "reason": "Shared Mistral files do not imply the same change; 46603 is a refactor/move of MistralConverter, whereas 46604 adds tekken tokenizer detection, conversion, and save utilities. They could not plausibly be merged into one PR as the same fix." } ] }, @@ -24558,7 +24489,7 @@ { "label": "pull_request", "component_index": 36, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24576,12 +24507,12 @@ "pull_request:46741|pull_request:46773", "pull_request:46741|pull_request:46781" ], - "prepared_review_unit_hash": "26d606a82b35a0e1ea0fba4a232190a1db92867f163426afd9dfcf96b73678d5", + "prepared_review_unit_hash": "7e7918e6821bece01ec0f5c28bd7eb2c2b45d76020477bfbec6cbcc38081cd55", "component_budget": { "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 8456, + "serialized_chars": 8455, "estimated_input_tokens": 2114, "estimated_eval_tokens": 4484 }, @@ -24589,7 +24520,7 @@ "node_count": 3, "item_count": 3, "soft_pair_count": 2, - "serialized_chars": 8456, + "serialized_chars": 8455, "estimated_input_tokens": 2114, "estimated_eval_tokens": 4484 }, @@ -24598,24 +24529,24 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "All three items are pull requests and they address different bugs: two model-forward-path fixes and one QuantizedCache hybrid-model fix. Despite some shared model filenames, they are not the same underlying change, so the soft links should be rejected.", - "confidence": 0.93, + "summary": "These PRs are thematically related because they touch overlapping model families and cache/padding paths, but they fix three different bugs: chunked-prefill/speculative decoding, padding-mask handling for batch size 1, and QuantizedCache construction for hybrid models.", + "confidence": 0.92, "canonical_issue_reason": null, - "canonical_pr_reason": "PR #46741 is the strongest canonical PR: it targets a concrete execution-path bug in chunked-prefill/speculative decoding across several linear-attention/hybrid models and has the clearest code-path-specific fix.", + "canonical_pr_reason": "No true duplicate PR exists here; if one must be chosen as the representative, 46741 is the broadest and most central model-path fix across the affected family.", "best_issue_reason": null, - "best_pr_reason": "PR #46741 is the best representative of the cluster because it fixes a specific runtime bug in model forward logic, is broader in model coverage than #46773, and is more directly related to the shared Bamba/Falcon/Granite/Nemotron/Zamba code paths than the cache-layer change in #46781.", + "best_pr_reason": "46741 is the strongest standalone candidate because it targets a concrete runtime bug in the core forward/cache path and spans the same affected model family more directly than the others.", "soft_edge_verdicts": [ { "left": "pull_request:46741", "right": "pull_request:46773", "accept": false, - "reason": "Shared filenames are not enough here: #46741 fixes chunked-prefill/speculative decoding state handling, while #46773 fixes padding-mask application for batch-size-1 linear-attention models. Different symptoms, different conditions, different code paths." + "reason": "Different root causes: 46741 changes precomputed-state handling for chunked prefill/speculative decoding, while 46773 fixes padding-mask application when batch size is 1. Shared files/models are not enough to make them the same bug." }, { "left": "pull_request:46741", "right": "pull_request:46781", "accept": false, - "reason": "#46741 is a model forward-pass fix; #46781 changes QuantizedCache construction for hybrid SSM+attention models in cache_utils. They touch unrelated failure modes and would not plausibly be merged as one PR." + "reason": "Different code paths and failures: 46741 is about cache reuse in model forward passes, whereas 46781 fixes QuantizedCache initialization for hybrid layer types. They are related to caching, but not the same underlying issue." } ] }, @@ -24629,7 +24560,7 @@ { "label": "pull_request", "component_index": 37, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24637,78 +24568,63 @@ "source": "llm", "cache_hit": false, "model": "gpt-5.4-mini?service_tier=flex", - "cluster_id": "cluster-46895-4", + "cluster_id": "cluster-46895-3", "nodes": [ "pull_request:46895", "pull_request:46898", - "pull_request:46903", "pull_request:46916" ], "soft_pairs": [ "pull_request:46895|pull_request:46898", "pull_request:46895|pull_request:46916", - "pull_request:46895|pull_request:46903", - "pull_request:46898|pull_request:46916", - "pull_request:46898|pull_request:46903" + "pull_request:46898|pull_request:46916" ], - "prepared_review_unit_hash": "39218a15dec2b2c04a7cc64e8a0b0d9d23cc19f5fa0296e1cf5a76353a701cd4", + "prepared_review_unit_hash": "45350dd8e9b6741bafb86181dbcd404cfd2b8f6d113a17b3ea3347554c11376b", "component_budget": { - "node_count": 4, - "item_count": 4, - "soft_pair_count": 5, - "serialized_chars": 10575, - "estimated_input_tokens": 2644, - "estimated_eval_tokens": 5544 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7714, + "estimated_input_tokens": 1929, + "estimated_eval_tokens": 4114 }, "budget": { - "node_count": 4, - "item_count": 4, - "soft_pair_count": 5, - "serialized_chars": 10575, - "estimated_input_tokens": 2644, - "estimated_eval_tokens": 5544 + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7714, + "estimated_input_tokens": 1929, + "estimated_eval_tokens": 4114 }, "overflow_policy": "truncate_then_skip", "trimmed": false, "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "This cluster is about a repeated training-loss double-shift pattern across several model-specific PRs. The Florence2 and CohereASR changes overlap with the broader roll-up PR 46916, but PPFormulaNet is a separate code path and the cross-model pairings should not be merged just because they share the same bug pattern.", - "confidence": 0.78, + "summary": "Three open PRs address the same family of training-loss double-shift bugs, but they split across two model code paths: CohereASR and Florence2. The combined PR 46916 is the best representative because it includes the CohereASR fix and also covers Florence2.", + "confidence": 0.79, "canonical_issue_reason": null, - "canonical_pr_reason": "46916 is the broadest representative: it covers the exact Florence2 and CohereASR loss-path changes in one PR, making it the best cluster anchor even though it bundles two model fixes.", + "canonical_pr_reason": "PR 46916 is the broadest and most central change: it includes the CohereASR loss fix from 46895 and the Florence2 fix from 46898, so it best represents the shared double-shift correction work in this cluster.", "best_issue_reason": null, - "best_pr_reason": "46916 best represents the shared training-loss double-shift fix because it includes both affected model paths and the same loss computation change, while the others are narrower model-specific variants.", + "best_pr_reason": "PR 46916 is the strongest cluster anchor because it subsumes one concrete fix from each of the two smaller PRs and directly reflects the common training-loss double-shift pattern.", "soft_edge_verdicts": [ { "left": "pull_request:46895", "right": "pull_request:46898", "accept": false, - "reason": "Same general symptom, but different models and code paths (CohereASR vs Florence2); too broad to treat as the same PR change." + "reason": "They fix the same kind of bug pattern, but in different model implementations and different concrete code paths (CohereASR vs Florence2), so they are not the same PR-level change." }, { "left": "pull_request:46895", "right": "pull_request:46916", "accept": true, - "reason": "Both fix the CohereASR training-loss double-shift on the same forward/loss path; 46916 just bundles an additional Florence2 fix." - }, - { - "left": "pull_request:46895", - "right": "pull_request:46903", - "accept": false, - "reason": "PPFormulaNet is a separate model and implementation; this is not the same concrete bug/change as the CohereASR fix." + "reason": "46916 includes the same CohereASR loss-fix code path as 46895, so this is the same underlying CohereASR change even though 46916 also adds Florence2." }, { "left": "pull_request:46898", "right": "pull_request:46916", "accept": true, - "reason": "Both touch the Florence2 training-loss double-shift path, so they address the same underlying bug in the same model." - }, - { - "left": "pull_request:46898", - "right": "pull_request:46903", - "accept": false, - "reason": "Different models and loss implementations; no shared concrete code-path bug to merge." + "reason": "46916 includes the same Florence2 double-shift fix as 46898, so they overlap on the same underlying Florence2 change." } ] }, @@ -24722,7 +24638,7 @@ { "label": "pull_request", "component_index": 38, - "component_count": 38, + "component_count": 39, "review_unit_index": 1, "review_unit_count": 1, "status": "reviewed", @@ -24738,12 +24654,12 @@ "soft_pairs": [ "pull_request:46907|pull_request:46908" ], - "prepared_review_unit_hash": "0ee4faccda813f7fe6cec612d05e150972cfe12710b67830e47c4b2b9d618541", + "prepared_review_unit_hash": "0b308c8fd0da32e7663075c1b69aa955e780fa85e3cfb54111be129f74c47547", "component_budget": { "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4383, + "serialized_chars": 4384, "estimated_input_tokens": 1096, "estimated_eval_tokens": 2448 }, @@ -24751,7 +24667,7 @@ "node_count": 2, "item_count": 2, "soft_pair_count": 1, - "serialized_chars": 4383, + "serialized_chars": 4384, "estimated_input_tokens": 1096, "estimated_eval_tokens": 2448 }, @@ -24760,18 +24676,81 @@ "aggressively_trimmed": false, "split": false, "analyst_result": { - "summary": "The two PRs are unrelated: one fixes EdgeTAM\u2019s forward pass by stripping an invalid processor kwarg, while the other changes DeepSeekV32 tests to provide an offload folder during loading.", - "confidence": 0.96, + "summary": "These two PRs are unrelated: one fixes EdgeTam\u2019s forward pass by dropping an invalid `original_sizes` kwarg, while the other adjusts DeepSeekV32 integration tests to supply an offload folder during `from_pretrained` loading.", + "confidence": 0.98, "canonical_issue_reason": null, - "canonical_pr_reason": null, + "canonical_pr_reason": "PR 46907 is the better canonical representative because it fixes a concrete runtime/code-path bug in the model implementation itself, whereas PR 46908 is a narrower test-only workaround for loading setup.", "best_issue_reason": null, - "best_pr_reason": "PR 46907 is the stronger standalone candidate because it fixes a concrete model forward-path bug in code, whereas PR 46908 is a test-only adjustment for loading setup and does not address the same kind of underlying issue.", + "best_pr_reason": "PR 46907 is the strongest standalone PR in the cluster: it addresses a specific integration failure at the source code boundary and is more broadly meaningful than the test harness change in PR 46908.", "soft_edge_verdicts": [ { "left": "pull_request:46907", "right": "pull_request:46908", "accept": false, - "reason": "Different models, different failure modes, and different fixes: EdgeTAM removes an unsupported kwarg in model forward, while DeepSeekV32 only changes test loading parameters. They do not look like the same underlying bug or mergeable code change." + "reason": "Different models, different failure modes, and different fix types. EdgeTam removes an unsupported kwarg before calling the backbone; DeepSeekV32 only adds `offload_folder` in tests for loading. They do not look like the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 39, + "component_count": 39, + "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-46915-2", + "nodes": [ + "pull_request:46915", + "pull_request:46923" + ], + "soft_pairs": [ + "pull_request:46915|pull_request:46923" + ], + "prepared_review_unit_hash": "8df4131cd112becff605df3293d82ab4153e757286944b903cecc76445bebdf8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4252, + "estimated_input_tokens": 1063, + "estimated_eval_tokens": 2382 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4252, + "estimated_input_tokens": 1063, + "estimated_eval_tokens": 2382 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests target the same DIA docs page and the same tracking issue, but they make different fixes to different code examples, so they look related rather than true duplicates.", + "confidence": 0.76, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46923 is the better canonical candidate because it appears to provide the more complete docs correction, covering multiple example blocks on the DIA page with a consistent fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 46923 is the strongest standalone choice for the cluster: it addresses the broken DIA documentation examples more broadly and looks like the fuller, cleaner fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46915", + "right": "pull_request:46923", + "accept": false, + "reason": "Same docs file and same issue target, but the diffs change different example snippets and use different remediation approaches, so they are not clearly the same underlying change and would not safely merge into one PR." } ] }, diff --git a/analysis/current/manifest.json b/analysis/current/manifest.json index b320030f158d42382bb7dab46ec5af85839b64bd..86b17b3fdd7e67fbe9feb639389ab95f0babcead 100644 --- a/analysis/current/manifest.json +++ b/analysis/current/manifest.json @@ -1,8 +1,8 @@ { - "analysis_id": "hybrid-model-20260626t120015z", + "analysis_id": "hybrid-model-20260626t180014z", "archived_artifacts": { - "hybrid": "snapshots/20260626T120015Z/analysis-runs/hybrid-model-20260626t120015z/analysis-report-hybrid.json", - "hybrid_reviews": "snapshots/20260626T120015Z/analysis-runs/hybrid-model-20260626t120015z/analysis-report-hybrid.llm-reviews.json" + "hybrid": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json" }, "artifacts": { "hybrid": "analysis/current/analysis-report-hybrid.json", @@ -10,9 +10,9 @@ }, "channel": "canonical", "model": null, - "published_at": "2026-06-26T12:13:39Z", + "published_at": "2026-06-26T18:10:13Z", "repo": "huggingface/transformers", "schema_version": 1, - "snapshot_id": "20260626T120015Z", + "snapshot_id": "20260626T180014Z", "variant": "hybrid" } diff --git a/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json new file mode 100644 index 0000000000000000000000000000000000000000..eec1171f783b643d2095924114fbb687e17a714f --- /dev/null +++ b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json @@ -0,0 +1,2545 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260626T180014Z", + "generated_at": "2026-06-26T18:10:11Z", + "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-46894-5", + "summary": "Cluster of 2 issues and 3 PRs centered on issue #46897.", + "status": "open", + "confidence": 0.8, + "canonical_issue_number": 46897, + "canonical_pr_number": 46895, + "issue_numbers": [ + 46894, + 46897 + ], + "pr_numbers": [ + 46895, + 46898, + 46916 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target", + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 46895, + "right_pr_number": 46898, + "code_similarity": 0.121, + "size_similarity": 0.483, + "file_overlap": 0.0, + "area_overlap": 0.0, + "patch_similarity": 0.164, + "shared_filenames": [], + "shared_file_areas": [] + }, + { + "left_pr_number": 46895, + "right_pr_number": 46916, + "code_similarity": 0.341, + "size_similarity": 0.644, + "file_overlap": 0.4, + "area_overlap": 0.194, + "patch_similarity": 0.162, + "shared_filenames": [ + "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "src/transformers/models/cohere_asr/modular_cohere_asr.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/cohere_asr/modeling_cohere_asr.py", + "left_ranges": [ + [ + 615, + 627 + ], + [ + 639, + 654 + ] + ], + "right_ranges": [ + [ + 637, + 644 + ] + ] + }, + { + "filename": "src/transformers/models/cohere_asr/modular_cohere_asr.py", + "left_ranges": [ + [ + 478, + 490 + ], + [ + 502, + 517 + ] + ], + "right_ranges": [ + [ + 500, + 507 + ] + ] + } + ] + }, + { + "left_pr_number": 46898, + "right_pr_number": 46916, + "code_similarity": 0.42, + "size_similarity": 0.75, + "file_overlap": 0.5, + "area_overlap": 0.194, + "patch_similarity": 0.35, + "shared_filenames": [ + "src/transformers/models/florence2/modeling_florence2.py", + "src/transformers/models/florence2/modular_florence2.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/florence2/modeling_florence2.py", + "left_ranges": [ + [ + 872, + 878 + ], + [ + 881, + 887 + ], + [ + 904, + 918 + ] + ], + "right_ranges": [ + [ + 902, + 909 + ] + ] + }, + { + "filename": "src/transformers/models/florence2/modular_florence2.py", + "left_ranges": [ + [ + 1580, + 1586 + ], + [ + 1589, + 1595 + ], + [ + 1612, + 1626 + ] + ], + "right_ranges": [ + [ + 1610, + 1617 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43240-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43240.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43240, + "canonical_pr_number": 43251, + "issue_numbers": [ + 43240 + ], + "pr_numbers": [ + 43251, + 43254 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43251, + "right_pr_number": 43254, + "code_similarity": 0.794, + "size_similarity": 0.64, + "file_overlap": 1.0, + "area_overlap": 0.667, + "patch_similarity": 0.882, + "shared_filenames": [ + "src/transformers/loss/loss_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/loss/loss_utils.py", + "left_ranges": [ + [ + 30, + 50 + ] + ], + "right_ranges": [ + [ + 30, + 43 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-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-46869-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46869.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 46869, + "canonical_pr_number": 46886, + "issue_numbers": [ + 46869 + ], + "pr_numbers": [ + 46886, + 46922 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 46886, + "right_pr_number": 46922, + "code_similarity": 0.732, + "size_similarity": 0.857, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.571, + "shared_filenames": [ + "src/transformers/models/yolos/modeling_yolos.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/yolos/modeling_yolos.py", + "left_ranges": [ + [ + 434, + 440 + ] + ], + "right_ranges": [ + [ + 434, + 440 + ], + [ + 446, + 452 + ] + ] + } + ] + } + ] + }, + { + "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-46909-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #46909.", + "status": "open", + "confidence": 0.55, + "canonical_issue_number": 46909, + "canonical_pr_number": 46923, + "issue_numbers": [ + 46909 + ], + "pr_numbers": [ + 46915, + 46923 + ], + "evidence_types": [ + "closing_reference" + ], + "pr_comparisons": [ + { + "left_pr_number": 46915, + "right_pr_number": 46923, + "code_similarity": 0.622, + "size_similarity": 0.477, + "file_overlap": 1.0, + "area_overlap": 0.5, + "patch_similarity": 0.341, + "shared_filenames": [ + "docs/source/en/model_doc/dia.md" + ], + "shared_file_areas": [ + { + "filename": "docs/source/en/model_doc/dia.md", + "left_ranges": [ + [ + 42, + 62 + ], + [ + 76, + 92 + ], + [ + 106, + 122 + ] + ], + "right_ranges": [ + [ + 47, + 56 + ], + [ + 75, + 84 + ], + [ + 102, + 110 + ], + [ + 114, + 119 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-43698-3", + "summary": "Cluster of 1 issues and 2 PRs centered on issue #43698.", + "status": "open", + "confidence": 0.75, + "canonical_issue_number": 43698, + "canonical_pr_number": 43779, + "issue_numbers": [ + 43698 + ], + "pr_numbers": [ + 43779, + 43816 + ], + "evidence_types": [ + "closing_reference", + "shared_issue_target" + ], + "pr_comparisons": [ + { + "left_pr_number": 43779, + "right_pr_number": 43816, + "code_similarity": 0.418, + "size_similarity": 0.538, + "file_overlap": 1.0, + "area_overlap": 0.02, + "patch_similarity": 0.02, + "shared_filenames": [ + "src/transformers/integrations/integration_utils.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/integrations/integration_utils.py", + "left_ranges": [ + [ + 2227, + 2241 + ], + [ + 2303, + 2309 + ] + ], + "right_ranges": [ + [ + 2278, + 2291 + ], + [ + 2309, + 2322 + ] + ] + } + ] + } + ] + }, + { + "cluster_id": "cluster-44018-2", + "summary": "Cluster of 2 related pull requests linked by soft_similarity.", + "status": "open", + "confidence": 0.5, + "canonical_issue_number": null, + "canonical_pr_number": 44068, + "issue_numbers": [], + "pr_numbers": [ + 44018, + 44068 + ], + "evidence_types": [ + "soft_similarity" + ], + "pr_comparisons": [ + { + "left_pr_number": 44018, + "right_pr_number": 44068, + "code_similarity": 0.766, + "size_similarity": 0.939, + "file_overlap": 1.0, + "area_overlap": 0.425, + "patch_similarity": 0.866, + "shared_filenames": [ + "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "tests/models/gpt_neo/test_modeling_gpt_neo.py" + ], + "shared_file_areas": [ + { + "filename": "src/transformers/models/gpt_neo/modeling_gpt_neo.py", + "left_ranges": [ + [ + 26, + 46 + ], + [ + 138, + 143 + ], + [ + 182, + 187 + ], + [ + 283, + 295 + ], + [ + 327, + 341 + ], + [ + 348, + 354 + ], + [ + 360, + 369 + ], + [ + 399, + 406 + ], + [ + 411, + 419 + ], + [ + 428, + 433 + ], + [ + 472, + 492 + ], + [ + 507, + 513 + ], + [ + 519, + 528 + ], + [ + 541, + 559 + ], + [ + 562, + 567 + ], + [ + 595, + 601 + ], + [ + 607, + 614 + ], + [ + 627, + 643 + ], + [ + 685, + 690 + ], + [ + 708, + 714 + ], + [ + 720, + 727 + ], + [ + 740, + 757 + ], + [ + 761, + 766 + ], + [ + 780, + 786 + ], + [ + 791, + 798 + ], + [ + 807, + 822 + ], + [ + 840, + 845 + ] + ], + "right_ranges": [ + [ + 26, + 31 + ], + [ + 34, + 44 + ], + [ + 136, + 141 + ], + [ + 180, + 185 + ], + [ + 281, + 293 + ], + [ + 325, + 339 + ], + [ + 346, + 352 + ], + [ + 358, + 367 + ], + [ + 397, + 404 + ], + [ + 409, + 417 + ], + [ + 426, + 434 + ], + [ + 463, + 483 + ], + [ + 498, + 504 + ], + [ + 510, + 519 + ], + [ + 532, + 550 + ], + [ + 553, + 564 + ], + [ + 586, + 592 + ], + [ + 598, + 605 + ], + [ + 618, + 634 + ], + [ + 676, + 688 + ], + [ + 699, + 705 + ], + [ + 711, + 718 + ], + [ + 731, + 748 + ], + [ + 752, + 762 + ], + [ + 771, + 777 + ], + [ + 782, + 789 + ], + [ + 798, + 813 + ], + [ + 831, + 836 + ] + ] + }, + { + "filename": "tests/models/gpt_neo/test_modeling_gpt_neo.py", + "left_ranges": [ + [ + 458, + 464 + ] + ], + "right_ranges": [ + [ + 458, + 464 + ] + ] + } + ] + } + ] + } + ], + "duplicate_issues": [ + { + "cluster_id": "cluster-46894-5", + "canonical_issue_number": 46897, + "duplicate_issue_numbers": [ + 46894 + ], + "reason": "Issues in cluster-46894-5 are treated as duplicates because they share closing_reference, shared_issue_target, soft_similarity evidence." + } + ], + "duplicate_prs": [ + { + "cluster_id": "cluster-41211-3", + "canonical_pr_number": 44339, + "duplicate_pr_numbers": [ + 41356 + ], + "target_issue_number": 41211, + "reason": "PRs in cluster-41211-3 are treated as duplicates because they converge on issue #41211 with closing_reference evidence." + }, + { + "cluster_id": "cluster-43240-3", + "canonical_pr_number": 43251, + "duplicate_pr_numbers": [ + 43254 + ], + "target_issue_number": 43240, + "reason": "PRs in cluster-43240-3 are treated as duplicates because they converge on issue #43240 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43698-3", + "canonical_pr_number": 43779, + "duplicate_pr_numbers": [ + 43816 + ], + "target_issue_number": 43698, + "reason": "PRs in cluster-43698-3 are treated as duplicates because they converge on issue #43698 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-43979-11", + "canonical_pr_number": 44007, + "duplicate_pr_numbers": [ + 43996, + 44013, + 44044, + 44066, + 44072, + 44085, + 44129, + 44154, + 44722 + ], + "target_issue_number": 43979, + "reason": "PRs in cluster-43979-11 are treated as duplicates because they converge on issue #43979 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-44018-2", + "canonical_pr_number": 44068, + "duplicate_pr_numbers": [ + 44018 + ], + "target_issue_number": null, + "reason": "PRs in cluster-44018-2 are treated as duplicates because they share soft_similarity evidence." + }, + { + "cluster_id": "cluster-45081-3", + "canonical_pr_number": 45317, + "duplicate_pr_numbers": [ + 45086 + ], + "target_issue_number": 45081, + "reason": "PRs in cluster-45081-3 are treated as duplicates because they converge on issue #45081 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-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." + }, + { + "cluster_id": "cluster-46869-3", + "canonical_pr_number": 46886, + "duplicate_pr_numbers": [ + 46922 + ], + "target_issue_number": 46869, + "reason": "PRs in cluster-46869-3 are treated as duplicates because they converge on issue #46869 with closing_reference, shared_issue_target evidence." + }, + { + "cluster_id": "cluster-46894-5", + "canonical_pr_number": 46895, + "duplicate_pr_numbers": [ + 46898, + 46916 + ], + "target_issue_number": 46897, + "reason": "PRs in cluster-46894-5 are treated as duplicates because they converge on issue #46897 with closing_reference, shared_issue_target, soft_similarity evidence." + }, + { + "cluster_id": "cluster-46909-3", + "canonical_pr_number": 46923, + "duplicate_pr_numbers": [ + 46915 + ], + "target_issue_number": 46909, + "reason": "PRs in cluster-46909-3 are treated as duplicates because they converge on issue #46909 with closing_reference evidence." + } + ], + "best_issue": { + "cluster_id": "cluster-46894-5", + "issue_number": 46897, + "reason": "Issue #46897 is the strongest global issue candidate because it is open, belongs to a cluster with 5 artifacts, and carries 0 discussion comments plus 9 inbound references.", + "score": 159.0 + }, + "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/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json new file mode 100644 index 0000000000000000000000000000000000000000..1c7c8fc5f2658f564744253dbac8734af8c539ec --- /dev/null +++ b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json @@ -0,0 +1,24765 @@ +{ + "schema_version": "1.0", + "repo": "huggingface/transformers", + "snapshot_id": "20260626T180014Z", + "generated_at": "2026-06-26T18:10:11Z", + "model": "gpt-5.4-mini?service_tier=flex", + "reviews": [ + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 1, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "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", + "issue:46869" + ], + "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:46165|issue:46869", + "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: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:43716|issue:46869", + "issue:24643|issue:30277", + "issue:34928|issue:44805", + "issue:45859|issue:45902", + "issue:45685|issue:46869", + "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:45800|issue:46515", + "issue:45823|issue:45874", + "issue:43425|issue:44292", + "issue:34928|issue:41720", + "issue:43828|issue:45161", + "issue:43541|issue:43716", + "issue:43828|issue:44292", + "issue:43716|issue:46468", + "issue:44292|issue:44918", + "issue:45507|issue:45685", + "issue:43716|issue:44292", + "issue:44918|issue:45161", + "issue:43716|issue:45685", + "issue:43039|issue:44918", + "issue:41762|issue:43638", + "issue:43827|issue:44512", + "issue:43828|issue:45685", + "issue:43039|issue:43828", + "issue:30277|issue:36331", + "issue:34928|issue:36010", + "issue:45237|issue:46015", + "issue:43039|issue:44805" + ], + "prepared_review_unit_hash": "e57caac5f04946b685b34c61db3aa01ef85311216039204de458db7e02c077ad", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 63, + "serialized_chars": 46576, + "estimated_input_tokens": 11644, + "estimated_eval_tokens": 23544 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous similarity cluster with one clear duplicate family around DeepSpeed ZeRO-3/BERT index-out-of-bounds errors, while most other pairs are false positives across unrelated model, docs, and hardware issues.", + "confidence": 0.63, + "canonical_issue_reason": "Issue 43638 is the broadest and clearest representative of the one obvious duplicate family: the DeepSpeed ZeRO-3 index-out-of-bounds failure during BERT loading.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43638 is the best issue representative here because it describes the shared concrete failure mode most directly and is the closest true duplicate hub in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44322", + "right": "issue:45464", + "accept": false, + "reason": "Different bug classes: Qwen3.5 config attribute missing vs streaming chat/completions failure." + }, + { + "left": "issue:45237", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated symptoms and platforms: AMD GPU failure vs MPS histogram int error." + }, + { + "left": "issue:43716", + "right": "issue:45237", + "accept": false, + "reason": "Different model/runtime problems: image dtype mismatch vs AMD GPU incompatibility." + }, + { + "left": "issue:43828", + "right": "issue:45237", + "accept": false, + "reason": "Autocast dtype mismatch and AMD GPU failure are not the same underlying bug." + }, + { + "left": "issue:45237", + "right": "issue:46468", + "accept": false, + "reason": "Different models and failures; no concrete shared code-path." + }, + { + "left": "issue:45237", + "right": "issue:45507", + "accept": false, + "reason": "AMD GPU issue vs GraniteMoEHybrid invalid method; unrelated." + }, + { + "left": "issue:30277", + "right": "issue:35545", + "accept": false, + "reason": "DeepSpeed collective mismatch and ONNX export error are different problems." + }, + { + "left": "issue:35545", + "right": "issue:42915", + "accept": false, + "reason": "ONNX export vs FineGrainedFP8 Qwen3Moe failure; unrelated." + }, + { + "left": "issue:35545", + "right": "issue:41720", + "accept": false, + "reason": "Different subsystem and symptom: export bug vs device-map cuda assert." + }, + { + "left": "issue:45901", + "right": "issue:46015", + "accept": false, + "reason": "Table QA crash and outdated RAG examples are separate issues; only broadly docs-related." + }, + { + "left": "issue:44509", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs cleanup, but they reference different removed content and would not be one PR." + }, + { + "left": "issue:35545", + "right": "issue:36010", + "accept": false, + "reason": "ONNX export error and GenerationMixin import failure are unrelated." + }, + { + "left": "issue:44918", + "right": "issue:46165", + "accept": false, + "reason": "Different topics: Qwen3.5 embedding unpacking vs resume-training remote-code problem." + }, + { + "left": "issue:46165", + "right": "issue:46869", + "accept": false, + "reason": "Resume-training remote-code issue and YOLO multi-GPU auto device-map issue are unrelated." + }, + { + "left": "issue:44805", + "right": "issue:45823", + "accept": false, + "reason": "Mask-shape index error and Gemma4 PLE device mismatch are different bugs." + }, + { + "left": "issue:43638", + "right": "issue:44805", + "accept": false, + "reason": "Both are index-style runtime errors, but the error modes and model contexts differ too much to be the same bug." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "DeepSpeed ZeRO-3 BERT loading issue vs remote-code resume training issue; unrelated." + }, + { + "left": "issue:43596", + "right": "issue:44805", + "accept": false, + "reason": "Different failure signatures and contexts; not the same underlying bug." + }, + { + "left": "issue:44805", + "right": "issue:46294", + "accept": false, + "reason": "Mask/index error and Qwen3.5 video token mismatch are unrelated." + }, + { + "left": "issue:44805", + "right": "issue:45161", + "accept": false, + "reason": "Different models and failures; no concrete shared fix." + }, + { + "left": "issue:46165", + "right": "issue:46468", + "accept": false, + "reason": "Remote-code resume issue and Nemotron example breakage are unrelated." + }, + { + "left": "issue:44560", + "right": "issue:46165", + "accept": false, + "reason": "Video StopIteration and training resume are unrelated." + }, + { + "left": "issue:34634", + "right": "issue:35545", + "accept": false, + "reason": "BarkProcessor voice preset bug is unrelated to ModernBERT ONNX export." + }, + { + "left": "issue:30277", + "right": "issue:43638", + "accept": false, + "reason": "Both involve DeepSpeed, but the concrete failures differ: collective mismatch vs zero-3 index-out-of-bounds." + }, + { + "left": "issue:43596", + "right": "issue:43638", + "accept": true, + "reason": "Same underlying DeepSpeed ZeRO-3 empty-tensor index-out-of-bounds failure during BERT loading; one is just a narrower variant of the other." + }, + { + "left": "issue:43039", + "right": "issue:45161", + "accept": false, + "reason": "Liger Kernel cross-entropy call vs GPT-OSS MoE TP failure are unrelated." + }, + { + "left": "issue:43975", + "right": "issue:44457", + "accept": false, + "reason": "DeepSeek detokenization issue and LoRA merge/save mismatch are different bugs." + }, + { + "left": "issue:44292", + "right": "issue:44457", + "accept": false, + "reason": "Qwen NVFP4 runtime error vs LoRA merge inconsistency are unrelated." + }, + { + "left": "issue:44841", + "right": "issue:46661", + "accept": false, + "reason": "Voxtral processor failure and remote-code model_class config_class error are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:46869", + "accept": false, + "reason": "Different model families and failure modes; no shared bug." + }, + { + "left": "issue:24643", + "right": "issue:30277", + "accept": false, + "reason": "Both are DeepSpeed-related, but one is a weight dimensionality error and the other is a collective mismatch; not the same bug." + }, + { + "left": "issue:34928", + "right": "issue:44805", + "accept": false, + "reason": "Activation-checkpointing/FSDP recompute mismatch is different from the mask index error." + }, + { + "left": "issue:45859", + "right": "issue:45902", + "accept": false, + "reason": "Same component family, but CPU offload incompatibility and uninitialized memory after to_empty() are distinct bugs." + }, + { + "left": "issue:45685", + "right": "issue:46869", + "accept": false, + "reason": "MPS MoE histogram issue and YOLO multi-GPU device-map issue are unrelated." + }, + { + "left": "issue:43638", + "right": "issue:45161", + "accept": false, + "reason": "Zero-3 BERT index error and GPT-OSS TP failure are different problems." + }, + { + "left": "issue:34928", + "right": "issue:36331", + "accept": false, + "reason": "FSDP checkpointing mismatch vs custom trainer signature change are unrelated." + }, + { + "left": "issue:43596", + "right": "issue:45161", + "accept": false, + "reason": "DeepSpeed ZeRO-3 BERT loading bug vs GPT-OSS MoE TP issue are unrelated." + }, + { + "left": "issue:43716", + "right": "issue:45507", + "accept": false, + "reason": "Different model paths and symptoms; not the same bug." + }, + { + "left": "issue:44292", + "right": "issue:45237", + "accept": false, + "reason": "Different models, hardware, and failure contexts; no shared fix." + }, + { + "left": "issue:45874", + "right": "issue:46294", + "accept": false, + "reason": "Gemma4 reverse-embedding issue and Qwen3.5 video token mismatch are unrelated." + }, + { + "left": "issue:39290", + "right": "issue:45081", + "accept": false, + "reason": "Gemma3 config attribute error and Mistral regex crash are different bugs." + }, + { + "left": "issue:45800", + "right": "issue:46515", + "accept": false, + "reason": "Both mention torch 2.4.1 incompatibility, but one is tied to transformers 5.8.0 and the underlying failure is not specific enough to treat as the same bug." + }, + { + "left": "issue:45823", + "right": "issue:45874", + "accept": false, + "reason": "Both are Gemma4-related, but one is a device mismatch and the other is an expensive reverse embedding lookup; different code paths." + }, + { + "left": "issue:43425", + "right": "issue:44292", + "accept": false, + "reason": "Generic Torch 2.10 incompatibility vs Qwen NVFP4 runtime error; unrelated." + }, + { + "left": "issue:34928", + "right": "issue:41720", + "accept": false, + "reason": "FSDP activation-checkpointing mismatch and Qwen3 device-map cuda assert are different issues." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:43716", + "accept": false, + "reason": "Torch dynamo grouped_mm error and image-preprocessor dtype mismatch are different bugs." + }, + { + "left": "issue:43828", + "right": "issue:44292", + "accept": false, + "reason": "Autocast dtype mismatch vs Qwen NVFP4 error is not the same bug." + }, + { + "left": "issue:43716", + "right": "issue:46468", + "accept": false, + "reason": "Mistral-3 dtype mismatch and Nemotron example breakage are unrelated." + }, + { + "left": "issue:44292", + "right": "issue:44918", + "accept": false, + "reason": "Both are Qwen-related, but one is an NVFP4 runtime error and the other is a TRL embedding unpacking issue." + }, + { + "left": "issue:44918", + "right": "issue:45161", + "accept": false, + "reason": "Different models and failure modes; not a duplicate." + }, + { + "left": "issue:43716", + "right": "issue:45685", + "accept": false, + "reason": "Dtype mismatch and MPS histogram error are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44918", + "accept": false, + "reason": "Liger Kernel cross-entropy call and TRL/Qwen embedding unpacking are unrelated." + }, + { + "left": "issue:41762", + "right": "issue:43638", + "accept": false, + "reason": "Same broad ZeRO-3 loading area, but different models and error context; not clearly the same underlying bug." + }, + { + "left": "issue:43827", + "right": "issue:44512", + "accept": false, + "reason": "Both are docs regressions from v5 removals, but they cover different removed references and should not be merged as one duplicate." + }, + { + "left": "issue:43828", + "right": "issue:45685", + "accept": false, + "reason": "Autocast dtype mismatch and MPS histogram int error are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44805", + "accept": false, + "reason": "Liger Kernel cross-entropy issue and mask shape IndexError are unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded overall. The summary stays conservative, and the one accepted soft-edge pair (43596 \u2194 43638) is well supported by the packet: both are near-identical DeepSpeed ZeRO-3/BERT index-out-of-bounds failures. The remaining rejections are mostly appropriately cautious given the differing models, subsystems, or hardware contexts." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 2, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:36010", + "issue:39290", + "issue:41093", + "issue:41720", + "issue:41762", + "issue:43039", + "issue:43296", + "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:45507", + "issue:45685", + "issue:45800", + "issue:45812", + "issue:45823", + "issue:45901", + "issue:45920", + "issue:46015", + "issue:46294", + "issue:46304", + "issue:46402", + "issue:46661", + "issue:46772", + "issue:46869" + ], + "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:43975|issue:46015", + "issue:43039|issue:44560", + "issue:45823|issue:46294", + "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:43541|issue:46869", + "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:45507|issue:46869", + "issue:41093|issue:41762", + "issue:45399|issue:46661", + "issue:43572|issue:45070", + "issue:45800|issue:45823" + ], + "prepared_review_unit_hash": "e106e8db2be6abc313510784bc8854f6951a540f6e9b415493e756b25bfbce50", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 42, + "serialized_chars": 40042, + "estimated_input_tokens": 10011, + "estimated_eval_tokens": 20278 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly a loose cluster of unrelated bug reports. The only clear duplicate-like soft edge is the AutoTokenizer v5 regression pair; the rest differ by model, API path, or failure mode and should stay separate.", + "confidence": 0.81, + "canonical_issue_reason": "issue:45920 is the best canonical issue because it describes the broader AutoTokenizer v5 token-ID regression across multiple model families, subsuming the narrower Granite-only report in issue:45812.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 is the strongest global representative: broad scope, clear regression framing, and likely shared root cause with related tokenizer reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43866", + "right": "issue:46402", + "accept": false, + "reason": "Different problems: corrupted checkpoint vs missing-loaded weights in a different model family." + }, + { + "left": "issue:45800", + "right": "issue:46304", + "accept": false, + "reason": "Both mention version compatibility, but they describe different concrete failures and release contexts." + }, + { + "left": "issue:43296", + "right": "issue:46304", + "accept": false, + "reason": "Different stack and symptom: vLLM/PaddleOCR-VL load failure vs PyTorch distributed support regression." + }, + { + "left": "issue:41762", + "right": "issue:43596", + "accept": false, + "reason": "Both are index errors under DeepSpeed ZeRO-3, but they affect different models and likely different code paths." + }, + { + "left": "issue:39290", + "right": "issue:41720", + "accept": false, + "reason": "Gemma3 config attribute error and Qwen3 auto device-map CUDA assert are unrelated failures." + }, + { + "left": "issue:43901", + "right": "issue:44512", + "accept": false, + "reason": "Both are documentation-related, but they cover different docs and different stale references." + }, + { + "left": "issue:44589", + "right": "issue:46304", + "accept": false, + "reason": "Unrelated runtime storage lookup error vs PyTorch distributed build/runtime incompatibility." + }, + { + "left": "issue:45237", + "right": "issue:45901", + "accept": false, + "reason": "Different model/runtime issues with no shared concrete bug." + }, + { + "left": "issue:43975", + "right": "issue:46015", + "accept": false, + "reason": "Wrong detokenization and outdated RAG examples are unrelated." + }, + { + "left": "issue:43039", + "right": "issue:44560", + "accept": false, + "reason": "Liger Kernel cross-entropy routing and Qwen3-vl video stop-iteration are different paths." + }, + { + "left": "issue:45823", + "right": "issue:46294", + "accept": false, + "reason": "Device-map mismatch in Gemma4 PLE is unrelated to video token/feature mismatch in Qwen3.5." + }, + { + "left": "issue:44841", + "right": "issue:45823", + "accept": false, + "reason": "Processor failure for Voxtral and Gemma4 device mismatch are different bugs." + }, + { + "left": "issue:43866", + "right": "issue:46772", + "accept": false, + "reason": "Checkpoint corruption and ModelOpt/NVFP4 compatibility are not the same issue." + }, + { + "left": "issue:44863", + "right": "issue:46015", + "accept": false, + "reason": "NemotronH checkpoint loading and outdated RAG docs are unrelated." + }, + { + "left": "issue:45020", + "right": "issue:46304", + "accept": false, + "reason": "Remote-code breakage and PyTorch USE_DISTRIBUTED=0 support regression are different compatibility problems." + }, + { + "left": "issue:45812", + "right": "issue:45920", + "accept": true, + "reason": "Same underlying AutoTokenizer v5 regression producing wrong token IDs; 45920 is the broader multi-model manifestation and 45812 is a narrower Granite-specific report." + }, + { + "left": "issue:43866", + "right": "issue:44863", + "accept": false, + "reason": "Corrupted checkpoint vs failed checkpoint loading for a different architecture." + }, + { + "left": "issue:43872", + "right": "issue:46304", + "accept": false, + "reason": "bitsandbytes parameter incompatibility is unrelated to distributed build support." + }, + { + "left": "issue:44292", + "right": "issue:46015", + "accept": false, + "reason": "Qwen-3 NVFP4 runtime failure and RAG docs issues are unrelated." + }, + { + "left": "issue:39290", + "right": "issue:43531", + "accept": false, + "reason": "Different Qwen/Gemma sliding-window and device-map failures." + }, + { + "left": "issue:41093", + "right": "issue:43638", + "accept": false, + "reason": "Both are IndexErrors, but the affected model/load setup differs enough that they are not the same bug." + }, + { + "left": "issue:43792", + "right": "issue:45901", + "accept": false, + "reason": "Whisper model load failure and table-question-answering crash are unrelated." + }, + { + "left": "issue:43854", + "right": "issue:43866", + "accept": false, + "reason": "Unit test load failure for GLM-4.7 and Ovis2 checkpoint corruption are different issues." + }, + { + "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": "Detokenization bug and table QA crash do not share a concrete code path." + }, + { + "left": "issue:43792", + "right": "issue:45070", + "accept": false, + "reason": "Whisper load failure is unrelated to the pydantic model-field regression." + }, + { + "left": "issue:43596", + "right": "issue:45800", + "accept": false, + "reason": "ZeRO-3 index error with BertModel is not the same as the Torch 2.4.1 compatibility report." + }, + { + "left": "issue:45362", + "right": "issue:45464", + "accept": false, + "reason": "Both mention Qwen3.5 and chat, but one is a model crash and the other is a streaming chat/completions API failure." + }, + { + "left": "issue:43541", + "right": "issue:46869", + "accept": false, + "reason": "Mixtral torch-dynamo grouped_mm tracing failure and YOLO multi-GPU device-map failure are different code paths." + }, + { + "left": "issue:44292", + "right": "issue:45901", + "accept": false, + "reason": "NVFP4 model runtime failure and table QA task crash are unrelated." + }, + { + "left": "issue:39290", + "right": "issue:45823", + "accept": false, + "reason": "Gemma3 config regression and Gemma4 auto device-map mismatch are different bugs." + }, + { + "left": "issue:45464", + "right": "issue:45685", + "accept": false, + "reason": "Chat/completions streaming failure and MPS histogram implementation error are unrelated." + }, + { + "left": "issue:36010", + "right": "issue:39290", + "accept": false, + "reason": "Importing GenerationMixin and Gemma3TextConfig attribute errors are unrelated." + }, + { + "left": "issue:43541", + "right": "issue:45685", + "accept": false, + "reason": "Torch dynamo tracing failure and MPS histogram type error are different issues." + }, + { + "left": "issue:43541", + "right": "issue:43828", + "accept": false, + "reason": "Grouped_mm tracing error and autocast dtype mismatch are distinct runtime problems." + }, + { + "left": "issue:44841", + "right": "issue:45084", + "accept": false, + "reason": "Processor failure for Voxtral and non-template-node compilation error are unrelated." + }, + { + "left": "issue:43404", + "right": "issue:43866", + "accept": false, + "reason": "Weight-tying bug in Mistral3 and Ovis2 checkpoint corruption are different model-specific issues." + }, + { + "left": "issue:45507", + "right": "issue:46869", + "accept": false, + "reason": "Invalid method call in GraniteMoEHybrid and YOLO multi-GPU device-map failure are unrelated." + }, + { + "left": "issue:41093", + "right": "issue:41762", + "accept": false, + "reason": "Both are index-related errors, but the models and failure contexts differ." + }, + { + "left": "issue:45399", + "right": "issue:46661", + "accept": false, + "reason": "Flash-attn fallback gating and remote-code config_class handling are different bugs." + }, + { + "left": "issue:43572", + "right": "issue:45070", + "accept": false, + "reason": "Missing pad_token_idx in StableLmConfig and pydantic PretrainedConfig regression are unrelated." + }, + { + "left": "issue:45800", + "right": "issue:45823", + "accept": false, + "reason": "Torch/transformers compatibility and Gemma4 device placement are different problems." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The clustering is grounded in the packet and stays conservative. The only clearly supported duplicate-like edge is issue:45812 \u2194 issue:45920, which both describe the AutoTokenizer v5 wrong-token-ID regression, with 45920 being the broader multi-model report. The remaining rejected pairs are appropriately treated as separate bugs based on differing model families, APIs, or 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": 23, + "review_unit_index": 3, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:39186", + "issue:40990", + "issue:41093", + "issue:42915", + "issue:43054", + "issue:43296", + "issue:43366", + "issue:43404", + "issue:43531", + "issue:43541", + "issue:43596", + "issue:43645", + "issue:43716", + "issue:43828", + "issue:43873", + "issue:43994", + "issue:44291", + "issue:44360", + "issue:44387", + "issue:44479", + "issue:44492", + "issue:44512", + "issue:44530", + "issue:44805", + "issue:44863", + "issue:44918", + "issue:44960", + "issue:45005", + "issue:45020", + "issue:45070", + "issue:45161", + "issue:45183", + "issue:45399", + "issue:45464", + "issue:45632", + "issue:45698", + "issue:45759", + "issue:45800", + "issue:45902", + "issue:46165", + "issue:46294", + "issue:46463", + "issue:46468", + "issue:46489", + "issue:46491", + "issue:46640", + "issue:46661", + "issue:46869" + ], + "soft_pairs": [ + "issue:43296|issue:43366", + "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:41093|issue:43596", + "issue:45020|issue:45800", + "issue:43716|issue:46294", + "issue:43054|issue:43994", + "issue:44479|issue:46294", + "issue:46468|issue:46869", + "issue:45759|issue:45902", + "issue:43596|issue:46165", + "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:43873|issue:45005", + "issue:44530|issue:45005", + "issue:44360|issue:44512", + "issue:44492|issue:44512", + "issue:44863|issue:45005", + "issue:45183|issue:46463", + "issue:46489|issue:46491" + ], + "prepared_review_unit_hash": "db2c189a91c1be95a2761d79df723dc3b41b003e196acf962b3712354668e83c", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36886, + "estimated_input_tokens": 9222, + "estimated_eval_tokens": 18700 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are only superficially similar; they span unrelated Transformers bugs across loading, quantization, MoE, tokenization, video, and docs. None of the soft pairs look like the same concrete underlying bug/change, so all should be rejected.", + "confidence": 0.91, + "canonical_issue_reason": "Issue 45005 is the broadest hub in this set and has the most similarity links, but it still does not truly subsume the others.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45005 is the most plausible representative by breadth and centrality among the provided issues, even though the cluster is not actually cohesive.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43296", + "right": "issue:43366", + "accept": false, + "reason": "Different model families and failure modes: PaddleOCR-VL load failure vs GGUF/gpt-oss architecture support." + }, + { + "left": "issue:44387", + "right": "issue:45005", + "accept": false, + "reason": "Int4 quantization OOM/reserved-memory growth is not the same bug as tied-weights handling on translation models." + }, + { + "left": "issue:44805", + "right": "issue:44918", + "accept": false, + "reason": "Both mention shape-related errors, but one is a generic mask indexing mismatch and the other is Qwen3.5 embedding unpacking with TRL SFT." + }, + { + "left": "issue:43541", + "right": "issue:45161", + "accept": false, + "reason": "Grouped_mm torch-dynamo tracing failure for Mixtral is a different code path from GPT-OSS tensor-parallel support." + }, + { + "left": "issue:43828", + "right": "issue:45464", + "accept": false, + "reason": "Autocast dtype mismatch on Phi-tiny-MoE is unrelated to Qwen3.5 streaming chat/completions failure." + }, + { + "left": "issue:43366", + "right": "issue:45464", + "accept": false, + "reason": "GGUF gpt-oss support and Qwen3.5 streaming inference are unrelated bugs." + }, + { + "left": "issue:44291", + "right": "issue:45070", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument in init_empty_weights is unrelated to the pydantic PretrainedConfig field regression." + }, + { + "left": "issue:43531", + "right": "issue:45070", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and pydantic PretrainedConfig serialization are different issues." + }, + { + "left": "issue:41093", + "right": "issue:43596", + "accept": false, + "reason": "Both are IndexErrors, but one is a mask-length mismatch and the other is zero-sized indexing during deepspeed zero3/BertModel init." + }, + { + "left": "issue:45020", + "right": "issue:45800", + "accept": false, + "reason": "Remote-code model breakage and torch/transformers version incompatibility are not the same concrete bug." + }, + { + "left": "issue:43716", + "right": "issue:46294", + "accept": false, + "reason": "Image preprocessor/model dtype mismatch is distinct from Qwen3.5 video token/feature count mismatch." + }, + { + "left": "issue:43054", + "right": "issue:43994", + "accept": false, + "reason": "Siglip2 poor text embeddings vs nonsensical AutoModel/pipeline output are related only at a very high level." + }, + { + "left": "issue:44479", + "right": "issue:46294", + "accept": false, + "reason": "A broad v5.3 video regression across multiple Qwen models is not the same as the specific apply_chat_template video mismatch in Qwen3.5-9B." + }, + { + "left": "issue:46468", + "right": "issue:46869", + "accept": false, + "reason": "Nemotron example breakage and YOLO multi-GPU device-map failure are unrelated subsystems." + }, + { + "left": "issue:45759", + "right": "issue:45902", + "accept": false, + "reason": "Composite Qwen config unwrapping and uninitialized inv_freq after meta->to_empty are different bugs." + }, + { + "left": "issue:43596", + "right": "issue:46165", + "accept": false, + "reason": "Zero3/BertModel initialization failure is unrelated to resuming training with local models and trust_remote_code." + }, + { + "left": "issue:46640", + "right": "issue:46661", + "accept": false, + "reason": "Flash Attention 2 problems and remote-code model_class/config_class AttributeError are different code paths." + }, + { + "left": "issue:43716", + "right": "issue:44918", + "accept": false, + "reason": "Dtype mismatch in image preprocessing is not the same as Qwen3.5 input-embedding unpacking with TRL SFT." + }, + { + "left": "issue:39186", + "right": "issue:44918", + "accept": false, + "reason": "FSDP 2-D weight error and Qwen3.5 embedding unpacking are unrelated." + }, + { + "left": "issue:40990", + "right": "issue:44960", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and a GLM5 issue are not the same bug." + }, + { + "left": "issue:43645", + "right": "issue:45399", + "accept": false, + "reason": "Custom-model initialization in Jupyter and flash-attn2 fallback blocking are unrelated." + }, + { + "left": "issue:45632", + "right": "issue:45698", + "accept": false, + "reason": "Both concern trust_remote_code/custom modules, but cache-path collision and loading the wrong module after save_pretrained are distinct bugs." + }, + { + "left": "issue:43404", + "right": "issue:45005", + "accept": false, + "reason": "Tied lm_head in Mistral3 image-text generation is not the same concrete issue as translation-model tied weights in v5." + }, + { + "left": "issue:42915", + "right": "issue:45399", + "accept": false, + "reason": "Qwen3Moe FineGrainedFP8Config failure and flash-attn2 fallback gating are unrelated." + }, + { + "left": "issue:43873", + "right": "issue:45005", + "accept": false, + "reason": "Quantization/offloading behavior is a different problem from tied-weights handling." + }, + { + "left": "issue:44530", + "right": "issue:45005", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash on Qwen3.5 is unrelated to translation-model tied weights." + }, + { + "left": "issue:44360", + "right": "issue:44512", + "accept": false, + "reason": "DSA indexer/ReLU discussion and docs mentioning a removed command are unrelated." + }, + { + "left": "issue:44492", + "right": "issue:44512", + "accept": false, + "reason": "A typo in cache strategies is not the same bug as outdated run-command documentation." + }, + { + "left": "issue:44863", + "right": "issue:45005", + "accept": false, + "reason": "NemotronH checkpoint loading failure is unrelated to tied-weights regressions." + }, + { + "left": "issue:45183", + "right": "issue:46463", + "accept": false, + "reason": "These are both validation-message bugs, but one is XOR input_ids/inputs_embeds handling and the other is T5 decoder-input validation; not the same concrete change." + }, + { + "left": "issue:46489", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but for different models and different symptoms." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 4, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:24643", + "issue:28530", + "issue:36331", + "issue:38175", + "issue:39692", + "issue:41553", + "issue:41762", + "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:44292", + "issue:44387", + "issue:44403", + "issue:44451", + "issue:44488", + "issue:44534", + "issue:44589", + "issue:44910", + "issue:44918", + "issue:44928", + "issue:44960", + "issue:44991", + "issue:45020", + "issue:45399", + "issue:45507", + "issue:45520", + "issue:45663", + "issue:45685", + "issue:45701", + "issue:45823", + "issue:45859", + "issue:45902", + "issue:45920", + "issue:45987", + "issue:46489", + "issue:46858" + ], + "soft_pairs": [ + "issue:43975|issue:44292", + "issue:44918|issue:45507", + "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:45987|issue:46858", + "issue:45920|issue:46489", + "issue:43054|issue:45701", + "issue:38175|issue:43054", + "issue:43950|issue:45902", + "issue:41553|issue:43883", + "issue:43054|issue:43493", + "issue:44910|issue:45859", + "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:43782|issue:44960", + "issue:43881|issue:43883", + "issue:45685|issue:45823" + ], + "prepared_review_unit_hash": "4d493b836e887b56bf853a2d28bd12de0a921099cb962288c3f130507c7b136d", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37032, + "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 heterogeneous, but there is one clear duplicate family around Transformers 5.x non-persistent buffer corruption, plus one around `cache.get_seq_length()` returning a tensor instead of an int. Most other pairs are only superficially similar by model/subsystem.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 43950 is the broadest/root-cause report for the non-persistent-buffer regression and best covers the related duplicate reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43950 is the most representative issue: it states the underlying regression clearly, is not overly model-specific, and matches the strongest duplicate candidates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43975", + "right": "issue:44292", + "accept": false, + "reason": "Both are tokenizer/output complaints, but they involve different models and different failure modes; not the same concrete bug." + }, + { + "left": "issue:44918", + "right": "issue:45507", + "accept": false, + "reason": "Different models and different code paths: TRL SFT unpacking vs an invalid model method call." + }, + { + "left": "issue:43950", + "right": "issue:44534", + "accept": true, + "reason": "Same underlying regression: Transformers 5.x mishandling non-persistent buffers leading to junk/uninitialized values." + }, + { + "left": "issue:39692", + "right": "issue:43054", + "accept": false, + "reason": "Doc-example errors and a model quality/performance complaint are not the same bug." + }, + { + "left": "issue:43644", + "right": "issue:43950", + "accept": true, + "reason": "Same non-persistent-buffer corruption regression in Transformers 5.x, described from the reverse angle." + }, + { + "left": "issue:36331", + "right": "issue:41762", + "accept": false, + "reason": "Unrelated training API mismatch vs DeepSpeed loading failure." + }, + { + "left": "issue:43872", + "right": "issue:44589", + "accept": false, + "reason": "Both are quantization/storage related, but the concrete errors and code paths differ." + }, + { + "left": "issue:43828", + "right": "issue:44928", + "accept": false, + "reason": "Different model-specific dtype/autocast vs RLHF/SDPA behavior; not the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46858", + "accept": true, + "reason": "Same API contract break: `cache.get_seq_length()` returns a tensor instead of an int, causing downstream failures." + }, + { + "left": "issue:45920", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but the reports target different model families and specific breakages." + }, + { + "left": "issue:43054", + "right": "issue:45701", + "accept": false, + "reason": "A model embedding-quality complaint is not the same as a tokenizer-version change issue." + }, + { + "left": "issue:38175", + "right": "issue:43054", + "accept": false, + "reason": "Both mention SigLIP2, but the observed symptoms and likely causes are different." + }, + { + "left": "issue:43950", + "right": "issue:45902", + "accept": true, + "reason": "Same family of non-persistent-buffer corruption/uninitialized-memory regression after materialization/loading." + }, + { + "left": "issue:41553", + "right": "issue:43883", + "accept": false, + "reason": "Bad error message for tokenizer loading is unrelated to missing model attribute behavior." + }, + { + "left": "issue:43054", + "right": "issue:43493", + "accept": false, + "reason": "Both are SigLIP2-related but describe different discrepancies and likely different causes." + }, + { + "left": "issue:44910", + "right": "issue:45859", + "accept": false, + "reason": "Both involve Qwen3.5, but one is flash-attn packed-sequence handling and the other is CPU-offload rotary embedding compatibility." + }, + { + "left": "issue:42947", + "right": "issue:44387", + "accept": false, + "reason": "LoRA gradient checkpointing and int4 quantization OOM are different optimization issues." + }, + { + "left": "issue:45663", + "right": "issue:45902", + "accept": false, + "reason": "FSDP2 per-layer state rebuild vs non-persistent-buffer materialization bug are different code paths." + }, + { + "left": "issue:41553", + "right": "issue:43881", + "accept": false, + "reason": "Different models and different loading failures." + }, + { + "left": "issue:44403", + "right": "issue:44451", + "accept": false, + "reason": "General loading noise vs a specific model loading failure are not duplicates." + }, + { + "left": "issue:43975", + "right": "issue:45520", + "accept": false, + "reason": "Tokenizer detokenization issue and import error on Python 3.13 are unrelated." + }, + { + "left": "issue:44403", + "right": "issue:44488", + "accept": false, + "reason": "Both are load-related, but the symptoms and likely fixes differ." + }, + { + "left": "issue:24643", + "right": "issue:28530", + "accept": false, + "reason": "DeepSpeed weight-dimension runtime error is unrelated to early stopping metric selection." + }, + { + "left": "issue:42947", + "right": "issue:43856", + "accept": false, + "reason": "Training-memory inefficiency and gradient-checkpointing behavior are not the same bug." + }, + { + "left": "issue:43643", + "right": "issue:45020", + "accept": false, + "reason": "Remote-code config omission is related to, but not the same as, the broader remote-code breakage report." + }, + { + "left": "issue:41762", + "right": "issue:42915", + "accept": false, + "reason": "Different model/loading failures with different quantization and distributed settings." + }, + { + "left": "issue:44991", + "right": "issue:45399", + "accept": false, + "reason": "Tokenizer loading failure and flash-attn2 fallback logic are unrelated." + }, + { + "left": "issue:43824", + "right": "issue:43872", + "accept": false, + "reason": "Model import failure and bitsandbytes parameter incompatibility are different problems." + }, + { + "left": "issue:43782", + "right": "issue:44960", + "accept": false, + "reason": "Different models and different failure modes." + }, + { + "left": "issue:43881", + "right": "issue:43883", + "accept": false, + "reason": "A model-loading failure and a missing attribute on another model are not the same underlying bug." + }, + { + "left": "issue:45685", + "right": "issue:45823", + "accept": false, + "reason": "MPS histogram limitation and Gemma4 device-mismatch are distinct issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet. The summary stays conservative, and the accepted soft-edge pairs are plausibly the same bug families from the titles: the Transformers 5.x non-persistent-buffer regression cluster and the `cache.get_seq_length()` contract regression. The rejected pairs are also cautious and generally supported by the issue titles." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 5, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:24643", + "issue:30064", + "issue:36331", + "issue:41093", + "issue:41553", + "issue:42491", + "issue:42947", + "issue:43257", + "issue:43404", + "issue:43792", + "issue:43827", + "issue:43828", + "issue:43854", + "issue:43872", + "issue:43940", + "issue:43950", + "issue:44220", + "issue:44291", + "issue:44368", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44560", + "issue:44661", + "issue:44805", + "issue:44863", + "issue:44928", + "issue:44991", + "issue:45020", + "issue:45081", + "issue:45305", + "issue:45313", + "issue:45356", + "issue:45399", + "issue:45488", + "issue:45588", + "issue:45715", + "issue:45759", + "issue:45823", + "issue:45923", + "issue:46315", + "issue:46402", + "issue:46489", + "issue:46661", + "issue:46869", + "issue:46894", + "issue:46897" + ], + "soft_pairs": [ + "issue:45081|issue:45399", + "issue:45020|issue:45823", + "issue:44661|issue:45020", + "issue:42491|issue:44368", + "issue:43828|issue:45823", + "issue:45020|issue:45356", + "issue:45715|issue:46315", + "issue:45488|issue:46489", + "issue:43827|issue:44509", + "issue:46894|issue:46897", + "issue:43872|issue:46869", + "issue:43404|issue:43950", + "issue:45759|issue:46661", + "issue:43872|issue:44291", + "issue:45020|issue:46661", + "issue:43792|issue:44220", + "issue:43257|issue:45305", + "issue:41553|issue:44991", + "issue:42947|issue:45923", + "issue:36331|issue:41093", + "issue:43854|issue:44863", + "issue:43940|issue:45313", + "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" + ], + "prepared_review_unit_hash": "bb65905f513d32be23c654c2367a945f8d6d3412137818a2d1fe690d9ac710d0", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 37084, + "estimated_input_tokens": 9271, + "estimated_eval_tokens": 18798 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is mostly unrelated single-issue regressions. The clearest duplicate clusters are the v5 docs cleanup, the `_is_hf_initialized`/`init_empty_weights` compatibility regression, and the repeated training-loss double-shift reports; most other soft edges are only loosely similar and should stay separate.", + "confidence": 0.72, + "canonical_issue_reason": "Issue 43872 is the strongest anchor: it is a concrete, widely-impacting runtime compatibility regression with a crisp failure signature, and it has the clearest close duplicate-style follow-up in 44291. It is more actionable and representative than the broader or more speculative matches in the set.", + "canonical_pr_reason": null, + "best_issue_reason": "43872 is the best issue to keep as the representative canonical bug in this batch because the symptom is specific, reproducible, and clearly tied to a single compatibility break.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45081", + "right": "issue:45399", + "accept": false, + "reason": "Both touch tokenizer/attention-related code, but one is a Mistral regex patch crash and the other is flash-attn fallback blocking; different code paths and fixes." + }, + { + "left": "issue:45020", + "right": "issue:45823", + "accept": false, + "reason": "Remote-code loading regressions and a device_map auto PLE device mismatch are different failure modes." + }, + { + "left": "issue:44661", + "right": "issue:45020", + "accept": false, + "reason": "`add-new-model-like` registry logic is unrelated to the broader remote-code breakage described here." + }, + { + "left": "issue:42491", + "right": "issue:44368", + "accept": false, + "reason": "Both mention Qwen3/LoRA, but one is a model-compatibility/loading break and the other is a warning about `tie_word_embeddings`; not the same bug." + }, + { + "left": "issue:43828", + "right": "issue:45823", + "accept": false, + "reason": "Autocast dtype mismatch and PLE device mismatch are distinct runtime errors." + }, + { + "left": "issue:45020", + "right": "issue:45356", + "accept": false, + "reason": "Remote-code model breakage and Kimi tokenizer codec/regex regression are separate issues." + }, + { + "left": "issue:45715", + "right": "issue:46315", + "accept": false, + "reason": "These are opposite-direction tokenizer performance regressions (`convert_ids_to_tokens` vs `convert_tokens_to_ids`); related area, but not the same bug." + }, + { + "left": "issue:45488", + "right": "issue:46489", + "accept": false, + "reason": "Both concern DeepSeek tokenizer behavior, but they target different model families and different symptom descriptions; too broad to merge." + }, + { + "left": "issue:43827", + "right": "issue:44509", + "accept": true, + "reason": "Both report stale docs still referencing pipeline tasks removed in v5; same documentation cleanup." + }, + { + "left": "issue:46894", + "right": "issue:46897", + "accept": true, + "reason": "Both are the same training-loss double-shift bug pattern reported for different model variants; one shared fix could plausibly cover both." + }, + { + "left": "issue:43872", + "right": "issue:46869", + "accept": false, + "reason": "The `_is_hf_initialized` bitsandbytes compatibility issue is unrelated to a YOLO multi-GPU device_map failure." + }, + { + "left": "issue:43404", + "right": "issue:43950", + "accept": false, + "reason": "A tied lm_head regression and silent non-persistent buffer corruption are different model-loading bugs." + }, + { + "left": "issue:45759", + "right": "issue:46661", + "accept": false, + "reason": "Composite multimodal config unwrapping and remote-code `config_class` lookup failures are separate loader problems." + }, + { + "left": "issue:43872", + "right": "issue:44291", + "accept": true, + "reason": "Both describe the same unexpected `_is_hf_initialized` kwarg regression during model initialization/loading." + }, + { + "left": "issue:45020", + "right": "issue:46661", + "accept": false, + "reason": "Generic remote-code breakage and the specific missing `config_class` attribute error are not the same underlying bug." + }, + { + "left": "issue:43792", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio-related, but one is a model that cannot run at all and the other is a specific `_torch_extract_fbank_features()` issue; not enough evidence of one root cause." + }, + { + "left": "issue:43257", + "right": "issue:45305", + "accept": false, + "reason": "DeepSpeed weight-loading failure and gradient averaging with GAS are different training-path bugs." + }, + { + "left": "issue:41553", + "right": "issue:44991", + "accept": false, + "reason": "Bad tokenizer error messaging for Voxtral and a tokenizer-loading regression for EMBEDDIA/est-roberta are unrelated." + }, + { + "left": "issue:42947", + "right": "issue:45923", + "accept": false, + "reason": "Gradient checkpointing being ineffective under PEFT LoRA is not the same as a missing `supports_gradient_checkpointing` flag in a remote-code variant." + }, + { + "left": "issue:36331", + "right": "issue:41093", + "accept": false, + "reason": "An unexpected `num_items_in_batch` kwarg and a mask/tensor shape mismatch are different failures." + }, + { + "left": "issue:43854", + "right": "issue:44863", + "accept": false, + "reason": "Different model-support issues: one is GLM-4.7 Flash loading in tests, the other is NemotronH checkpoint loading." + }, + { + "left": "issue:43940", + "right": "issue:45313", + "accept": false, + "reason": "Both are Qwen3-family ZeRO-3 load failures, but the model variants and missing-weight manifestations differ enough that they should stay separate." + }, + { + "left": "issue:41553", + "right": "issue:44488", + "accept": false, + "reason": "Voxtral tokenizer error messaging and `cjvt/sleng-bert` loading failure are unrelated." + }, + { + "left": "issue:44928", + "right": "issue:45588", + "accept": false, + "reason": "Gradient explosion from 3D position_ids/SDPA fallback and a flash_attention `s_aux=None` AttributeError are different code-path problems." + }, + { + "left": "issue:41093", + "right": "issue:44805", + "accept": false, + "reason": "Both are shape-mismatch IndexErrors, but there is no evidence they share the same root cause or fix." + }, + { + "left": "issue:44479", + "right": "issue:44560", + "accept": false, + "reason": "Both are video regressions, but they affect different model families and have different symptoms." + }, + { + "left": "issue:24643", + "right": "issue:30064", + "accept": false, + "reason": "DeepSpeed training weight shape error and void segmentation-map preprocessing are unrelated." + }, + { + "left": "issue:44530", + "right": "issue:44863", + "accept": false, + "reason": "PagedAttentionCache linear_attention crash and NemotronH checkpoint loading are separate model/runtime issues." + }, + { + "left": "issue:43872", + "right": "issue:46402", + "accept": false, + "reason": "The `_is_hf_initialized` compatibility regression is unrelated to Sam3Model text-encoder weights not loading." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the accepted soft edges are well-supported by near-identical titles or clearly shared cleanup patterns. The rest of the pair verdicts stay appropriately conservative." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 6, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:36010", + "issue:38175", + "issue:39692", + "issue:42491", + "issue:42617", + "issue:42915", + "issue:43317", + "issue:43366", + "issue:43531", + "issue:43550", + "issue:43596", + "issue:43632", + "issue:43638", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43720", + "issue:43824", + "issue:43828", + "issue:43950", + "issue:43957", + "issue:43994", + "issue:44155", + "issue:44534", + "issue:44560", + "issue:44589", + "issue:44661", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44877", + "issue:44910", + "issue:44918", + "issue:45030", + "issue:45325", + "issue:45584", + "issue:45663", + "issue:45672", + "issue:45685", + "issue:45812", + "issue:45823", + "issue:45834", + "issue:45902", + "issue:46489", + "issue:46515", + "issue:46858", + "issue:46899" + ], + "soft_pairs": [ + "issue:43366|issue:43828", + "issue:43596|issue:46515", + "issue:45325|issue:45902", + "issue:43366|issue:43531", + "issue:43638|issue:46515", + "issue:45663|issue:45823", + "issue:44155|issue:45584", + "issue:43366|issue:44589", + "issue:43824|issue:43957", + "issue:43950|issue:45834", + "issue:42915|issue:43957", + "issue:44910|issue:46858", + "issue:43716|issue:44560", + "issue:43645|issue:43646", + "issue:44560|issue:45685", + "issue:43550|issue:44910", + "issue:42617|issue:43366", + "issue:44683|issue:46515", + "issue:44560|issue:44918", + "issue:36010|issue:42915", + "issue:45812|issue:46489", + "issue:45823|issue:46515", + "issue:44661|issue:45823", + "issue:38175|issue:43994", + "issue:44560|issue:44805", + "issue:45672|issue:46899", + "issue:44877|issue:45030", + "issue:43632|issue:45834", + "issue:43638|issue:44661", + "issue:43596|issue:44661", + "issue:42491|issue:43720", + "issue:44661|issue:44805", + "issue:43317|issue:44756", + "issue:38175|issue:39692", + "issue:43644|issue:44534" + ], + "prepared_review_unit_hash": "603d013f23123eb51c09e7cad2d49fa886bfefcf1b71dfbbd513f58ddbe1d77c", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 35, + "serialized_chars": 38348, + "estimated_input_tokens": 9587, + "estimated_eval_tokens": 19430 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous issue cluster with a few clear duplicate subclusters, but no single theme spans everything. The strongest duplicate groups are the Transformers v5 non-persistent-buffer regression, SigLIP2 bad-output reports, Gemma4 4-bit dtype-cast failures, and a custom-model initialization regression.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 43950 is the clearest anchor for the strongest duplicate subcluster: it precisely describes the v5 `from_pretrained()` corruption of `register_buffer(persistent=False)`, which also subsumes the shorter reports in 43644/44534.", + "canonical_pr_reason": null, + "best_issue_reason": "43950 is the best representative issue overall because it states the underlying regression most explicitly and is the most actionable/diagnostic of the closely related duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43366", + "right": "issue:43828", + "accept": false, + "reason": "Different problems: GGUF support for gpt-oss vs a Phi-tiny-MoE autocast dtype mismatch." + }, + { + "left": "issue:43596", + "right": "issue:46515", + "accept": false, + "reason": "Unrelated failures: DeepSpeed Zero3/Bert index error vs a broad torch-version incompatibility report." + }, + { + "left": "issue:45325", + "right": "issue:45902", + "accept": false, + "reason": "Different code paths: Qwen2.5-VL rope/position-id scaling vs Qwen3-MoE meta materialization of `inv_freq`." + }, + { + "left": "issue:43366", + "right": "issue:43531", + "accept": false, + "reason": "No overlap beyond model family names; GGUF support request vs a sliding-window bug in Qwen3-MoE." + }, + { + "left": "issue:43638", + "right": "issue:46515", + "accept": false, + "reason": "Different underlying bugs: Zero3 Bert training index error vs torch 2.4.1 incompatibility." + }, + { + "left": "issue:45663", + "right": "issue:45823", + "accept": false, + "reason": "Same model family, but distinct issues: per-layer `shared_kv_states` rebuild vs `device_map=auto` device mismatch." + }, + { + "left": "issue:44155", + "right": "issue:45584", + "accept": false, + "reason": "Different subsystems: AudioFlamingo3 batched-track leakage vs Whisper empty-transcription handling." + }, + { + "left": "issue:43366", + "right": "issue:44589", + "accept": false, + "reason": "Unrelated: gpt-oss GGUF support vs a Float8 storage lookup error." + }, + { + "left": "issue:43824", + "right": "issue:43957", + "accept": false, + "reason": "Import error for a specific Qwen class vs a generic meta-device loading regression." + }, + { + "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": "Different failures: Qwen3-MoE FineGrainedFP8Config issue vs meta-loading breakage across some models." + }, + { + "left": "issue:44910", + "right": "issue:46858", + "accept": false, + "reason": "Both touch cache/position logic, but the concrete bugs are different and not mergeable as one fix." + }, + { + "left": "issue:43716", + "right": "issue:44560", + "accept": false, + "reason": "Different multimodal bugs: image preprocessor dtype mismatch vs video StopIteration." + }, + { + "left": "issue:43645", + "right": "issue:43646", + "accept": true, + "reason": "These describe the same custom-model initialization regression; one is just notebook-specific wording of the same breakage." + }, + { + "left": "issue:44560", + "right": "issue:45685", + "accept": false, + "reason": "Unrelated: Qwen3-VL video input failure vs an MPS histogram kernel missing for Int tensors." + }, + { + "left": "issue:43550", + "right": "issue:44910", + "accept": false, + "reason": "Different failures in different models/backends: Bamba torch.compile+SDPA vs Qwen3.5 flash-attention position-id handling." + }, + { + "left": "issue:42617", + "right": "issue:43366", + "accept": false, + "reason": "No shared bug: a 3d_parallel.py execution problem vs gpt-oss GGUF support." + }, + { + "left": "issue:44683", + "right": "issue:46515", + "accept": false, + "reason": "Both mention torch versions, but one is a compiled flex_attention regression on torch>=2.9 and the other is a compatibility report for torch 2.4.1." + }, + { + "left": "issue:44560", + "right": "issue:44918", + "accept": false, + "reason": "Different Qwen3-related bugs: video StopIteration vs SFT trainer embedding unpacking." + }, + { + "left": "issue:36010", + "right": "issue:42915", + "accept": false, + "reason": "Unrelated import-path issue for `GenerationMixin` vs a Qwen3-MoE FP8 failure." + }, + { + "left": "issue:45812", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different model families and likely different code paths; not the same concrete bug." + }, + { + "left": "issue:45823", + "right": "issue:46515", + "accept": false, + "reason": "Gemma4 forward-device mismatch is unrelated to the torch 2.4.1 incompatibility report." + }, + { + "left": "issue:44661", + "right": "issue:45823", + "accept": false, + "reason": "Tokenizer-mapping edge case vs Gemma4 PLE device placement bug." + }, + { + "left": "issue:38175", + "right": "issue:43994", + "accept": true, + "reason": "Same SigLIP2 model/processor failure mode: one report shows zero probabilities and the other nonsensical outputs from the same model family." + }, + { + "left": "issue:44560", + "right": "issue:44805", + "accept": false, + "reason": "Different bugs: video StopIteration vs mask-shape index mismatch." + }, + { + "left": "issue:45672", + "right": "issue:46899", + "accept": true, + "reason": "Same Gemma4 4-bit multimodal quantization regression; both point to unsafe dtype handling in audio/vision embedders." + }, + { + "left": "issue:44877", + "right": "issue:45030", + "accept": false, + "reason": "Both involve config validation, but they are for different model families and different validation failures." + }, + { + "left": "issue:43632", + "right": "issue:45834", + "accept": false, + "reason": "Unrelated: `_is_hf_initialized` regression vs Kosmos2.5 OCR indexing." + }, + { + "left": "issue:43638", + "right": "issue:44661", + "accept": false, + "reason": "Different issues: Zero3/Bert index error vs tokenizer-mapping behavior in add-new-model-like." + }, + { + "left": "issue:43596", + "right": "issue:44661", + "accept": false, + "reason": "Different code paths and symptoms; these are not the same underlying bug." + }, + { + "left": "issue:42491", + "right": "issue:43720", + "accept": false, + "reason": "Distinct loading regressions in different model stacks; no clear shared root cause." + }, + { + "left": "issue:44661", + "right": "issue:44805", + "accept": false, + "reason": "Tokenizer registration edge case vs a tensor shape/indexing bug." + }, + { + "left": "issue:43317", + "right": "issue:44756", + "accept": false, + "reason": "Different memory/loading issues: gpu+cpu offload load failure vs disabling mmap to avoid OOM on Strix Halo." + }, + { + "left": "issue:38175", + "right": "issue:39692", + "accept": true, + "reason": "Same underlying SigLIP2 regression around model/processor mismatch; 39692 adds docs-example failures on top of the same bad setup." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The identified subclusters are supported by the titles/body cues, and the accepted soft-edge pairs are reasonably mergeable as the same underlying bugfix rather than just thematically similar issues." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 7, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:39692", + "issue:41720", + "issue:42491", + "issue:42915", + "issue:43257", + "issue:43425", + "issue:43644", + "issue:43645", + "issue:43708", + "issue:43854", + "issue:43856", + "issue:43873", + "issue:43931", + "issue:43957", + "issue:43994", + "issue:44075", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44488", + "issue:44756", + "issue:44863", + "issue:44910", + "issue:44960", + "issue:44977", + "issue:44991", + "issue:45276", + "issue:45292", + "issue:45313", + "issue:45335", + "issue:45357", + "issue:45362", + "issue:45406", + "issue:45488", + "issue:45507", + "issue:45588", + "issue:45812", + "issue:46050", + "issue:46459", + "issue:46468", + "issue:46512", + "issue:46846", + "issue:46897", + "issue:46901" + ], + "soft_pairs": [ + "issue:42491|issue:45276", + "issue:41720|issue:42915", + "issue:42915|issue:44756", + "issue:44451|issue:44488", + "issue:44960|issue:44977", + "issue:44863|issue:46846", + "issue:46459|issue:46512", + "issue:45313|issue:45357", + "issue:45507|issue:46468", + "issue:44488|issue:44991", + "issue:43644|issue:43645", + "issue:42491|issue:45292", + "issue:39692|issue:43994", + "issue:43257|issue:43854", + "issue:43856|issue:43873", + "issue:43856|issue:44387", + "issue:43854|issue:43931", + "issue:43708|issue:44075", + "issue:42491|issue:45335", + "issue:44960|issue:46050", + "issue:46897|issue:46901", + "issue:45488|issue:45812", + "issue:45276|issue:45335", + "issue:45362|issue:45406", + "issue:44164|issue:44291", + "issue:44910|issue:45588", + "issue:44402|issue:44488", + "issue:39692|issue:44960", + "issue:43425|issue:43957", + "issue:41720|issue:44155" + ], + "prepared_review_unit_hash": "fac7adefd08e96300735fdaf953459089415767200680769ed268cba6ba686d0", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "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": "The cluster is mostly heterogeneous, with several unrelated model-loading, quantization, tokenizer, and training bugs. The 44451 \u2194 44488 pair is the closest match, but the evidence only supports \u201csimilar BERT-loading failures,\u201d not a confident duplicate/root-cause match.", + "confidence": 0.61, + "canonical_issue_reason": "No strong cluster-wide canonical issue: the packet spans many unrelated regressions. If a representative anchor is needed, 44488 is the broadest BERT-loading report, but the evidence is not strong enough to treat it as a confirmed duplicate target for 44451.", + "canonical_pr_reason": null, + "best_issue_reason": "44488 is the best representative issue here because it is the broader, later BERT-loading report and has the stronger discussion footprint, but only as a loose anchor rather than a proven duplicate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42491", + "right": "issue:45276", + "accept": false, + "reason": "Different bugs: Qwen3 MoE LoRA compatibility vs Gemma4 token-embedding resize behavior." + }, + { + "left": "issue:41720", + "right": "issue:42915", + "accept": false, + "reason": "Same model family, but different failures: device-mapping cuda assert vs FineGrainedFP8Config loading." + }, + { + "left": "issue:42915", + "right": "issue:44756", + "accept": false, + "reason": "Unrelated problems: Qwen3 MoE FP8 config vs Strix Halo mmap/OOM behavior." + }, + { + "left": "issue:44451", + "right": "issue:44488", + "accept": false, + "reason": "Very similar BERT-loading symptoms, but the packet does not \u0581\u0578\u0582\u0575\u0581 a shared root cause or deterministic duplicate signal." + }, + { + "left": "issue:44960", + "right": "issue:44977", + "accept": false, + "reason": "Different models and failure modes: GLM5 vs Qwen3.5 flash-attention generation." + }, + { + "left": "issue:44863", + "right": "issue:46846", + "accept": false, + "reason": "Different code paths: NemotronH checkpoint loading vs Qwen3.5 tensor-parallel plan/OOM." + }, + { + "left": "issue:46459", + "right": "issue:46512", + "accept": false, + "reason": "Both involve dtype handling, but one is composite-checkpoint loading and the other is config state consistency." + }, + { + "left": "issue:45313", + "right": "issue:45357", + "accept": false, + "reason": "Same model family, but separate bugs: ZeRO-3 loading vs save_pretrained visual-key serialization." + }, + { + "left": "issue:45507", + "right": "issue:46468", + "accept": false, + "reason": "Different artifacts and symptoms: GraniteMoEHybrid invalid method vs broken Nemotron example." + }, + { + "left": "issue:44488", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading-related, but 44991 is a distinct tokenizer-loading regression for another model family." + }, + { + "left": "issue:43644", + "right": "issue:43645", + "accept": false, + "reason": "Both are Transformers 5.x issues, but one is buffer initialization and the other is custom-model init in Jupyter." + }, + { + "left": "issue:42491", + "right": "issue:45292", + "accept": false, + "reason": "Same API surface, but different bugs and model paths; not the same concrete issue." + }, + { + "left": "issue:39692", + "right": "issue:43994", + "accept": false, + "reason": "Different SigLIP2 failures: docs example mismatch/quantization vs nonsensical AutoModel and pipeline results." + }, + { + "left": "issue:43257", + "right": "issue:43854", + "accept": false, + "reason": "Different Qwen3-related failures: accelerate+deepspeed MoE conversion vs unit-test loading error." + }, + { + "left": "issue:43856", + "right": "issue:43873", + "accept": false, + "reason": "Both mention memory, but the underlying issues are different: training efficiency vs quantization offloading behavior." + }, + { + "left": "issue:43856", + "right": "issue:44387", + "accept": false, + "reason": "Different memory regressions: Qwen3 MoE training memory vs int4 quantization reserved-memory OOM." + }, + { + "left": "issue:43854", + "right": "issue:43931", + "accept": false, + "reason": "Different model-loading failures on different models; no shared code-path evidence." + }, + { + "left": "issue:43708", + "right": "issue:44075", + "accept": false, + "reason": "Trainer checkpoint-resume/max_steps bug vs optimizer args being ignored." + }, + { + "left": "issue:42491", + "right": "issue:45335", + "accept": false, + "reason": "Different model families and embedding paths; too model-specific to merge." + }, + { + "left": "issue:44960", + "right": "issue:46050", + "accept": false, + "reason": "Different features: flash-attention generation failure vs quantization save/load corruption." + }, + { + "left": "issue:46897", + "right": "issue:46901", + "accept": false, + "reason": "Same broad symptom pattern, but different model implementations and no evidence of a shared code path." + }, + { + "left": "issue:45488", + "right": "issue:45812", + "accept": false, + "reason": "Different tokenizer regressions in different stacks: pre-tokenizer override vs Granite token ID mismatch." + }, + { + "left": "issue:45276", + "right": "issue:45335", + "accept": false, + "reason": "Same API, but different model families and embedding wiring; not the same bug." + }, + { + "left": "issue:45362", + "right": "issue:45406", + "accept": false, + "reason": "Unrelated crashes: Qwen3.5 chat generation vs Gemma4Processor missing _tokenizer in serve." + }, + { + "left": "issue:44164", + "right": "issue:44291", + "accept": false, + "reason": "Different save/load bugs: extra_state handling vs unexpected _is_hf_initialized argument." + }, + { + "left": "issue:44910", + "right": "issue:45588", + "accept": false, + "reason": "Both touch flash attention, but they fail on different inputs and distinct code paths." + }, + { + "left": "issue:44402", + "right": "issue:44488", + "accept": false, + "reason": "Pad/unk vocab-size mismatch is a different tokenizer/model-config issue from the BERT loading failure." + }, + { + "left": "issue:39692", + "right": "issue:44960", + "accept": false, + "reason": "Completely different models and symptoms: SigLIP2 docs/example issues vs GLM5." + }, + { + "left": "issue:43425", + "right": "issue:43957", + "accept": false, + "reason": "Different compatibility failures: Torch version incompatibility vs torch.device('meta') loading breakage." + }, + { + "left": "issue:41720", + "right": "issue:44155", + "accept": false, + "reason": "Different failures: Qwen3 device-mapping assert vs AudioFlamingo3 batched inference leak." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The summary overstates the evidence around 44451 \u2194 44488. The packet only shows two similar BERT-loading issues with no deterministic duplicate signal and no explicit shared root cause; calling it the \"only clear duplicate pair\" and the same underlying regression is stronger than the evidence supports. The rest of the pairwise rejections are mostly conservative, but the canonical/summary claim is not sufficiently grounded." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 8, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:36010", + "issue:39404", + "issue:42796", + "issue:42947", + "issue:43335", + "issue:43381", + "issue:43404", + "issue:43425", + "issue:43454", + "issue:43645", + "issue:43646", + "issue:43708", + "issue:43828", + "issue:43854", + "issue:43873", + "issue:43957", + "issue:44079", + "issue:44155", + "issue:44164", + "issue:44291", + "issue:44292", + "issue:44387", + "issue:44402", + "issue:44451", + "issue:44479", + "issue:44589", + "issue:44756", + "issue:44863", + "issue:44877", + "issue:44912", + "issue:44928", + "issue:45216", + "issue:45276", + "issue:45292", + "issue:45335", + "issue:45588", + "issue:45663", + "issue:45684", + "issue:45698", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45859", + "issue:46326", + "issue:46468", + "issue:46829", + "issue:46894", + "issue:46901" + ], + "soft_pairs": [ + "issue:45663|issue:45759", + "issue:44079|issue:46829", + "issue:43404|issue:43454", + "issue:44291|issue:44756", + "issue:43873|issue:44387", + "issue:42947|issue:45704", + "issue:44292|issue:44912", + "issue:36010|issue:39404", + "issue:45663|issue:45859", + "issue:44387|issue:44756", + "issue:42947|issue:43381", + "issue:44155|issue:45823", + "issue:43957|issue:44756", + "issue:43335|issue:45663", + "issue:45276|issue:45292", + "issue:44863|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:46894|issue:46901", + "issue:43828|issue:43957", + "issue:44402|issue:44451", + "issue:43646|issue:43708", + "issue:43645|issue:44164", + "issue:44589|issue:44756", + "issue:43425|issue:43828" + ], + "prepared_review_unit_hash": "1a6de54c84c3c3d05b6695eded916a1100d706169f6f29423b063768ab07639d", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37131, + "estimated_input_tokens": 9283, + "estimated_eval_tokens": 18822 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly heterogeneous; the only strong duplicate-like family is the resize_token_embeddings propagation bug around output embeddings / decoder embeddings. Most other pairs are same subsystem at best, but different concrete bugs and should stay separate.", + "confidence": 0.74, + "canonical_issue_reason": "Issue 45292 is the cleanest generic statement of the resize_token_embeddings/output_embeddings bug and best serves as the umbrella for the closely related 45276 and 45335 reports. The rest of the cluster is unrelated enough that no single issue can represent the whole set.", + "canonical_pr_reason": null, + "best_issue_reason": "45292 is the best overall issue candidate because it is the broadest, least model-specific description of the shared embedding-resize failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45663", + "right": "issue:45759", + "accept": false, + "reason": "Different bugs: FSDP2 KeyError in Gemma4 attention vs from_config not unwrapping text_config for Qwen multimodal configs." + }, + { + "left": "issue:44079", + "right": "issue:46829", + "accept": false, + "reason": "Both involve outputs, but one is a generic ModelOutput key assignment bug and the other is a specific model returning the wrong tensor in hidden_states." + }, + { + "left": "issue:43404", + "right": "issue:43454", + "accept": false, + "reason": "Both are lm_head tying problems, but in different model classes with different likely code paths and fixes." + }, + { + "left": "issue:44291", + "right": "issue:44756", + "accept": false, + "reason": "Unrelated: init_empty_weights argument handling vs mmap-related OOM on Strix Halo." + }, + { + "left": "issue:43873", + "right": "issue:44387", + "accept": false, + "reason": "Both mention quantization and memory, but one is offloading behavior and the other is int4 reserved-memory OOM; not the same bug." + }, + { + "left": "issue:42947", + "right": "issue:45704", + "accept": false, + "reason": "Gradient checkpointing/LoRA issue is unrelated to the apex.FusedRMSNorm memory leak in T5." + }, + { + "left": "issue:44292", + "right": "issue:44912", + "accept": false, + "reason": "Different quantization failures: NVFP4 loading error vs MXFP4 load fallback to bf16." + }, + { + "left": "issue:36010", + "right": "issue:39404", + "accept": false, + "reason": "Import error for GenerationMixin is unrelated to Whisper pipeline return_language regression." + }, + { + "left": "issue:45663", + "right": "issue:45859", + "accept": false, + "reason": "Different model families and different offload/device behavior bugs." + }, + { + "left": "issue:44387", + "right": "issue:44756", + "accept": false, + "reason": "Both can lead to OOM, but the causes are different: int4 reserved memory vs mmap on Strix Halo." + }, + { + "left": "issue:42947", + "right": "issue:43381", + "accept": false, + "reason": "Both mention gradient checkpointing, but one is ineffective with PEFT LoRA while the other forbids checkpointing in eval mode." + }, + { + "left": "issue:44155", + "right": "issue:45823", + "accept": false, + "reason": "AudioFlamingo batching leak is unrelated to Gemma4 device mismatch under device_map=auto." + }, + { + "left": "issue:43957", + "right": "issue:44756", + "accept": false, + "reason": "Meta-device model loading regression is unrelated to Strix Halo mmap OOM." + }, + { + "left": "issue:43335", + "right": "issue:45663", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction bug is unrelated to Gemma4 FSDP2 KeyError." + }, + { + "left": "issue:45276", + "right": "issue:45292", + "accept": true, + "reason": "Same resize_token_embeddings propagation bug: both report output embeddings not being updated, with 45276 adding a Gemma4-specific manifestation." + }, + { + "left": "issue:44863", + "right": "issue:46468", + "accept": false, + "reason": "NemotronH checkpoint loading failure is separate from the broken official Nemotron example." + }, + { + "left": "issue:44479", + "right": "issue:44877", + "accept": false, + "reason": "Video-input regression for Qwen VL models is unrelated to strict config blocking granite_speech." + }, + { + "left": "issue:43381", + "right": "issue:44928", + "accept": false, + "reason": "Checkpointing-in-eval-mode and Qwen3.5 RLHF NaN explosion are different problems." + }, + { + "left": "issue:44164", + "right": "issue:44756", + "accept": false, + "reason": "save/from_pretrained extra_state handling is unrelated to mmap OOM behavior." + }, + { + "left": "issue:45684", + "right": "issue:45698", + "accept": false, + "reason": "Both involve custom-model save/load, but one is read-only permission propagation and the other is loading the wrong custom module." + }, + { + "left": "issue:45588", + "right": "issue:45663", + "accept": false, + "reason": "Flash-attention crash on s_aux=None is unrelated to Gemma4 FSDP2 KeyError." + }, + { + "left": "issue:43854", + "right": "issue:45216", + "accept": false, + "reason": "Different serialization/load regressions affecting different models and APIs." + }, + { + "left": "issue:42796", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer BPE marker handling is not the same as continue_final_message behavior for mistral_common tokenizers." + }, + { + "left": "issue:45292", + "right": "issue:45335", + "accept": true, + "reason": "Same resize_token_embeddings bug family: both say resized token embeddings do not propagate to model output/input embeddings." + }, + { + "left": "issue:46894", + "right": "issue:46901", + "accept": false, + "reason": "Same broad double-shift pattern, but different model-specific training-loss implementations and not a single concrete code-path." + }, + { + "left": "issue:43828", + "right": "issue:43957", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to meta-device model loading." + }, + { + "left": "issue:44402", + "right": "issue:44451", + "accept": false, + "reason": "Tokenizer vocab-size mismatch is unrelated to inability to load ScandiBERT." + }, + { + "left": "issue:43646", + "right": "issue:43708", + "accept": false, + "reason": "Custom model initialization breakage and Trainer max_steps resume miscalculation are unrelated." + }, + { + "left": "issue:43645", + "right": "issue:44164", + "accept": false, + "reason": "Jupyter custom-model initialization bug is unrelated to save/from_pretrained extra_state handling." + }, + { + "left": "issue:44589", + "right": "issue:44756", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to Strix Halo mmap OOM." + }, + { + "left": "issue:43425", + "right": "issue:43828", + "accept": false, + "reason": "Torch 2.10 compatibility and autocast dtype mismatch are separate issues." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only clearly duplicate-like family supported by the titles is the resize_token_embeddings/output_embeddings group around 45292, 45276, and 45335, and the rejected soft pairs are framed as likely different bugs rather than asserted duplicates. 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": 23, + "review_unit_index": 9, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:36032", + "issue:39404", + "issue:43299", + "issue:43425", + "issue:43452", + "issue:43596", + "issue:43611", + "issue:43644", + "issue:43645", + "issue:43716", + "issue:43856", + "issue:43901", + "issue:43906", + "issue:43931", + "issue:43994", + "issue:44079", + "issue:44164", + "issue:44265", + "issue:44479", + "issue:44488", + "issue:44509", + "issue:44530", + "issue:44589", + "issue:44671", + "issue:44683", + "issue:44756", + "issue:44805", + "issue:44898", + "issue:44912", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45479", + "issue:45701", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:46050", + "issue:46315", + "issue:46326", + "issue:46491", + "issue:46665", + "issue:46821", + "issue:46869", + "issue:46894", + "issue:46897", + "issue:46899" + ], + "soft_pairs": [ + "issue:46315|issue:46491", + "issue:45812|issue:46315", + "issue:39404|issue:45701", + "issue:39404|issue:46665", + "issue:44530|issue:45084", + "issue:43611|issue:43645", + "issue:44683|issue:45823", + "issue:43425|issue:46869", + "issue:43425|issue:43716", + "issue:43425|issue:45823", + "issue:44589|issue:45084", + "issue:44479|issue:44805", + "issue:43906|issue:44530", + "issue:43901|issue:44509", + "issue:44671|issue:44991", + "issue:43994|issue:44079", + "issue:44265|issue:44898", + "issue:36032|issue:43452", + "issue:39404|issue:44912", + "issue:46050|issue:46326", + "issue:44164|issue:45759", + "issue:43906|issue:46326", + "issue:44991|issue:45081", + "issue:43299|issue:43931", + "issue:45823|issue:46899", + "issue:45479|issue:46894", + "issue:43856|issue:44756", + "issue:43596|issue:44683", + "issue:45479|issue:46897", + "issue:44164|issue:46821", + "issue:44488|issue:44987", + "issue:44488|issue:45081", + "issue:43611|issue:43644" + ], + "prepared_review_unit_hash": "5dd70508918d855533bf2ae4e6203be8da2b3e0b961a689c686194101e55d7c9", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37737, + "estimated_input_tokens": 9435, + "estimated_eval_tokens": 19126 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of mostly unrelated Transformers v5 regressions across tokenizers, model loading, multimodal inputs, quantization, and docs. No soft-edge pair looks like the same concrete bug/change, so all candidates are rejected. The broadest representative issue is the Granite tokenizer regression.", + "confidence": 0.84, + "canonical_issue_reason": "Issue 45812 is the strongest representative: it describes a broad, user-visible tokenizer regression across all Granite models, has inbound references, and is more central than the many highly model-specific failures.", + "canonical_pr_reason": null, + "best_issue_reason": "Best overall issue candidate because it is the clearest high-impact regression with broad scope and likely duplicate potential, rather than a narrow one-off model bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46315", + "right": "issue:46491", + "accept": false, + "reason": "One is a tokenizer speed regression in convert_tokens_to_ids; the other is an encode/decode space-roundtrip correctness bug in CodeLlama. Different symptoms and code paths." + }, + { + "left": "issue:45812", + "right": "issue:46315", + "accept": false, + "reason": "Granite token IDs are wrong vs slow convert_tokens_to_ids for slow tokenizers. Different failure mode: correctness regression vs performance regression." + }, + { + "left": "issue:39404", + "right": "issue:45701", + "accept": false, + "reason": "Whisper pipeline return_language broke; the other is a broad tokenization version-change complaint. Too general and not the same concrete bug." + }, + { + "left": "issue:39404", + "right": "issue:46665", + "accept": false, + "reason": "Whisper pipeline behavior and from_pretrained inside Python classes are unrelated issues." + }, + { + "left": "issue:44530", + "right": "issue:45084", + "accept": false, + "reason": "Qwen3.5 PagedAttentionCache crashes on linear_attention; the other is a template compilation error. Different subsystems and symptoms." + }, + { + "left": "issue:43611", + "right": "issue:43645", + "accept": false, + "reason": "Both involve loading/custom models, but one is base_model_prefix loading breakage and the other is Jupyter custom-model initialization. Not the same bug." + }, + { + "left": "issue:44683", + "right": "issue:45823", + "accept": false, + "reason": "Torch>=2.9 flex_attention compilation failure is unrelated to Gemma4 device-map forward mismatches." + }, + { + "left": "issue:43425", + "right": "issue:46869", + "accept": false, + "reason": "Torch 2.10 incompatibility is not the same as a YOLO multi-GPU device_map auto failure." + }, + { + "left": "issue:43425", + "right": "issue:43716", + "accept": false, + "reason": "Torch version incompatibility and Mistral image-preprocessor dtype mismatch are different problems." + }, + { + "left": "issue:43425", + "right": "issue:45823", + "accept": false, + "reason": "General torch incompatibility vs Gemma4 device mismatch during forward; not the same underlying bug." + }, + { + "left": "issue:44589", + "right": "issue:45084", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to the template-node compilation error." + }, + { + "left": "issue:44479", + "right": "issue:44805", + "accept": false, + "reason": "Video-input regression for Qwen VL models vs a tensor mask shape mismatch. Different code paths." + }, + { + "left": "issue:43906", + "right": "issue:44530", + "accept": false, + "reason": "An isolated reproduction of an earlier issue is not the same as a Qwen3.5 cache crash." + }, + { + "left": "issue:43901", + "right": "issue:44509", + "accept": false, + "reason": "Both are docs-related, but they cover different pipeline-task deprecations and behavior mismatches." + }, + { + "left": "issue:44671", + "right": "issue:44991", + "accept": false, + "reason": "CamemBERT masked-LM prediction error and tokenizer load failure for est-roberta are distinct." + }, + { + "left": "issue:43994", + "right": "issue:44079", + "accept": false, + "reason": "Siglip2 produces bad outputs; ModelOutput key assignment is a separate data-structure bug." + }, + { + "left": "issue:44265", + "right": "issue:44898", + "accept": false, + "reason": "torch.export failure with torch_compilable_check is unrelated to Perceiver interpolation failures." + }, + { + "left": "issue:36032", + "right": "issue:43452", + "accept": false, + "reason": "T5 tokenizer add_special_tokens conflict and gguf_file loading breakage are different loading issues." + }, + { + "left": "issue:39404", + "right": "issue:44912", + "accept": false, + "reason": "Whisper pipeline return_language regression is unrelated to MXFP4 quantization fallback on git-oss-20b." + }, + { + "left": "issue:46050", + "right": "issue:46326", + "accept": false, + "reason": "Quantized-model save/load breakage via sinq is not the same as continue_final_message for mistral_common tokenizers." + }, + { + "left": "issue:44164", + "right": "issue:45759", + "accept": false, + "reason": "extra_state save/from_pretrained handling and unwrapping text_config in from_config are different model-construction problems." + }, + { + "left": "issue:43906", + "right": "issue:46326", + "accept": false, + "reason": "Repro of a prior issue and a mistral_common tokenizer behavior bug are unrelated." + }, + { + "left": "issue:44991", + "right": "issue:45081", + "accept": false, + "reason": "Both are tokenizer loading regressions, but for different models and different code paths (generic tokenizer load vs Mistral regex patching)." + }, + { + "left": "issue:43299", + "right": "issue:43931", + "accept": false, + "reason": "Both are Qwen3-VL-family loading failures, but one is a MoE-model regression and the other is a specific weight-shape mismatch. Not enough to treat as one bug." + }, + { + "left": "issue:45823", + "right": "issue:46899", + "accept": false, + "reason": "Gemma4 device-map mismatch during forward and Gemma4 4-bit multimodal dtype-cast breakage are separate issues." + }, + { + "left": "issue:45479", + "right": "issue:46894", + "accept": false, + "reason": "Single-label classification zero-loss bug and CohereASR training-loss double-shift bug are unrelated model-loss errors." + }, + { + "left": "issue:43856", + "right": "issue:44756", + "accept": false, + "reason": "Qwen3 MoE memory usage and Strix Halo mmap OOM are different memory-related problems." + }, + { + "left": "issue:43596", + "right": "issue:44683", + "accept": false, + "reason": "Deepspeed zero3/BertModel index error is unrelated to compiled flex_attention on torch>=2.9." + }, + { + "left": "issue:45479", + "right": "issue:46897", + "accept": false, + "reason": "Different training-loss bugs in different model families; not the same underlying defect." + }, + { + "left": "issue:44164", + "right": "issue:46821", + "accept": false, + "reason": "Handling extra_state in save/load is unrelated to a window_size AttributeError in TimesFM 2.5." + }, + { + "left": "issue:44488", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading failures, but for different models and likely different tokenizer/config issues." + }, + { + "left": "issue:44488", + "right": "issue:45081", + "accept": false, + "reason": "Different tokenizer/model load failures; no evidence they share the same root cause." + }, + { + "left": "issue:43611", + "right": "issue:43644", + "accept": false, + "reason": "base_model_prefix loading breakage and non-persistent buffer junk are separate Transformers v5 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": 23, + "review_unit_index": 10, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:32090", + "issue:34567", + "issue:36032", + "issue:39404", + "issue:40444", + "issue:40990", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43262", + "issue:43317", + "issue:43377", + "issue:43404", + "issue:43421", + "issue:43452", + "issue:43482", + "issue:43541", + "issue:43582", + "issue:43632", + "issue:43643", + "issue:43644", + "issue:43645", + "issue:43646", + "issue:43716", + "issue:43746", + "issue:43883", + "issue:44206", + "issue:44220", + "issue:44451", + "issue:44462", + "issue:44488", + "issue:44530", + "issue:44534", + "issue:44661", + "issue:44683", + "issue:44737", + "issue:44811", + "issue:44987", + "issue:44991", + "issue:45084", + "issue:45507", + "issue:45584", + "issue:45859", + "issue:46178", + "issue:46285", + "issue:46326", + "issue:46402", + "issue:46489" + ], + "soft_pairs": [ + "issue:44462|issue:46489", + "issue:42796|issue:43404", + "issue:46285|issue:46402", + "issue:36032|issue:42915", + "issue:32090|issue:39404", + "issue:43716|issue:45084", + "issue:44737|issue:45859", + "issue:44661|issue:44683", + "issue:43643|issue:44530", + "issue:44451|issue:44987", + "issue:40990|issue:43421", + "issue:42915|issue:44530", + "issue:43541|issue:45084", + "issue:43644|issue:43646", + "issue:43582|issue:43632", + "issue:34567|issue:40444", + "issue:42175|issue:46178", + "issue:43883|issue:44488", + "issue:40444|issue:46326", + "issue:44206|issue:44220", + "issue:44811|issue:45584", + "issue:43452|issue:43482", + "issue:36032|issue:43645", + "issue:44987|issue:44991", + "issue:43317|issue:43482", + "issue:45084|issue:45507", + "issue:43262|issue:43377", + "issue:43746|issue:46326", + "issue:43883|issue:44991", + "issue:43646|issue:44534" + ], + "prepared_review_unit_hash": "fe7f134db625d6a87dca485f09b5b689c309f53f627b47ed4db8a4a3307a3057", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "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 a loose bucket of Transformers v5 regression reports across many unrelated subsystems; the suggested soft pairs mostly share broad theme similarity but not the same concrete bug, so I reject all of them.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43646 is the broadest, most representative report here: a generic Transformers 5.0.0 custom-model initialization regression that fits the cluster\u2019s overall v5-breakage theme better than the more niche issues.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43646 is the best single anchor for this cluster because it is general enough to represent the repeated v5 initialization regressions, even though the cluster itself is too heterogeneous to treat as duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44462", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer-loading regressions, but one is AutoTokenizer ignoring tokenizer.json while the other is a DeepSeek-Coder v1 output bug tied to PR #44801; different failure modes." + }, + { + "left": "issue:42796", + "right": "issue:43404", + "accept": false, + "reason": "Ministral-3 tokenizer BPE-marker handling and Mistral3 lm_head weight tying are unrelated bugs in different code paths." + }, + { + "left": "issue:46285", + "right": "issue:46402", + "accept": false, + "reason": "CLIP text-model structure inconsistency and Sam3 text-encoder weight loading are different model-specific regressions." + }, + { + "left": "issue:36032", + "right": "issue:42915", + "accept": false, + "reason": "T5 tokenizer method/name conflict and Qwen3Moe FineGrainedFP8Config failure are unrelated issues." + }, + { + "left": "issue:32090", + "right": "issue:39404", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error and Whisper return_language pipeline regression are different bugs." + }, + { + "left": "issue:43716", + "right": "issue:45084", + "accept": false, + "reason": "Mistral-3 image preprocessor dtype mismatch and template-node compilation failure are unrelated." + }, + { + "left": "issue:44737", + "right": "issue:45859", + "accept": false, + "reason": "XLNet CPU arange device placement and Qwen3_5Moe rotary embedding CPU-offload incompatibility are distinct model/runtime problems." + }, + { + "left": "issue:44661", + "right": "issue:44683", + "accept": false, + "reason": "Tokenizer-mapping failure in add-new-model-like and compiled flex_attention failure on torch 2.9 are unrelated." + }, + { + "left": "issue:43643", + "right": "issue:44530", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss and PagedAttentionCache linear_attention crash are different layers and bugs." + }, + { + "left": "issue:44451", + "right": "issue:44987", + "accept": false, + "reason": "Both are model-loading failures, but they affect different repositories and appear to have different root causes." + }, + { + "left": "issue:40990", + "right": "issue:43421", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and runtime post-processor refresh for special tokens are not the same underlying bug." + }, + { + "left": "issue:42915", + "right": "issue:44530", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and PagedAttentionCache linear_attention crash are separate issues." + }, + { + "left": "issue:43541", + "right": "issue:45084", + "accept": false, + "reason": "Mixtral grouped_mm torch-dynamo tracing and non-template-node compilation errors are unrelated failures." + }, + { + "left": "issue:43644", + "right": "issue:43646", + "accept": false, + "reason": "Both are v5 regressions, but one is junk non-persistent buffers and the other is custom model initialization; same theme, different bug." + }, + { + "left": "issue:43582", + "right": "issue:43632", + "accept": false, + "reason": "AppleSilicon caching_allocator_warmup TypeError and the _is_hf_initialized flag regression are unrelated." + }, + { + "left": "issue:34567", + "right": "issue:40444", + "accept": false, + "reason": "TrainerState token counter not updating and Qwen2.5-VL multi-image iterable finetuning failure are different bugs." + }, + { + "left": "issue:42175", + "right": "issue:46178", + "accept": false, + "reason": "Missing tensorflow in a torch install and an unknown tensorflow_text backend mapping error are separate packaging/backend issues." + }, + { + "left": "issue:43883", + "right": "issue:44488", + "accept": false, + "reason": "Missing all_tied_weights_keys and loading cjvt/sleng-bert are different problems." + }, + { + "left": "issue:40444", + "right": "issue:46326", + "accept": false, + "reason": "Qwen2.5-VL iterable multi-image finetuning and mistral_common continue_final_message behavior are unrelated." + }, + { + "left": "issue:44206", + "right": "issue:44220", + "accept": false, + "reason": "Both are audio-related, but one is unsupported center arg in LasrFeatureExtractor and the other is _torch_extract_fbank_features; different API layers and failures." + }, + { + "left": "issue:44811", + "right": "issue:45584", + "accept": false, + "reason": "Whisper batch_decode skip_special_tokens handling and empty-transcription failure after align_special_tokens are related only by model area, not by the same bug." + }, + { + "left": "issue:43452", + "right": "issue:43482", + "accept": false, + "reason": "GGUF from_pretrained breakage and Qwen2.5-GGUF loading failure in v5 are both loading issues but not the same concrete defect." + }, + { + "left": "issue:36032", + "right": "issue:43645", + "accept": false, + "reason": "T5 tokenizer add_special_tokens conflict and notebook custom model initialization breakage are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:44991", + "accept": false, + "reason": "Loading physical-intelligence/fast and loading EMBEDDIA/est-roberta are separate repository-specific failures." + }, + { + "left": "issue:43317", + "right": "issue:43482", + "accept": false, + "reason": "Dequantized model device_map offload failure and Qwen2.5-GGUF loading failure are different loading paths and bugs." + }, + { + "left": "issue:45084", + "right": "issue:45507", + "accept": false, + "reason": "Non-template-node compilation error and GraniteMoEHybrid invalid method call are unrelated." + }, + { + "left": "issue:43262", + "right": "issue:43377", + "accept": false, + "reason": "Audio chat-template sampling-rate defaulting and MIMI batch padding-mask behavior are different issues." + }, + { + "left": "issue:43746", + "right": "issue:46326", + "accept": false, + "reason": "GraniteSpeech PEFT checkpoint loading and mistral_common continue_final_message are unrelated." + }, + { + "left": "issue:43883", + "right": "issue:44991", + "accept": false, + "reason": "Missing all_tied_weights_keys and tokenizer loading for EMBEDDIA/est-roberta are not the same bug." + }, + { + "left": "issue:43646", + "right": "issue:44534", + "accept": false, + "reason": "Both are Transformers v5 regressions, but they concern different internals: custom model init versus non-persistent buffer initialization." + } + ] + }, + "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": 23, + "review_unit_index": 11, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28530", + "issue:34567", + "issue:36032", + "issue:41720", + "issue:42915", + "issue:43089", + "issue:43296", + "issue:43322", + "issue:43377", + "issue:43404", + "issue:43425", + "issue:43531", + "issue:43540", + "issue:43611", + "issue:43632", + "issue:43645", + "issue:43742", + "issue:43792", + "issue:43819", + "issue:43994", + "issue:44008", + "issue:44062", + "issue:44220", + "issue:44265", + "issue:44291", + "issue:44403", + "issue:44451", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44534", + "issue:44589", + "issue:44610", + "issue:44683", + "issue:44849", + "issue:44877", + "issue:44987", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45216", + "issue:45325", + "issue:45357", + "issue:45987", + "issue:46315", + "issue:46821", + "issue:46869", + "issue:46899" + ], + "soft_pairs": [ + "issue:45084|issue:46869", + "issue:43792|issue:46869", + "issue:44403|issue:45081", + "issue:43645|issue:44534", + "issue:44849|issue:45325", + "issue:43377|issue:43819", + "issue:43404|issue:44466", + "issue:44464|issue:45987", + "issue:44987|issue:45081", + "issue:44008|issue:46899", + "issue:43994|issue:44610", + "issue:44451|issue:45081", + "issue:28530|issue:34567", + "issue:42915|issue:44291", + "issue:36032|issue:43742", + "issue:44062|issue:44987", + "issue:44265|issue:44610", + "issue:42915|issue:44589", + "issue:43089|issue:45987", + "issue:43322|issue:43540", + "issue:44479|issue:45325", + "issue:43632|issue:46821", + "issue:36032|issue:41720", + "issue:45216|issue:45357", + "issue:44991|issue:46315", + "issue:43531|issue:44877", + "issue:43425|issue:44403", + "issue:43296|issue:43531", + "issue:44220|issue:44683", + "issue:43611|issue:44534" + ], + "prepared_review_unit_hash": "eff9080efc3371a2363bd51c3798ef32f2311ae5cccbd5867d3ec4504a2ad979", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36746, + "estimated_input_tokens": 9187, + "estimated_eval_tokens": 18630 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose semantic group of unrelated Transformers regression reports (model loading, multimodal inputs, quantization, generation, tokenizer issues). Only one soft edge looks like a true duplicate: the Qwen3.5 `save_pretrained` regression pair.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45216 is the broader, earlier report about Qwen3.5 `save_pretrained` producing an incorrect checkpoint; issue:45357 reads like a narrower follow-up about one specific symptom (wrong visual encoder keys) of the same save-path bug.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45216 is the best representative because it describes the general regression, while issue:45357 is a more specific manifestation of that same checkpoint-saving problem.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45084", + "right": "issue:46869", + "accept": false, + "reason": "Different failures and code paths: template-node compilation vs YOLO multi-GPU auto device mapping." + }, + { + "left": "issue:43792", + "right": "issue:46869", + "accept": false, + "reason": "Unrelated model/runtime failures (Whisper inference vs YOLO device-map auto)." + }, + { + "left": "issue:44403", + "right": "issue:45081", + "accept": false, + "reason": "Generic loading noise report vs a specific Mistral tokenizer regex crash; not the same bug." + }, + { + "left": "issue:43645", + "right": "issue:44534", + "accept": false, + "reason": "Custom model init in notebooks vs non-persistent buffer serialization; different underlying issues." + }, + { + "left": "issue:44849", + "right": "issue:45325", + "accept": false, + "reason": "Qwen3.5 hidden-states bug vs Qwen2.5-VL temporal position-id scaling; related family, different bug." + }, + { + "left": "issue:43377", + "right": "issue:43819", + "accept": false, + "reason": "MIMI batching/padding-mask mismatch is unrelated to DAC latent/STE behavior." + }, + { + "left": "issue:43404", + "right": "issue:44466", + "accept": false, + "reason": "One is missing weight tying in a model class; the other is serialization inconsistency of tied weights. Same area, not the same bug." + }, + { + "left": "issue:44464", + "right": "issue:45987", + "accept": false, + "reason": "Chunked generation with compiled forward vs cache length return type; unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45081", + "accept": false, + "reason": "Different failure modes: loading a specific model vs tokenizer regex patching." + }, + { + "left": "issue:44008", + "right": "issue:46899", + "accept": false, + "reason": "Different Gemma models and different defects: variable-name collision vs unsafe dtype casting." + }, + { + "left": "issue:43994", + "right": "issue:44610", + "accept": false, + "reason": "Both are multimodal inference bugs, but one is SigLIP output quality and the other is processor/model input-size mismatch." + }, + { + "left": "issue:44451", + "right": "issue:45081", + "accept": false, + "reason": "Different model loading issue vs Mistral tokenizer patch crash." + }, + { + "left": "issue:28530", + "right": "issue:34567", + "accept": false, + "reason": "Early stopping metric selection is unrelated to token-count tracking." + }, + { + "left": "issue:42915", + "right": "issue:44291", + "accept": false, + "reason": "FP8 config failure vs `_is_hf_initialized` init-empty-weights incompatibility; different problems." + }, + { + "left": "issue:36032", + "right": "issue:43742", + "accept": false, + "reason": "Tokenizer method name conflict vs model key error on load; not the same bug." + }, + { + "left": "issue:44062", + "right": "issue:44987", + "accept": false, + "reason": "AddedToken constructor argument duplication is unrelated to model loading failure for a specific checkpoint." + }, + { + "left": "issue:44265", + "right": "issue:44610", + "accept": false, + "reason": "torch.export/torch_compilable_check failure is unrelated to OmDet processor image-size regression." + }, + { + "left": "issue:42915", + "right": "issue:44589", + "accept": false, + "reason": "Both involve FP8/quantization, but one is a model/config issue and the other is a storage deserialization issue." + }, + { + "left": "issue:43089", + "right": "issue:45987", + "accept": false, + "reason": "Generation performance overhead vs cache-length return type bug; different code paths." + }, + { + "left": "issue:43322", + "right": "issue:43540", + "accept": false, + "reason": "LLaVA model load segfault vs Qwen3OmniMoe video-input ValueError; both multimodal, not same underlying bug." + }, + { + "left": "issue:44479", + "right": "issue:45325", + "accept": false, + "reason": "Both concern Qwen vision/video inputs, but one is a broad video-input regression and the other is a specific still-image temporal-position bug; not clearly the same defect." + }, + { + "left": "issue:43632", + "right": "issue:46821", + "accept": false, + "reason": "`_is_hf_initialized` regression is unrelated to TimesFM `window_size` attribute handling." + }, + { + "left": "issue:36032", + "right": "issue:41720", + "accept": false, + "reason": "Tokenizer class conflict vs Qwen3 auto device-map CUDA assert; unrelated." + }, + { + "left": "issue:45216", + "right": "issue:45357", + "accept": true, + "reason": "Both report the same Qwen3.5 `save_pretrained` regression; 45357 is a more specific symptom of the broken checkpoint save path." + }, + { + "left": "issue:44991", + "right": "issue:46315", + "accept": false, + "reason": "Tokenizer loading failure vs slow `convert_tokens_to_ids` performance regression; not the same issue." + }, + { + "left": "issue:43531", + "right": "issue:44877", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior vs granite_speech strict-config loading; different bugs." + }, + { + "left": "issue:43425", + "right": "issue:44403", + "accept": false, + "reason": "Torch version incompatibility is unrelated to generic loading noise." + }, + { + "left": "issue:43296", + "right": "issue:43531", + "accept": false, + "reason": "PaddleOCR-VL load failure vs Qwen3-MoE sliding-window issue; different model families and failures." + }, + { + "left": "issue:44220", + "right": "issue:44683", + "accept": false, + "reason": "Audio feature extraction bug vs compiled flex_attention torch-version failure; unrelated." + }, + { + "left": "issue:43611", + "right": "issue:44534", + "accept": false, + "reason": "Base-model-prefix loading breakage vs non-persistent buffer junk values; different mechanisms." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative about the soft edges. The only accepted pair, issue:45216 vs issue:45357, is a plausible duplicate/follow-up on the same Qwen3.5 save_pretrained checkpoint bug. The rejected pairs are framed narrowly enough and do not overclaim equivalence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 12, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:33666", + "issue:35707", + "issue:40444", + "issue:40990", + "issue:41553", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:43296", + "issue:43317", + "issue:43377", + "issue:43381", + "issue:43404", + "issue:43421", + "issue:43482", + "issue:43525", + "issue:43531", + "issue:43582", + "issue:43632", + "issue:43716", + "issue:43746", + "issue:43792", + "issue:43856", + "issue:44079", + "issue:44291", + "issue:44403", + "issue:44462", + "issue:44509", + "issue:44534", + "issue:44568", + "issue:44589", + "issue:44756", + "issue:44811", + "issue:44991", + "issue:45081", + "issue:45325", + "issue:45491", + "issue:45538", + "issue:45542", + "issue:45663", + "issue:45812", + "issue:45923", + "issue:46491", + "issue:46710", + "issue:46821", + "issue:46899", + "issue:46909" + ], + "soft_pairs": [ + "issue:43716|issue:46899", + "issue:40444|issue:45325", + "issue:45812|issue:46491", + "issue:44403|issue:44991", + "issue:43381|issue:45923", + "issue:43632|issue:44756", + "issue:42796|issue:43746", + "issue:43792|issue:44589", + "issue:33666|issue:40444", + "issue:43404|issue:46899", + "issue:40990|issue:44811", + "issue:44291|issue:44589", + "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:43421|issue:44568", + "issue:43632|issue:44534", + "issue:43531|issue:44291", + "issue:42175|issue:45542", + "issue:44462|issue:45812", + "issue:41720|issue:45081", + "issue:43317|issue:43856", + "issue:44079|issue:44534", + "issue:44509|issue:46909", + "issue:43296|issue:46821", + "issue:43377|issue:45491" + ], + "prepared_review_unit_hash": "e7b9db8ac4eb059a164ceb5cec662fc6315bc7abdbf91ed52d0a74c7ea841209", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35906, + "estimated_input_tokens": 8977, + "estimated_eval_tokens": 18210 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab bag of mostly unrelated Transformers v5-era bugs: tokenizer regressions, multimodal/model-loading issues, config/API breakages, and a few training/runtime edge cases. The soft pairs only show surface similarity, not the same underlying defect, so none should be merged as duplicates.", + "confidence": 0.86, + "canonical_issue_reason": "Issue 45812 is the best representative anchor: it describes a broad, high-impact tokenizer regression affecting all Granite models, with inbound references and a clear v4\u2192v5 breakage pattern.", + "canonical_pr_reason": null, + "best_issue_reason": "45812 is the strongest global issue candidate because it is cross-model, user-visible, and regression-shaped; among these issues it is the most canonical-sounding and broadly impactful.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43716", + "right": "issue:46899", + "accept": false, + "reason": "Different bugs: Mistral-3 image preprocessor/model dtype mismatch vs Gemma4 4-bit quantization unsafe cast in multimodal embedders." + }, + { + "left": "issue:40444", + "right": "issue:45325", + "accept": false, + "reason": "One is multi-image finetuning with IterableDataset; the other is Qwen2.5-VL rope/index scaling for still-image positions." + }, + { + "left": "issue:45812", + "right": "issue:46491", + "accept": false, + "reason": "Granite-wide wrong token IDs regression and CodeLlama leading-space round-trip loss are distinct tokenizer bugs." + }, + { + "left": "issue:44403", + "right": "issue:44991", + "accept": false, + "reason": "Loading noise vs tokenizer loading failure for a specific model; same subsystem, different failure modes and fixes." + }, + { + "left": "issue:43381", + "right": "issue:45923", + "accept": false, + "reason": "Eval-mode gradient checkpointing restriction is not the same as a missing supports_gradient_checkpointing flag on a trust_remote_code model." + }, + { + "left": "issue:43632", + "right": "issue:44756", + "accept": false, + "reason": "_is_hf_initialized compatibility break and Strix Halo mmap OOM mitigation are unrelated problems." + }, + { + "left": "issue:42796", + "right": "issue:43746", + "accept": false, + "reason": "Tokenizer BPE-marker handling and PEFT local-checkpoint loading are separate code paths." + }, + { + "left": "issue:43792", + "right": "issue:44589", + "accept": false, + "reason": "Whisper runtime failure and Float8 storage lookup TypeError do not share the same root cause." + }, + { + "left": "issue:33666", + "right": "issue:40444", + "accept": false, + "reason": "Multi-GPU training support and iterable-dataset multi-image failures are different training issues." + }, + { + "left": "issue:43404", + "right": "issue:46899", + "accept": false, + "reason": "Mistral3 lm_head weight tying and Gemma4 quantized multimodal casting are unrelated model bugs." + }, + { + "left": "issue:40990", + "right": "issue:44811", + "accept": false, + "reason": "High perplexity on gpt-oss-20b and Whisper batch_decode skip_special_tokens behavior are unrelated." + }, + { + "left": "issue:44291", + "right": "issue:44589", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized argument breakage is distinct from Float8 storage resolution." + }, + { + "left": "issue:42796", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer BPE marker handling and DeepSeek accuracy regression are not the same underlying issue." + }, + { + "left": "issue:43482", + "right": "issue:43525", + "accept": false, + "reason": "GGUF loading failure in transformers v5 and missing pad_token_id on Llama4Config are separate compatibility bugs." + }, + { + "left": "issue:35707", + "right": "issue:40444", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen2.5-VL multi-image IterableDataset training failures." + }, + { + "left": "issue:44079", + "right": "issue:45663", + "accept": false, + "reason": "ModelOutput key assignment and Gemma4 FSDP shared_kv_states rebuilding are different implementation bugs." + }, + { + "left": "issue:41553", + "right": "issue:43582", + "accept": false, + "reason": "Bad tokenizer error for Voxtral and Apple Silicon caching warmup TypeError are unrelated." + }, + { + "left": "issue:43482", + "right": "issue:45538", + "accept": false, + "reason": "GGUF loading failure and CLIPTokenizer max_length misconfiguration are different issues." + }, + { + "left": "issue:43421", + "right": "issue:44568", + "accept": false, + "reason": "Runtime post-processor updates for special tokens and add_special_tokens BOS/EOS regression are separate tokenizer behaviors." + }, + { + "left": "issue:43632", + "right": "issue:44534", + "accept": false, + "reason": "_is_hf_initialized flag breakage and non-persistent buffer junk are different Transformers v5 regressions." + }, + { + "left": "issue:43531", + "right": "issue:44291", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and init_empty_weights argument incompatibility do not match." + }, + { + "left": "issue:42175", + "right": "issue:45542", + "accept": false, + "reason": "Missing TensorFlow in torch-only install and undefined tf backend from tensorboard-only install are related only at a broad packaging level, not the same bug." + }, + { + "left": "issue:44462", + "right": "issue:45812", + "accept": false, + "reason": "Ignoring tokenizer.json and wrong token IDs for Granite models are different tokenizer-loading regressions." + }, + { + "left": "issue:41720", + "right": "issue:45081", + "accept": false, + "reason": "Auto device mapping cudaErrorAssert and Mistral regex patch backend_tokenizer AttributeError are unrelated failures." + }, + { + "left": "issue:43317", + "right": "issue:43856", + "accept": false, + "reason": "Dequantized model offload loading failure and Qwen3 MoE memory inefficiency are different performance/load issues." + }, + { + "left": "issue:44079", + "right": "issue:44534", + "accept": false, + "reason": "ModelOutput key assignment and non-persistent buffer initialization are separate internal state bugs." + }, + { + "left": "issue:44509", + "right": "issue:46909", + "accept": false, + "reason": "Removed pipeline docs references and a broken DIA docs example are both documentation issues but not the same artifact." + }, + { + "left": "issue:43296", + "right": "issue:46821", + "accept": false, + "reason": "PaddleOCR-VL load failure in vLLM and TimesFM window_size AttributeError are distinct model/runtime bugs." + }, + { + "left": "issue:43377", + "right": "issue:45491", + "accept": false, + "reason": "Missing padding-mask support in MIMI encoder and Gemma3 NaN embeddings from sliding-window all-padding windows are different root causes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 13, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:34634", + "issue:35707", + "issue:39692", + "issue:40444", + "issue:40990", + "issue:41720", + "issue:42175", + "issue:42796", + "issue:42915", + "issue:43262", + "issue:43296", + "issue:43454", + "issue:43482", + "issue:43493", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43644", + "issue:43646", + "issue:43701", + "issue:43746", + "issue:43818", + "issue:43819", + "issue:43828", + "issue:43873", + "issue:43881", + "issue:43994", + "issue:44079", + "issue:44186", + "issue:44265", + "issue:44291", + "issue:44466", + "issue:44479", + "issue:44488", + "issue:44829", + "issue:44877", + "issue:45072", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45584", + "issue:45701", + "issue:45823", + "issue:45834", + "issue:46402", + "issue:46821" + ], + "soft_pairs": [ + "issue:43611|issue:43646", + "issue:43818|issue:46402", + "issue:44829|issue:45479", + "issue:43644|issue:44079", + "issue:42175|issue:43296", + "issue:43818|issue:45357", + "issue:44265|issue:45072", + "issue:43493|issue:43994", + "issue:40444|issue:45381", + "issue:45325|issue:45381", + "issue:45325|issue:45834", + "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:40990|issue:45584", + "issue:43262|issue:43746", + "issue:44479|issue:45823", + "issue:43881|issue:44488", + "issue:35707|issue:41720", + "issue:43828|issue:46821", + "issue:34634|issue:35707", + "issue:39692|issue:40444", + "issue:43454|issue:43550", + "issue:43606|issue:43819", + "issue:44186|issue:44265", + "issue:41720|issue:45823" + ], + "prepared_review_unit_hash": "58ffa82905bf794f62ac6f3e620bef82027f29e1ba24c880f88e4b2d6cf63884", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36224, + "estimated_input_tokens": 9056, + "estimated_eval_tokens": 18368 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly heterogeneous: the soft pairs share broad model families or regression themes, but not the same concrete bug/change. I would not merge any of the suggested links as duplicates.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45325 is the most concrete, actionable report in the set and the closest anchor to one of the few semantically adjacent Qwen2.5-VL position-id reports, even though the overall cluster is too mixed to treat it as a true duplicate hub.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45325 is the best issue to anchor triage because it names a specific helper-path regression with a clear symptom and is narrowly scoped compared with the broader, less directly comparable reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43611", + "right": "issue:43646", + "accept": false, + "reason": "Both are Transformers 5.0 regressions, but one is loading with base_model_prefix and the other is custom initialization; different concrete failure paths." + }, + { + "left": "issue:43818", + "right": "issue:46402", + "accept": false, + "reason": "Completely different models and bugs: Video-LLaVA temporal attention/weights vs Sam3Model text encoder loading." + }, + { + "left": "issue:44829", + "right": "issue:45479", + "accept": false, + "reason": "Unrelated symptoms: flash_attention_3 training regression vs num_labels=1 single-label loss degeneracy." + }, + { + "left": "issue:43644", + "right": "issue:44079", + "accept": false, + "reason": "One is a 5.0 buffer serialization/loading issue; the other is ModelOutput key assignment behavior. Same framework, different bug." + }, + { + "left": "issue:42175", + "right": "issue:43296", + "accept": false, + "reason": "TensorFlow extra/install packaging problem vs PaddleOCR-VL model load failure in vLLM; no shared code path." + }, + { + "left": "issue:43818", + "right": "issue:45357", + "accept": false, + "reason": "Both touch multimodal model internals, but one is Video-LLaVA temporal attention and the other is Qwen3.5 visual encoder save_pretrained keys." + }, + { + "left": "issue:44265", + "right": "issue:45072", + "accept": false, + "reason": "torch.export with torch_compilable_check is unrelated to dtype mismatches in bfloat16 inference." + }, + { + "left": "issue:43493", + "right": "issue:43994", + "accept": false, + "reason": "Both mention SigLIP2, but one is an implementation-vs-JAX discrepancy and the other is bad AutoModel/pipeline outputs; not clearly the same defect." + }, + { + "left": "issue:40444", + "right": "issue:45381", + "accept": false, + "reason": "Qwen2.5-VL multi-image finetuning failure vs Qwen2.5-VL video position-id regression; related family, different concrete bug." + }, + { + "left": "issue:45325", + "right": "issue:45381", + "accept": false, + "reason": "Both are Qwen2.5-VL position-id/rope issues, but one is still-image temporal scaling and the other is video vision_position_ids; not enough evidence they are the same bug." + }, + { + "left": "issue:45325", + "right": "issue:45834", + "accept": false, + "reason": "Different problems in the same model family: rope/position-index logic vs OCR long-input index error." + }, + { + "left": "issue:43454", + "right": "issue:44466", + "accept": false, + "reason": "One is missing lm_head weight tying causing bad generation; the other is inconsistent serialization depending on device. Related area, different bug." + }, + { + "left": "issue:42915", + "right": "issue:43482", + "accept": false, + "reason": "Different models and failure modes: FineGrainedFP8Config on Qwen3Moe vs GGUF loading on Qwen2.5." + }, + { + "left": "issue:44291", + "right": "issue:44877", + "accept": false, + "reason": "init_empty_weights/_is_hf_initialized TypeError is unrelated to strict config loading for granite_speech." + }, + { + "left": "issue:43701", + "right": "issue:45325", + "accept": false, + "reason": "resume_from_checkpoint key mismatch is a training checkpoint issue, not a Qwen2.5-VL position-id regression." + }, + { + "left": "issue:43873", + "right": "issue:45701", + "accept": false, + "reason": "Quantization/offloading behavior vs tokenization changes across versions; no concrete shared bug." + }, + { + "left": "issue:42796", + "right": "issue:45381", + "accept": false, + "reason": "Tokenizer BPE marker handling is unrelated to Qwen2.5-VL video vision_position_ids." + }, + { + "left": "issue:43606", + "right": "issue:44265", + "accept": false, + "reason": "CPU offload device mismatch in bark-small is unrelated to torch.export failures with torch_compilable_check." + }, + { + "left": "issue:40990", + "right": "issue:45584", + "accept": false, + "reason": "High perplexity on gpt-oss-20b is unrelated to Whisper empty-transcription generation failure." + }, + { + "left": "issue:43262", + "right": "issue:43746", + "accept": false, + "reason": "Audio processor chat-template sample-rate defaulting is unrelated to PEFT/local checkpoint loading." + }, + { + "left": "issue:44479", + "right": "issue:45823", + "accept": false, + "reason": "Qwen video-input regression and Gemma4 PLE device_map mismatch are different model-specific forward-path bugs." + }, + { + "left": "issue:43881", + "right": "issue:44488", + "accept": false, + "reason": "Different model-loading failures for unrelated models; no evidence of a shared root cause." + }, + { + "left": "issue:35707", + "right": "issue:41720", + "accept": false, + "reason": "Progressive generation with inputs_embeds/past_key_values is unrelated to Qwen3 auto device-mapping cudaErrorAssert." + }, + { + "left": "issue:43828", + "right": "issue:46821", + "accept": false, + "reason": "Autocast dtype mismatch in Phi-tiny-MoE is unrelated to TimesFM window_size AttributeError." + }, + { + "left": "issue:34634", + "right": "issue:35707", + "accept": false, + "reason": "BarkProcessor voice_preset and progressive generation with embeddings/past KV are unrelated bugs." + }, + { + "left": "issue:39692", + "right": "issue:40444", + "accept": false, + "reason": "SigLIP2 doc-example errors are unrelated to Qwen2.5-VL multi-image finetuning failures." + }, + { + "left": "issue:43454", + "right": "issue:43550", + "accept": false, + "reason": "lm_head weight tying/serialization issues are unrelated to Bamba-9B-v2 torch.compile+SDPA failures." + }, + { + "left": "issue:43606", + "right": "issue:43819", + "accept": false, + "reason": "CPU offload device mismatch and DAC.from_latents/STE mismatch are different code paths and symptoms." + }, + { + "left": "issue:44186", + "right": "issue:44265", + "accept": false, + "reason": "Tokenizer padding/truncation crashes are unrelated to torch.export with torch_compilable_check." + }, + { + "left": "issue:41720", + "right": "issue:45823", + "accept": false, + "reason": "Qwen3 auto device-map assert and Gemma4 PLE device mismatch are separate forward/device-placement 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": 23, + "review_unit_index": 14, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:26413", + "issue:28530", + "issue:36331", + "issue:40990", + "issue:41720", + "issue:42617", + "issue:42796", + "issue:42915", + "issue:42947", + "issue:43065", + "issue:43257", + "issue:43317", + "issue:43322", + "issue:43377", + "issue:43450", + "issue:43454", + "issue:43526", + "issue:43572", + "issue:43720", + "issue:43749", + "issue:43873", + "issue:43881", + "issue:43883", + "issue:44186", + "issue:44387", + "issue:44466", + "issue:44534", + "issue:44568", + "issue:44610", + "issue:44756", + "issue:44877", + "issue:44898", + "issue:44977", + "issue:44991", + "issue:45072", + "issue:45357", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45823", + "issue:45920", + "issue:45923", + "issue:46050", + "issue:46209", + "issue:46305", + "issue:46315", + "issue:46402", + "issue:46821" + ], + "soft_pairs": [ + "issue:43377|issue:44610", + "issue:46305|issue:46315", + "issue:43257|issue:45923", + "issue:44568|issue:45920", + "issue:42796|issue:46209", + "issue:43454|issue:45072", + "issue:45823|issue:46821", + "issue:43377|issue:44898", + "issue:43454|issue:43720", + "issue:40990|issue:42947", + "issue:42617|issue:42915", + "issue:43749|issue:44534", + "issue:43317|issue:43873", + "issue:43572|issue:45759", + "issue:43526|issue:43883", + "issue:43322|issue:46050", + "issue:41720|issue:43322", + "issue:43881|issue:44991", + "issue:36331|issue:42617", + "issue:44387|issue:45704", + "issue:41720|issue:44977", + "issue:36331|issue:42915", + "issue:44466|issue:45357", + "issue:43450|issue:45701", + "issue:44877|issue:44991", + "issue:43065|issue:46402", + "issue:40990|issue:45701", + "issue:44186|issue:45072", + "issue:26413|issue:28530", + "issue:41720|issue:42617", + "issue:43873|issue:44756" + ], + "prepared_review_unit_hash": "541e3372519f8cc48f4ed5e75a599af9febdd5213a6d29fa6f3af20ea1ecf38e", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36984, + "estimated_input_tokens": 9246, + "estimated_eval_tokens": 18748 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the issues span unrelated regressions in tokenizers, model loading, quantization, offloading, training, and model-specific bugs. None of the soft pairs look like the same underlying bug/change, so all are rejected.", + "confidence": 0.96, + "canonical_issue_reason": "No strong canonical issue: the cluster does not form a single duplicate family, and the items cover many different subsystems and failure modes.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43873 is the most broadly framed and active representative among the set, but only as a loose cluster anchor\u2014not as a true duplicate target.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43377", + "right": "issue:44610", + "accept": false, + "reason": "Both mention batched-input behavior, but they affect different models and different code paths: MIMI encoder padding masks vs OmDet-Turbo processor image sizing." + }, + { + "left": "issue:46305", + "right": "issue:46315", + "accept": false, + "reason": "Different regressions: a sliding-window mask off-by-one in streaming audio vs a tokenizer performance regression in convert_tokens_to_ids." + }, + { + "left": "issue:43257", + "right": "issue:45923", + "accept": false, + "reason": "Unrelated bugs: Qwen3 MoE weight conversion during accelerate/deepspeed loading vs a missing gradient-checkpointing flag on a trust_remote_code variant." + }, + { + "left": "issue:44568", + "right": "issue:45920", + "accept": false, + "reason": "Both are tokenizer regressions, but one is a specific BOS/EOS special-token bug while the other is wrong token IDs across several models." + }, + { + "left": "issue:42796", + "right": "issue:46209", + "accept": false, + "reason": "Completely different failures: tokenizer BPE-marker handling vs FP8 inference failing on B200 hardware." + }, + { + "left": "issue:43454", + "right": "issue:45072", + "accept": false, + "reason": "Different model paths and symptoms: tied lm_head weights for AyaVision vs dtype mismatches in SwitchTransformers/TimmWrapperModel bf16 inference." + }, + { + "left": "issue:45823", + "right": "issue:46821", + "accept": false, + "reason": "Unrelated issues: Gemma4 device placement during forward vs a TimesFM window_size AttributeError." + }, + { + "left": "issue:43377", + "right": "issue:44898", + "accept": false, + "reason": "Both are batching/resolution-related, but they affect different vision models and different bugs: missing padding-mask support vs Perceiver resolution handling." + }, + { + "left": "issue:43454", + "right": "issue:43720", + "accept": false, + "reason": "Different loading bugs in different model families: weight tying vs packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:40990", + "right": "issue:42947", + "accept": false, + "reason": "Different areas entirely: perplexity on a specific model/dataset vs gradient checkpointing behavior with PEFT LoRA." + }, + { + "left": "issue:42617", + "right": "issue:42915", + "accept": false, + "reason": "No shared underlying defect: a 3d_parallel.py execution problem vs a Qwen3Moe FineGrainedFP8Config failure." + }, + { + "left": "issue:43749", + "right": "issue:44534", + "accept": false, + "reason": "Not the same bug: FSDP CPU RAM-efficient loading breakage vs non-persistent buffers being filled with junk in v5." + }, + { + "left": "issue:43317", + "right": "issue:43873", + "accept": false, + "reason": "Both involve offloading/quantization, but one is a device_map load failure for dequantized models and the other is a broader quantization offloading issue." + }, + { + "left": "issue:43572", + "right": "issue:45759", + "accept": false, + "reason": "Different config/model-resolution problems: missing pad_token_idx in StableLmConfig vs failing to unwrap text_config for composite Qwen multimodal configs." + }, + { + "left": "issue:43526", + "right": "issue:43883", + "accept": false, + "reason": "Different bugs: BeitImageProcessorFast reduce_labels returning one label vs Molmo missing all_tied_weights_keys during loading." + }, + { + "left": "issue:43322", + "right": "issue:46050", + "accept": false, + "reason": "Different model/load paths: Llava Next segfault on load vs sinq quantized models breaking after save_pretrained/from_pretrained." + }, + { + "left": "issue:41720", + "right": "issue:43322", + "accept": false, + "reason": "Both are model loading failures, but one is Qwen3 auto device mapping cuda assert and the other is Llava Next segfault; no concrete shared code-path." + }, + { + "left": "issue:43881", + "right": "issue:44991", + "accept": false, + "reason": "Different components: a glm-4v-9b loading failure vs tokenizer loading regression for EMBEDDIA/est-roberta." + }, + { + "left": "issue:36331", + "right": "issue:42617", + "accept": false, + "reason": "Unrelated: CustomTrainer compute_loss signature change vs failure to run 3d_parallel.py." + }, + { + "left": "issue:44387", + "right": "issue:45704", + "accept": false, + "reason": "Different performance/memory issues in different subsystems: int4 quantization OOM vs an apex FusedRMSNorm memory leak." + }, + { + "left": "issue:41720", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen, but the failures differ: device-mapping cuda assert vs flash-attention generation malfunction." + }, + { + "left": "issue:36331", + "right": "issue:42915", + "accept": false, + "reason": "No relation: trainer compute_loss API break vs Qwen3Moe FP8 configuration failure." + }, + { + "left": "issue:44466", + "right": "issue:45357", + "accept": false, + "reason": "Both involve save_pretrained regressions, but they affect different serialized artifacts and different model families." + }, + { + "left": "issue:43450", + "right": "issue:45701", + "accept": false, + "reason": "Different areas: batched video processor output shape vs a tokenization behavior change across versions." + }, + { + "left": "issue:44877", + "right": "issue:44991", + "accept": false, + "reason": "Completely different bugs: strict config blocking granite_speech loading vs tokenizer loading failure for est-roberta." + }, + { + "left": "issue:43065", + "right": "issue:46402", + "accept": false, + "reason": "Same broad family (Sam3) but different failures and code paths: dummy Conv2d in pixel decoder vs text encoder weights not loading." + }, + { + "left": "issue:40990", + "right": "issue:45701", + "accept": false, + "reason": "Different regression classes: perplexity evaluation anomaly vs tokenization output changes between versions." + }, + { + "left": "issue:44186", + "right": "issue:45072", + "accept": false, + "reason": "Unrelated: LayoutLMv2Tokenizer crashes with NER/batched padding vs bf16 dtype mismatches in model inference." + }, + { + "left": "issue:26413", + "right": "issue:28530", + "accept": false, + "reason": "Different training loop issues: resume_from_checkpoint epoch iterator failure vs early stopping/metric_for_best_model disabling." + }, + { + "left": "issue:41720", + "right": "issue:42617", + "accept": false, + "reason": "Different failures in different workflows: auto device mapping cuda assert vs inability to run 3d_parallel.py." + }, + { + "left": "issue:43873", + "right": "issue:44756", + "accept": false, + "reason": "Not the same bug: quantization/offloading behavior vs mmap being disabled on Strix Halo to avoid OOM." + } + ] + }, + "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": 23, + "review_unit_index": 15, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28530", + "issue:33666", + "issue:36010", + "issue:36331", + "issue:41720", + "issue:42617", + "issue:42915", + "issue:43295", + "issue:43316", + "issue:43317", + "issue:43322", + "issue:43404", + "issue:43425", + "issue:43526", + "issue:43531", + "issue:43550", + "issue:43572", + "issue:43577", + "issue:43606", + "issue:43646", + "issue:43708", + "issue:43720", + "issue:43881", + "issue:44008", + "issue:44186", + "issue:44361", + "issue:44423", + "issue:44464", + "issue:44610", + "issue:44671", + "issue:44734", + "issue:44743", + "issue:44898", + "issue:44913", + "issue:45161", + "issue:45357", + "issue:45406", + "issue:45632", + "issue:45684", + "issue:45685", + "issue:45701", + "issue:45823", + "issue:46291", + "issue:46612", + "issue:46665", + "issue:46821", + "issue:46858", + "issue:46869" + ], + "soft_pairs": [ + "issue:43550|issue:43720", + "issue:46612|issue:46858", + "issue:43577|issue:43708", + "issue:43295|issue:45701", + "issue:45685|issue:46821", + "issue:43550|issue:43606", + "issue:44186|issue:44610", + "issue:43646|issue:46665", + "issue:43404|issue:45357", + "issue:44734|issue:46858", + "issue:44610|issue:44898", + "issue:42915|issue:43317", + "issue:36010|issue:42617", + "issue:44423|issue:45406", + "issue:44008|issue:45823", + "issue:46821|issue:46869", + "issue:45632|issue:45684", + "issue:44464|issue:46858", + "issue:41720|issue:45161", + "issue:28530|issue:33666", + "issue:44671|issue:44743", + "issue:43531|issue:43572", + "issue:42915|issue:46291", + "issue:36331|issue:41720", + "issue:44186|issue:44361", + "issue:43526|issue:43881", + "issue:43316|issue:44913", + "issue:43425|issue:46821", + "issue:42915|issue:43322" + ], + "prepared_review_unit_hash": "0f49742dd20ee0fa60a3c50c051a8e4a133cd0ad8af820cbc79c4479ff2ce332", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36487, + "estimated_input_tokens": 9122, + "estimated_eval_tokens": 18500 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is not a true duplicate cluster: the items span unrelated training, loading, tokenizer, processor, and generation bugs. All soft edges are false positives. If a single issue must anchor the set, 46858 is the most concrete low-level generation/cache bug, but it does not represent the rest well.", + "confidence": 0.97, + "canonical_issue_reason": "46858 is the clearest, most actionable low-level bug report in the set: it has a specific cache/arithmetic failure surface and a concrete generation code path. It is still only a weak anchor because the other issues are largely unrelated.", + "canonical_pr_reason": null, + "best_issue_reason": "46858 has the strongest problem statement and most explicit fix surface among these otherwise heterogeneous reports, so it is the best representative issue only in a very loose sense.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:43720", + "accept": false, + "reason": "Different model families and bugs: torch.compile/SDPA vs packed-weight loading." + }, + { + "left": "issue:46612", + "right": "issue:46858", + "accept": false, + "reason": "Both are cache/generation issues, but one is missing cache reorder support and the other is a seq-length/device-tensor arithmetic bug." + }, + { + "left": "issue:43577", + "right": "issue:43708", + "accept": false, + "reason": "Unrelated: dtype propagation on load vs resume_from_checkpoint max_steps." + }, + { + "left": "issue:43295", + "right": "issue:45701", + "accept": false, + "reason": "Both touch tokenization, but one is a regression in processor.tokenizer/images handling and the other is a version-dependent tokenization change." + }, + { + "left": "issue:45685", + "right": "issue:46821", + "accept": false, + "reason": "Different subsystems: MPS histogram on Int vs TimesFM window_size AttributeError." + }, + { + "left": "issue:43550", + "right": "issue:43606", + "accept": false, + "reason": "Both are device/compile-related, but the failure modes and code paths differ." + }, + { + "left": "issue:44186", + "right": "issue:44610", + "accept": false, + "reason": "Both are preprocessing bugs, but one is tokenizer/NER padding and the other is image-size mismatch." + }, + { + "left": "issue:43646", + "right": "issue:46665", + "accept": false, + "reason": "Both involve model loading, but one is custom initialization breaking in v5 and the other is from_pretrained semantics in Python classes." + }, + { + "left": "issue:43404", + "right": "issue:45357", + "accept": false, + "reason": "Different models and distinct state-management bugs: weight tying vs save_pretrained key mapping." + }, + { + "left": "issue:44734", + "right": "issue:46858", + "accept": false, + "reason": "Both mention cache continuation, but one is wrong tensor indexing in serve and the other is static-cache slice arithmetic on non-CUDA backends." + }, + { + "left": "issue:44610", + "right": "issue:44898", + "accept": false, + "reason": "Both are image input-size issues, but model/processor expectations and failures are unrelated." + }, + { + "left": "issue:42915", + "right": "issue:43317", + "accept": false, + "reason": "Different loading failures: FineGrainedFP8Config vs dequantized device_map offload." + }, + { + "left": "issue:36010", + "right": "issue:42617", + "accept": false, + "reason": "Completely unrelated: import error vs 3d_parallel.py runtime failure." + }, + { + "left": "issue:44423", + "right": "issue:45406", + "accept": false, + "reason": "Both are serve/multimodal crashes, but different processors and attribute failures." + }, + { + "left": "issue:44008", + "right": "issue:45823", + "accept": false, + "reason": "Gemma-family only by name; one is a variable collision, the other is a device-mapping mismatch." + }, + { + "left": "issue:46821", + "right": "issue:46869", + "accept": false, + "reason": "Unrelated: TimesFM config attribute vs YOLO multi-GPU device mapping." + }, + { + "left": "issue:45632", + "right": "issue:45684", + "accept": false, + "reason": "Both involve save/load paths, but one is cache-path collision and the other is file permission propagation." + }, + { + "left": "issue:44464", + "right": "issue:46858", + "accept": false, + "reason": "Both involve compiled generation, but one is inconsistent outputs and the other is a seq-length tensor arithmetic bug." + }, + { + "left": "issue:41720", + "right": "issue:45161", + "accept": false, + "reason": "Different distributed/device-map failures on different model families." + }, + { + "left": "issue:28530", + "right": "issue:33666", + "accept": false, + "reason": "Unrelated training UX issue vs Qwen2-VL multi-GPU training request." + }, + { + "left": "issue:44671", + "right": "issue:44743", + "accept": false, + "reason": "Different models and code paths; masked LM predictions vs recurrent state reset under cache." + }, + { + "left": "issue:43531", + "right": "issue:43572", + "accept": false, + "reason": "Unrelated model-specific sliding_window bug vs config default missing after 5.0 update." + }, + { + "left": "issue:42915", + "right": "issue:46291", + "accept": false, + "reason": "Different failures entirely: FP8 model loading vs pipelines import dependency check." + }, + { + "left": "issue:36331", + "right": "issue:41720", + "accept": false, + "reason": "Unrelated trainer API break vs CUDA assert during device mapping." + }, + { + "left": "issue:44186", + "right": "issue:44361", + "accept": false, + "reason": "Both are tokenizer/task AttributeErrors, but they affect different tokenizers and inputs." + }, + { + "left": "issue:43526", + "right": "issue:43881", + "accept": false, + "reason": "Unrelated: image processor reduce_labels bug vs glm-4v-9b loading failure." + }, + { + "left": "issue:43316", + "right": "issue:44913", + "accept": false, + "reason": "Both are config consistency issues, but one is Gemma3TextConfig API disparity and the other is GPTNeoX rotary_pct reload." + }, + { + "left": "issue:43425", + "right": "issue:46821", + "accept": false, + "reason": "Unrelated Torch version compatibility vs TimesFM window_size AttributeError." + }, + { + "left": "issue:42915", + "right": "issue:43322", + "accept": false, + "reason": "Different bug classes: fp8 config load failure vs Llava Next segmentation fault." + } + ] + }, + "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": 23, + "review_unit_index": 16, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28530", + "issue:29159", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34634", + "issue:35707", + "issue:36032", + "issue:41720", + "issue:42490", + "issue:43329", + "issue:43388", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43606", + "issue:43644", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43873", + "issue:44062", + "issue:44077", + "issue:44387", + "issue:44451", + "issue:44492", + "issue:44509", + "issue:44534", + "issue:44589", + "issue:44625", + "issue:44898", + "issue:44987", + "issue:44991", + "issue:45042", + "issue:45059", + "issue:45072", + "issue:45357", + "issue:45406", + "issue:45464", + "issue:45701", + "issue:45704", + "issue:45759", + "issue:45859", + "issue:46402", + "issue:46665", + "issue:46821", + "issue:46858" + ], + "soft_pairs": [ + "issue:28530|issue:30064", + "issue:33357|issue:35707", + "issue:29159|issue:36032", + "issue:43540|issue:43782", + "issue:43550|issue:43653", + "issue:44387|issue:45701", + "issue:43873|issue:45704", + "issue:30064|issue:34634", + "issue:44492|issue:44509", + "issue:43606|issue:43720", + "issue:44534|issue:46665", + "issue:43531|issue:45759", + "issue:41720|issue:43746", + "issue:42490|issue:43329", + "issue:43388|issue:45059", + "issue:45406|issue:45464", + "issue:43606|issue:43653", + "issue:34567|issue:35707", + "issue:43531|issue:46821", + "issue:34634|issue:41720", + "issue:43540|issue:46858", + "issue:44451|issue:44991", + "issue:45759|issue:45859", + "issue:45357|issue:46402", + "issue:43531|issue:44589", + "issue:44077|issue:44625", + "issue:43644|issue:46665", + "issue:44062|issue:44451", + "issue:44898|issue:45072", + "issue:44987|issue:45042" + ], + "prepared_review_unit_hash": "ced30994b2e1895074275556351f15d8a97e6757c639817b25d946a9290b52c6", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36794, + "estimated_input_tokens": 9199, + "estimated_eval_tokens": 18654 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Loose cluster of mostly unrelated issue reports spanning tokenization, model loading, multimodal processors, offload/quantization, and training-state bugs; none of the soft pairs look like the same underlying defect.", + "confidence": 0.78, + "canonical_issue_reason": "issue:45701 is the broadest umbrella report in this otherwise heterogeneous set and best represents the cluster\u2019s recurring version-regression/tokenization-loading theme, even though it is not a true duplicate hub.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45701 is the most representative issue artifact because it is broad and central rather than model-specific, making it the closest thing to a canonical issue for this mixed cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:28530", + "right": "issue:30064", + "accept": false, + "reason": "Early-stopping metric selection and void segmentation-map processing are unrelated bugs." + }, + { + "left": "issue:33357", + "right": "issue:35707", + "accept": false, + "reason": "MacOS CLIP bus error and progressive generation with inputs_embeds/past_key_values affect different code paths." + }, + { + "left": "issue:29159", + "right": "issue:36032", + "accept": false, + "reason": "Both are tokenizer-related, but one is fast/slow behavioral inconsistency and the other is a T5 special-method load error." + }, + { + "left": "issue:43540", + "right": "issue:43782", + "accept": false, + "reason": "Both touch Qwen multimodal loading, but video input validation and weight_only loading failure are distinct problems." + }, + { + "left": "issue:43550", + "right": "issue:43653", + "accept": false, + "reason": "torch.compile/SDPA failure in Bamba is unrelated to BigBirdTokenizer special-token registration." + }, + { + "left": "issue:44387", + "right": "issue:45701", + "accept": false, + "reason": "int4 memory growth under quantization is a runtime/memory bug, not the tokenization change reported in 45701." + }, + { + "left": "issue:43873", + "right": "issue:45704", + "accept": false, + "reason": "Quantization offloading and a T5 RMSNorm memory leak are different failure modes." + }, + { + "left": "issue:30064", + "right": "issue:34634", + "accept": false, + "reason": "Image processor segmentation-map handling and Bark voice_preset loading are unrelated." + }, + { + "left": "issue:44492", + "right": "issue:44509", + "accept": false, + "reason": "A cache-strategy typo and removed pipeline-task docs are both docs-ish, but not the same change or bug." + }, + { + "left": "issue:43606", + "right": "issue:43720", + "accept": false, + "reason": "CPU offload device mismatch in Bark is separate from BitNet packed-weight unpacking during accelerate loading." + }, + { + "left": "issue:44534", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer initialization and from_pretrained behavior inside Python classes are different loading issues." + }, + { + "left": "issue:43531", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and composite Qwen text_config unwrapping are different config bugs." + }, + { + "left": "issue:41720", + "right": "issue:43746", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to GraniteSpeech PEFT checkpoint loading." + }, + { + "left": "issue:42490", + "right": "issue:43329", + "accept": false, + "reason": "Trailing-slash local model path warnings and undefined multimodal token counting helpers are unrelated." + }, + { + "left": "issue:43388", + "right": "issue:45059", + "accept": false, + "reason": "Batch gathering label truncation and SAM3 text/box behavior are different evaluation vs inference bugs." + }, + { + "left": "issue:45406", + "right": "issue:45464", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Qwen3.5 streaming chat/completions failures are separate service/runtime issues." + }, + { + "left": "issue:43606", + "right": "issue:43653", + "accept": false, + "reason": "Offload device mismatch in a model is not the same as tokenizer special-token registration." + }, + { + "left": "issue:34567", + "right": "issue:35707", + "accept": false, + "reason": "Trainer token-count bookkeeping and generation with cached key values are unrelated training/runtime bugs." + }, + { + "left": "issue:43531", + "right": "issue:46821", + "accept": false, + "reason": "Qwen3-MoE sliding_window and TimesFM window_size AttributeError do not share a code path." + }, + { + "left": "issue:34634", + "right": "issue:41720", + "accept": false, + "reason": "Bark voice_preset loading and Qwen3 device-mapping failures are unrelated model-specific bugs." + }, + { + "left": "issue:43540", + "right": "issue:46858", + "accept": false, + "reason": "Qwen3OmniMoe video-input processing and cache.get_seq_length tensor arithmetic are different failures." + }, + { + "left": "issue:44451", + "right": "issue:44991", + "accept": false, + "reason": "Both are loading regressions for specific models, but the reported tokenizer/model failures are not the same underlying bug." + }, + { + "left": "issue:45759", + "right": "issue:45859", + "accept": false, + "reason": "Composite Qwen config unwrapping and Qwen3_5Moe rotary embedding CPU-offload compatibility are distinct issues." + }, + { + "left": "issue:45357", + "right": "issue:46402", + "accept": false, + "reason": "Incorrect visual encoder save keys and missing text encoder load weights are separate save/load problems." + }, + { + "left": "issue:44077", + "right": "issue:44625", + "accept": false, + "reason": "PatchTSMixer post_init policy and Qwen3.5 num_labels propagation are unrelated model-config bugs." + }, + { + "left": "issue:43644", + "right": "issue:46665", + "accept": false, + "reason": "Non-persistent buffer junk values and inconsistent from_pretrained-in-class behavior are different initialization/loading defects." + }, + { + "left": "issue:44062", + "right": "issue:44451", + "accept": false, + "reason": "AddedToken special-argument conflict is a tokenizer constructor bug, not the ScandiBERT loading failure." + }, + { + "left": "issue:44898", + "right": "issue:45072", + "accept": false, + "reason": "Perceiver interpolation at non-default resolution and bfloat16 dtype mismatches in inference are unrelated." + }, + { + "left": "issue:44987", + "right": "issue:45042", + "accept": false, + "reason": "A model-loading failure and PIL image processors requiring torchvision are separate v5 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": 23, + "review_unit_index": 17, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:30064", + "issue:36010", + "issue:39692", + "issue:41720", + "issue:42886", + "issue:43295", + "issue:43381", + "issue:43421", + "issue:43493", + "issue:43653", + "issue:43720", + "issue:43746", + "issue:43782", + "issue:43819", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43994", + "issue:44186", + "issue:44368", + "issue:44514", + "issue:44545", + "issue:44610", + "issue:44625", + "issue:44671", + "issue:44843", + "issue:44918", + "issue:44936", + "issue:44995", + "issue:45081", + "issue:45356", + "issue:45357", + "issue:45381", + "issue:45479", + "issue:45704", + "issue:45759", + "issue:45834", + "issue:45907", + "issue:45923", + "issue:45987", + "issue:46032", + "issue:46050", + "issue:46463", + "issue:46655", + "issue:46693", + "issue:46829", + "issue:46832", + "issue:46858" + ], + "soft_pairs": [ + "issue:44625|issue:45479", + "issue:43381|issue:43844", + "issue:44368|issue:45357", + "issue:39692|issue:44610", + "issue:46655|issue:46693", + "issue:43421|issue:43927", + "issue:41720|issue:45704", + "issue:43746|issue:46050", + "issue:43381|issue:45704", + "issue:42886|issue:44843", + "issue:45704|issue:45923", + "issue:44918|issue:44936", + "issue:44514|issue:44545", + "issue:45759|issue:45923", + "issue:43746|issue:43873", + "issue:43421|issue:43653", + "issue:43994|issue:44671", + "issue:39692|issue:43493", + "issue:46032|issue:46463", + "issue:43421|issue:44186", + "issue:43782|issue:45381", + "issue:44995|issue:46858", + "issue:45081|issue:45356", + "issue:30064|issue:36010", + "issue:45356|issue:45907", + "issue:39692|issue:43873", + "issue:43720|issue:43819", + "issue:45987|issue:46829", + "issue:46829|issue:46832", + "issue:43653|issue:43720", + "issue:43295|issue:45834" + ], + "prepared_review_unit_hash": "894dfd93e6fb41971280b3a19c1e385e2a4503280e9eb5fa7023fcdd4a341460", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 37562, + "estimated_input_tokens": 9391, + "estimated_eval_tokens": 19038 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Hugging Face issues; the only clear duplicate pair is the Qwen2_5_VLProcessor batched padding=False crash. Most other soft links are broad topical similarities, not the same concrete bug.", + "confidence": 0.91, + "canonical_issue_reason": "issue 44514 is the best canonical issue: it is the earlier, more detailed report and is an exact duplicate of 44545.", + "canonical_pr_reason": null, + "best_issue_reason": "issue 44514 is the strongest cluster representative because it cleanly captures the only confirmed duplicate theme and has the fuller problem description.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44625", + "right": "issue:45479", + "accept": false, + "reason": "Different sequence-classification bugs: one is config propagation, the other is a single-label zero-loss regression." + }, + { + "left": "issue:43381", + "right": "issue:43844", + "accept": false, + "reason": "Both mention gradients/training behavior, but the concrete failure modes and code paths are different." + }, + { + "left": "issue:44368", + "right": "issue:45357", + "accept": false, + "reason": "One is a warning about tie_word_embeddings, the other is a save_pretrained key regression for a multimodal model." + }, + { + "left": "issue:39692", + "right": "issue:44610", + "accept": false, + "reason": "Different models and failures: documentation/example mismatch vs processor/model input-size mismatch." + }, + { + "left": "issue:46655", + "right": "issue:46693", + "accept": false, + "reason": "Video padding bug versus FA2 long-context performance regression; not the same underlying change." + }, + { + "left": "issue:43421", + "right": "issue:43927", + "accept": false, + "reason": "Both involve tokenizer internals, but one is runtime post-processor updates and the other is save/load config corruption." + }, + { + "left": "issue:41720", + "right": "issue:45704", + "accept": false, + "reason": "Unrelated issues: Qwen3 device-mapping cuda assert vs T5 RMSNorm memory leak." + }, + { + "left": "issue:43746", + "right": "issue:46050", + "accept": false, + "reason": "Local checkpoint loading with PEFT adapters is a different path from quantized sinq save/load breakage." + }, + { + "left": "issue:43381", + "right": "issue:45704", + "accept": false, + "reason": "Eval-mode gradient checkpointing and T5 apex leak are separate defects." + }, + { + "left": "issue:42886", + "right": "issue:44843", + "accept": false, + "reason": "Both are offline-related tokenizer problems, but 44843 is specifically a _patch_mistral_regex model_info call and 42886 is broader cache-loading failure." + }, + { + "left": "issue:45704", + "right": "issue:45923", + "accept": false, + "reason": "T5 apex memory leak and missing supports_gradient_checkpointing are unrelated." + }, + { + "left": "issue:44918", + "right": "issue:44936", + "accept": false, + "reason": "One is an embedding-unpacking failure in TRL SFT, the other is trainer.evaluate breaking after trainer.train()." + }, + { + "left": "issue:44514", + "right": "issue:44545", + "accept": true, + "reason": "Same exact crash: Qwen2_5_VLProcessor.apply_chat_template fails on batched input when padding=False." + }, + { + "left": "issue:45759", + "right": "issue:45923", + "accept": false, + "reason": "Different multimodal/config issue versus a missing gradient-checkpointing flag." + }, + { + "left": "issue:43746", + "right": "issue:43873", + "accept": false, + "reason": "Local PEFT checkpoint loading and quantization/offloading behavior are not the same bug." + }, + { + "left": "issue:43421", + "right": "issue:43653", + "accept": false, + "reason": "Related to special tokens, but one is a backend post-processor update issue and the other is a BigBirdTokenizer registration bug." + }, + { + "left": "issue:43994", + "right": "issue:44671", + "accept": false, + "reason": "Different model families and symptoms: SigLIP2 inference weirdness vs CamemBERT MLM regression." + }, + { + "left": "issue:39692", + "right": "issue:43493", + "accept": false, + "reason": "Both mention SigLIP2, but one is docs/example errors and the other is an implementation discrepancy against JAX." + }, + { + "left": "issue:46032", + "right": "issue:46463", + "accept": false, + "reason": "Cache/use_with_seq_len bug and a T5 error-message bug are unrelated." + }, + { + "left": "issue:43421", + "right": "issue:44186", + "accept": false, + "reason": "Special-token post-processor updates versus LayoutLMv2 tokenizer crashes are different tokenizer bugs." + }, + { + "left": "issue:43782", + "right": "issue:45381", + "accept": false, + "reason": "Different Qwen multimodal failures: weight_only loading error versus incorrect video vision_position_ids." + }, + { + "left": "issue:44995", + "right": "issue:46858", + "accept": false, + "reason": "Both involve cache, but the affected models and failure mechanisms are different." + }, + { + "left": "issue:45081", + "right": "issue:45356", + "accept": false, + "reason": "Both touch fix_mistral_regex, but one is an AttributeError in the patch function and the other is a tokenizer regression with codec handling." + }, + { + "left": "issue:30064", + "right": "issue:36010", + "accept": false, + "reason": "Completely unrelated: image segmentation map processing versus a GenerationMixin import error." + }, + { + "left": "issue:45356", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer regression and repository listing API forwarding are unrelated." + }, + { + "left": "issue:39692", + "right": "issue:43873", + "accept": false, + "reason": "SigLIP2 docs/example errors are unrelated to quantization offloading behavior." + }, + { + "left": "issue:43720", + "right": "issue:43819", + "accept": false, + "reason": "BitNet accelerate-loading failure and DAC from_latents/STE mismatch are different code paths." + }, + { + "left": "issue:45987", + "right": "issue:46829", + "accept": false, + "reason": "StaticCache seq-length typing bug and VideoPrism output-field bug are unrelated." + }, + { + "left": "issue:46829", + "right": "issue:46832", + "accept": false, + "reason": "Different model implementations and failure modes; no shared concrete bug." + }, + { + "left": "issue:43653", + "right": "issue:43720", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and BitNet packed-weight loading are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:45834", + "accept": false, + "reason": "Custom checkpoint loading with processor/tokenizer access and Kosmos2.5 OCR index errors are different bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is well grounded in the packet: 44514/44545 is an exact duplicate pair by title, and the canonical choice for 44514 is conservative and supported by created_at/body length. The rest of the soft-edge rejections stay cautious and are consistent with the titles, without overstating duplicate evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 18, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:30064", + "issue:34634", + "issue:36010", + "issue:36331", + "issue:39692", + "issue:41720", + "issue:43335", + "issue:43531", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43611", + "issue:43618", + "issue:43643", + "issue:43644", + "issue:43653", + "issue:43697", + "issue:43749", + "issue:43756", + "issue:43819", + "issue:43927", + "issue:44008", + "issue:44062", + "issue:44186", + "issue:44220", + "issue:44361", + "issue:44442", + "issue:44448", + "issue:44488", + "issue:44661", + "issue:44671", + "issue:44821", + "issue:44843", + "issue:44877", + "issue:44898", + "issue:44991", + "issue:45081", + "issue:45084", + "issue:45183", + "issue:45381", + "issue:45479", + "issue:45584", + "issue:45834", + "issue:46294", + "issue:46665", + "issue:46693", + "issue:46858" + ], + "soft_pairs": [ + "issue:43550|issue:44361", + "issue:43550|issue:43927", + "issue:44062|issue:44488", + "issue:43596|issue:44220", + "issue:45381|issue:46693", + "issue:43540|issue:46294", + "issue:44008|issue:46858", + "issue:43531|issue:43643", + "issue:43611|issue:46665", + "issue:43653|issue:43819", + "issue:44220|issue:44661", + "issue:44186|issue:44898", + "issue:43618|issue:43697", + "issue:41720|issue:44991", + "issue:30064|issue:36331", + "issue:43643|issue:44877", + "issue:43550|issue:43819", + "issue:43335|issue:43756", + "issue:44843|issue:45081", + "issue:44821|issue:45084", + "issue:44186|issue:44442", + "issue:39692|issue:45834", + "issue:44448|issue:44671", + "issue:43540|issue:44008", + "issue:43644|issue:43749", + "issue:34634|issue:36010", + "issue:45183|issue:45479", + "issue:45584|issue:45834" + ], + "prepared_review_unit_hash": "4d7e57b94e0a09dede32c7b6abae0c620240f17d8b8dba413f44c221721559f3", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35675, + "estimated_input_tokens": 8919, + "estimated_eval_tokens": 18094 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is highly heterogeneous, with no clear duplicate cluster and no PRs present. All soft edges are rejected as they point to different concrete bugs or failure modes, even when they touch similar subsystems.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:44361", + "accept": false, + "reason": "Different models and failures: Bamba torch.compile/SDPA vs MLuke tokenizer AttributeError." + }, + { + "left": "issue:43550", + "right": "issue:43927", + "accept": false, + "reason": "Different code paths: torch.compile/SDPA crash vs DiaConfig save/load token ID loss." + }, + { + "left": "issue:44062", + "right": "issue:44488", + "accept": false, + "reason": "AddedToken special-arg duplication is unrelated to loading EMBEDDIA/est-roberta tokenizer failure." + }, + { + "left": "issue:43596", + "right": "issue:44220", + "accept": false, + "reason": "Deepspeed ZeRO3/BertModel index error is unrelated to `_torch_extract_fbank_features()`." + }, + { + "left": "issue:45381", + "right": "issue:46693", + "accept": false, + "reason": "Qwen2.5-VL video position IDs bug is unrelated to FA2 long-context performance regression." + }, + { + "left": "issue:43540", + "right": "issue:46294", + "accept": false, + "reason": "Both are video-related Qwen bugs, but one is a Qwen3OmniMoe processing ValueError and the other is a chat-template token/feature mismatch." + }, + { + "left": "issue:44008", + "right": "issue:46858", + "accept": false, + "reason": "Gemma3n variable-name collision is unrelated to `_prefill` cache seq-length device tensor arithmetic." + }, + { + "left": "issue:43531", + "right": "issue:43643", + "accept": false, + "reason": "Sliding-window handling in Qwen3-MoE is unrelated to missing fields from `trust_remote_code=True` config loading." + }, + { + "left": "issue:43611", + "right": "issue:46665", + "accept": false, + "reason": "Transformers 5 base_model_prefix breakage is unrelated to inconsistent `from_pretrained` behavior inside Python classes." + }, + { + "left": "issue:43653", + "right": "issue:43819", + "accept": false, + "reason": "BigBirdTokenizer special-token registration is unrelated to DAC `from_latents` vs forward mismatch." + }, + { + "left": "issue:44220", + "right": "issue:44661", + "accept": false, + "reason": "Audio fbank extraction bug is unrelated to `add-new-model-like` failing on `TOKENIZER_MAPPING_NAMES`." + }, + { + "left": "issue:44186", + "right": "issue:44898", + "accept": false, + "reason": "LayoutLMv2 tokenizer crashes on NER padding/truncation are unrelated to Perceiver image classification with non-default resolution." + }, + { + "left": "issue:43618", + "right": "issue:43697", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to RTDetr producing different outputs in v5." + }, + { + "left": "issue:41720", + "right": "issue:44991", + "accept": false, + "reason": "Qwen3 auto device mapping CUDA assert is unrelated to loading the EMBEDDIA/est-roberta tokenizer." + }, + { + "left": "issue:30064", + "right": "issue:36331", + "accept": false, + "reason": "Void segmentation-map processing is unrelated to `compute_loss(num_items_in_batch)` signature mismatch." + }, + { + "left": "issue:43643", + "right": "issue:44877", + "accept": false, + "reason": "Both are config-loading issues, but one is missing fields from `trust_remote_code` while the other is strict config rejection for `granite_speech`." + }, + { + "left": "issue:43550", + "right": "issue:43819", + "accept": false, + "reason": "Bamba torch.compile/SDPA crash and DAC `from_latents` mismatch are unrelated bugs." + }, + { + "left": "issue:43335", + "right": "issue:43756", + "accept": false, + "reason": "Sparse-layer creation with zero encoder layers is unrelated to Smollm3 RoPE layer dropping." + }, + { + "left": "issue:44843", + "right": "issue:45081", + "accept": false, + "reason": "Same helper name, but different root causes: offline `model_info()` call vs missing `backend_tokenizer` attribute in `_patch_mistral_regex`." + }, + { + "left": "issue:44821", + "right": "issue:45084", + "accept": false, + "reason": "AutoImageProcessor URL loading is unrelated to Jinja/template compilation errors." + }, + { + "left": "issue:44186", + "right": "issue:44442", + "accept": false, + "reason": "LayoutLMv2 NER/padding crash is unrelated to AutoTokenizer failing to load FastSpeech2ConformerTokenizer." + }, + { + "left": "issue:39692", + "right": "issue:45834", + "accept": false, + "reason": "SigLIP2 doc example errors are unrelated to Kosmos2.5 long OCR index errors." + }, + { + "left": "issue:44448", + "right": "issue:44671", + "accept": false, + "reason": "Pegasus v4/v5 output drift is unrelated to CamemBERT masked LM prediction changes." + }, + { + "left": "issue:43540", + "right": "issue:44008", + "accept": false, + "reason": "Qwen3OmniMoe video processing ValueError is unrelated to Gemma3n audio_mel_mask variable collision." + }, + { + "left": "issue:43644", + "right": "issue:43749", + "accept": false, + "reason": "Non-persistent buffer initialization junk is unrelated to broken FSDP CPU RAM-efficient loading." + }, + { + "left": "issue:34634", + "right": "issue:36010", + "accept": false, + "reason": "Bark voice_preset and GenerationMixin import errors are different components and failures." + }, + { + "left": "issue:45183", + "right": "issue:45479", + "accept": false, + "reason": "Input/inputs_embeds validation messaging bug is unrelated to degenerate single-label classification loss." + }, + { + "left": "issue:45584", + "right": "issue:45834", + "accept": false, + "reason": "Whisper empty transcription after special-token alignment is unrelated to Kosmos2.5 long OCR index 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": 23, + "review_unit_index": 19, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28218", + "issue:29159", + "issue:42175", + "issue:43322", + "issue:43388", + "issue:43421", + "issue:43540", + "issue:43550", + "issue:43577", + "issue:43582", + "issue:43606", + "issue:43645", + "issue:43716", + "issue:43720", + "issue:43749", + "issue:43782", + "issue:43818", + "issue:43825", + "issue:43827", + "issue:43931", + "issue:43950", + "issue:43994", + "issue:44008", + "issue:44291", + "issue:44479", + "issue:44561", + "issue:44610", + "issue:44671", + "issue:44805", + "issue:44811", + "issue:44936", + "issue:44977", + "issue:44991", + "issue:45020", + "issue:45072", + "issue:45081", + "issue:45381", + "issue:45685", + "issue:45701", + "issue:45834", + "issue:45987", + "issue:46032", + "issue:46291", + "issue:46294", + "issue:46463", + "issue:46512", + "issue:46665", + "issue:46846" + ], + "soft_pairs": [ + "issue:43606|issue:45072", + "issue:43322|issue:45685", + "issue:43550|issue:45072", + "issue:43388|issue:46032", + "issue:44479|issue:45834", + "issue:43825|issue:43994", + "issue:43782|issue:44977", + "issue:43582|issue:46512", + "issue:43577|issue:43818", + "issue:44991|issue:46291", + "issue:43421|issue:44811", + "issue:44561|issue:45020", + "issue:43950|issue:45987", + "issue:43716|issue:44008", + "issue:43931|issue:46846", + "issue:46294|issue:46463", + "issue:43825|issue:43827", + "issue:28218|issue:29159", + "issue:44805|issue:44936", + "issue:44610|issue:45072", + "issue:43540|issue:45381", + "issue:43720|issue:45072", + "issue:43749|issue:44291", + "issue:43577|issue:43582", + "issue:43645|issue:46665", + "issue:44671|issue:45701", + "issue:42175|issue:45081" + ], + "prepared_review_unit_hash": "cb1855bd669dc602a989e98a009aeefe9d62ea8d97fdeeadfd18bd022ca1a70b", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36219, + "estimated_input_tokens": 9055, + "estimated_eval_tokens": 18366 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a broad mix of Transformers v5 regression reports: tokenization, remote-code loading, multimodal/video bugs, dtype/device mismatches, and trainer/cache edge cases. The soft pairs are mostly thematically similar but do not look like the same concrete bug or change.", + "confidence": 0.66, + "canonical_issue_reason": "issue:45020 is the broadest representative issue here, covering the recurring 'recent transformers versions break remote-code/models' regression theme that many other reports in this cluster loosely resemble.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the best global issue to anchor the cluster because it is the most umbrella-like report and can plausibly subsume several nearby v5 compatibility breakages, even though the candidates are not true duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43606", + "right": "issue:45072", + "accept": false, + "reason": "Both mention inference/device mismatch, but one is a CPU offload bug for bark-small and the other is a bfloat16 dtype mismatch; different models and code paths." + }, + { + "left": "issue:43322", + "right": "issue:45685", + "accept": false, + "reason": "Segfault while loading Llava Next is unrelated to the MPS histogram Int error in moe code; no shared concrete failure path." + }, + { + "left": "issue:43550", + "right": "issue:45072", + "accept": false, + "reason": "torch.compile + SDPA failure in Bamba is a different bug from the dtype mismatch issue in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43388", + "right": "issue:46032", + "accept": false, + "reason": "Label truncation in gather_for_metrics and Mamba2 use_cache producing wrong results are separate training/runtime bugs." + }, + { + "left": "issue:44479", + "right": "issue:45834", + "accept": false, + "reason": "Both are vision/video-ish regressions, but one is a Qwen video-input regression and the other is an OCR index error in Kosmos2.5." + }, + { + "left": "issue:43825", + "right": "issue:43994", + "accept": false, + "reason": "One is an error-message/docs regression around pipeline translation support; the other is a model output quality problem with AutoModel/pipeline." + }, + { + "left": "issue:43782", + "right": "issue:44977", + "accept": false, + "reason": "Qwen3VL weight_only loading failure and Qwen3.5 flash-attention generation corruption are different symptoms with different triggers." + }, + { + "left": "issue:43582", + "right": "issue:46512", + "accept": false, + "reason": "Apple Silicon TypeError in caching warmup is unrelated to from_config leaving config.dtype stale." + }, + { + "left": "issue:43577", + "right": "issue:43818", + "accept": false, + "reason": "BLIP2 dtype persistence and Video-LLaVA temporal-attention/shared-weights issues are separate model-loading/model-architecture bugs." + }, + { + "left": "issue:44991", + "right": "issue:46291", + "accept": false, + "reason": "Tokenizer loading failure for a specific model is not the same as importing pipelines failing due to kernels package requirements." + }, + { + "left": "issue:43421", + "right": "issue:44811", + "accept": false, + "reason": "Special-token post-processor sync in TokenizersBackend and Whisper batch_decode ignoring skip_special_tokens are different tokenizer API bugs." + }, + { + "left": "issue:44561", + "right": "issue:45020", + "accept": false, + "reason": "Both concern remote-code breakage, but one is specifically about removing is_torch_fx_available while the other is a broader report; not the same concrete cause." + }, + { + "left": "issue:43950", + "right": "issue:45987", + "accept": false, + "reason": "Non-persistent buffer corruption in from_pretrained is unrelated to StaticCache.get_seq_length returning the wrong type/shape." + }, + { + "left": "issue:43716", + "right": "issue:44008", + "accept": false, + "reason": "A dtype mismatch between preprocessor and model is different from a variable-name collision causing AttributeError in Gemma3nModel.forward()." + }, + { + "left": "issue:43931", + "right": "issue:46846", + "accept": false, + "reason": "Weight-shape mismatch loading Qwen3-VL-30B and missing linear_attn entries in TP plan are distinct model-configuration bugs." + }, + { + "left": "issue:46294", + "right": "issue:46463", + "accept": false, + "reason": "Qwen3.5 video chat-template mismatch and T5 decoder-input validation error message are unrelated issues." + }, + { + "left": "issue:43825", + "right": "issue:43827", + "accept": false, + "reason": "One is a pipeline error-message regression and the other is stale documentation; same area, different artifact and fix." + }, + { + "left": "issue:28218", + "right": "issue:29159", + "accept": false, + "reason": "Both are tokenizer-related, but one is an added-token spacing bug and the other is a broad slow-vs-fast tokenizer inconsistency report." + }, + { + "left": "issue:44805", + "right": "issue:44936", + "accept": false, + "reason": "Mask shape mismatch in indexing and trainer.evaluate failing after train are different runtime failures." + }, + { + "left": "issue:44610", + "right": "issue:45072", + "accept": false, + "reason": "Processor output size mismatch for OmDet-Turbo is unrelated to the bfloat16 dtype mismatch in SwitchTransformers/TimmWrapperModel." + }, + { + "left": "issue:43540", + "right": "issue:45381", + "accept": false, + "reason": "Both involve video input, but they target different Qwen variants and different bugs: ValueError processing video inputs vs wrong vision_position_ids." + }, + { + "left": "issue:43720", + "right": "issue:45072", + "accept": false, + "reason": "Packed-weight unpacking during accelerate loading is a different failure mode from the dtype mismatch reports." + }, + { + "left": "issue:43749", + "right": "issue:44291", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and init_empty_weights rejecting _is_hf_initialized are separate loading regressions." + }, + { + "left": "issue:43577", + "right": "issue:43582", + "accept": false, + "reason": "BLIP2 dtype handling and Apple Silicon caching allocator warmup are unrelated bugs." + }, + { + "left": "issue:43645", + "right": "issue:46665", + "accept": false, + "reason": "Custom models in Jupyter notebooks and inconsistent from_pretrained behavior inside Python classes are related only at a very high level." + }, + { + "left": "issue:44671", + "right": "issue:45701", + "accept": false, + "reason": "CamemBERT masked LM prediction regression and version-dependent tokenization changes are not the same underlying bug." + }, + { + "left": "issue:42175", + "right": "issue:45081", + "accept": false, + "reason": "Missing tensorflow in extras install and the Mistral regex patch crash are unrelated packaging vs tokenizer-initialization 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": 23, + "review_unit_index": 20, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:34634", + "issue:36010", + "issue:36331", + "issue:42175", + "issue:42490", + "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:43720", + "issue:43742", + "issue:43782", + "issue:43819", + "issue:43906", + "issue:43927", + "issue:44060", + "issue:44368", + "issue:44554", + "issue:44589", + "issue:44610", + "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:46209", + "issue:46285", + "issue:46291", + "issue:46326", + "issue:46697", + "issue:46728" + ], + "soft_pairs": [ + "issue:34634|issue:36331", + "issue:45003|issue:46285", + "issue:46697|issue:46728", + "issue:43653|issue:44912", + "issue:43262|issue:43819", + "issue:43782|issue:45325", + "issue:45081|issue:45406", + "issue:45372|issue:46326", + "issue:43582|issue:44857", + "issue:44589|issue:45042", + "issue:43550|issue:44912", + "issue:43742|issue:45020", + "issue:43329|issue:45600", + "issue:45084|issue:45538", + "issue:44060|issue:44368", + "issue:43452|issue:44991", + "issue:43526|issue:43697", + "issue:43257|issue:43329", + "issue:43526|issue:44554", + "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:42490|issue:43257", + "issue:43720|issue:44610" + ], + "prepared_review_unit_hash": "3ac680d38e8e1c248cffd2aa3bd5d10754f99a689e85ceb96e22631994508261", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36521, + "estimated_input_tokens": 9131, + "estimated_eval_tokens": 18518 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the items span unrelated bugs across audio, tokenizer loading, training, multimodal processors, quantization, and backend-specific regressions. None of the suggested soft edges look like the same concrete bug/change, so all are rejected.", + "confidence": 0.92, + "canonical_issue_reason": "No single issue clearly represents a duplicate set here; if a representative anchor is needed, issue:45020 is the broadest systemic regression report, but the cluster is too mixed to treat as one bug family.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45020 is the most broadly scoped and could serve as a weak representative anchor, but it does not actually unify the other issues into a duplicate cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:34634", + "right": "issue:36331", + "accept": false, + "reason": "Different subsystems and failures: Bark voice_preset handling vs CustomTrainer compute_loss signature mismatch." + }, + { + "left": "issue:45003", + "right": "issue:46285", + "accept": false, + "reason": "One is unsafe sys.modules access in modeling_utils; the other is a CLIP class structure inconsistency. Related theme only, not same bug." + }, + { + "left": "issue:46697", + "right": "issue:46728", + "accept": false, + "reason": "Both are image-processor padding/patch-count issues, but they affect different models and different code paths." + }, + { + "left": "issue:43653", + "right": "issue:44912", + "accept": false, + "reason": "BigBirdTokenizer special-token decode bug is unrelated to MXFP4 quantization fallback on git-oss-20b." + }, + { + "left": "issue:43262", + "right": "issue:43819", + "accept": false, + "reason": "Audio chat-template sample-rate default bug vs DAC latent/forward mismatch; no shared concrete code path." + }, + { + "left": "issue:43782", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen/VL bugs: weight_only loading failure vs rope index temporal scaling." + }, + { + "left": "issue:45081", + "right": "issue:45406", + "accept": false, + "reason": "Tokenizer regex patching crash and Gemma4Processor missing _tokenizer are separate processor-loading failures." + }, + { + "left": "issue:45372", + "right": "issue:46326", + "accept": false, + "reason": "Both involve mistral_common-adjacent processor/tokenizer behavior, but one is a missing import and the other is continue_final_message logic." + }, + { + "left": "issue:43582", + "right": "issue:44857", + "accept": false, + "reason": "AppleSilicon caching allocator warmup TypeError and LwDetrImageLoss AMP crash are unrelated runtime errors." + }, + { + "left": "issue:44589", + "right": "issue:45042", + "accept": false, + "reason": "Float8 storage lookup failure is unrelated to PIL image processors incorrectly requiring torchvision." + }, + { + "left": "issue:43550", + "right": "issue:44912", + "accept": false, + "reason": "Bamba torch.compile SDPA failure and git-oss-20b MXFP4 loading fallback are different model/runtime problems." + }, + { + "left": "issue:43742", + "right": "issue:45020", + "accept": false, + "reason": "MobileLLM key error during load is a model-specific mapping bug, not the broad remote_code regression described in 45020." + }, + { + "left": "issue:43329", + "right": "issue:45600", + "accept": false, + "reason": "Video-token counting bug in multimodal processors is unrelated to stale auto_mappings.py references breaking CI." + }, + { + "left": "issue:45084", + "right": "issue:45538", + "accept": false, + "reason": "Chat template compilation TypeError is unrelated to CLIPTokenizer's extreme model_max_length value." + }, + { + "left": "issue:44060", + "right": "issue:44368", + "accept": false, + "reason": "Both mention Qwen3-Next/Qwen3.5 tie-weight warnings, but one is a specific incorrect tie mapping and the other is a user-facing warning from fine-tuning." + }, + { + "left": "issue:43452", + "right": "issue:44991", + "accept": false, + "reason": "gguf_file handling and EMBEDDIA tokenizer loading failure are separate from different loader paths and symptoms." + }, + { + "left": "issue:43526", + "right": "issue:43697", + "accept": false, + "reason": "BeitImageProcessorFast reduce_labels bug is unrelated to RTDetrV2 output drift in Transformers v5." + }, + { + "left": "issue:43257", + "right": "issue:43329", + "accept": false, + "reason": "Qwen3 MoE accelerate/deepspeed conversion issue is unrelated to multimodal video token counting." + }, + { + "left": "issue:43526", + "right": "issue:44554", + "accept": false, + "reason": "reduce_labels returning one label and MPS attention correctness with mismatched heads are unrelated bugs." + }, + { + "left": "issue:43906", + "right": "issue:46209", + "accept": false, + "reason": "An isolated repro of issue 38071 is not the same as FP8 inference failing on B200." + }, + { + "left": "issue:36010", + "right": "issue:36331", + "accept": false, + "reason": "GenerationMixin import regression and compute_loss signature change are different breakages." + }, + { + "left": "issue:43819", + "right": "issue:43927", + "accept": false, + "reason": "DAC latent reconstruction mismatch is unrelated to DiaConfig losing custom token IDs on save/load." + }, + { + "left": "issue:36010", + "right": "issue:43645", + "accept": false, + "reason": "GenerationMixin import error is unrelated to Jupyter custom-model initialization problems." + }, + { + "left": "issue:43531", + "right": "issue:45042", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and PIL backend requiring torchvision are unrelated." + }, + { + "left": "issue:43638", + "right": "issue:44936", + "accept": false, + "reason": "Zero3 index error with non-pretrained Bert and trainer.evaluate() after train() are different training failures." + }, + { + "left": "issue:42175", + "right": "issue:46291", + "accept": false, + "reason": "Torch-only install missing TensorFlow backend is unrelated to pipelines import failing because kernels now needs version/revision." + }, + { + "left": "issue:42490", + "right": "issue:43257", + "accept": false, + "reason": "Trailing-slash local model import warning and Qwen3 MOE weight conversion are different loading issues." + }, + { + "left": "issue:43720", + "right": "issue:44610", + "accept": false, + "reason": "BitNet packed-weight unpacking during accelerate loading and OmDet-Turbo 640px-vs-224px preprocessing mismatch 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": 23, + "review_unit_index": 21, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28282", + "issue:29159", + "issue:29870", + "issue:34567", + "issue:38175", + "issue:39692", + "issue:40990", + "issue:42915", + "issue:43257", + "issue:43262", + "issue:43329", + "issue:43450", + "issue:43454", + "issue:43525", + "issue:43540", + "issue:43577", + "issue:43606", + "issue:43643", + "issue:43650", + "issue:43696", + "issue:43742", + "issue:43746", + "issue:43749", + "issue:43854", + "issue:43906", + "issue:43927", + "issue:43931", + "issue:44291", + "issue:44387", + "issue:44479", + "issue:44568", + "issue:44610", + "issue:44617", + "issue:44756", + "issue:44811", + "issue:44821", + "issue:44912", + "issue:44977", + "issue:45042", + "issue:45356", + "issue:45459", + "issue:45759", + "issue:45803", + "issue:45859", + "issue:46291", + "issue:46421", + "issue:46424", + "issue:46701" + ], + "soft_pairs": [ + "issue:39692|issue:44912", + "issue:43746|issue:44811", + "issue:43854|issue:44977", + "issue:43643|issue:45759", + "issue:38175|issue:44610", + "issue:43257|issue:43696", + "issue:43262|issue:44811", + "issue:44387|issue:44617", + "issue:43257|issue:43931", + "issue:45459|issue:45803", + "issue:43329|issue:43450", + "issue:46421|issue:46424", + "issue:43329|issue:44479", + "issue:42915|issue:43742", + "issue:43749|issue:44756", + "issue:34567|issue:40990", + "issue:44291|issue:45042", + "issue:43650|issue:43906", + "issue:42915|issue:44821", + "issue:29159|issue:34567", + "issue:43606|issue:45859", + "issue:43927|issue:44912", + "issue:43540|issue:46701", + "issue:28282|issue:29870", + "issue:43454|issue:43577", + "issue:43525|issue:46291", + "issue:44568|issue:45356" + ], + "prepared_review_unit_hash": "b1a06a8903c377e64e3719af326c3ba4daa05899cb244bb4db6f5d59d925b1b2", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 36016, + "estimated_input_tokens": 9004, + "estimated_eval_tokens": 18264 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: the soft pairs mostly share only broad themes (tokenizers, multimodal loading, quantization, offload, or video processing) but not the same underlying bug. I would not merge any of the candidate pairs.", + "confidence": 0.98, + "canonical_issue_reason": "No single canonical issue fits the cluster; the items cover unrelated failures across different models, processors, and code paths, so there is no true duplicate set here.", + "canonical_pr_reason": null, + "best_issue_reason": "No strong global best issue. If forced, the most concrete code-path bug is issue 43329, but it does not represent the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:39692", + "right": "issue:44912", + "accept": false, + "reason": "Both involve model/loading quality, but one is a SigLIP2 docs/example mismatch and the other is MXFP4 quantization fallback for gpt-oss-20b; different bugs." + }, + { + "left": "issue:43746", + "right": "issue:44811", + "accept": false, + "reason": "GraniteSpeech PEFT checkpoint loading and Whisper batch_decode skip_special_tokens are unrelated code paths." + }, + { + "left": "issue:43854", + "right": "issue:44977", + "accept": false, + "reason": "GLM-4.7-Flash test loading error and Qwen3.5 flash-attention generation regression are different model-specific issues." + }, + { + "left": "issue:43643", + "right": "issue:45759", + "accept": false, + "reason": "Both concern config handling, but one is AutoConfig trust_remote_code field loss and the other is AutoModelForCausalLM text_config unwrapping for composite Qwen configs." + }, + { + "left": "issue:38175", + "right": "issue:44610", + "accept": false, + "reason": "SigLIP2 zero probabilities and OmDet-Turbo processor/model size mismatch are distinct failures." + }, + { + "left": "issue:43257", + "right": "issue:43696", + "accept": false, + "reason": "Qwen3 MoE weight conversion with accelerate+deepspeed is unrelated to GPT-oss-20b CUDA OOM." + }, + { + "left": "issue:43262", + "right": "issue:44811", + "accept": false, + "reason": "Audio chat-template sampling-rate behavior and Whisper batch_decode token skipping are different processor bugs." + }, + { + "left": "issue:44387", + "right": "issue:44617", + "accept": false, + "reason": "Both are OOM-related, but one is an int4 reserved-memory regression and the other is Sam3Video running out of memory; not the same underlying defect." + }, + { + "left": "issue:43257", + "right": "issue:43931", + "accept": false, + "reason": "Qwen3 MoE conversion failure and Qwen3-VL weight-shape mismatch are separate loading problems." + }, + { + "left": "issue:45459", + "right": "issue:45803", + "accept": false, + "reason": "Both involve exception handling, but the affected functions and failure modes differ; they are not the same bug." + }, + { + "left": "issue:43329", + "right": "issue:43450", + "accept": false, + "reason": "Undefined helpers in _get_num_multimodal_tokens versus wrong batched video processor output shape are different video-path issues." + }, + { + "left": "issue:46421", + "right": "issue:46424", + "accept": false, + "reason": "Chunked-prefill recompilation and max_cache_length only applying to Quantized Cache are separate cache features." + }, + { + "left": "issue:43329", + "right": "issue:44479", + "accept": false, + "reason": "The first is a specific undefined-variable bug in multimodal token counting; the second is a broader v5.3.0 video-input regression across several VLMs." + }, + { + "left": "issue:42915", + "right": "issue:43742", + "accept": false, + "reason": "FineGrainedFP8Config failure for Qwen3Moe and MobileLLM-125M key error are unrelated model-loading bugs." + }, + { + "left": "issue:43749", + "right": "issue:44756", + "accept": false, + "reason": "FSDP CPU RAM efficient loading and Strix Halo mmap OOM are different loading/memory issues." + }, + { + "left": "issue:34567", + "right": "issue:40990", + "accept": false, + "reason": "TrainerState token counting and gpt-oss-20b perplexity are unrelated training/evaluation problems." + }, + { + "left": "issue:44291", + "right": "issue:45042", + "accept": false, + "reason": "init_empty_weights argument handling and PIL image processor torchvision dependency are unrelated issues." + }, + { + "left": "issue:43650", + "right": "issue:43906", + "accept": false, + "reason": "Generic 'ADD THE DATA' report and isolated reproduction of another issue are not duplicates." + }, + { + "left": "issue:42915", + "right": "issue:44821", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and AutoImageProcessor-from-URL loading are unrelated." + }, + { + "left": "issue:29159", + "right": "issue:34567", + "accept": false, + "reason": "Tokenizer slow/fast inconsistency and TrainerState num_input_tokens_seen not updating are different components and bugs." + }, + { + "left": "issue:43606", + "right": "issue:45859", + "accept": false, + "reason": "Both mention CPU offload, but they affect different models and different code paths." + }, + { + "left": "issue:43927", + "right": "issue:44912", + "accept": false, + "reason": "DiaConfig token IDs lost on save/load and MXFP4 quantization fallback are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3OmniMoe video-processing ValueError and DiffusionGemma pipeline failure are different multimodal problems." + }, + { + "left": "issue:28282", + "right": "issue:29870", + "accept": false, + "reason": "Missing PyTorch import for AutoModel and Bart generate() decoder_attention_mask error are unrelated." + }, + { + "left": "issue:43454", + "right": "issue:43577", + "accept": false, + "reason": "AyaVision lm_head tying and Blip2 dtype propagation are separate model-loading bugs." + }, + { + "left": "issue:43525", + "right": "issue:46291", + "accept": false, + "reason": "Llama4Config missing pad_token_id and transformers.pipelines import failure with kernels v0.15 are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45356", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different models and different tokenizer behaviors; 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": 23, + "review_unit_index": 22, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:24643", + "issue:27957", + "issue:28282", + "issue:29127", + "issue:29159", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:41720", + "issue:42175", + "issue:42915", + "issue:43404", + "issue:43425", + "issue:43452", + "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:44792", + "issue:44849", + "issue:44857", + "issue:44936", + "issue:44964", + "issue:44987", + "issue:44991", + "issue:45127", + "issue:45137", + "issue:45357", + "issue:45405", + "issue:45488", + "issue:45538", + "issue:45701", + "issue:45907", + "issue:46291", + "issue:46512", + "issue:46693", + "issue:46821", + "issue:46832", + "issue:46869" + ], + "soft_pairs": [ + "issue:44987|issue:45538", + "issue:28282|issue:34689", + "issue:45538|issue:46869", + "issue:44857|issue:46832", + "issue:43697|issue:45357", + "issue:43425|issue:46291", + "issue:33357|issue:34567", + "issue:43404|issue:45127", + "issue:42915|issue:45538", + "issue:45137|issue:45405", + "issue:42175|issue:43824", + "issue:28282|issue:29127", + "issue:44062|issue:44991", + "issue:44479|issue:46693", + "issue:43742|issue:44291", + "issue:44936|issue:46821", + "issue:43577|issue:46512", + "issue:42915|issue:43749", + "issue:44849|issue:45701", + "issue:43452|issue:43582", + "issue:27957|issue:28282", + "issue:24643|issue:29159", + "issue:44987|issue:45907", + "issue:41720|issue:43540", + "issue:42175|issue:44488", + "issue:44568|issue:45488", + "issue:43540|issue:44560", + "issue:44792|issue:44964" + ], + "prepared_review_unit_hash": "5828c21ec554c57af9a86fe82dc79bc3b7c615ed4db46a88e3474ded5f9194ad", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36012, + "estimated_input_tokens": 9003, + "estimated_eval_tokens": 18262 + }, + "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 broad thematic overlaps (tokenizers, loading, multimodal models, training) rather than true duplicates. I would not merge any of the candidate pairs.", + "confidence": 0.95, + "canonical_issue_reason": "No single canonical issue emerges from this set; the items describe unrelated bugs and regressions, so any one issue would be only an arbitrary representative.", + "canonical_pr_reason": null, + "best_issue_reason": "If a representative issue is needed, issue #45488 is the strongest candidate because it is a detailed, high-signal regression report with active discussion and broader impact on DeepSeek V3/R1 tokenization.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44987", + "right": "issue:45538", + "accept": false, + "reason": "Both are version-related regressions, but one is a model-loading failure and the other is a tokenizer max-length bug; different code paths and symptoms." + }, + { + "left": "issue:28282", + "right": "issue:34689", + "accept": false, + "reason": "These are distinct loading failures for different models and causes; one is a missing PyTorch dependency error, the other is a specific model-loading regression." + }, + { + "left": "issue:45538", + "right": "issue:46869", + "accept": false, + "reason": "Unrelated problems: tokenizer max_length behavior versus multi-GPU device-map failure in YOLO." + }, + { + "left": "issue:44857", + "right": "issue:46832", + "accept": false, + "reason": "Different RT-DETR issues: one is an AMP/CUDA crash in the loss, the other is a forward-pass shape/feature-level mismatch." + }, + { + "left": "issue:43697", + "right": "issue:45357", + "accept": false, + "reason": "Both involve model behavior regressions, but one is RT-DETR output differences and the other is Qwen3.5 save_pretrained key serialization." + }, + { + "left": "issue:43425", + "right": "issue:46291", + "accept": false, + "reason": "One is a Torch-version compatibility complaint, the other is an import-time kernels API requirement; different failure modes." + }, + { + "left": "issue:33357", + "right": "issue:34567", + "accept": false, + "reason": "A MacOS bus error for CLIP community models is unrelated to TrainerState token-count tracking." + }, + { + "left": "issue:43404", + "right": "issue:45127", + "accept": false, + "reason": "Both mention tying weights, but they are different bugs: one is a missing lm_head tie in Mistral3, the other is collapse after LoRA merge with extended vocab." + }, + { + "left": "issue:42915", + "right": "issue:45538", + "accept": false, + "reason": "Fine-grained FP8 failure in Qwen3Moe is unrelated to CLIPTokenizer max_length behavior." + }, + { + "left": "issue:45137", + "right": "issue:45405", + "accept": false, + "reason": "DeepSpeed ZeRO3 deque error and a PEFT version pin issue are unrelated." + }, + { + "left": "issue:42175", + "right": "issue:43824", + "accept": false, + "reason": "Packaging/backend selection for torch install does not match an import error for Qwen2_5_VL." + }, + { + "left": "issue:28282", + "right": "issue:29127", + "accept": false, + "reason": "A missing PyTorch dependency error is unrelated to LayoutLMv3's box-content validation/error-message issue." + }, + { + "left": "issue:44062", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-related, but one is an AddedToken argument collision and the other is a model-specific tokenizer loading regression." + }, + { + "left": "issue:44479", + "right": "issue:46693", + "accept": false, + "reason": "A video-input regression in Qwen VL models is not the same as an FA2 long-context performance regression." + }, + { + "left": "issue:43742", + "right": "issue:44291", + "accept": false, + "reason": "Loading MobileLLM-125M fails for a different reason than init_empty_weights rejecting an unexpected internal argument." + }, + { + "left": "issue:44936", + "right": "issue:46821", + "accept": false, + "reason": "Trainer evaluate-after-train behavior is unrelated to the TimesFM window_size attribute error." + }, + { + "left": "issue:43577", + "right": "issue:46512", + "accept": false, + "reason": "One bug is incorrect module dtypes in BLIP2 loading; the other is stale config dtype after from_config. Similar topic, different defect." + }, + { + "left": "issue:42915", + "right": "issue:43749", + "accept": false, + "reason": "Qwen3Moe FP8 config failure and FSDP CPU RAM efficient loading are separate training/serialization issues." + }, + { + "left": "issue:44849", + "right": "issue:45701", + "accept": false, + "reason": "Qwen3.5 output_hidden_states regression is not the same as a tokenizer-version change issue." + }, + { + "left": "issue:43452", + "right": "issue:43582", + "accept": false, + "reason": "GGUF loading problems and an Apple Silicon caching_allocator_warmup TypeError do not share the same underlying bug." + }, + { + "left": "issue:27957", + "right": "issue:28282", + "accept": false, + "reason": "XLMRoberta Flash Attention 2 support and a missing PyTorch import error are unrelated." + }, + { + "left": "issue:24643", + "right": "issue:29159", + "accept": false, + "reason": "DeepSpeed training weight-shape runtime error is unrelated to tokenizer fast/slow inconsistency." + }, + { + "left": "issue:44987", + "right": "issue:45907", + "accept": false, + "reason": "The first is a model loading regression; the second is a Hub utility bug about forwarding revision. Same ecosystem, different code paths." + }, + { + "left": "issue:41720", + "right": "issue:43540", + "accept": false, + "reason": "Qwen3 auto device-map CUDA assert and Qwen3OmniMoe video processing ValueError are different failures." + }, + { + "left": "issue:42175", + "right": "issue:44488", + "accept": false, + "reason": "Backend-install packaging issue and a model-loading failure for cjvt/sleng-bert are unrelated." + }, + { + "left": "issue:44568", + "right": "issue:45488", + "accept": false, + "reason": "Both are tokenizer regressions, but one is missing BOS/EOS on add_special_tokens and the other is overriding tokenizer.json pre-tokenizer behavior; not the same bug." + }, + { + "left": "issue:43540", + "right": "issue:44560", + "accept": false, + "reason": "Different Qwen multimodal failures: one is video-input processing in Qwen3OmniMoe, the other is a StopIteration in Qwen3-vl-embedding." + }, + { + "left": "issue:44792", + "right": "issue:44964", + "accept": false, + "reason": "Janus image-generation test failure and Phi-4 multimodal model loading failure are not the same 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": 23, + "review_unit_index": 23, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:24643", + "issue:29870", + "issue:34567", + "issue:36331", + "issue:38175", + "issue:43257", + "issue:43452", + "issue:43531", + "issue:43540", + "issue:43577", + "issue:43582", + "issue:43697", + "issue:43716", + "issue:43742", + "issue:43756", + "issue:43906", + "issue:44062", + "issue:44291", + "issue:44479", + "issue:44488", + "issue:44568", + "issue:44589", + "issue:44625", + "issue:44737", + "issue:44821", + "issue:44877", + "issue:44898", + "issue:45042", + "issue:45072", + "issue:45081", + "issue:45325", + "issue:45357", + "issue:45381", + "issue:45405", + "issue:45600", + "issue:45672", + "issue:45674", + "issue:45823", + "issue:45920", + "issue:46399", + "issue:46463", + "issue:46468", + "issue:46491", + "issue:46693", + "issue:46701", + "issue:46821", + "issue:46829" + ], + "soft_pairs": [ + "issue:44479|issue:45381", + "issue:38175|issue:44898", + "issue:44291|issue:44821", + "issue:43540|issue:44479", + "issue:34567|issue:36331", + "issue:43257|issue:43756", + "issue:44062|issue:45081", + "issue:45042|issue:45081", + "issue:43582|issue:46821", + "issue:45672|issue:45674", + "issue:43742|issue:44589", + "issue:43756|issue:46399", + "issue:43756|issue:45325", + "issue:44568|issue:46491", + "issue:45405|issue:45600", + "issue:43716|issue:46701", + "issue:43577|issue:45072", + "issue:43531|issue:44821", + "issue:43906|issue:46468", + "issue:24643|issue:29870", + "issue:44737|issue:46463", + "issue:43906|issue:46693", + "issue:45823|issue:46463", + "issue:43452|issue:44488", + "issue:43697|issue:46829", + "issue:45357|issue:45920", + "issue:44625|issue:44877" + ], + "prepared_review_unit_hash": "8aac51839931c7cf4cf902b3d3eb82ea265b88f377945f6089428636dc7b4dde", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 27, + "serialized_chars": 35593, + "estimated_input_tokens": 8899, + "estimated_eval_tokens": 18054 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of semantically related but mostly distinct Transformers bugs spanning tokenizers, model/config loading, trainer behavior, dtype/device mismatches, and Qwen video regressions. None of the soft pairs look like the same concrete underlying bug or fix.", + "confidence": 0.91, + "canonical_issue_reason": "If a single issue must anchor the cluster, issue:44479 is the strongest representative: it is a concrete regression report, names the affected models and version, and has the clearest bug scope. That said, the overall set is heterogeneous rather than a true duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue:44479 is the best issue candidate overall because it is regression-focused, well-scoped, and actionable, with explicit affected models and version context.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44479", + "right": "issue:45381", + "accept": false, + "reason": "Both concern Qwen video behavior, but 44479 is a broader v5.3.0 regression across several Qwen families while 45381 is a specific qwen2.5-vl position_ids bug; different concrete failures." + }, + { + "left": "issue:38175", + "right": "issue:44898", + "accept": false, + "reason": "Different model families and failure modes: zero probabilities in SigLIP2 vs non-default-resolution Perceiver classification failure." + }, + { + "left": "issue:44291", + "right": "issue:44821", + "accept": false, + "reason": "One is an init_empty_weights / unexpected argument error, the other is AutoImageProcessor URL loading; unrelated code paths." + }, + { + "left": "issue:43540", + "right": "issue:44479", + "accept": false, + "reason": "Both involve video inputs, but they target different model families and different bugs; not the same fixable issue." + }, + { + "left": "issue:34567", + "right": "issue:36331", + "accept": false, + "reason": "Both are trainer-related, but one is token counting not updating and the other is a compute_loss signature break; distinct problems." + }, + { + "left": "issue:43257", + "right": "issue:43756", + "accept": false, + "reason": "Different domains: DeepSpeed/accelerate weight conversion vs Smollm3 RoPE layer omission." + }, + { + "left": "issue:44062", + "right": "issue:45081", + "accept": false, + "reason": "Tokenizer AddedToken keyword collision and Mistral regex patch crash are separate tokenizer-loading bugs." + }, + { + "left": "issue:45042", + "right": "issue:45081", + "accept": false, + "reason": "PIL backend torchvision dependency regression is unrelated to the Mistral regex AttributeError." + }, + { + "left": "issue:43582", + "right": "issue:46821", + "accept": false, + "reason": "Apple Silicon allocator warmup TypeError and TimesFM window_size AttributeError are different API/runtime failures." + }, + { + "left": "issue:45672", + "right": "issue:45674", + "accept": false, + "reason": "Both touch 4-bit inference, but one is Gemma4 audio-module dtype handling and the other is BitsAndBytes skip_modules clearing a default exclusion; different bugs." + }, + { + "left": "issue:43742", + "right": "issue:44589", + "accept": false, + "reason": "Separate model-loading failures: key error for MobileLLM-125M versus missing Float8 storage type." + }, + { + "left": "issue:43756", + "right": "issue:46399", + "accept": false, + "reason": "RoPE-layer dropping in Smollm3 and Qwen3.5 missing decorator for apply_rotary_pos_emb are unrelated." + }, + { + "left": "issue:43756", + "right": "issue:45325", + "accept": false, + "reason": "Different Qwen branches and different defects: Smollm3 RoPE-layer count vs Qwen2.5-VL still-image temporal position scaling." + }, + { + "left": "issue:44568", + "right": "issue:46491", + "accept": false, + "reason": "Tokenizer regressions, but not the same one: missing BOS/EOS insertion versus CodeLlama leading-space round-trip loss." + }, + { + "left": "issue:45405", + "right": "issue:45600", + "accept": false, + "reason": "One is a dependency version bump problem; the other is stale auto-mapping references breaking CI." + }, + { + "left": "issue:43716", + "right": "issue:46701", + "accept": false, + "reason": "Mistral-3 image preprocessor dtype mismatch and DiffusionGemma image-text-to-text pipeline failure are different paths and symptoms." + }, + { + "left": "issue:43577", + "right": "issue:45072", + "accept": false, + "reason": "Both are dtype mismatches, but they affect different model architectures and loading/inference flows; not one shared bug." + }, + { + "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": "The first is an isolated reproduction of an existing issue, while the Nemotron example breakage is a different model/example failure." + }, + { + "left": "issue:24643", + "right": "issue:29870", + "accept": false, + "reason": "DeepSpeed training weight-shape runtime error and BART generate decoder_attention_mask initialization bug are different code paths." + }, + { + "left": "issue:44737", + "right": "issue:46463", + "accept": false, + "reason": "XLNet CPU device placement bug and T5 decoder-input validation error message are not the same underlying issue." + }, + { + "left": "issue:43906", + "right": "issue:46693", + "accept": false, + "reason": "One is a reproduction of another issue, the other is a long-context FA2 performance regression; different bugs." + }, + { + "left": "issue:45823", + "right": "issue:46463", + "accept": false, + "reason": "Gemma4 PLE device mismatch during forward and T5 decoder validation messaging are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:44488", + "accept": false, + "reason": "gguf_file breaking tokenizer/model.from_pretrained and a specific sleng-bert load failure are different loading issues." + }, + { + "left": "issue:43697", + "right": "issue:46829", + "accept": false, + "reason": "RTDetrV2 output divergence in v5 and VideoPrism hidden_states/last_hidden_state mix-up are distinct output-shape/semantics bugs." + }, + { + "left": "issue:45357", + "right": "issue:45920", + "accept": false, + "reason": "Qwen3.5 visual encoder save_pretrained key bug and AutoTokenizer wrong token IDs across many models are unrelated regressions." + }, + { + "left": "issue:44625", + "right": "issue:44877", + "accept": false, + "reason": "num_labels propagation in Qwen3.5 config and strict config loading for granite_speech are different configuration 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": 23, + "review_unit_index": 24, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28282", + "issue:32090", + "issue:38175", + "issue:41762", + "issue:43295", + "issue:43452", + "issue:43525", + "issue:43526", + "issue:43540", + "issue:43643", + "issue:43653", + "issue:43688", + "issue:43697", + "issue:43720", + "issue:43749", + "issue:43761", + "issue:43819", + "issue:43994", + "issue:44361", + "issue:44442", + "issue:44466", + "issue:44493", + "issue:44737", + "issue:44811", + "issue:44843", + "issue:44898", + "issue:44913", + "issue:44936", + "issue:45005", + "issue:45200", + "issue:45216", + "issue:45356", + "issue:45372", + "issue:45375", + "issue:45479", + "issue:45542", + "issue:45588", + "issue:45663", + "issue:46050", + "issue:46121", + "issue:46170", + "issue:46178", + "issue:46463", + "issue:46693", + "issue:46701", + "issue:46829", + "issue:46846", + "issue:46899" + ], + "soft_pairs": [ + "issue:45663|issue:46846", + "issue:43540|issue:46693", + "issue:44913|issue:46121", + "issue:44737|issue:45588", + "issue:28282|issue:32090", + "issue:46463|issue:46899", + "issue:43697|issue:45216", + "issue:43526|issue:45479", + "issue:43720|issue:44898", + "issue:43643|issue:45375", + "issue:43761|issue:46829", + "issue:45200|issue:46463", + "issue:44466|issue:45005", + "issue:44361|issue:44442", + "issue:44493|issue:46050", + "issue:43653|issue:44361", + "issue:41762|issue:44936", + "issue:43452|issue:44843", + "issue:43688|issue:43749", + "issue:43819|issue:44811", + "issue:38175|issue:43295", + "issue:43994|issue:44898", + "issue:43525|issue:46701", + "issue:45372|issue:46170", + "issue:45542|issue:46178", + "issue:43295|issue:45356" + ], + "prepared_review_unit_hash": "f6f331295973e96c9f0aea7e9449a134502b6ef46d2c345c539f115a0070e7ac", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 36199, + "estimated_input_tokens": 9050, + "estimated_eval_tokens": 18356 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a grab bag of unrelated bug reports spanning models, tokenizers, training, loading, and backend issues. None of the soft pairs look like the same underlying defect, so all candidate duplicate edges are rejected.", + "confidence": 0.73, + "canonical_issue_reason": "issue:43749 is the strongest standalone report here: it has the most discussion/inbound refs and a concrete, high-impact training failure, making it the best anchor if one issue must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43749 is the clearest and most actionable issue in the set, with broad training impact and enough context to serve as the representative item.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45663", + "right": "issue:46846", + "accept": false, + "reason": "Different models and different failure modes: Gemma-4 FSDP2 attention state vs Qwen3.5 TP plan entries." + }, + { + "left": "issue:43540", + "right": "issue:46693", + "accept": false, + "reason": "One is a video-input processing ValueError; the other is a long-context FA2 performance regression." + }, + { + "left": "issue:44913", + "right": "issue:46121", + "accept": false, + "reason": "Both are rope-related, but one is config reload persistence and the other is a validation helper type error." + }, + { + "left": "issue:44737", + "right": "issue:45588", + "accept": false, + "reason": "XLNet CPU tensor creation and flash-attention aux-state handling are unrelated code paths." + }, + { + "left": "issue:28282", + "right": "issue:32090", + "accept": false, + "reason": "PyTorch import availability vs Trainer GPU broadcast NoneType handling are completely different failures." + }, + { + "left": "issue:46463", + "right": "issue:46899", + "accept": false, + "reason": "Incorrect T5 error messaging is unrelated to Gemma-4 4-bit multimodal dtype casting." + }, + { + "left": "issue:43697", + "right": "issue:45216", + "accept": false, + "reason": "Different issues in different paths: RTDetrV2 output mismatch vs Qwen3.5 save_pretrained regression." + }, + { + "left": "issue:43526", + "right": "issue:45479", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to sequence-classification zero-loss behavior." + }, + { + "left": "issue:43720", + "right": "issue:44898", + "accept": false, + "reason": "Packed-weight loading in BitNet and Perceiver positional interpolation are distinct problems." + }, + { + "left": "issue:43643", + "right": "issue:45375", + "accept": false, + "reason": "A generic AutoConfig trust_remote_code field-loss issue is not the same as a specific Qwen3_5MoeVisionConfig field omission." + }, + { + "left": "issue:43761", + "right": "issue:46829", + "accept": false, + "reason": "Both mention hidden states, but the models and concrete bugs differ; one omits hidden_states, the other misassigns last_hidden_state." + }, + { + "left": "issue:45200", + "right": "issue:46463", + "accept": false, + "reason": "Gemma 4 mm_token_type_ids defaulting is unrelated to T5 decoder-input validation messaging." + }, + { + "left": "issue:44466", + "right": "issue:45005", + "accept": false, + "reason": "Both touch tied weights, but one is device-dependent serialization and the other is a broader translation-model tied-weights issue." + }, + { + "left": "issue:44361", + "right": "issue:44442", + "accept": false, + "reason": "Different tokenizer failures affecting different models and code paths." + }, + { + "left": "issue:44493", + "right": "issue:46050", + "accept": false, + "reason": "Unexpected position-id keys and sinq quantization save/load breakage are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44361", + "accept": false, + "reason": "BigBird special-token registration and MLukeTokenizer task AttributeError are not the same defect." + }, + { + "left": "issue:41762", + "right": "issue:44936", + "accept": false, + "reason": "Gemma3 ZeRO-3 loading IndexError and trainer.evaluate() post-train failure are different training/runtime issues." + }, + { + "left": "issue:43452", + "right": "issue:44843", + "accept": false, + "reason": "Both concern from_pretrained loading, but one is gguf_file support and the other is an unconditional model_info() offline-mode bug." + }, + { + "left": "issue:43688", + "right": "issue:43749", + "accept": false, + "reason": "Auxiliary-loss normalization in MoE models is unrelated to FSDP CPU RAM efficient loading." + }, + { + "left": "issue:43819", + "right": "issue:44811", + "accept": false, + "reason": "DAC latent reconstruction mismatch and Whisper batch_decode skip_special_tokens behavior are unrelated." + }, + { + "left": "issue:38175", + "right": "issue:43295", + "accept": false, + "reason": "Both mention Siglip/processor flows, but one is zero probabilities from a model and the other is a processor.tokenizer regression." + }, + { + "left": "issue:43994", + "right": "issue:44898", + "accept": false, + "reason": "Siglip2 nonsensical outputs and Perceiver non-default-resolution failures are different model-specific bugs." + }, + { + "left": "issue:43525", + "right": "issue:46701", + "accept": false, + "reason": "Llama4Config pad_token_id and DiffusionGemma image-text pipeline failures are unrelated." + }, + { + "left": "issue:45372", + "right": "issue:46170", + "accept": false, + "reason": "Gemma 4 processor import breakage from mistral_common is unrelated to Windows ROCm torch.distributed import/type issues." + }, + { + "left": "issue:45542", + "right": "issue:46178", + "accept": false, + "reason": "Different backend resolution failures: tensorboard-without-TF vs tensorflow_text backend mapping." + }, + { + "left": "issue:43295", + "right": "issue:45356", + "accept": false, + "reason": "Processor.tokenizer/image passing regression and Kimi-K2.5 codec/fix_mistral_regex regression are different bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 25, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29159", + "issue:32090", + "issue:33357", + "issue:33666", + "issue:36010", + "issue:41720", + "issue:42898", + "issue:43452", + "issue:43504", + "issue:43531", + "issue:43550", + "issue:43577", + "issue:43643", + "issue:43646", + "issue:43653", + "issue:43688", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43756", + "issue:43927", + "issue:44464", + "issue:44466", + "issue:44479", + "issue:44484", + "issue:44554", + "issue:44561", + "issue:44568", + "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:46491", + "issue:46612", + "issue:46693", + "issue:46821" + ], + "soft_pairs": [ + "issue:43531|issue:43742", + "issue:42898|issue:46491", + "issue:29159|issue:33357", + "issue:43653|issue:44568", + "issue:43646|issue:43749", + "issue:43452|issue:43577", + "issue:43688|issue:45704", + "issue:43756|issue:46402", + "issue:44843|issue:44991", + "issue:43653|issue:43927", + "issue:44464|issue:44743", + "issue:43688|issue:45701", + "issue:44484|issue:44704", + "issue:45491|issue:46002", + "issue:44479|issue:44625", + "issue:44821|issue:46821", + "issue:43688|issue:45440", + "issue:33666|issue:36010", + "issue:32090|issue:33666", + "issue:44734|issue:46612", + "issue:44466|issue:46050", + "issue:43504|issue:43720", + "issue:44977|issue:46693", + "issue:43504|issue:43550", + "issue:44464|issue:44977", + "issue:41720|issue:44464", + "issue:43643|issue:44561", + "issue:43577|issue:44554" + ], + "prepared_review_unit_hash": "7b5177315b6760faf1d22060e592e245537f8ae6e83df0876ac66755cc388465", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35540, + "estimated_input_tokens": 8885, + "estimated_eval_tokens": 18026 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a grab-bag cluster of mostly unrelated issue reports; the soft-linked pairs are only superficially similar, so none should be merged as duplicates.", + "confidence": 0.89, + "canonical_issue_reason": "issue:44464 is the strongest representative by connectivity and clarity: it describes a concrete, reproducible generation inconsistency in compiled-forward chunked generation, whereas the other items are more specialized or clearly orthogonal.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44464 is the best issue candidate overall because it is a precise, high-signal bug report with a concrete code-path regression and the strongest soft-link presence in the set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43531", + "right": "issue:43742", + "accept": false, + "reason": "Different failures and code paths: Qwen3-MoE `sliding_window` behavior vs a MobileLLM pretrained-loading key error." + }, + { + "left": "issue:42898", + "right": "issue:46491", + "accept": false, + "reason": "Both are tokenizer regressions, but one is `clean_up_tokenization_spaces` semantics in v5 and the other is CodeLlama leading-space round-trip loss." + }, + { + "left": "issue:29159", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer inconsistency vs MacOS CLIP bus error; unrelated symptoms and root causes." + }, + { + "left": "issue:43653", + "right": "issue:44568", + "accept": false, + "reason": "Both involve special tokens, but BigBird mask-token registration and mDeBERTa BOS/EOS insertion are different bugs." + }, + { + "left": "issue:43646", + "right": "issue:43749", + "accept": false, + "reason": "`trust_remote_code` config-field loss is unrelated to FSDP CPU RAM-efficient loading." + }, + { + "left": "issue:43452", + "right": "issue:43577", + "accept": false, + "reason": "GGUF/AutoTokenizer-AutoModel loading failure and BLIP2 dtype propagation are separate subsystems." + }, + { + "left": "issue:43688", + "right": "issue:45704", + "accept": false, + "reason": "Aux-loss normalization is unrelated to T5's `apex.FusedRMSNorm` memory leak." + }, + { + "left": "issue:43756", + "right": "issue:46402", + "accept": false, + "reason": "Smollm3 RoPE-layer selection and Sam3 text-encoder weight loading are different model-specific issues." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Both are tokenizer-loading problems, but offline `model_info()` access and EMBEDDIA tokenizer breakage have different causes." + }, + { + "left": "issue:43653", + "right": "issue:43927", + "accept": false, + "reason": "Special-token registration vs custom token IDs lost on save/load; not the same bug." + }, + { + "left": "issue:44464", + "right": "issue:44743", + "accept": false, + "reason": "Both affect generation/cache behavior, but compiled-forward chunked generation mismatch and recurrent-state reset are distinct code paths." + }, + { + "left": "issue:43688", + "right": "issue:45701", + "accept": false, + "reason": "Aux-loss normalization and broad tokenization-behavior change are unrelated." + }, + { + "left": "issue:44484", + "right": "issue:44704", + "accept": false, + "reason": "Checkpoint shard-size policy is unrelated to `AutoProcessor` kwargs forwarding." + }, + { + "left": "issue:45491", + "right": "issue:46002", + "accept": false, + "reason": "Both mention NaNs, but Gemma3 sliding-window attention and classification-head init/checkout are different failures." + }, + { + "left": "issue:44479", + "right": "issue:44625", + "accept": false, + "reason": "Qwen3.5 video-input regression and `num_labels` propagation are separate issues." + }, + { + "left": "issue:44821", + "right": "issue:46821", + "accept": false, + "reason": "AutoImageProcessor URL loading and TimesFM `window_size` attribute error are unrelated." + }, + { + "left": "issue:43688", + "right": "issue:45440", + "accept": false, + "reason": "Aux-loss normalization is unrelated to Native DeepseekV3MoE divergence." + }, + { + "left": "issue:33666", + "right": "issue:36010", + "accept": false, + "reason": "Multi-GPU training and a `GenerationMixin` import error are different problems." + }, + { + "left": "issue:32090", + "right": "issue:33666", + "accept": false, + "reason": "Trainer GPU broadcast NoneType error is unrelated to Qwen2-VL multi-GPU training." + }, + { + "left": "issue:44734", + "right": "issue:46612", + "accept": false, + "reason": "Both involve generation state, but server KV-cache continuation indexing and beam-search cache reorder are not the same bug." + }, + { + "left": "issue:44466", + "right": "issue:46050", + "accept": false, + "reason": "Device-dependent tied-weight serialization differs from quantized-model save/load corruption." + }, + { + "left": "issue:43504", + "right": "issue:43720", + "accept": false, + "reason": "Beit legacy-field loading and BitNet packed-weight unpacking during accelerate loading are distinct issues." + }, + { + "left": "issue:44977", + "right": "issue:46693", + "accept": false, + "reason": "Qwen3.5 flash-attention correctness and FA2 long-context performance regression are different failure modes." + }, + { + "left": "issue:43504", + "right": "issue:43550", + "accept": false, + "reason": "Beit pretrained loading and Bamba torch.compile/SDPA failure are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:44977", + "accept": false, + "reason": "Chunked-generation inconsistency and Qwen3.5 flash-attention regression are different code paths." + }, + { + "left": "issue:41720", + "right": "issue:44464", + "accept": false, + "reason": "Auto device mapping cuda assert on A800 is unrelated to compiled-forward chunked generation output mismatch." + }, + { + "left": "issue:43643", + "right": "issue:44561", + "accept": false, + "reason": "`trust_remote_code` missing-fields behavior and removal of `is_torch_fx_available` are distinct compatibility issues." + }, + { + "left": "issue:43577", + "right": "issue:44554", + "accept": false, + "reason": "BLIP2 dtype propagation and MPS attention correctness are unrelated." + }, + { + "left": "issue:44843", + "right": "issue:44991", + "accept": false, + "reason": "Tokenizer offline-path failure and EMBEDDIA tokenizer loading regression are not the same root cause." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 26, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28218", + "issue:33357", + "issue:33666", + "issue:36331", + "issue:39692", + "issue:42796", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43452", + "issue:43493", + "issue:43504", + "issue:43525", + "issue:43572", + "issue:43606", + "issue:43611", + "issue:43646", + "issue:43688", + "issue:43701", + "issue:43720", + "issue:43782", + "issue:43825", + "issue:43844", + "issue:43927", + "issue:43994", + "issue:44361", + "issue:44451", + "issue:44466", + "issue:44514", + "issue:44554", + "issue:44568", + "issue:44610", + "issue:44625", + "issue:44849", + "issue:44964", + "issue:44977", + "issue:45290", + "issue:45405", + "issue:45584", + "issue:45823", + "issue:46050", + "issue:46192", + "issue:46209", + "issue:46365", + "issue:46489", + "issue:46693", + "issue:46710", + "issue:46846" + ], + "soft_pairs": [ + "issue:43611|issue:46846", + "issue:43720|issue:43927", + "issue:43504|issue:44610", + "issue:33666|issue:36331", + "issue:43606|issue:43927", + "issue:43278|issue:43688", + "issue:43782|issue:46693", + "issue:43262|issue:43295", + "issue:43720|issue:44361", + "issue:44514|issue:45290", + "issue:44964|issue:45405", + "issue:43493|issue:43825", + "issue:44554|issue:46365", + "issue:43525|issue:43572", + "issue:44977|issue:46209", + "issue:44625|issue:45584", + "issue:43452|issue:44451", + "issue:28218|issue:33357", + "issue:45823|issue:46050", + "issue:42796|issue:44977", + "issue:43994|issue:46192", + "issue:43493|issue:43844", + "issue:39692|issue:43295", + "issue:43278|issue:43844", + "issue:43646|issue:44466", + "issue:46209|issue:46710", + "issue:33357|issue:33666", + "issue:44849|issue:46209", + "issue:44568|issue:46489", + "issue:43278|issue:43701" + ], + "prepared_review_unit_hash": "1df35fef2f8a05068eb4ff352756d858ff2d629f3851027c611600476e3fb5f9", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36787, + "estimated_input_tokens": 9197, + "estimated_eval_tokens": 18650 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly distinct bug reports that are only loosely related by broad themes like tokenization, model loading, and multimodal regressions. I did not find any soft pair that looks like the same concrete underlying bug/change. Best umbrella issue candidate is the generic model-loading regression in issue:43611.", + "confidence": 0.86, + "canonical_issue_reason": "issue:43611 is the broadest, most representative regression report here: it describes a concrete loading failure affecting models that rely on `base_model_prefix`, which makes it a good umbrella issue compared with the more model-specific reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43611 is the best issue candidate because it is a clearly scoped, general regression with likely wider impact and a clearer path to a shared fix than the other, more specialized reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43611", + "right": "issue:46846", + "accept": false, + "reason": "Different failures: generic model-loading regression vs Qwen3.5 tensor-parallel `linear_attn` OOM/shape issue." + }, + { + "left": "issue:43720", + "right": "issue:43927", + "accept": false, + "reason": "Unrelated bugs: packed-weight loading in BitNet vs DiaConfig token-ID persistence after save/load." + }, + { + "left": "issue:43504", + "right": "issue:44610", + "accept": false, + "reason": "Different model paths and symptoms: Beit legacy-field load failure vs OmDet-Turbo processor/model input-size mismatch." + }, + { + "left": "issue:33666", + "right": "issue:36331", + "accept": false, + "reason": "One is multi-GPU Qwen2-VL training behavior; the other is a custom trainer signature mismatch. Not the same bug." + }, + { + "left": "issue:43606", + "right": "issue:43927", + "accept": false, + "reason": "CPU offload device mismatch in Bark is unrelated to DiaConfig token-ID loss on serialization." + }, + { + "left": "issue:43278", + "right": "issue:43688", + "accept": false, + "reason": "Different issues: dtype drift between train/eval vs auxiliary-loss normalization." + }, + { + "left": "issue:43782", + "right": "issue:46693", + "accept": false, + "reason": "Different concrete problems: Qwen3VL `weight_only` load error vs FA2 long-context performance regression." + }, + { + "left": "issue:43262", + "right": "issue:43295", + "accept": false, + "reason": "Both involve processors, but the bugs differ: audio sampling-rate default vs `processor.tokenizer`/image handling regression." + }, + { + "left": "issue:43720", + "right": "issue:44361", + "accept": false, + "reason": "BitNet packed-weight loading and MLukeTokenizer task AttributeError are unrelated." + }, + { + "left": "issue:44514", + "right": "issue:45290", + "accept": false, + "reason": "Both touch chat templates, but the failure modes differ: batched padding behavior vs assistant tool-call messages with no content." + }, + { + "left": "issue:44964", + "right": "issue:45405", + "accept": false, + "reason": "Phi-4 multimodal loading and an unreleased PEFT version bump are unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43825", + "accept": false, + "reason": "SigLIP2 implementation mismatch is not the same as the pipeline translation-task error message regression." + }, + { + "left": "issue:44554", + "right": "issue:46365", + "accept": false, + "reason": "MPS attention correctness and Trainer import failure via datasets are unrelated." + }, + { + "left": "issue:43525", + "right": "issue:43572", + "accept": false, + "reason": "Similar missing-field shape, but different configs and fields (`pad_token_id` vs `pad_token_idx`) and no shared code-path shown." + }, + { + "left": "issue:44977", + "right": "issue:46209", + "accept": false, + "reason": "Different hardware/model regressions: Qwen3.5 flash-attention generation vs FP8 inference on B200." + }, + { + "left": "issue:44625", + "right": "issue:45584", + "accept": false, + "reason": "Qwen3.5 `num_labels` propagation bug is unrelated to Whisper empty-transcription generation failure." + }, + { + "left": "issue:43452", + "right": "issue:44451", + "accept": false, + "reason": "General `gguf_file` loading breakage and a specific ScandiBERT load failure are not the same concrete bug." + }, + { + "left": "issue:28218", + "right": "issue:33357", + "accept": false, + "reason": "Tokenizer spacing after added tokens is unrelated to the MacOS bus error on a CLIP model." + }, + { + "left": "issue:45823", + "right": "issue:46050", + "accept": false, + "reason": "Gemma4 PLE device mismatch during forward is different from sinq quantization breakage after save/load." + }, + { + "left": "issue:42796", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer BPE-marker handling and Qwen3.5 flash-attention generation are unrelated." + }, + { + "left": "issue:43994", + "right": "issue:46192", + "accept": false, + "reason": "SigLIP2 incorrect outputs and `return_tensors` being ignored are separate API bugs." + }, + { + "left": "issue:43493", + "right": "issue:43844", + "accept": false, + "reason": "SigLIP2 discrepancy vs JAX and a DeepSpeed gradient regression are different problems." + }, + { + "left": "issue:39692", + "right": "issue:43295", + "accept": false, + "reason": "Documentation/example errors for SigLIP2 do not match the processor/tokenizer regression." + }, + { + "left": "issue:43278", + "right": "issue:43844", + "accept": false, + "reason": "Embedding dtype change and the ZeRO-3 gradient increase regression are distinct training issues." + }, + { + "left": "issue:43646", + "right": "issue:44466", + "accept": false, + "reason": "Custom model initialization breakage is not the same as device-dependent `lm_head.weight` serialization." + }, + { + "left": "issue:46209", + "right": "issue:46710", + "accept": false, + "reason": "FP8 inference failure on B200 and an accuracy regression after upgrade are different bugs." + }, + { + "left": "issue:33357", + "right": "issue:33666", + "accept": false, + "reason": "MacOS bus error in CLIP and Qwen2-VL multi-GPU training are unrelated." + }, + { + "left": "issue:44849", + "right": "issue:46209", + "accept": false, + "reason": "Qwen3.5 hidden-state output bug and FP8 inference failure are different issues." + }, + { + "left": "issue:44568", + "right": "issue:46489", + "accept": false, + "reason": "Both are tokenizer bugs, but they affect different models and behaviors; no shared concrete code-path is evident." + }, + { + "left": "issue:43278", + "right": "issue:43701", + "accept": false, + "reason": "Embedding dtype drift and checkpoint key mismatch 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": 23, + "review_unit_index": 27, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29942", + "issue:30333", + "issue:43257", + "issue:43299", + "issue:43334", + "issue:43504", + "issue:43572", + "issue:43575", + "issue:43606", + "issue:43653", + "issue:43824", + "issue:43844", + "issue:43873", + "issue:43927", + "issue:43937", + "issue:44112", + "issue:44291", + "issue:44361", + "issue:44464", + "issue:44488", + "issue:44493", + "issue:44568", + "issue:44589", + "issue:44792", + "issue:44843", + "issue:44849", + "issue:44869", + "issue:44964", + "issue:45072", + "issue:45200", + "issue:45229", + "issue:45245", + "issue:45357", + "issue:45405", + "issue:45520", + "issue:45759", + "issue:45812", + "issue:46050", + "issue:46285", + "issue:46399", + "issue:46424", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821", + "issue:46832", + "issue:46846", + "issue:46858" + ], + "soft_pairs": [ + "issue:44488|issue:44843", + "issue:44792|issue:45405", + "issue:43927|issue:44361", + "issue:44568|issue:45812", + "issue:43824|issue:44589", + "issue:43575|issue:45405", + "issue:44849|issue:45357", + "issue:43504|issue:43606", + "issue:43257|issue:43299", + "issue:46399|issue:46846", + "issue:45245|issue:46710", + "issue:46821|issue:46832", + "issue:44493|issue:46285", + "issue:44589|issue:45520", + "issue:43606|issue:44112", + "issue:43334|issue:43572", + "issue:43334|issue:45759", + "issue:45072|issue:46832", + "issue:43504|issue:43653", + "issue:43572|issue:44291", + "issue:44464|issue:46424", + "issue:43937|issue:46701", + "issue:46832|issue:46858", + "issue:43844|issue:44964", + "issue:45200|issue:45229", + "issue:43873|issue:46050", + "issue:44869|issue:46752", + "issue:29942|issue:30333" + ], + "prepared_review_unit_hash": "a56984f95d5d1043eb13f5685653846153579374271b37dfd6c2d2237f113946", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36302, + "estimated_input_tokens": 9076, + "estimated_eval_tokens": 18408 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is not a true duplicate set: the issues span unrelated tokenizer, loading, generation, quantization, and CI failures. All soft-pair candidates are only superficially similar and should stay separate.", + "confidence": 0.95, + "canonical_issue_reason": "No single issue is a reliable canonical because the set does not share one underlying bug; the apparent similarities are mostly broad theme overlap (model loading/tokenization/version regressions) rather than duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue cleanly represents the whole cluster. If forced, a broad regression report could be a weak proxy, but none of the items subsume the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44488", + "right": "issue:44843", + "accept": false, + "reason": "Both involve loading/tokenizer behavior, but one is a specific model load failure and the other is an HF Hub offline regression in Mistral regex patching." + }, + { + "left": "issue:44792", + "right": "issue:45405", + "accept": false, + "reason": "Janus image generation test failure and an unreleased PEFT version bump are unrelated bugs." + }, + { + "left": "issue:43927", + "right": "issue:44361", + "accept": false, + "reason": "Different models and failure modes: DiaConfig save/load token IDs vs MLukeTokenizer task AttributeError." + }, + { + "left": "issue:44568", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but one is mDeBERTa special-token handling and the other is a broad Granite token-ID regression; not the same code-path bug." + }, + { + "left": "issue:43824", + "right": "issue:44589", + "accept": false, + "reason": "Qwen2.5-VL import error and Float8 storage lookup failure are unrelated." + }, + { + "left": "issue:43575", + "right": "issue:45405", + "accept": false, + "reason": "Qwen2-57B-A14B-Instruct TP OOM is unrelated to a PEFT dependency version bump." + }, + { + "left": "issue:44849", + "right": "issue:45357", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is an output_hidden_states bug and the other is incorrect save_pretrained visual-encoder keys." + }, + { + "left": "issue:43504", + "right": "issue:43606", + "accept": false, + "reason": "BEiT legacy-field loading and bark-small CPU offload device mismatch are different subsystems and bugs." + }, + { + "left": "issue:43257", + "right": "issue:43299", + "accept": false, + "reason": "Both mention Qwen MoE, but one is accelerate+deepspeed conversion and the other is a Qwen3VL loading break in dev transformers." + }, + { + "left": "issue:46399", + "right": "issue:46846", + "accept": false, + "reason": "Qwen3.5 kernel registration crash and missing TP plan entries are separate problems with different failure points." + }, + { + "left": "issue:45245", + "right": "issue:46710", + "accept": false, + "reason": "Category-count runtime limit and an accuracy regression after upgrade are not the same bug." + }, + { + "left": "issue:46821", + "right": "issue:46832", + "accept": false, + "reason": "TimesFM window_size AttributeError and RT-DETR forward TypeError are unrelated model-specific failures." + }, + { + "left": "issue:44493", + "right": "issue:46285", + "accept": false, + "reason": "Unexpected position-id keys across many models is not the same as the CLIP text-model structure change." + }, + { + "left": "issue:44589", + "right": "issue:45520", + "accept": false, + "reason": "Float8 storage lookup failure and Python 3.13 flash_attn KeyError are different import/runtime issues." + }, + { + "left": "issue:43606", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/device-related, but bark offload mismatch and GraniteSpeech stale device override are not the same defect." + }, + { + "left": "issue:43334", + "right": "issue:43572", + "accept": false, + "reason": "Different missing-config fields on different models: Qwen3-VL pad_token_id vs StableLm pad_token_idx." + }, + { + "left": "issue:43334", + "right": "issue:45759", + "accept": false, + "reason": "Qwen3-VL loading error and composite Qwen multimodal text_config unwrapping are distinct load-path issues." + }, + { + "left": "issue:45072", + "right": "issue:46832", + "accept": false, + "reason": "bfloat16 dtype mismatch in inference and RT-DETR feature-level mismatch are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:43653", + "accept": false, + "reason": "A legacy-field pretrained-load bug and a BigBirdTokenizer special-token registration bug are not the same issue." + }, + { + "left": "issue:43572", + "right": "issue:44291", + "accept": false, + "reason": "Missing pad_token_idx after 5.0 and unexpected _is_hf_initialized argument are different 5.0 migration problems." + }, + { + "left": "issue:44464", + "right": "issue:46424", + "accept": false, + "reason": "Chunked generation inconsistency with compiled forward and cache.max_cache_length behavior are different cache/generation bugs." + }, + { + "left": "issue:43937", + "right": "issue:46701", + "accept": false, + "reason": "GLM-5 GenerationConfig validation and DiffusionGemma pipeline failure are unrelated generation-path issues." + }, + { + "left": "issue:46832", + "right": "issue:46858", + "accept": false, + "reason": "RT-DETR forward error and static-cache slice arithmetic breakage on non-CUDA backends are different code paths." + }, + { + "left": "issue:43844", + "right": "issue:44964", + "accept": false, + "reason": "ZeRO-3 gradient growth and Phi-4 multimodal loading failure are unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45229", + "accept": false, + "reason": "Gemma4 text-only fine-tuning token IDs and multi-GPU inference OOM are different problems." + }, + { + "left": "issue:43873", + "right": "issue:46050", + "accept": false, + "reason": "Quantization offloading behavior and sinq save/load breakage are related only broadly by quantization, not the same bug." + }, + { + "left": "issue:44869", + "right": "issue:46752", + "accept": false, + "reason": "Whisper word-timestamp decode crash and empty-list chat-template IndexError are unrelated." + }, + { + "left": "issue:29942", + "right": "issue:30333", + "accept": false, + "reason": "Flash Attention 2 test failures and MLFlow reporting on failed training jobs are unrelated CI/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": 23, + "review_unit_index": 28, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:15354", + "issue:28218", + "issue:29159", + "issue:29870", + "issue:29942", + "issue:32090", + "issue:34567", + "issue:37428", + "issue:38175", + "issue:41720", + "issue:42490", + "issue:43065", + "issue:43295", + "issue:43450", + "issue:43493", + "issue:43540", + "issue:43572", + "issue:43696", + "issue:43701", + "issue:43873", + "issue:43957", + "issue:43994", + "issue:44186", + "issue:44423", + "issue:44493", + "issue:44554", + "issue:44610", + "issue:44734", + "issue:44977", + "issue:45072", + "issue:45084", + "issue:45198", + "issue:45200", + "issue:45229", + "issue:45406", + "issue:45446", + "issue:45520", + "issue:45588", + "issue:45759", + "issue:45823", + "issue:46050", + "issue:46121", + "issue:46192", + "issue:46369", + "issue:46701", + "issue:46710", + "issue:46821" + ], + "soft_pairs": [ + "issue:45229|issue:45823", + "issue:43994|issue:44493", + "issue:45588|issue:46050", + "issue:42490|issue:43696", + "issue:29159|issue:32090", + "issue:44977|issue:46701", + "issue:45084|issue:45520", + "issue:28218|issue:34567", + "issue:43065|issue:46369", + "issue:44554|issue:45446", + "issue:44423|issue:44734", + "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:43295|issue:43450", + "issue:37428|issue:43957", + "issue:45588|issue:45759", + "issue:43540|issue:45200", + "issue:43295|issue:43873", + "issue:44186|issue:46192", + "issue:43572|issue:46821", + "issue:43295|issue:44610", + "issue:45823|issue:46192", + "issue:45200|issue:45823", + "issue:29870|issue:29942" + ], + "prepared_review_unit_hash": "8a4371ffabb603dbd12462e72d1ca407f12c5182eb7f51da02ec47d9d4934ce1", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35563, + "estimated_input_tokens": 8891, + "estimated_eval_tokens": 18038 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is heterogeneous: the soft links are mostly title-level similarity across unrelated bugs (tokenizers, Gemma4, serving, flash-attention, quantization, device mapping, and assorted model-specific regressions). I would not merge any of the candidate soft edges.", + "confidence": 0.96, + "canonical_issue_reason": "No single canonical issue fits the cluster; the items span unrelated bug families rather than one underlying defect.", + "canonical_pr_reason": null, + "best_issue_reason": "No globally best issue: 43873 is the broadest/open item, but it does not actually unify the other issues into one duplicate set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45229", + "right": "issue:45823", + "accept": false, + "reason": "Both are Gemma4-related, but one is CUDA OOM in multi-GPU inference and the other is a device-mismatch forward bug; different failure modes." + }, + { + "left": "issue:43994", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 output-quality issue vs generic position-id key warning; no shared code-path bug." + }, + { + "left": "issue:45588", + "right": "issue:46050", + "accept": false, + "reason": "Flash-attention sink-less crash vs sinq save/load corruption; unrelated problems." + }, + { + "left": "issue:42490", + "right": "issue:43696", + "accept": false, + "reason": "Import warning on trailing-slash model paths vs GPT-oss CUDA OOM; not the same bug." + }, + { + "left": "issue:29159", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer fast/slow inconsistency is unrelated to the Trainer broadcast NoneType error." + }, + { + "left": "issue:44977", + "right": "issue:46701", + "accept": false, + "reason": "Qwen3.5 flash-attention generation bug vs DiffusionGemma pipeline failure; different models and code paths." + }, + { + "left": "issue:45084", + "right": "issue:45520", + "accept": false, + "reason": "Template-node compilation error vs Python 3.13 flash_attn import KeyError; different root causes." + }, + { + "left": "issue:28218", + "right": "issue:34567", + "accept": false, + "reason": "Added-token spacing bug is unrelated to TrainerState token counting." + }, + { + "left": "issue:43065", + "right": "issue:46369", + "accept": false, + "reason": "Dummy Conv2d in Sam3PixelDecoder vs SAM3 litetext memory overflow; not duplicates." + }, + { + "left": "issue:44554", + "right": "issue:45446", + "accept": false, + "reason": "MPS attention correctness bug vs flex_attention version-check import bug; separate issues." + }, + { + "left": "issue:44423", + "right": "issue:44734", + "accept": false, + "reason": "Both are serve crashes, but one is multimodal .to() handling and the other is KV-cache indexing; distinct bugs." + }, + { + "left": "issue:15354", + "right": "issue:38175", + "accept": false, + "reason": "TorchScript GeneratorExp export failure vs SigLIP2 zero-probabilities behavior; unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43701", + "accept": false, + "reason": "SigLIP2 implementation discrepancy vs resume_from_checkpoint key mismatch; no common underlying defect." + }, + { + "left": "issue:41720", + "right": "issue:46710", + "accept": false, + "reason": "Qwen3 auto device-mapping CUDA assert vs DeepSeek accuracy regression; different symptoms and causes." + }, + { + "left": "issue:45406", + "right": "issue:46121", + "accept": false, + "reason": "Gemma4Processor _tokenizer attribute crash vs rope-validation TypeError; unrelated." + }, + { + "left": "issue:44977", + "right": "issue:46710", + "accept": false, + "reason": "Flash-attention generation bug and post-upgrade accuracy regression are not the same issue." + }, + { + "left": "issue:45072", + "right": "issue:45198", + "accept": false, + "reason": "bfloat16 dtype mismatch in inference vs Wav2Vec2 save/tokenization failure; different subsystems." + }, + { + "left": "issue:43295", + "right": "issue:43450", + "accept": false, + "reason": "Processor.tokenizer regression vs batched video shape bug; both processor-related but not the same code-path." + }, + { + "left": "issue:37428", + "right": "issue:43957", + "accept": false, + "reason": "Flash-support import failure vs meta-device model loading regression; different failures." + }, + { + "left": "issue:45588", + "right": "issue:45759", + "accept": false, + "reason": "Flash-attention sink-less crash vs composite Qwen multimodal config unwrapping; unrelated." + }, + { + "left": "issue:43540", + "right": "issue:45200", + "accept": false, + "reason": "Video-input processing error vs Gemma4 mm_token_type_ids defaulting issue; distinct multimodal bugs." + }, + { + "left": "issue:43295", + "right": "issue:43873", + "accept": false, + "reason": "Processor.tokenizer regression vs quantization offloading issue; no shared underlying bug." + }, + { + "left": "issue:44186", + "right": "issue:46192", + "accept": false, + "reason": "LayoutLMv2 tokenizer crash vs return_tensors being ignored when text_kwargs is passed; different tokenizer behaviors." + }, + { + "left": "issue:43572", + "right": "issue:46821", + "accept": false, + "reason": "Missing StableLmConfig pad_token_idx after update vs TimesFM window_size AttributeError; unrelated." + }, + { + "left": "issue:43295", + "right": "issue:44610", + "accept": false, + "reason": "Processor/tokenizer regression vs OmDet-Turbo image-size mismatch; separate model-specific issues." + }, + { + "left": "issue:45823", + "right": "issue:46192", + "accept": false, + "reason": "Gemma4 device mismatch vs return_tensors kwarg handling; unrelated." + }, + { + "left": "issue:45200", + "right": "issue:45823", + "accept": false, + "reason": "Text-only mm_token_type_ids defaulting bug vs Gemma4 PLE device mismatch; different concrete problems." + }, + { + "left": "issue:29870", + "right": "issue:29942", + "accept": false, + "reason": "BART decoder_attention_mask initialization bug vs Flash Attention 2 test failures; not the same defect." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 29, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28218", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:42915", + "issue:43232", + "issue:43262", + "issue:43278", + "issue:43295", + "issue:43334", + "issue:43450", + "issue:43452", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43572", + "issue:43575", + "issue:43653", + "issue:43824", + "issue:43844", + "issue:43856", + "issue:43950", + "issue:43975", + "issue:44112", + "issue:44442", + "issue:44466", + "issue:44521", + "issue:44554", + "issue:44737", + "issue:44792", + "issue:44849", + "issue:44987", + "issue:45127", + "issue:45198", + "issue:45200", + "issue:45290", + "issue:45356", + "issue:45381", + "issue:45520", + "issue:45563", + "issue:46178", + "issue:46192", + "issue:46209", + "issue:46285", + "issue:46512", + "issue:46894", + "issue:46899" + ], + "soft_pairs": [ + "issue:43550|issue:43844", + "issue:45520|issue:46178", + "issue:43450|issue:43575", + "issue:43572|issue:44987", + "issue:44554|issue:45356", + "issue:42490|issue:43493", + "issue:37428|issue:43824", + "issue:43232|issue:44849", + "issue:43295|issue:43493", + "issue:44466|issue:45127", + "issue:39692|issue:46192", + "issue:44521|issue:45290", + "issue:43653|issue:44112", + "issue:45381|issue:46209", + "issue:44737|issue:46894", + "issue:28218|issue:35141", + "issue:43278|issue:43856", + "issue:42915|issue:43824", + "issue:43950|issue:46285", + "issue:43824|issue:43975", + "issue:43452|issue:46512", + "issue:43262|issue:43450", + "issue:43334|issue:43525", + "issue:44442|issue:45198", + "issue:45200|issue:46899", + "issue:44792|issue:45563" + ], + "prepared_review_unit_hash": "330ac7c6775d7a7be0b42f42c798c80e8767c8e6b073cb6cbc63b2ee9177e0ca", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 26, + "serialized_chars": 35887, + "estimated_input_tokens": 8972, + "estimated_eval_tokens": 18200 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly heterogeneous issues across unrelated subsystems. The only plausible true duplicate is the pair about missing `pad_token_id` on config objects during model loading; the rest are distinct bugs and should stay separate.", + "confidence": 0.66, + "canonical_issue_reason": "`issue:43525` is the broadest formulation of the only likely duplicate theme here: a load-time `AttributeError` caused by configs missing `pad_token_id`.", + "canonical_pr_reason": null, + "best_issue_reason": "`issue:44521` is the strongest standalone anchor in this mixed set: it has the richest discussion/inbound references and a clear, reproducible multimodal `apply_chat_template` bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:43844", + "accept": false, + "reason": "Different failure modes in different code paths: SDPA/torch.compile on one side, ZeRO-3 gradient growth on the other." + }, + { + "left": "issue:45520", + "right": "issue:46178", + "accept": false, + "reason": "Both are import/backend errors, but they involve different missing symbols and different backend mapping problems." + }, + { + "left": "issue:43450", + "right": "issue:43575", + "accept": false, + "reason": "One is a batched video-processor shape bug; the other is a tensor-parallel OOM when loading a model." + }, + { + "left": "issue:43572", + "right": "issue:44987", + "accept": false, + "reason": "Both are load-time regressions, but the concrete symptoms and affected paths are different." + }, + { + "left": "issue:44554", + "right": "issue:45356", + "accept": false, + "reason": "MPS attention correctness and Kimi tokenizer codec handling are unrelated bugs." + }, + { + "left": "issue:42490", + "right": "issue:43493", + "accept": false, + "reason": "A trailing-slash local-path warning is unrelated to a SigLIP2 implementation discrepancy." + }, + { + "left": "issue:37428", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they are for different missing names/modules and different fixes." + }, + { + "left": "issue:43232", + "right": "issue:44849", + "accept": false, + "reason": "Both touch generation, but one is a sync_gpus kwargs update bug and the other is an output_hidden_states issue." + }, + { + "left": "issue:43295", + "right": "issue:43493", + "accept": false, + "reason": "A processor.tokenizer regression is not the same as the SigLIP2 JAX-vs-HF discrepancy." + }, + { + "left": "issue:44466", + "right": "issue:45127", + "accept": false, + "reason": "Both mention tied weights, but one is serialization behavior and the other is LoRA merging/model collapse." + }, + { + "left": "issue:39692", + "right": "issue:46192", + "accept": false, + "reason": "Documentation/example errors are not the same as `return_tensors` being ignored when `text_kwargs` is passed." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Same API family, but different concrete bugs: wrong assistant masks vs a crash on tool-call messages with no content." + }, + { + "left": "issue:43653", + "right": "issue:44112", + "accept": false, + "reason": "Tokenizer special-token registration and a stale CI test are unrelated." + }, + { + "left": "issue:45381", + "right": "issue:46209", + "accept": false, + "reason": "A Qwen2.5-VL position-id bug is unrelated to an FP8 inference failure on B200." + }, + { + "left": "issue:44737", + "right": "issue:46894", + "accept": false, + "reason": "Different model families and different bugs: CPU placement in XLNet vs a training-loss shift issue." + }, + { + "left": "issue:28218", + "right": "issue:35141", + "accept": false, + "reason": "Tokenizer spacing after added tokens is not the same bug as embedding reinitialization after resizing." + }, + { + "left": "issue:43278", + "right": "issue:43856", + "accept": false, + "reason": "A dtype drift bug during eval is not the same as an MoE training memory inefficiency." + }, + { + "left": "issue:42915", + "right": "issue:43824", + "accept": false, + "reason": "A FineGrainedFP8Config failure and an unrelated model import error are different issues." + }, + { + "left": "issue:43950", + "right": "issue:46285", + "accept": false, + "reason": "Non-persistent buffer corruption from `from_pretrained()` is distinct from CLIPTextModel structure inconsistency." + }, + { + "left": "issue:43824", + "right": "issue:43975", + "accept": false, + "reason": "Missing model import and incorrect detokenization are unrelated." + }, + { + "left": "issue:43452", + "right": "issue:46512", + "accept": false, + "reason": "Broken `gguf_file` loading and stale `config.dtype` after `from_config()` are different loading behaviors." + }, + { + "left": "issue:43262", + "right": "issue:43450", + "accept": false, + "reason": "Audio chat-template sampling-rate defaults and batched video-shape bugs are unrelated modality-specific issues." + }, + { + "left": "issue:43334", + "right": "issue:43525", + "accept": true, + "reason": "Both report the same load-time `AttributeError` for missing `pad_token_id` on config objects, likely the same regression across model configs." + }, + { + "left": "issue:44442", + "right": "issue:45198", + "accept": false, + "reason": "AutoTokenizer load failure for one speech tokenizer is not the same as save_pretrained/tokenization failure for Wav2Vec2." + }, + { + "left": "issue:44521", + "right": "issue:45290", + "accept": false, + "reason": "Both are chat-template issues, but one is incorrect masks for multimodal inputs and the other is a crash on tool-call assistant messages." + }, + { + "left": "issue:44792", + "right": "issue:45563", + "accept": false, + "reason": "A failed image-generation test for Janus is unrelated to a stale paged-generate warning about `num_return_sequences`." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet, and the soft-edge verdicts stay conservative. The lone accepted duplicate pair (43334/43525) is well supported by the shared missing-`pad_token_id` load-time AttributeError pattern, while the rest are reasonably treated as distinct bugs." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 30, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28282", + "issue:29942", + "issue:32090", + "issue:33357", + "issue:37428", + "issue:38175", + "issue:41720", + "issue:43278", + "issue:43335", + "issue:43388", + "issue:43425", + "issue:43504", + "issue:43526", + "issue:43550", + "issue:43606", + "issue:43611", + "issue:43653", + "issue:43819", + "issue:43825", + "issue:43827", + "issue:43901", + "issue:43931", + "issue:44112", + "issue:44206", + "issue:44448", + "issue:44561", + "issue:44610", + "issue:44743", + "issue:44855", + "issue:44857", + "issue:44898", + "issue:45083", + "issue:45200", + "issue:45325", + "issue:45412", + "issue:45520", + "issue:45584", + "issue:45593", + "issue:45759", + "issue:46170", + "issue:46192", + "issue:46209", + "issue:46234", + "issue:46285", + "issue:46424", + "issue:46512", + "issue:46821", + "issue:46858" + ], + "soft_pairs": [ + "issue:43819|issue:44112", + "issue:44898|issue:45412", + "issue:43526|issue:46192", + "issue:45200|issue:45325", + "issue:43827|issue:43901", + "issue:41720|issue:46209", + "issue:43335|issue:43550", + "issue:44561|issue:46234", + "issue:43388|issue:43526", + "issue:43819|issue:45083", + "issue:44857|issue:45593", + "issue:44610|issue:44857", + "issue:46424|issue:46858", + "issue:45520|issue:46170", + "issue:45759|issue:46285", + "issue:46512|issue:46821", + "issue:43606|issue:45083", + "issue:38175|issue:45412", + "issue:44206|issue:45584", + "issue:43335|issue:43653", + "issue:37428|issue:43425", + "issue:29942|issue:32090", + "issue:43550|issue:44112", + "issue:28282|issue:33357", + "issue:43611|issue:43931", + "issue:44448|issue:44743", + "issue:43335|issue:43504", + "issue:44855|issue:45083", + "issue:43278|issue:43825" + ], + "prepared_review_unit_hash": "1bab095a14491838123eb899db599fdece0a0855d62743b8ab242d1cc0a431a5", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36747, + "estimated_input_tokens": 9187, + "estimated_eval_tokens": 18630 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab-bag of unrelated Transformers bugs, regressions, and docs issues rather than a duplicate family. None of the soft pairs look like the same underlying defect or change.", + "confidence": 0.97, + "canonical_issue_reason": "No single duplicate family spans the cluster. If forced to pick an anchor, issue 43825 is the most representative concrete regression with references and a clear repro.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43825 is the best representative issue because it is concrete, user-facing, and well-scoped; the rest are largely unrelated bugs or documentation regressions.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43819", + "right": "issue:44112", + "accept": false, + "reason": "DAC latent reconstruction mismatch and a GraniteSpeech CI device-override failure are different code paths and symptoms." + }, + { + "left": "issue:44898", + "right": "issue:45412", + "accept": false, + "reason": "Perceiver resolution failure and RT-DETR memory release behavior are unrelated model bugs." + }, + { + "left": "issue:43526", + "right": "issue:46192", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and silent return_tensors handling are different preprocessing issues." + }, + { + "left": "issue:45200", + "right": "issue:45325", + "accept": false, + "reason": "Gemma 4 mm_token_type_ids defaults and Qwen2.5-VL rope index scaling are separate multimodal configuration bugs." + }, + { + "left": "issue:43827", + "right": "issue:43901", + "accept": false, + "reason": "Both are docs-related, but one is about removed pipeline support and the other about a changed text-classification flag; not the same issue." + }, + { + "left": "issue:41720", + "right": "issue:46209", + "accept": false, + "reason": "A800 cudaErrorAssert with auto device mapping and B200 FP8 inference failure are different hardware/model-path problems." + }, + { + "left": "issue:43335", + "right": "issue:43550", + "accept": false, + "reason": "SwitchTransformers sparse-layer construction and Bamba torch.compile+SDPA failures affect different models and execution paths." + }, + { + "left": "issue:44561", + "right": "issue:46234", + "accept": false, + "reason": "Missing is_torch_fx_available and missing is_falcon_mamba_ssm_available are different import_utils symbols, not one bug." + }, + { + "left": "issue:43388", + "right": "issue:43526", + "accept": false, + "reason": "gather_for_metrics truncating tuple labels and BeitImageProcessorFast reduce_labels behavior are unrelated label-processing bugs." + }, + { + "left": "issue:43819", + "right": "issue:45083", + "accept": false, + "reason": "DAC missing STE mismatch and qwen3_omni_moe feature-length helper behavior are different model-specific bugs." + }, + { + "left": "issue:44857", + "right": "issue:45593", + "accept": false, + "reason": "LwDetrImageLoss AMP crash and D-FINE auxiliary-loss omission are distinct training-path problems." + }, + { + "left": "issue:44610", + "right": "issue:44857", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch and LwDetrImageLoss float16 crash are unrelated components." + }, + { + "left": "issue:46424", + "right": "issue:46858", + "accept": false, + "reason": "Quantized-cache max length handling and static-cache device-tensor slice arithmetic are different cache bugs." + }, + { + "left": "issue:45520", + "right": "issue:46170", + "accept": false, + "reason": "flash_attn import lookup on Python 3.13 and Windows ROCm torch.distributed dependency issues are not the same failure." + }, + { + "left": "issue:45759", + "right": "issue:46285", + "accept": false, + "reason": "Composite Qwen multimodal config unwrapping and CLIP text-model structure changes are separate model-loading regressions." + }, + { + "left": "issue:46512", + "right": "issue:46821", + "accept": false, + "reason": "Stale config.dtype after from_config and TimesFM window_size AttributeError are unrelated API bugs." + }, + { + "left": "issue:43606", + "right": "issue:45083", + "accept": false, + "reason": "CPU offload device mismatch in bark-small and qwen3_omni_moe helper output-length behavior are different issues." + }, + { + "left": "issue:38175", + "right": "issue:45412", + "accept": false, + "reason": "Siglip2 zero probabilities and RT-DETR memory leak are different symptoms in different models." + }, + { + "left": "issue:44206", + "right": "issue:45584", + "accept": false, + "reason": "Unsupported feature-extractor arg and Whisper empty-transcription generation failure are unrelated pipeline bugs." + }, + { + "left": "issue:43335", + "right": "issue:43653", + "accept": false, + "reason": "Sparse-layer config creation and BigBirdTokenizer special-token registration are different components and failure modes." + }, + { + "left": "issue:37428", + "right": "issue:43425", + "accept": false, + "reason": "Missing flash-attention helper import and Torch 2.10 compatibility are not the same underlying problem." + }, + { + "left": "issue:29942", + "right": "issue:32090", + "accept": false, + "reason": "Flash Attention 2 test failures and Trainer _gpu_broadcast_one NoneType errors are unrelated test/runtime problems." + }, + { + "left": "issue:43550", + "right": "issue:44112", + "accept": false, + "reason": "Bamba SDPA+torch.compile and GraniteSpeech stale device-override CI failures do not share a concrete code path." + }, + { + "left": "issue:28282", + "right": "issue:33357", + "accept": false, + "reason": "PyTorch missing ImportError for AutoModel and MacOS bus error loading a community CLIP model are different environment/model issues." + }, + { + "left": "issue:43611", + "right": "issue:43931", + "accept": false, + "reason": "base_model_prefix loading breakage and Qwen3-VL weight-shape mismatch are distinct model-loading failures." + }, + { + "left": "issue:44448", + "right": "issue:44743", + "accept": false, + "reason": "Pegasus v4/v5 output differences and qwen3_5 recurrent-state reset with cache are unrelated behavior changes." + }, + { + "left": "issue:43335", + "right": "issue:43504", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation and BeitForSemanticSegmentation legacy-field loading are different model/config bugs." + }, + { + "left": "issue:44855", + "right": "issue:45083", + "accept": false, + "reason": "Python 3.13 IndentationError in DebertaV2Model import and qwen3_omni_moe helper behavior are unrelated." + }, + { + "left": "issue:43278", + "right": "issue:43825", + "accept": false, + "reason": "BF16-to-FP32 dtype drift during evaluate and pipeline() misleading translation support messaging are different 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": 23, + "review_unit_index": 31, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:26413", + "issue:28218", + "issue:29127", + "issue:31501", + "issue:32090", + "issue:33357", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:39692", + "issue:42490", + "issue:43335", + "issue:43526", + "issue:43575", + "issue:43606", + "issue:43701", + "issue:43720", + "issue:43819", + "issue:43854", + "issue:44112", + "issue:44206", + "issue:44351", + "issue:44360", + "issue:44448", + "issue:44493", + "issue:44514", + "issue:44521", + "issue:44561", + "issue:44617", + "issue:44779", + "issue:44857", + "issue:45072", + "issue:45083", + "issue:45127", + "issue:45245", + "issue:45290", + "issue:45310", + "issue:45459", + "issue:45478", + "issue:45588", + "issue:45593", + "issue:45823", + "issue:45859", + "issue:46234", + "issue:46326", + "issue:46431", + "issue:46566", + "issue:46697" + ], + "soft_pairs": [ + "issue:29127|issue:32090", + "issue:43335|issue:45083", + "issue:43720|issue:44112", + "issue:37428|issue:44351", + "issue:44514|issue:44521", + "issue:43335|issue:45593", + "issue:43335|issue:44112", + "issue:45290|issue:46326", + "issue:45127|issue:45245", + "issue:45588|issue:45859", + "issue:34689|issue:35141", + "issue:42490|issue:43854", + "issue:46431|issue:46566", + "issue:45310|issue:45478", + "issue:44448|issue:44779", + "issue:43819|issue:44493", + "issue:39692|issue:43526", + "issue:26413|issue:31501", + "issue:43606|issue:44857", + "issue:28218|issue:32090", + "issue:45459|issue:46234", + "issue:39692|issue:44493", + "issue:43701|issue:45083", + "issue:44617|issue:46697", + "issue:44206|issue:45823", + "issue:32090|issue:33357", + "issue:42490|issue:43575", + "issue:44360|issue:44561", + "issue:44857|issue:45072" + ], + "prepared_review_unit_hash": "ee9699a6e05df51cefcd3ff99a59772b5d85ecf2fa9c41f77f9b4ee6d0558a79", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36468, + "estimated_input_tokens": 9117, + "estimated_eval_tokens": 18490 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Nearly all soft pairs are false positives from broad lexical similarity. The only clear duplicate is the two identical Qwen3.5-MoE from_pretrained reports (issue:45310 and issue:45478).", + "confidence": 0.89, + "canonical_issue_reason": "issue:45310 is the better canonical: it is the earlier report of the exact same Qwen3.5-MoE from_pretrained regression, while issue:45478 appears to be a later duplicate.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the strongest representative because it is a clean, exact duplicate target with the earliest context and same concrete failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:29127", + "right": "issue:32090", + "accept": false, + "reason": "Different subsystems and failures: LayoutLMv3 error clarity vs Trainer _gpu_broadcast_one(NoneType)." + }, + { + "left": "issue:43335", + "right": "issue:45083", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug vs qwen3_omni_moe feature-length helper behavior; unrelated code paths." + }, + { + "left": "issue:43720", + "right": "issue:44112", + "accept": false, + "reason": "BitNet accelerate-loading unpacking bug vs GraniteSpeech stale device override CI failure; not the same issue." + }, + { + "left": "issue:37428", + "right": "issue:44351", + "accept": false, + "reason": "Missing flash-attention utility import vs missing HybridCache import; distinct API breakages." + }, + { + "left": "issue:44514", + "right": "issue:44521", + "accept": false, + "reason": "Both involve chat templates, but one is a batched padding=False crash and the other is zero assistant masks for multimodal inputs." + }, + { + "left": "issue:43335", + "right": "issue:45593", + "accept": false, + "reason": "SwitchTransformers sparse-layer creation vs D-FINE auxiliary-loss omission; different model families and bugs." + }, + { + "left": "issue:43335", + "right": "issue:44112", + "accept": false, + "reason": "Sparse-layer config issue vs GraniteSpeech CI device-override failure; no shared concrete bug." + }, + { + "left": "issue:45290", + "right": "issue:46326", + "accept": false, + "reason": "Both are chat-template related, but the tool-call/no-content crash is not the same as continue_final_message behavior for mistral_common tokenizers." + }, + { + "left": "issue:45127", + "right": "issue:45245", + "accept": false, + "reason": "LoRA/extended-vocab collapse vs category-cardinality runtime limit; unrelated." + }, + { + "left": "issue:45588", + "right": "issue:45859", + "accept": false, + "reason": "Flash-attention s_aux=None AttributeError vs CPU-offload incompatibility in Qwen3_5MoeTextRotaryEmbedding; different code paths." + }, + { + "left": "issue:34689", + "right": "issue:35141", + "accept": false, + "reason": "Model-loading regression for Llama 3.2 vision vs output embedding reinit after resize; not the same bug." + }, + { + "left": "issue:42490", + "right": "issue:43854", + "accept": false, + "reason": "Local-path trailing-slash warning vs GLM-4.7-Flash loading failure; different symptoms and causes." + }, + { + "left": "issue:46431", + "right": "issue:46566", + "accept": false, + "reason": "Missing chat template error vs DiffusionGemma meta-device placement bug; unrelated." + }, + { + "left": "issue:45310", + "right": "issue:45478", + "accept": true, + "reason": "Exact same title and same concrete Qwen3.5-MoE from_pretrained failure; clear duplicate pair." + }, + { + "left": "issue:44448", + "right": "issue:44779", + "accept": false, + "reason": "Pegasus v4/v5 output mismatch vs Deepseek tokenizer regression; similar area but different models and bugs." + }, + { + "left": "issue:43819", + "right": "issue:44493", + "accept": false, + "reason": "DAC from_latents/STE mismatch vs unexpected position-id key warnings across many models; different problems." + }, + { + "left": "issue:39692", + "right": "issue:43526", + "accept": false, + "reason": "SigLIP2 documentation example issues vs BeitImageProcessorFast reduce_labels bug; docs vs processor logic." + }, + { + "left": "issue:26413", + "right": "issue:31501", + "accept": false, + "reason": "resume_from_checkpoint epoch_iterator failure vs FineWeb training start failure; not the same root cause." + }, + { + "left": "issue:43606", + "right": "issue:44857", + "accept": false, + "reason": "CPU offload device mismatch in bark-small vs float16 AMP crash in LwDetrImageLoss; unrelated." + }, + { + "left": "issue:28218", + "right": "issue:32090", + "accept": false, + "reason": "Tokenizer added-space bug vs Trainer broadcast NoneType error; different components." + }, + { + "left": "issue:45459", + "right": "issue:46234", + "accept": false, + "reason": "Tokenizer protobuf error masking vs missing import_utils availability helper; unrelated." + }, + { + "left": "issue:39692", + "right": "issue:44493", + "accept": false, + "reason": "SigLIP2 documentation example bugs vs many models showing unexpected position-id keys; not the same issue." + }, + { + "left": "issue:43701", + "right": "issue:45083", + "accept": false, + "reason": "resume_from_checkpoint key mismatch vs qwen3_omni_moe helper behavior; different failure modes." + }, + { + "left": "issue:44617", + "right": "issue:46697", + "accept": false, + "reason": "Sam3Video OOM vs Swin2SR over-padding; unrelated." + }, + { + "left": "issue:44206", + "right": "issue:45823", + "accept": false, + "reason": "Unsupported center arg in LasrFeatureExtractor vs Gemma4 PLE device mismatch; different bugs." + }, + { + "left": "issue:32090", + "right": "issue:33357", + "accept": false, + "reason": "Trainer broadcast NoneType vs MacOS bus error on CLIP model load; unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43575", + "accept": false, + "reason": "Local-path import warning vs Qwen2-57B-A14B-Instruct tensor-parallel OOM; different issues." + }, + { + "left": "issue:44360", + "right": "issue:44561", + "accept": false, + "reason": "DSA indexer ReLU discussion vs removal of is_torch_fx_available breaking trust_remote_code models; not the same bug." + }, + { + "left": "issue:44857", + "right": "issue:45072", + "accept": false, + "reason": "LwDetrImageLoss AMP/CUDA crash vs SwitchTransformers/TimmWrapperModel bfloat16 dtype mismatch; different model paths." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: issue:45310 and issue:45478 have the same title and clearly look like the same Qwen3.5-MoE from_pretrained regression, with 45310 earlier, so the canonical choice is well supported. The soft-pair rejections are conservative and the reasons stay within the evidence in the titles/body summaries; no pair is overstated as a duplicate. No mergeability concern applies here since this is an issue-dedup/canonicalization judgment, not a PR-pair merge decision." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 32, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:10000", + "issue:15354", + "issue:25251", + "issue:27957", + "issue:29870", + "issue:29942", + "issue:32090", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:37428", + "issue:38175", + "issue:40444", + "issue:42175", + "issue:42490", + "issue:43278", + "issue:43329", + "issue:43381", + "issue:43526", + "issue:43550", + "issue:43575", + "issue:43873", + "issue:43927", + "issue:43976", + "issue:44448", + "issue:44466", + "issue:44493", + "issue:44561", + "issue:44568", + "issue:44779", + "issue:44855", + "issue:44857", + "issue:44871", + "issue:44977", + "issue:45083", + "issue:45229", + "issue:45290", + "issue:45381", + "issue:45440", + "issue:45491", + "issue:45588", + "issue:45698", + "issue:45920", + "issue:46192", + "issue:46399", + "issue:46421", + "issue:46655", + "issue:46858" + ], + "soft_pairs": [ + "issue:43550|issue:45083", + "issue:27957|issue:34689", + "issue:29942|issue:35141", + "issue:38175|issue:44493", + "issue:29870|issue:32090", + "issue:43278|issue:43329", + "issue:44448|issue:44977", + "issue:45920|issue:46192", + "issue:45290|issue:45381", + "issue:10000|issue:15354", + "issue:44977|issue:45229", + "issue:43278|issue:43381", + "issue:29870|issue:33290", + "issue:43927|issue:45698", + "issue:44855|issue:46399", + "issue:45290|issue:46655", + "issue:40444|issue:43526", + "issue:44857|issue:45588", + "issue:37428|issue:42175", + "issue:32090|issue:33290", + "issue:46421|issue:46858", + "issue:44561|issue:44779", + "issue:44448|issue:44466", + "issue:43873|issue:45229", + "issue:25251|issue:32090", + "issue:43976|issue:45440", + "issue:42490|issue:43873", + "issue:44568|issue:45491", + "issue:43278|issue:43575", + "issue:44871|issue:44977" + ], + "prepared_review_unit_hash": "f1bbd6991142d658798e50be6bd82ce3a623da2d2e5a3fb0b06e7daeb36b9ff4", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 37074, + "estimated_input_tokens": 9269, + "estimated_eval_tokens": 18794 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a loose collection of mostly unrelated Transformers bug reports; none of the soft pairs look like true duplicates. The broadest representative issue is the v5 tokenizer regression in 45920.", + "confidence": 0.74, + "canonical_issue_reason": "issue:45920 is the broadest, most central regression report here: it describes a v5 tokenizer bug affecting multiple model families with clear user impact and strong evidence.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45920 is the strongest standalone anchor for this set because it is broad, actionable, and tied to a concrete regression spanning several models.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:45083", + "accept": false, + "reason": "Different models and different failure modes: SDPA/torch.compile vs a helper-function bug in qwen3_omni_moe." + }, + { + "left": "issue:27957", + "right": "issue:34689", + "accept": false, + "reason": "Both mention model/attention loading issues, but they are distinct model-specific regressions with different symptoms." + }, + { + "left": "issue:29942", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention 2 test failures are not the same bug as embeddings being reinitialized after resize." + }, + { + "left": "issue:38175", + "right": "issue:44493", + "accept": false, + "reason": "One is zero probabilities in a specific model; the other is an unexpected position-id key regression across many models." + }, + { + "left": "issue:29870", + "right": "issue:32090", + "accept": false, + "reason": "UnboundLocalError in generate and a NoneType broadcast error in Trainer are unrelated code paths." + }, + { + "left": "issue:43278", + "right": "issue:43329", + "accept": false, + "reason": "Embedding dtype drift during evaluate is unrelated to undefined multimodal-token variables in the video branch." + }, + { + "left": "issue:44448", + "right": "issue:44977", + "accept": false, + "reason": "Both are v5 regressions, but one is Pegasus output drift and the other is Qwen3.5 flash-attention generation failure." + }, + { + "left": "issue:45920", + "right": "issue:46192", + "accept": false, + "reason": "Tokenizer ID corruption across models is a different bug from return_tensors being ignored when text_kwargs is passed." + }, + { + "left": "issue:45290", + "right": "issue:45381", + "accept": false, + "reason": "Chat-template tool-call handling and Qwen2.5-VL video position IDs are separate features and failure modes." + }, + { + "left": "issue:10000", + "right": "issue:15354", + "accept": false, + "reason": "German DistilBertModel error and GeneratorExp torch.jit.script export failure are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45229", + "accept": false, + "reason": "Flash-attention generation issues are not the same as a multi-GPU CUDA OOM on Gemma4." + }, + { + "left": "issue:43278", + "right": "issue:43381", + "accept": false, + "reason": "Evaluate-time dtype change and gradient-checkpointing-in-eval are different problems." + }, + { + "left": "issue:29870", + "right": "issue:33290", + "accept": false, + "reason": "These are different runtime issues: generate-time decoder mask bug vs optimizer/deepspeed OOM." + }, + { + "left": "issue:43927", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load behavior, but one loses custom token IDs while the other loads the wrong custom module." + }, + { + "left": "issue:44855", + "right": "issue:46399", + "accept": false, + "reason": "Python 3.13 parsing/IndentationError in DebertaV2 is unrelated to the Qwen3.5 kernel registration error." + }, + { + "left": "issue:45290", + "right": "issue:46655", + "accept": false, + "reason": "Chat templating/tool calls and SmolVLM video padding are distinct bugs in different subsystems." + }, + { + "left": "issue:40444", + "right": "issue:43526", + "accept": false, + "reason": "Multi-image IterableDataset finetuning failure is unrelated to BEiT image label reduction behavior." + }, + { + "left": "issue:44857", + "right": "issue:45588", + "accept": false, + "reason": "Loss instability under float16 AMP is not the same as flash_attention.py crashing on s_aux=None." + }, + { + "left": "issue:37428", + "right": "issue:42175", + "accept": false, + "reason": "A missing flash-attention utility import and a packaging/backend install issue are unrelated." + }, + { + "left": "issue:32090", + "right": "issue:33290", + "accept": false, + "reason": "Trainer broadcast TypeError and deepspeed/adafactor OOM are different failures." + }, + { + "left": "issue:46421", + "right": "issue:46858", + "accept": false, + "reason": "Both touch cache/prefill, but one is recompilation churn and the other is a device-tensor arithmetic bug in static cache." + }, + { + "left": "issue:44561", + "right": "issue:44779", + "accept": false, + "reason": "Removing is_torch_fx_available and DeepSeek tokenizer regressions are unrelated v5 breakages." + }, + { + "left": "issue:44448", + "right": "issue:44466", + "accept": false, + "reason": "Different concrete bugs: Pegasus output mismatch vs inconsistent lm_head.weight serialization tied to device." + }, + { + "left": "issue:43873", + "right": "issue:45229", + "accept": false, + "reason": "Quantization/offloading behavior and Gemma4 multi-GPU OOM are not the same underlying issue." + }, + { + "left": "issue:25251", + "right": "issue:32090", + "accept": false, + "reason": "Pipeline top_k output shape change and Trainer NoneType broadcast error are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45440", + "accept": false, + "reason": "Python version compatibility and DeepseekV3MoE divergence are different classes of bugs." + }, + { + "left": "issue:42490", + "right": "issue:43873", + "accept": false, + "reason": "Local-path import warning with trailing slash is unrelated to quantization offloading." + }, + { + "left": "issue:44568", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizer special-token behavior and Gemma3 NaN embeddings are distinct problems." + }, + { + "left": "issue:44871", + "right": "issue:44977", + "accept": false, + "reason": "EOS-token config inconsistency and Qwen3.5 flash-attention generation failure do not share the same underlying cause." + } + ] + }, + "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": 23, + "review_unit_index": 33, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29127", + "issue:29870", + "issue:29942", + "issue:30064", + "issue:33357", + "issue:34567", + "issue:34689", + "issue:39692", + "issue:41720", + "issue:42490", + "issue:42915", + "issue:42994", + "issue:43065", + "issue:43066", + "issue:43381", + "issue:43412", + "issue:43575", + "issue:43611", + "issue:43653", + "issue:43701", + "issue:43756", + "issue:43994", + "issue:44186", + "issue:44206", + "issue:44265", + "issue:44361", + "issue:44442", + "issue:44462", + "issue:44479", + "issue:44743", + "issue:44977", + "issue:45198", + "issue:45229", + "issue:45245", + "issue:45372", + "issue:45406", + "issue:45412", + "issue:45440", + "issue:45459", + "issue:45491", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45712", + "issue:45803", + "issue:45823", + "issue:46121", + "issue:46693" + ], + "soft_pairs": [ + "issue:43381|issue:43701", + "issue:45588|issue:45823", + "issue:33357|issue:39692", + "issue:29127|issue:34567", + "issue:41720|issue:45229", + "issue:44977|issue:45563", + "issue:45198|issue:45803", + "issue:44361|issue:45198", + "issue:44186|issue:45198", + "issue:43653|issue:44743", + "issue:43611|issue:46121", + "issue:29870|issue:34689", + "issue:45372|issue:45588", + "issue:42915|issue:42994", + "issue:43756|issue:45440", + "issue:43653|issue:44442", + "issue:29942|issue:30064", + "issue:45372|issue:45406", + "issue:44206|issue:44479", + "issue:43412|issue:45588", + "issue:45459|issue:45712", + "issue:45491|issue:46693", + "issue:44265|issue:45198", + "issue:29127|issue:33357", + "issue:43065|issue:43994", + "issue:43065|issue:45412", + "issue:45245|issue:45563", + "issue:43575|issue:43701", + "issue:43066|issue:44462", + "issue:45440|issue:45593", + "issue:45372|issue:45459", + "issue:42490|issue:43381" + ], + "prepared_review_unit_hash": "e54f594dfcd80adaff1f6027f8321b58b810dbe4919e2c1dc72e818870e7c2a1", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 32, + "serialized_chars": 37289, + "estimated_input_tokens": 9323, + "estimated_eval_tokens": 18902 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is heterogeneous: the soft pairs share at most a broad subsystem or model family, but none look like the same underlying bug/change. I would not merge any of the proposed pairs.", + "confidence": 0.93, + "canonical_issue_reason": "Issue 46693 is the broadest active regression report here and has inbound references; the rest are mostly unrelated one-off bugs.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46693 has the clearest current user impact, explicit regression framing, and ongoing activity, so it is the best representative issue if one must be chosen.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43381", + "right": "issue:43701", + "accept": false, + "reason": "Different problems: eval-mode gradient checkpointing vs checkpoint resume key naming mismatch." + }, + { + "left": "issue:45588", + "right": "issue:45823", + "accept": false, + "reason": "Both touch Gemma/Gemma4 paths, but one is a flash-attention crash and the other is a device-mapping mismatch; different root causes." + }, + { + "left": "issue:33357", + "right": "issue:39692", + "accept": false, + "reason": "CLIP MacOS bus error and SigLIP2 doc/example errors are unrelated failures." + }, + { + "left": "issue:29127", + "right": "issue:34567", + "accept": false, + "reason": "LayoutLMv3 error messaging and TrainerState token counting are separate features/bugs." + }, + { + "left": "issue:41720", + "right": "issue:45229", + "accept": false, + "reason": "Both are memory/runtime failures, but Qwen3 device-map cuda assert and Gemma4 multi-GPU OOM are different code paths." + }, + { + "left": "issue:44977", + "right": "issue:45563", + "accept": false, + "reason": "Flash-attention generation failure and stale warning text are unrelated issues." + }, + { + "left": "issue:45198", + "right": "issue:45803", + "accept": false, + "reason": "Wav2Vec2 save/tokenization failure is unrelated to FuyuBatchFeature's bare except swallowing errors." + }, + { + "left": "issue:44361", + "right": "issue:45198", + "accept": false, + "reason": "MLukeTokenizer AttributeError and Wav2Vec2 save/tokenization failure do not share the same underlying bug." + }, + { + "left": "issue:44186", + "right": "issue:45198", + "accept": false, + "reason": "LayoutLMv2 tokenizer padding/NER crashes and Wav2Vec2 serialization/tokenization failures are different defects." + }, + { + "left": "issue:43653", + "right": "issue:44743", + "accept": false, + "reason": "BigBirdTokenizer special-token registration and Qwen3.5 recurrent-state reset are unrelated." + }, + { + "left": "issue:43611", + "right": "issue:46121", + "accept": false, + "reason": "base_model_prefix loading behavior and rope-parameter dict conversion are separate model/config issues." + }, + { + "left": "issue:29870", + "right": "issue:34689", + "accept": false, + "reason": "Bart generate decoder mask bug and Llama 3.2 vision model-loading regression are not the same issue." + }, + { + "left": "issue:45372", + "right": "issue:45588", + "accept": false, + "reason": "Gemma4 processor import dependency failure and flash-attention crash on s_aux=None are different failure points." + }, + { + "left": "issue:42915", + "right": "issue:42994", + "accept": false, + "reason": "Both are quantization-related, but one is FP8 config loading and the other is quantized saving; not the same bug." + }, + { + "left": "issue:43756", + "right": "issue:45440", + "accept": false, + "reason": "Smollm3 RoPE-layer behavior and DeepseekV3 implementation divergence are unrelated model issues." + }, + { + "left": "issue:43653", + "right": "issue:44442", + "accept": false, + "reason": "BigBird tokenizer special-token bug and AutoTokenizer loading FastSpeech2ConformerTokenizer are different tokenizer failures." + }, + { + "left": "issue:29942", + "right": "issue:30064", + "accept": false, + "reason": "Flash Attention 2 test failures and void segmentation map processing are unrelated." + }, + { + "left": "issue:45372", + "right": "issue:45406", + "accept": false, + "reason": "Gemma4 processor import failure and Gemma4Processor missing _tokenizer in serve are distinct bugs." + }, + { + "left": "issue:44206", + "right": "issue:44479", + "accept": false, + "reason": "Unsupported center argument in LasrFeatureExtractor and Qwen video-input regression are separate regressions." + }, + { + "left": "issue:43412", + "right": "issue:45588", + "accept": false, + "reason": "Gemma3n executorch export issues and flash-attention s_aux=None crash are unrelated." + }, + { + "left": "issue:45459", + "right": "issue:45712", + "accept": false, + "reason": "Tokenizer error masking from protobuf import handling and leftover dummy classes in dummy_pt_objects.py are unrelated maintenance bugs." + }, + { + "left": "issue:45491", + "right": "issue:46693", + "accept": false, + "reason": "Both involve Gemma/FA2 generation behavior, but one is NaN embeddings from sliding-window padding and the other is a performance regression; different symptoms and likely fixes." + }, + { + "left": "issue:44265", + "right": "issue:45198", + "accept": false, + "reason": "torch.export.export failure with torch_compilable_check is unrelated to Wav2Vec2 save/tokenization failure." + }, + { + "left": "issue:29127", + "right": "issue:33357", + "accept": false, + "reason": "LayoutLMv3 box-content error messaging and CLIP MacOS bus error are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:43994", + "accept": false, + "reason": "Sam3PixelDecoder dummy Conv2d cleanup and SigLIP2 nonsensical outputs are different models and bugs." + }, + { + "left": "issue:43065", + "right": "issue:45412", + "accept": false, + "reason": "Dummy class cleanup in Sam3PixelDecoder and RT-DETR memory retention are unrelated." + }, + { + "left": "issue:45245", + "right": "issue:45563", + "accept": false, + "reason": "Category-cardinality runtime error and stale warning text are unrelated." + }, + { + "left": "issue:43575", + "right": "issue:43701", + "accept": false, + "reason": "Tensor-parallel OOM on Qwen2-57B and checkpoint resume key mismatch are different problems." + }, + { + "left": "issue:43066", + "right": "issue:44462", + "accept": false, + "reason": "Wrong tokenizer decoder type in v5 and AutoTokenizer ignoring tokenizer.json are both tokenizer-related but not the same root cause." + }, + { + "left": "issue:45440", + "right": "issue:45593", + "accept": false, + "reason": "DeepseekV3 implementation divergence and D-FINE auxiliary-loss behavior are unrelated model bugs." + }, + { + "left": "issue:45372", + "right": "issue:45459", + "accept": false, + "reason": "Gemma4 processor import failure and protobuf exception masking for tokenizers are unrelated." + }, + { + "left": "issue:42490", + "right": "issue:43381", + "accept": false, + "reason": "Local-path import warning and eval-mode gradient checkpointing 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": 23, + "review_unit_index": 34, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:27957", + "issue:29127", + "issue:33290", + "issue:34689", + "issue:35141", + "issue:42490", + "issue:42915", + "issue:43262", + "issue:43334", + "issue:43381", + "issue:43421", + "issue:43526", + "issue:43531", + "issue:43575", + "issue:43646", + "issue:43761", + "issue:43824", + "issue:43873", + "issue:43931", + "issue:43994", + "issue:44077", + "issue:44186", + "issue:44206", + "issue:44279", + "issue:44387", + "issue:44442", + "issue:44464", + "issue:44514", + "issue:44743", + "issue:44811", + "issue:44898", + "issue:44977", + "issue:45072", + "issue:45245", + "issue:45250", + "issue:45381", + "issue:45412", + "issue:45491", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45834", + "issue:45859", + "issue:46121", + "issue:46369", + "issue:46399", + "issue:46640", + "issue:46693" + ], + "soft_pairs": [ + "issue:29127|issue:35141", + "issue:43421|issue:45491", + "issue:43994|issue:45412", + "issue:45412|issue:46369", + "issue:43575|issue:44387", + "issue:44442|issue:45072", + "issue:44464|issue:45584", + "issue:42915|issue:43931", + "issue:43526|issue:45834", + "issue:43526|issue:46693", + "issue:43824|issue:44279", + "issue:27957|issue:29127", + "issue:27957|issue:33290", + "issue:45250|issue:46640", + "issue:43994|issue:46369", + "issue:33290|issue:34689", + "issue:43526|issue:44186", + "issue:33290|issue:35141", + "issue:43334|issue:43531", + "issue:46121|issue:46399", + "issue:45588|issue:45600", + "issue:43381|issue:43873", + "issue:44514|issue:45381", + "issue:43646|issue:44077", + "issue:43761|issue:45859", + "issue:44977|issue:45245", + "issue:44743|issue:44811", + "issue:44442|issue:44898", + "issue:43262|issue:44206", + "issue:42490|issue:44743" + ], + "prepared_review_unit_hash": "612aa02ee5b7363d4ded75780d6f1193c3fc26bbda03b66a304ae0aa5f71ffcb", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "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 edges are mostly false positives driven by broad keyword overlap (Flash Attention, OOM, Qwen, loading, tokenizer, etc.), but they refer to different concrete bugs or features. I do not see a true duplicate cluster here.", + "confidence": 0.94, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:29127", + "right": "issue:35141", + "accept": false, + "reason": "Different problems: LayoutLMv3 error-message clarity vs token embedding resize/post_init reinitialization." + }, + { + "left": "issue:43421", + "right": "issue:45491", + "accept": false, + "reason": "Tokenizers post-processor sync issue is unrelated to Gemma3 NaN embeddings from sliding-window/all-padding behavior." + }, + { + "left": "issue:43994", + "right": "issue:45412", + "accept": false, + "reason": "SIGLIP2 bad outputs and RT-DETR memory release are unrelated model failures." + }, + { + "left": "issue:45412", + "right": "issue:46369", + "accept": false, + "reason": "Both mention memory, but one is a deletion/GC leak and the other is a model-call memory overflow; different code paths." + }, + { + "left": "issue:43575", + "right": "issue:44387", + "accept": false, + "reason": "Both are OOMs, but one is tensor-parallel load OOM and the other is int4/Transformers 5.x reserved-memory regression." + }, + { + "left": "issue:44442", + "right": "issue:45072", + "accept": false, + "reason": "Tokenizer loading failure and dtype mismatch in inference are unrelated." + }, + { + "left": "issue:44464", + "right": "issue:45584", + "accept": false, + "reason": "Compiled-forward chunked generation mismatch is not the same as Whisper empty-transcription handling." + }, + { + "left": "issue:42915", + "right": "issue:43931", + "accept": false, + "reason": "Different Qwen model families and different failure modes: FineGrainedFP8Config vs weight shape mismatch." + }, + { + "left": "issue:43526", + "right": "issue:45834", + "accept": false, + "reason": "BeitImageProcessorFast label reduction bug is unrelated to Kosmos2.5 long-OCR indexing." + }, + { + "left": "issue:43526", + "right": "issue:46693", + "accept": false, + "reason": "Different processors/models and different bug classes; only superficial image/model overlap." + }, + { + "left": "issue:43824", + "right": "issue:44279", + "accept": false, + "reason": "ImportError for a missing symbol is not the same as a generic dependency issue." + }, + { + "left": "issue:27957", + "right": "issue:29127", + "accept": false, + "reason": "Flash Attention support for XLM-Roberta and LayoutLMv3 error-message clarity are unrelated." + }, + { + "left": "issue:27957", + "right": "issue:33290", + "accept": false, + "reason": "Model support for FA2 and Deepspeed Adafactor OOM are different issues." + }, + { + "left": "issue:45250", + "right": "issue:46640", + "accept": false, + "reason": "Both mention Flash Attention 2, but the titles do not indicate the same concrete bug or fix." + }, + { + "left": "issue:43994", + "right": "issue:46369", + "accept": false, + "reason": "SIGLIP2 inference correctness and sam3-litetext memory overflow are unrelated." + }, + { + "left": "issue:33290", + "right": "issue:34689", + "accept": false, + "reason": "Deepspeed Adafactor OOM and Llama 3.2 Vision model loading regression are different failures." + }, + { + "left": "issue:43526", + "right": "issue:44186", + "accept": false, + "reason": "Different components: BeitImageProcessorFast label reduction vs LayoutLMv2Tokenizer crashes." + }, + { + "left": "issue:33290", + "right": "issue:35141", + "accept": false, + "reason": "OOM in optimizer/offload path is unrelated to embedding resize reinitialization." + }, + { + "left": "issue:43334", + "right": "issue:43531", + "accept": false, + "reason": "Qwen3-VL pad_token_id loading error and Qwen3-MoE sliding_window behavior are different bugs." + }, + { + "left": "issue:46121", + "right": "issue:46399", + "accept": false, + "reason": "RoPE validation TypeError and missing decorator for apply_rotary_pos_emb are distinct implementation issues." + }, + { + "left": "issue:45588", + "right": "issue:45600", + "accept": false, + "reason": "Flash-attention crash on s_aux=None is unrelated to stale auto_mappings entries." + }, + { + "left": "issue:43381", + "right": "issue:43873", + "accept": false, + "reason": "Gradient checkpointing eval-mode restriction and quantization offloading behavior are different concerns." + }, + { + "left": "issue:44514", + "right": "issue:45381", + "accept": false, + "reason": "Batched chat-template padding crash and wrong video vision_position_ids in Qwen2.5-VL are different bugs." + }, + { + "left": "issue:43646", + "right": "issue:44077", + "accept": false, + "reason": "Custom model initialization breakage in v5 is not the same as a patchtsmixer post_init policy issue." + }, + { + "left": "issue:43761", + "right": "issue:45859", + "accept": false, + "reason": "CLIP hidden_states regression and Qwen3_5Moe CPU offload incompatibility are unrelated." + }, + { + "left": "issue:44977", + "right": "issue:45245", + "accept": false, + "reason": "Flash-attention generation regression and category-cardinality RuntimeError are unrelated." + }, + { + "left": "issue:44743", + "right": "issue:44811", + "accept": false, + "reason": "Qwen3.5 recurrent-state reset and Whisper batch_decode skip_special_tokens handling are different code paths." + }, + { + "left": "issue:44442", + "right": "issue:44898", + "accept": false, + "reason": "Tokenizer loading and Perceiver interpolation failure are unrelated." + }, + { + "left": "issue:43262", + "right": "issue:44206", + "accept": false, + "reason": "Audio chat-template sampling-rate default and LASR feature extractor center-arg crash are different issues." + }, + { + "left": "issue:42490", + "right": "issue:44743", + "accept": false, + "reason": "Trailing-slash import warning is unrelated to Qwen3 recurrent-state reset." + } + ] + }, + "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": 23, + "review_unit_index": 35, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:27957", + "issue:29870", + "issue:32090", + "issue:42175", + "issue:42617", + "issue:42907", + "issue:43232", + "issue:43316", + "issue:43388", + "issue:43493", + "issue:43525", + "issue:43550", + "issue:43575", + "issue:43606", + "issue:43643", + "issue:43701", + "issue:43720", + "issue:43742", + "issue:43749", + "issue:43761", + "issue:43827", + "issue:43873", + "issue:43931", + "issue:44062", + "issue:44077", + "issue:44079", + "issue:44188", + "issue:44190", + "issue:44442", + "issue:44479", + "issue:44496", + "issue:44855", + "issue:44871", + "issue:44913", + "issue:44987", + "issue:45003", + "issue:45137", + "issue:45216", + "issue:45325", + "issue:45440", + "issue:45698", + "issue:45759", + "issue:45803", + "issue:45907", + "issue:46178", + "issue:46234", + "issue:46506", + "issue:46697" + ], + "soft_pairs": [ + "issue:45003|issue:45440", + "issue:43720|issue:44442", + "issue:42175|issue:46234", + "issue:27957|issue:29870", + "issue:43493|issue:43827", + "issue:43316|issue:43525", + "issue:44062|issue:45907", + "issue:44077|issue:44479", + "issue:45325|issue:46697", + "issue:43873|issue:44871", + "issue:44855|issue:45137", + "issue:43643|issue:44913", + "issue:43749|issue:43931", + "issue:43316|issue:45759", + "issue:43761|issue:44079", + "issue:44190|issue:44496", + "issue:43761|issue:44188", + "issue:44987|issue:46178", + "issue:43931|issue:45216", + "issue:44442|issue:45803", + "issue:43232|issue:43388", + "issue:42907|issue:45698", + "issue:43701|issue:46506", + "issue:42617|issue:43575", + "issue:27957|issue:32090", + "issue:43550|issue:43701", + "issue:43575|issue:43742", + "issue:43606|issue:43701" + ], + "prepared_review_unit_hash": "06a6a0a421fe613969915c50bc3704e520f8aa8d22dad4a9bc2ee50439379f11", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36218, + "estimated_input_tokens": 9055, + "estimated_eval_tokens": 18366 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is a broad false-positive mix of unrelated issue reports spanning model loading, generation, config, backend, and docs regressions. None of the proposed soft pairs look like the same underlying bug or change, so no pair should be merged as duplicates.", + "confidence": 0.96, + "canonical_issue_reason": "No strong duplicate anchor stands out; if a representative is needed, issue 45698 is the broadest and most active save/load regression here, but the set is too heterogeneous to treat any item as a true canonical.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45698 is the closest global representative because it is open, recent, and has the most discussion/inbound references, but it still does not unify the rest of the reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45003", + "right": "issue:45440", + "accept": false, + "reason": "One is a sys.modules safety issue in modeling_utils; the other is a DeepSeekV3MoE implementation divergence. Different code paths and bug classes." + }, + { + "left": "issue:43720", + "right": "issue:44442", + "accept": false, + "reason": "BitNet accelerate-loading unpacking bug vs tokenizer auto-loading failure. Not the same underlying defect." + }, + { + "left": "issue:42175", + "right": "issue:46234", + "accept": false, + "reason": "Package/backend dependency omission vs a missing import_utils availability helper. Different targets." + }, + { + "left": "issue:27957", + "right": "issue:29870", + "accept": false, + "reason": "Flash Attention 2 support for XLM-Roberta vs a BART generate() decoder mask initialization error. Unrelated." + }, + { + "left": "issue:43493", + "right": "issue:43827", + "accept": false, + "reason": "SigLIP2 implementation mismatch vs docs still referencing removed pipeline() APIs. Different issue types." + }, + { + "left": "issue:43316", + "right": "issue:43525", + "accept": false, + "reason": "Both are config-field discrepancies, but for different models and fields (`Gemma3TextConfig` API vs missing `pad_token_id` on Llama4Config)." + }, + { + "left": "issue:44062", + "right": "issue:45907", + "accept": false, + "reason": "Tokenizer AddedToken constructor error vs Hub `list_repo_files` revision forwarding. Unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44479", + "accept": false, + "reason": "PatchTSMixer post_init validation vs a Qwen video-input regression. Different subsystems and failures." + }, + { + "left": "issue:45325", + "right": "issue:46697", + "accept": false, + "reason": "Qwen2.5-VL rope/position-id scaling bug vs Swin2SR image over-padding. Not the same preprocessing logic." + }, + { + "left": "issue:43873", + "right": "issue:44871", + "accept": false, + "reason": "Quantization/offloading behavior issue vs Gemma-3 eos_token_id config mismatch. Different bugs." + }, + { + "left": "issue:44855", + "right": "issue:45137", + "accept": false, + "reason": "Python 3.13 torch.jit parsing error vs DeepSpeed ZeRO3 deque underflow. Different layers." + }, + { + "left": "issue:43643", + "right": "issue:44913", + "accept": false, + "reason": "AutoConfig trust_remote_code field loss vs GPTNeoX rotary_pct persistence on reload. Both config-related, but distinct failures." + }, + { + "left": "issue:43749", + "right": "issue:43931", + "accept": false, + "reason": "FSDP CPU RAM-efficient loading regression vs Qwen3-VL weight-shape mismatch. Not the same root cause." + }, + { + "left": "issue:43316", + "right": "issue:45759", + "accept": false, + "reason": "Both touch config handling, but one is a Gemma3TextConfig API discrepancy and the other is AutoModelForCausalLM.from_config not unwrapping text_config for composite Qwen configs." + }, + { + "left": "issue:43761", + "right": "issue:44079", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression vs ModelOutput key assignment bug. Unrelated." + }, + { + "left": "issue:44190", + "right": "issue:44496", + "accept": false, + "reason": "Local dataset loading in a training script vs an unrecognized model/config error. Different failure domains." + }, + { + "left": "issue:43761", + "right": "issue:44188", + "accept": false, + "reason": "CLIPVisionModel hidden_states regression vs attention-kernel branching under torch.compile. Different modules and symptoms." + }, + { + "left": "issue:44987", + "right": "issue:46178", + "accept": false, + "reason": "A model-loading failure for `physical-intelligence/fast` vs a backend mapping error for `tensorflow_text`. Different backend issues." + }, + { + "left": "issue:43931", + "right": "issue:45216", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch on load vs a Qwen3.5 save_pretrained regression. Different phases and causes." + }, + { + "left": "issue:44442", + "right": "issue:45803", + "accept": false, + "reason": "Tokenizer loading failure vs bare-except error swallowing in FuyuBatchFeature. Different code paths." + }, + { + "left": "issue:43232", + "right": "issue:43388", + "accept": false, + "reason": "Generation kwargs sync after sync_gpus vs gather_for_metrics dropping tuple labels. Different training utilities." + }, + { + "left": "issue:42907", + "right": "issue:45698", + "accept": false, + "reason": "Both involve save/load, but one is dequantized-model saving and the other reloads the wrong custom module after save_pretrained. Not mergeable." + }, + { + "left": "issue:43701", + "right": "issue:46506", + "accept": false, + "reason": "Checkpoint resume key mismatch vs CPU/CUDA output divergence in PPDocLayoutV3. Unrelated." + }, + { + "left": "issue:42617", + "right": "issue:43575", + "accept": false, + "reason": "3d_parallel.py runtime failure vs tensor-parallel OOM when loading a large model. Different symptoms and code paths." + }, + { + "left": "issue:27957", + "right": "issue:32090", + "accept": false, + "reason": "XLMRoberta Flash Attention support vs Trainer GPU broadcast NoneType error. Unrelated." + }, + { + "left": "issue:43550", + "right": "issue:43701", + "accept": false, + "reason": "torch.compile SDPA failure in Bamba-9B-v2 vs checkpoint resume key mismatch. Different issues." + }, + { + "left": "issue:43575", + "right": "issue:43742", + "accept": false, + "reason": "Large-model TP OOM vs MobileLLM config key error. Not the same defect." + }, + { + "left": "issue:43606", + "right": "issue:43701", + "accept": false, + "reason": "CPU offload device mismatch vs checkpoint resume key mismatch. Different subsystems." + } + ] + }, + "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": 23, + "review_unit_index": 36, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:16998", + "issue:22355", + "issue:25251", + "issue:28282", + "issue:42175", + "issue:43295", + "issue:43316", + "issue:43352", + "issue:43404", + "issue:43441", + "issue:43452", + "issue:43504", + "issue:43540", + "issue:43572", + "issue:43646", + "issue:43650", + "issue:43716", + "issue:43761", + "issue:43825", + "issue:44077", + "issue:44162", + "issue:44242", + "issue:44464", + "issue:44479", + "issue:44496", + "issue:44514", + "issue:44534", + "issue:44610", + "issue:44792", + "issue:44871", + "issue:44913", + "issue:45542", + "issue:45584", + "issue:45632", + "issue:45698", + "issue:45803", + "issue:46050", + "issue:46170", + "issue:46178", + "issue:46202", + "issue:46234", + "issue:46291", + "issue:46365", + "issue:46369", + "issue:46421", + "issue:46424", + "issue:46640", + "issue:46894" + ], + "soft_pairs": [ + "issue:25251|issue:28282", + "issue:46234|issue:46291", + "issue:43295|issue:43825", + "issue:16998|issue:22355", + "issue:43295|issue:44871", + "issue:44871|issue:45584", + "issue:44464|issue:46421", + "issue:43316|issue:43716", + "issue:44077|issue:44534", + "issue:42175|issue:43316", + "issue:44871|issue:46050", + "issue:43646|issue:44913", + "issue:45803|issue:46365", + "issue:44162|issue:44792", + "issue:43452|issue:44496", + "issue:43404|issue:43441", + "issue:44242|issue:46894", + "issue:45632|issue:46424", + "issue:45803|issue:46170", + "issue:44610|issue:46369", + "issue:43650|issue:46202", + "issue:43504|issue:45698", + "issue:43761|issue:44479", + "issue:45542|issue:46234", + "issue:43540|issue:44514", + "issue:43352|issue:46640", + "issue:43572|issue:46178" + ], + "prepared_review_unit_hash": "e44b95a263a5811dfa9aff9aaf49390c57dd1da3a9aec52b1c2b90e94ae988a2", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35666, + "estimated_input_tokens": 8917, + "estimated_eval_tokens": 18090 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly a loose thematic grouping of distinct Transformers regressions and import/load issues, not true duplicates. The closest representative issue is the custom model loading/saving regression, but none of the soft pairs look like the same underlying bug/change.", + "confidence": 0.73, + "canonical_issue_reason": "issue:45698 is the broadest and most central report around custom module resolution during from_pretrained/save_pretrained, which overlaps with several nearby loading-related reports in this set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45698 is the best overall representative because it is active, high-discussion, and captures the most reusable umbrella theme in the cluster: broken custom-model loading after serialization.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:25251", + "right": "issue:28282", + "accept": false, + "reason": "One is a pipeline output-shape bug; the other is an AutoModel backend import error. Different code paths and symptoms." + }, + { + "left": "issue:46234", + "right": "issue:46291", + "accept": false, + "reason": "Both are import-related, but one is a missing utility symbol and the other is a kernels dependency/version requirement. Not the same bug." + }, + { + "left": "issue:43295", + "right": "issue:43825", + "accept": false, + "reason": "Both involve v5 pipeline/processor regressions, but one is about processor.tokenizer behavior and the other is an error-message/task-support claim. Different underlying issues." + }, + { + "left": "issue:16998", + "right": "issue:22355", + "accept": false, + "reason": "DeBERTa model_max_length handling and missing transformers.onnx are unrelated." + }, + { + "left": "issue:43295", + "right": "issue:44871", + "accept": false, + "reason": "Different models and different failures: processor.tokenizer regression versus Gemma-3 eos_token_id config mismatch." + }, + { + "left": "issue:44871", + "right": "issue:45584", + "accept": false, + "reason": "Gemma-3 EOS config inconsistency and Whisper empty-transcription failure are separate model-specific bugs." + }, + { + "left": "issue:44464", + "right": "issue:46421", + "accept": false, + "reason": "Both mention chunked generation/cache, but one is inconsistent outputs from compiled forward while the other is lazy cache initialization causing recompilation. Not the same defect." + }, + { + "left": "issue:43316", + "right": "issue:43716", + "accept": false, + "reason": "Gemma3TextConfig API inconsistency and Mistral-3 dtype mismatch are unrelated model integration issues." + }, + { + "left": "issue:44077", + "right": "issue:44534", + "accept": false, + "reason": "One is about a model class still allowing optional post_init; the other is a v5 buffer initialization bug. Different changes and failures." + }, + { + "left": "issue:42175", + "right": "issue:43316", + "accept": false, + "reason": "TensorFlow backend packaging and Gemma3TextConfig API discrepancy are unrelated." + }, + { + "left": "issue:44871", + "right": "issue:46050", + "accept": false, + "reason": "Gemma-3 EOS config mismatch and sinq quantization save/load breakage are distinct serialization/config bugs." + }, + { + "left": "issue:43646", + "right": "issue:44913", + "accept": false, + "reason": "Both are config/initialization related, but one is a broad v5 custom initialization breakage and the other is GPTNeoX rotary_pct not persisting on reload. Not the same issue." + }, + { + "left": "issue:45803", + "right": "issue:46365", + "accept": false, + "reason": "Bare except swallowing interrupts in Fuyu conversion and Trainer import failing because of datasets are unrelated." + }, + { + "left": "issue:44162", + "right": "issue:44792", + "accept": false, + "reason": "ESM2 runtime breakage and a janus image-generation test failure are different model-specific problems." + }, + { + "left": "issue:43452", + "right": "issue:44496", + "accept": false, + "reason": "Both are model-loading failures, but one is a gguf/from_pretrained breakage and the other is an unrecognized custom model config.json issue. Different root causes." + }, + { + "left": "issue:43404", + "right": "issue:43441", + "accept": false, + "reason": "Mistral3 weight tying and Ministral FlashAttention support are separate model implementation issues." + }, + { + "left": "issue:44242", + "right": "issue:46894", + "accept": false, + "reason": "Both touch training loss, but one is MoE load-balancing loss gating and the other is a CohereASR double-shift bug. Not the same code-path problem." + }, + { + "left": "issue:45632", + "right": "issue:46424", + "accept": false, + "reason": "Both are cache-related, but one is a trust_remote_code path collision and the other is max_cache_length handling. Different bugs." + }, + { + "left": "issue:45803", + "right": "issue:46170", + "accept": false, + "reason": "One is an exception-handling bug in tensor conversion; the other is a Windows/ROCm distributed import dependency issue." + }, + { + "left": "issue:44610", + "right": "issue:46369", + "accept": false, + "reason": "Processor image-size mismatch and memory overflow in sam3-litetext are unrelated." + }, + { + "left": "issue:43650", + "right": "issue:46202", + "accept": false, + "reason": "A data request and a Gemma pipeline roundtrip failure are unrelated." + }, + { + "left": "issue:43504", + "right": "issue:45698", + "accept": false, + "reason": "Both involve loading pretrained/custom models, but one is a legacy field problem in Beit segmentation preset loading and the other is wrong custom module resolution after save/load. Different underlying bugs." + }, + { + "left": "issue:43761", + "right": "issue:44479", + "accept": false, + "reason": "CLIPVision hidden_states regression and Qwen video-input regression are separate v5 issues." + }, + { + "left": "issue:45542", + "right": "issue:46234", + "accept": false, + "reason": "TensorBoard/TensorFlow backend packaging and missing is_falcon_mamba_ssm_available are unrelated." + }, + { + "left": "issue:43540", + "right": "issue:44514", + "accept": false, + "reason": "Both are Qwen multimodal issues, but one is video-input processing in Qwen3OmniMoe and the other is batched chat-template padding behavior in Qwen2_5_VLProcessor. Not the same bug." + }, + { + "left": "issue:43352", + "right": "issue:46640", + "accept": false, + "reason": "One is an explicit lack of Flash Attention 2 support for NemotronHForCausalLM; the other is a generic Flash Attention issue. Different specificity and likely different fixes." + }, + { + "left": "issue:43572", + "right": "issue:46178", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression and an undefined tensorflow_text backend 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": 23, + "review_unit_index": 37, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:25251", + "issue:27957", + "issue:29870", + "issue:35141", + "issue:41950", + "issue:42490", + "issue:43065", + "issue:43388", + "issue:43475", + "issue:43525", + "issue:43531", + "issue:43644", + "issue:43673", + "issue:43688", + "issue:43827", + "issue:43931", + "issue:43976", + "issue:43994", + "issue:44060", + "issue:44190", + "issue:44242", + "issue:44534", + "issue:44617", + "issue:44734", + "issue:44743", + "issue:44792", + "issue:44855", + "issue:44869", + "issue:44913", + "issue:44933", + "issue:44964", + "issue:44977", + "issue:45127", + "issue:45200", + "issue:45245", + "issue:45362", + "issue:45406", + "issue:45584", + "issue:45588", + "issue:45600", + "issue:45632", + "issue:46082", + "issue:46421", + "issue:46617", + "issue:46697", + "issue:46832", + "issue:46858", + "issue:46897" + ], + "soft_pairs": [ + "issue:44792|issue:45600", + "issue:43065|issue:44617", + "issue:44792|issue:44977", + "issue:43388|issue:43688", + "issue:45127|issue:45200", + "issue:43673|issue:44743", + "issue:41950|issue:43475", + "issue:44190|issue:44964", + "issue:43525|issue:43531", + "issue:44534|issue:44913", + "issue:35141|issue:42490", + "issue:44734|issue:45406", + "issue:46082|issue:46697", + "issue:45632|issue:46617", + "issue:44869|issue:46832", + "issue:45245|issue:46858", + "issue:44855|issue:45588", + "issue:43827|issue:43994", + "issue:43673|issue:46421", + "issue:43931|issue:44060", + "issue:44242|issue:46897", + "issue:29870|issue:35141", + "issue:44617|issue:45584", + "issue:43644|issue:44913", + "issue:44933|issue:45362", + "issue:25251|issue:27957", + "issue:43976|issue:44855" + ], + "prepared_review_unit_hash": "715c6dab26a80eca016bb9c7427e5886c3155b0812d6a563f47a9ad5a1055063", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35944, + "estimated_input_tokens": 8986, + "estimated_eval_tokens": 18228 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is mostly noisy and spans unrelated Transformers issues. The only soft pair that looks like a true duplicate is the trust_remote_code cache-loading pair; the rest are adjacent themes but different bugs or code paths.", + "confidence": 0.58, + "canonical_issue_reason": "issue:45632 is the clearest root-cause report: it names the cache-path collision for trust_remote_code local models, which also explains the FileNotFoundError symptom in issue:46617.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45632 is the best representative because it describes the concrete underlying cache bug, not just one downstream failure mode.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44792", + "right": "issue:45600", + "accept": false, + "reason": "Janus test failure vs removed Sam3Lite config references; different subsystems and failure modes." + }, + { + "left": "issue:43065", + "right": "issue:44617", + "accept": false, + "reason": "Both involve SAM3, but one is a dummy Conv2d placeholder and the other is an OOM report; not the same bug." + }, + { + "left": "issue:44792", + "right": "issue:44977", + "accept": false, + "reason": "Unrelated models and symptoms; no shared concrete code path." + }, + { + "left": "issue:43388", + "right": "issue:43688", + "accept": false, + "reason": "Metric-gathering tuple truncation vs auxiliary-loss normalization; different logic and fixes." + }, + { + "left": "issue:45127", + "right": "issue:45200", + "accept": false, + "reason": "LoRA merge/vocab collapse and Gemma4 token-type-id defaults are distinct issues." + }, + { + "left": "issue:43673", + "right": "issue:44743", + "accept": false, + "reason": "Both cache-related, but one is missing cache initialization in chunked prefill and the other is recurrent-state reset in a modular model; not the same bug." + }, + { + "left": "issue:41950", + "right": "issue:43475", + "accept": false, + "reason": "Same SAM3/video area, but pipeline processor lookup and missing vision-encoder attribute are different failures." + }, + { + "left": "issue:44190", + "right": "issue:44964", + "accept": false, + "reason": "Local dataset loading in a training script vs a multimodal model loading regression; different paths." + }, + { + "left": "issue:43525", + "right": "issue:43531", + "accept": false, + "reason": "Llama4 pad_token_id config error vs Qwen3-MoE sliding_window behavior; unrelated bugs." + }, + { + "left": "issue:44534", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk and rotary_pct reload regression are separate state/config persistence problems." + }, + { + "left": "issue:35141", + "right": "issue:42490", + "accept": false, + "reason": "Embedding reinitialization after resize vs import warning on a trailing-slash local path; not the same issue." + }, + { + "left": "issue:44734", + "right": "issue:45406", + "accept": false, + "reason": "KV-cache continuation indexing crash vs Gemma4Processor missing _tokenizer; different code paths." + }, + { + "left": "issue:46082", + "right": "issue:46697", + "accept": false, + "reason": "Config validation for head_dim vs image over-padding; unrelated." + }, + { + "left": "issue:45632", + "right": "issue:46617", + "accept": true, + "reason": "Both report trust_remote_code cache-loading failures; 45632 describes the underlying cache-path collision that can surface as the FileNotFoundError in 46617." + }, + { + "left": "issue:44869", + "right": "issue:46832", + "accept": false, + "reason": "Whisper timestamp decoding crash and RT-DETR feature-level TypeError are unrelated." + }, + { + "left": "issue:45245", + "right": "issue:46858", + "accept": false, + "reason": "Category-cardinality limit vs device-tensor slice arithmetic in cache handling; different bugs." + }, + { + "left": "issue:44855", + "right": "issue:45588", + "accept": false, + "reason": "Python 3.13 indentation/parsing problem vs flash-attention None-attribute crash; unrelated." + }, + { + "left": "issue:43827", + "right": "issue:43994", + "accept": false, + "reason": "Docs still referencing removed pipelines vs a model producing bad outputs with AutoModel/pipeline; not the same change." + }, + { + "left": "issue:43673", + "right": "issue:46421", + "accept": false, + "reason": "Both mention chunked prefill/cache, but one is missing cache initialization and the other is spurious recompilation from lazy init; distinct symptoms and fixes." + }, + { + "left": "issue:43931", + "right": "issue:44060", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch vs a tied-weights warning in Qwen3-Next; different model/config bugs." + }, + { + "left": "issue:44242", + "right": "issue:46897", + "accept": false, + "reason": "Load-balancing loss omission and Florence2 double-shift are different training-loss bugs." + }, + { + "left": "issue:29870", + "right": "issue:35141", + "accept": false, + "reason": "BART generate() decoder mask bug vs post_init reinitializing output embeddings; unrelated." + }, + { + "left": "issue:44617", + "right": "issue:45584", + "accept": false, + "reason": "SAM3 video OOM vs Whisper empty-transcription failure; no shared underlying bug." + }, + { + "left": "issue:43644", + "right": "issue:44913", + "accept": false, + "reason": "Non-persistent buffer junk on v5 and rotary_pct resetting on reload are separate regressions." + }, + { + "left": "issue:44933", + "right": "issue:45362", + "accept": false, + "reason": "Nonexistent import from image_utils vs Qwen3.5 chat crash; unrelated." + }, + { + "left": "issue:25251", + "right": "issue:27957", + "accept": false, + "reason": "Pipeline top_k nesting bug vs Flash Attention 2 support issue; different code paths." + }, + { + "left": "issue:43976", + "right": "issue:44855", + "accept": false, + "reason": "Python version compatibility problem vs DebertaV2 import parsing error; different failures." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "Grounded and conservative overall. The summary correctly treats the cluster as noisy, and the one accepted soft pair (issue:45632 / issue:46617) is plausibly the same trust_remote_code cache bug rather than an overreach. The other negative verdicts also avoid claiming duplicates where the packet only shows thematic similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 38, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:25251", + "issue:29870", + "issue:39692", + "issue:41720", + "issue:43065", + "issue:43295", + "issue:43335", + "issue:43526", + "issue:43540", + "issue:43550", + "issue:43673", + "issue:43854", + "issue:44077", + "issue:44112", + "issue:44190", + "issue:44462", + "issue:44464", + "issue:44479", + "issue:44521", + "issue:44610", + "issue:44617", + "issue:44734", + "issue:44737", + "issue:44779", + "issue:44877", + "issue:44938", + "issue:44964", + "issue:45127", + "issue:45245", + "issue:45276", + "issue:45290", + "issue:45357", + "issue:45381", + "issue:45542", + "issue:45563", + "issue:45584", + "issue:45600", + "issue:45632", + "issue:45803", + "issue:45907", + "issue:46234", + "issue:46369", + "issue:46489", + "issue:46655", + "issue:46697", + "issue:46752", + "issue:46858" + ], + "soft_pairs": [ + "issue:45563|issue:45584", + "issue:43550|issue:44190", + "issue:45381|issue:46369", + "issue:44077|issue:44877", + "issue:44479|issue:44734", + "issue:41720|issue:45245", + "issue:43335|issue:44190", + "issue:43065|issue:44610", + "issue:44464|issue:45563", + "issue:39692|issue:44190", + "issue:45127|issue:45357", + "issue:45381|issue:46655", + "issue:44112|issue:45600", + "issue:44779|issue:46489", + "issue:43295|issue:45276", + "issue:44617|issue:46369", + "issue:43526|issue:46697", + "issue:45290|issue:46752", + "issue:44462|issue:45907", + "issue:45276|issue:45584", + "issue:44737|issue:45803", + "issue:43540|issue:44734", + "issue:44938|issue:45542", + "issue:25251|issue:29870", + "issue:44521|issue:45381", + "issue:43673|issue:46858", + "issue:43854|issue:44964", + "issue:45632|issue:46234" + ], + "prepared_review_unit_hash": "057543853b638f3b7f3d6ea6b62810cba729617d34871874659a03b65ca0a2b6", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35688, + "estimated_input_tokens": 8922, + "estimated_eval_tokens": 18100 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Mostly unrelated closed issues with one plausible duplicate pair around the DeepSeek tokenizer regression in v5; the rest are different models, subsystems, or failure modes and should stay separate.", + "confidence": 0.79, + "canonical_issue_reason": "issue:46489 is the more specific and actionable report: it names the DeepSeek-Coder v1 tokenizer regression in transformers v5+ and points to the gap in PR #44801's fix, making it a better canonical anchor than the broader symptom report.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:46489 has the clearest root-cause framing and strongest linkage to an existing fix gap, so it is the best representative issue in this set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45563", + "right": "issue:45584", + "accept": false, + "reason": "Both are generation bugs, but one is a stale warning for num_return_sequences in paged generate while the other is Whisper empty-transcription failure after align_special_tokens; different code paths and fixes." + }, + { + "left": "issue:43550", + "right": "issue:44190", + "accept": false, + "reason": "Unrelated subsystems: torch.compile/SDPA model failure vs local dataset loading in a training script." + }, + { + "left": "issue:45381", + "right": "issue:46369", + "accept": false, + "reason": "Different models and problems: Qwen2.5-VL video position ids vs Sam3 LiteText memory overflow." + }, + { + "left": "issue:44077", + "right": "issue:44877", + "accept": false, + "reason": "Both involve config handling, but one is PatchTSMixer post_init validation and the other is Granite Speech strict config loading; not the same bug." + }, + { + "left": "issue:44479", + "right": "issue:44734", + "accept": false, + "reason": "Separate areas: multimodal video-input regression in Qwen models vs KV-cache continuation tensor indexing in transformers serve." + }, + { + "left": "issue:41720", + "right": "issue:45245", + "accept": false, + "reason": "Completely different failures: Qwen3 auto device mapping CUDA assert vs category-count limit RuntimeError." + }, + { + "left": "issue:43335", + "right": "issue:44190", + "accept": false, + "reason": "SwitchTransformers sparse-layer config bug is unrelated to local dataset loading in the example script." + }, + { + "left": "issue:43065", + "right": "issue:44610", + "accept": false, + "reason": "Both are vision/model-input issues, but the specific bugs differ: dummy Conv2d in Sam3PixelDecoder vs OmDet-Turbo processor/model size mismatch." + }, + { + "left": "issue:44464", + "right": "issue:45563", + "accept": false, + "reason": "Chunked generation inconsistency with compiled forward is not the same as a stale warning about num_return_sequences." + }, + { + "left": "issue:39692", + "right": "issue:44190", + "accept": false, + "reason": "SigLIP2 doc example errors are unrelated to image-classification script dataset loading." + }, + { + "left": "issue:45127", + "right": "issue:45357", + "accept": false, + "reason": "Different features and failures: LoRA merging with tied embeddings vs incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:45381", + "right": "issue:46655", + "accept": false, + "reason": "Both concern video processing, but one is wrong vision_position_ids and the other is incorrect padding calculation in SmolVLM; not the same concrete bug." + }, + { + "left": "issue:44112", + "right": "issue:45600", + "accept": false, + "reason": "Both are CI-related, but the root causes differ: stale GraniteSpeech device override test vs removed Sam3LiteText configs in auto_mappings." + }, + { + "left": "issue:44779", + "right": "issue:46489", + "accept": true, + "reason": "These describe the same DeepSeek tokenizer regression in transformers v5+: incorrect output/behavior in DeepSeek tokenizer families, with 46489 explicitly identifying the gap in PR #44801's fix." + }, + { + "left": "issue:43295", + "right": "issue:45276", + "accept": false, + "reason": "Different regressions: processor.tokenizer/image handling breakage vs resize_token_embeddings not updating tied embeddings." + }, + { + "left": "issue:44617", + "right": "issue:46369", + "accept": false, + "reason": "Both mention memory pressure in Sam3 variants, but one is Sam3Video OOM and the other is Sam3 LiteText memory overflow; distinct model paths." + }, + { + "left": "issue:43526", + "right": "issue:46697", + "accept": false, + "reason": "Different image processor bugs: BEiT label reduction vs Swin2SR padding on divisible dimensions." + }, + { + "left": "issue:45290", + "right": "issue:46752", + "accept": false, + "reason": "Both crash in apply_chat_template, but one is tool-call assistant messages with no content and the other is an empty list edge case; separate triggers." + }, + { + "left": "issue:44462", + "right": "issue:45907", + "accept": false, + "reason": "AutoTokenizer ignoring tokenizer.json is unrelated to list_repo_files not forwarding revision." + }, + { + "left": "issue:45276", + "right": "issue:45584", + "accept": false, + "reason": "Unrelated issues: gemma4 embedding resize bug vs Whisper empty transcription generation failure." + }, + { + "left": "issue:44737", + "right": "issue:45803", + "accept": false, + "reason": "Different code paths: XLNet relative positional encoding device placement vs bare except in FuyuBatchFeature.convert_to_tensors." + }, + { + "left": "issue:43540", + "right": "issue:44734", + "accept": false, + "reason": "Unrelated failures: Qwen3OmniMoe video-input ValueError vs serving KV-cache continuation tensor indexing." + }, + { + "left": "issue:44938", + "right": "issue:45542", + "accept": false, + "reason": "Python 3.14 import/load failure is unrelated to the TensorBoard-without-TensorFlow backend error." + }, + { + "left": "issue:25251", + "right": "issue:29870", + "accept": false, + "reason": "Both are generation-related, but one is pipeline top_k nesting and the other is Bart decoder_attention_mask initialization; different bugs." + }, + { + "left": "issue:44521", + "right": "issue:45381", + "accept": false, + "reason": "Different multimodal issues: all-zero assistant masks vs wrong vision_position_ids for video inputs." + }, + { + "left": "issue:43673", + "right": "issue:46858", + "accept": false, + "reason": "Both concern cache/prefill, but one is missing cache in v5.0.0 chunked_prefill and the other is a device-tensor seq_length bug in static cache on non-CUDA backends." + }, + { + "left": "issue:43854", + "right": "issue:44964", + "accept": false, + "reason": "Separate model-loading failures for different models and test contexts; no shared concrete bug." + }, + { + "left": "issue:45632", + "right": "issue:46234", + "accept": false, + "reason": "Different concerns: trust_remote_code cache-path collision vs missing import_utils helper for Falcon Mamba availability." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is well grounded in the packet. The DeepSeek pair (issue:44779, issue:46489) is the only clearly plausible duplicate, and the rest of the soft-edge rejections are conservative and supported by the titles/summaries. Minor wording note: some items in the set are open, so 'mostly unrelated closed issues' could be phrased a bit more generally, but this does not affect the core judgment." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 39, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:27957", + "issue:32090", + "issue:35141", + "issue:36246", + "issue:38617", + "issue:41628", + "issue:41950", + "issue:42175", + "issue:42796", + "issue:42907", + "issue:43295", + "issue:43381", + "issue:43404", + "issue:43441", + "issue:43479", + "issue:43526", + "issue:43531", + "issue:43643", + "issue:43650", + "issue:43673", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44112", + "issue:44220", + "issue:44246", + "issue:44279", + "issue:44295", + "issue:44336", + "issue:44496", + "issue:44521", + "issue:44683", + "issue:44792", + "issue:44964", + "issue:44987", + "issue:45003", + "issue:45042", + "issue:45083", + "issue:45216", + "issue:45335", + "issue:45375", + "issue:45446", + "issue:45593", + "issue:45600", + "issue:46082", + "issue:46121", + "issue:46424", + "issue:46506" + ], + "soft_pairs": [ + "issue:44246|issue:45003", + "issue:44964|issue:45600", + "issue:43643|issue:44496", + "issue:44112|issue:44964", + "issue:27957|issue:35141", + "issue:42175|issue:44279", + "issue:46082|issue:46121", + "issue:36246|issue:41628", + "issue:42796|issue:44521", + "issue:36246|issue:38617", + "issue:43404|issue:45216", + "issue:32090|issue:35141", + "issue:44295|issue:46506", + "issue:41950|issue:44016", + "issue:43531|issue:45600", + "issue:43650|issue:44220", + "issue:43673|issue:46424", + "issue:43976|issue:44987", + "issue:43295|issue:45335", + "issue:44016|issue:44220", + "issue:45083|issue:46506", + "issue:42907|issue:43441", + "issue:44683|issue:45446", + "issue:43976|issue:44279", + "issue:43526|issue:43901", + "issue:43381|issue:45593", + "issue:43479|issue:45375", + "issue:44496|issue:45042", + "issue:43901|issue:44496", + "issue:43673|issue:44336", + "issue:44792|issue:46424" + ], + "prepared_review_unit_hash": "37b409e7070621fd54ae3ae64041e4097e971e8fbfecccfa20a31b90e86c2b4a", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36930, + "estimated_input_tokens": 9233, + "estimated_eval_tokens": 18722 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is broadly Transformer/model-loading/regression related, but the individual issues are mostly about different models, APIs, and failure modes. None of the soft pairs look like the same underlying bug/change, so they should stay separate.", + "confidence": 0.71, + "canonical_issue_reason": "issue:44964 is the broadest user-facing representative of the cluster\u2019s dominant theme (latest-transformers model loading regressions), and it has some inbound discussion.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44964 is the most representative of the cluster overall, but it still covers only one slice of the heterogeneous set.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44246", + "right": "issue:45003", + "accept": false, + "reason": "One is slow imports; the other is a sys.modules access safety bug. Same area, different problems." + }, + { + "left": "issue:44964", + "right": "issue:45600", + "accept": false, + "reason": "Both are load-time regressions, but they affect different models and have different root causes." + }, + { + "left": "issue:43643", + "right": "issue:44496", + "accept": false, + "reason": "Both involve config/model loading, but one drops fields with trust_remote_code while the other rejects an unknown model type." + }, + { + "left": "issue:44112", + "right": "issue:44964", + "accept": false, + "reason": "CI test failure in GraniteSpeech vs a model-loading failure for Phi-4 multimodal; unrelated bugs." + }, + { + "left": "issue:27957", + "right": "issue:35141", + "accept": false, + "reason": "Flash Attention support for XLMRoberta and embedding reinit on resize are unrelated code paths." + }, + { + "left": "issue:42175", + "right": "issue:44279", + "accept": false, + "reason": "Packaging/backend dependency complaint vs a generic dependency issue; too vague and not the same defect." + }, + { + "left": "issue:46082", + "right": "issue:46121", + "accept": false, + "reason": "Both are validation-related, but one is about head_dim acceptance and the other about JSON-loaded list handling in rope validation." + }, + { + "left": "issue:36246", + "right": "issue:41628", + "accept": false, + "reason": "Both are import errors, but for different symbols and modules; not the same missing-export bug." + }, + { + "left": "issue:42796", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer BPE marker handling and multimodal assistant masks are different tokenization/template bugs." + }, + { + "left": "issue:36246", + "right": "issue:38617", + "accept": false, + "reason": "Different import failures for different names; no shared concrete fix." + }, + { + "left": "issue:43404", + "right": "issue:45216", + "accept": false, + "reason": "Both touch weight tying/save behavior, but they involve different models and different failure modes." + }, + { + "left": "issue:32090", + "right": "issue:35141", + "accept": false, + "reason": "Trainer broadcast NoneType error and embedding reinitialization are unrelated." + }, + { + "left": "issue:44295", + "right": "issue:46506", + "accept": false, + "reason": "Reading a registered buffer vs CPU/CUDA output divergence are different bugs." + }, + { + "left": "issue:41950", + "right": "issue:44016", + "accept": false, + "reason": "Pipeline image-processor lookup bug vs a notebook syntax error; not duplicates." + }, + { + "left": "issue:43531", + "right": "issue:45600", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and auto_mappings config cleanup are unrelated." + }, + { + "left": "issue:43650", + "right": "issue:44220", + "accept": false, + "reason": "The titles are generic, but the underlying issues are different and provide no concrete overlap." + }, + { + "left": "issue:43673", + "right": "issue:46424", + "accept": false, + "reason": "Both concern cache behavior, but one is chunked_prefill GenerationMixin cache and the other is max_cache_length handling." + }, + { + "left": "issue:43976", + "right": "issue:44987", + "accept": false, + "reason": "Python version compatibility vs a model-loading failure on a specific checkpoint; different root cause." + }, + { + "left": "issue:43295", + "right": "issue:45335", + "accept": false, + "reason": "Both are regressions around tokenization/weights, but one is processor.tokenizer/image passthrough and the other is decoder embedding resize." + }, + { + "left": "issue:44016", + "right": "issue:44220", + "accept": false, + "reason": "Notebook syntax error and audio feature extraction issue are unrelated." + }, + { + "left": "issue:45083", + "right": "issue:46506", + "accept": false, + "reason": "Helper length computation in qwen3_omni_moe vs CPU/CUDA inconsistency in layout model; not the same bug." + }, + { + "left": "issue:42907", + "right": "issue:43441", + "accept": false, + "reason": "Both mention Ministral/FlashAttention, but one is save/load of dequantized weights and the other is FlashAttention inference failure." + }, + { + "left": "issue:44683", + "right": "issue:45446", + "accept": false, + "reason": "Both are flex_attention-related, but one is a compiled-kernel failure on torch>=2.9 and the other is an incorrect version check for AuxRequest import." + }, + { + "left": "issue:43976", + "right": "issue:44279", + "accept": false, + "reason": "A Python compatibility bug and a generic dependency issue are not the same concrete defect." + }, + { + "left": "issue:43526", + "right": "issue:43901", + "accept": false, + "reason": "Image processor label reduction bug vs documentation drift for return_all_scores; one is runtime behavior, the other docs." + }, + { + "left": "issue:43381", + "right": "issue:45593", + "accept": false, + "reason": "Eval-mode gradient checkpointing and D-FINE auxiliary losses are unrelated training behaviors." + }, + { + "left": "issue:43479", + "right": "issue:45375", + "accept": false, + "reason": "Different config-initialization bugs in different multimodal model families." + }, + { + "left": "issue:44496", + "right": "issue:45042", + "accept": false, + "reason": "Unknown model type loading failure vs PIL image processors incorrectly requiring torchvision; different loading layers." + }, + { + "left": "issue:43901", + "right": "issue:44496", + "accept": false, + "reason": "Docs mismatch for a pipeline option vs an unknown-model config error; not duplicates." + }, + { + "left": "issue:43673", + "right": "issue:44336", + "accept": false, + "reason": "Generation cache bug and ANSI output formatting in loading_report are unrelated." + }, + { + "left": "issue:44792", + "right": "issue:46424", + "accept": false, + "reason": "Janus image-generation test failure vs cache max length behavior; different subsystems and symptoms." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 40, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:39692", + "issue:41950", + "issue:42796", + "issue:43295", + "issue:43299", + "issue:43441", + "issue:43577", + "issue:43650", + "issue:43653", + "issue:43824", + "issue:43854", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44062", + "issue:44075", + "issue:44242", + "issue:44279", + "issue:44327", + "issue:44351", + "issue:44368", + "issue:44466", + "issue:44496", + "issue:44521", + "issue:44779", + "issue:44933", + "issue:45020", + "issue:45200", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45335", + "issue:45357", + "issue:45405", + "issue:45584", + "issue:45588", + "issue:45593", + "issue:45759", + "issue:45812", + "issue:45823", + "issue:46082", + "issue:46123", + "issue:46369", + "issue:46399", + "issue:46512", + "issue:46515", + "issue:46693", + "issue:46752" + ], + "soft_pairs": [ + "issue:44016|issue:44062", + "issue:43824|issue:44933", + "issue:44496|issue:45759", + "issue:43441|issue:45588", + "issue:46123|issue:46369", + "issue:44327|issue:46752", + "issue:44521|issue:46693", + "issue:44279|issue:46515", + "issue:45335|issue:45584", + "issue:43299|issue:45216", + "issue:44279|issue:45405", + "issue:46082|issue:46399", + "issue:43295|issue:44779", + "issue:44075|issue:44368", + "issue:39692|issue:43901", + "issue:43976|issue:45405", + "issue:43577|issue:45216", + "issue:44351|issue:45278", + "issue:46082|issue:46512", + "issue:42796|issue:43650", + "issue:44779|issue:45812", + "issue:41950|issue:45020", + "issue:43299|issue:45357", + "issue:44466|issue:45276", + "issue:43299|issue:43854", + "issue:43653|issue:45593", + "issue:41950|issue:45823", + "issue:44242|issue:45200" + ], + "prepared_review_unit_hash": "92a3ca9e3b90919a88f381040ba9e8881e53d1720307cb139c6d76dab4100c59", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 36216, + "estimated_input_tokens": 9054, + "estimated_eval_tokens": 18364 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a loose set of unrelated Transformers issues with some superficial similarity around multimodal/tokenizer/load-path regressions; none of the soft pairs look like true duplicates.", + "confidence": 0.37, + "canonical_issue_reason": "Issue 43295 is the best anchor because it is a concrete, reproducible regression in multimodal processor/tokenizer handling and is broader than the other highly specific model- or docs-only reports.", + "canonical_pr_reason": null, + "best_issue_reason": "43295 is the strongest standalone issue in the set: clear regression scope, concrete failure mode, and good triage value as a representative multimodal API break.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44016", + "right": "issue:44062", + "accept": false, + "reason": "Different failures: notebook syntax error vs AddedToken keyword-argument error; no shared code path." + }, + { + "left": "issue:43824", + "right": "issue:44933", + "accept": false, + "reason": "Both are import errors, but they concern different missing symbols/modules and different APIs." + }, + { + "left": "issue:44496", + "right": "issue:45759", + "accept": false, + "reason": "Generic model-config recognition failure vs composite Qwen text_config unwrapping bug; related area, not same bug." + }, + { + "left": "issue:43441", + "right": "issue:45588", + "accept": false, + "reason": "Distinct flash-attention crashes with different triggers and stack paths." + }, + { + "left": "issue:46123", + "right": "issue:46369", + "accept": false, + "reason": "Batch-finalization bug in MaskGenerationPipeline vs memory overflow in sam3-litetext; unrelated." + }, + { + "left": "issue:44327", + "right": "issue:46752", + "accept": false, + "reason": "QA span top-k bounds error vs empty-list chat-template IndexError; different pipeline code." + }, + { + "left": "issue:44521", + "right": "issue:46693", + "accept": false, + "reason": "Assistant-mask bug in chat templates vs long-context FA2 performance regression; not the same underlying issue." + }, + { + "left": "issue:44279", + "right": "issue:46515", + "accept": false, + "reason": "Dependency complaint vs torch-version incompatibility; too broad and not one concrete bug." + }, + { + "left": "issue:45335", + "right": "issue:45584", + "accept": false, + "reason": "resize_token_embeddings propagation bug vs Whisper empty-transcription failure after token alignment." + }, + { + "left": "issue:43299", + "right": "issue:45216", + "accept": false, + "reason": "Loading regression for Qwen3VL MoE vs save_pretrained checkpoint serialization regression; different code paths." + }, + { + "left": "issue:43577", + "right": "issue:45216", + "accept": false, + "reason": "Blip2 dtype propagation bug vs Qwen3.5 save_pretrained regression; unrelated models and behavior." + }, + { + "left": "issue:44351", + "right": "issue:45278", + "accept": false, + "reason": "Single missing import vs broad post-update import failures; same symptom class, not same bug." + }, + { + "left": "issue:46082", + "right": "issue:46399", + "accept": false, + "reason": "Config validation around head_dim vs missing decorator for rotary-pos-emb kernel registration." + }, + { + "left": "issue:43295", + "right": "issue:44779", + "accept": false, + "reason": "Processor.tokenizer access/passing images regression vs Deepseek tokenizer output regression; different functionality." + }, + { + "left": "issue:44075", + "right": "issue:44368", + "accept": false, + "reason": "SGD args ignored vs warning about tie_word_embeddings; not the same issue." + }, + { + "left": "issue:39692", + "right": "issue:43901", + "accept": false, + "reason": "SigLIP2 example/model mismatch and quantization failure vs outdated docs mentioning return_all_scores." + }, + { + "left": "issue:43976", + "right": "issue:45405", + "accept": false, + "reason": "Python-version compatibility issue vs unreleased PEFT version bump; unrelated." + }, + { + "left": "issue:44279", + "right": "issue:45405", + "accept": false, + "reason": "Generic dependency problem vs specific unreleased dependency version; not duplicate." + }, + { + "left": "issue:46082", + "right": "issue:46512", + "accept": false, + "reason": "Head_dim validation failure vs stale config.dtype after from_config(dtype=...); distinct config bugs." + }, + { + "left": "issue:42796", + "right": "issue:43650", + "accept": false, + "reason": "Tokenizer BPE-marker bug vs nonspecific request to 'ADD THE DATA'; no concrete match." + }, + { + "left": "issue:44779", + "right": "issue:45812", + "accept": false, + "reason": "Different tokenizer regressions affecting different model families and behaviors." + }, + { + "left": "issue:41950", + "right": "issue:45020", + "accept": false, + "reason": "Video-classification processor lookup bug vs remote_code model breakage; too broad to merge." + }, + { + "left": "issue:43299", + "right": "issue:45357", + "accept": false, + "reason": "Loading failure vs incorrect visual-encoder keys on save_pretrained; different regression surfaces." + }, + { + "left": "issue:44466", + "right": "issue:45276", + "accept": false, + "reason": "lm_head serialization/device-dependent tied-weights issue vs resize_token_embeddings not updating Gemma4 embeddings." + }, + { + "left": "issue:43299", + "right": "issue:43854", + "accept": false, + "reason": "Both involve Qwen VL models, but one is a loading regression and the other an import/export issue for a different class." + }, + { + "left": "issue:43653", + "right": "issue:45593", + "accept": false, + "reason": "BigBird mask-token registration bug vs D-FINE auxiliary-loss omission; unrelated." + }, + { + "left": "issue:41950", + "right": "issue:45823", + "accept": false, + "reason": "Pipeline processor-type bug vs Gemma4 PLE device mismatch during forward; different subsystems." + }, + { + "left": "issue:44242", + "right": "issue:45200", + "accept": false, + "reason": "MoE load-balancing loss gating vs Gemma4 mm_token_type_ids defaulting; 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": 23, + "review_unit_index": 41, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:33290", + "issue:42175", + "issue:42796", + "issue:42994", + "issue:43317", + "issue:43441", + "issue:43479", + "issue:43525", + "issue:43650", + "issue:43673", + "issue:43723", + "issue:43761", + "issue:43824", + "issue:43881", + "issue:43901", + "issue:43976", + "issue:44016", + "issue:44162", + "issue:44246", + "issue:44261", + "issue:44263", + "issue:44291", + "issue:44297", + "issue:44336", + "issue:44361", + "issue:44462", + "issue:44623", + "issue:44671", + "issue:44683", + "issue:44749", + "issue:44779", + "issue:45081", + "issue:45103", + "issue:45372", + "issue:45397", + "issue:45542", + "issue:45593", + "issue:45600", + "issue:45656", + "issue:45712", + "issue:45800", + "issue:45834", + "issue:45859", + "issue:45907", + "issue:46209", + "issue:46291", + "issue:46399", + "issue:46851" + ], + "soft_pairs": [ + "issue:43650|issue:46209", + "issue:44297|issue:44462", + "issue:43479|issue:43525", + "issue:44683|issue:45542", + "issue:44016|issue:44291", + "issue:44361|issue:45593", + "issue:45081|issue:45907", + "issue:33290|issue:45656", + "issue:43441|issue:45834", + "issue:42994|issue:43317", + "issue:44623|issue:44749", + "issue:44261|issue:44263", + "issue:45103|issue:45372", + "issue:44336|issue:46851", + "issue:45600|issue:45712", + "issue:45859|issue:46851", + "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:43824|issue:43976", + "issue:45542|issue:45800", + "issue:42796|issue:43441", + "issue:43525|issue:43881", + "issue:46291|issue:46399" + ], + "prepared_review_unit_hash": "2e6a5c7ae4ea778c30bc60102d573343b54855644eabc32351e69a29f8ed9a62", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 27, + "serialized_chars": 35701, + "estimated_input_tokens": 8926, + "estimated_eval_tokens": 18108 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Transformers regressions and environment/package issues; none of the soft pairs look like true duplicates.", + "confidence": 0.94, + "canonical_issue_reason": "No single canonical issue stands out for the whole set because the reports span different subsystems and failure modes. If a representative issue is needed, issue:44462 is the strongest candidate because it is a concrete, reproducible tokenizer-loading regression with broad user impact.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44462 is the best representative issue overall: it describes a clear v5 tokenizer-loading bug, has a specific failure mode, and is the most plausible anchor for nearby tokenizer-related reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43650", + "right": "issue:46209", + "accept": false, + "reason": "Unrelated: a vague data request versus an FP8 inference failure on B200." + }, + { + "left": "issue:44297", + "right": "issue:44462", + "accept": false, + "reason": "Both involve tokenizer save/load, but one is save_pretrained metadata mismatch while the other is AutoTokenizer ignoring repository tokenizer.json; different code paths." + }, + { + "left": "issue:43479", + "right": "issue:43525", + "accept": false, + "reason": "Different configs and failures: Phi4Multimodal default sub-config initialization vs Llama4 missing pad_token_id." + }, + { + "left": "issue:44683", + "right": "issue:45542", + "accept": false, + "reason": "Different subsystems: compiled flex_attention on torch>=2.9 versus TensorFlow backend packaging with tensorboard." + }, + { + "left": "issue:44016", + "right": "issue:44291", + "accept": false, + "reason": "Notebook syntax error versus init_empty_weights/_is_hf_initialized argument mismatch; not the same bug." + }, + { + "left": "issue:44361", + "right": "issue:45593", + "accept": false, + "reason": "Different model/features: MLukeTokenizer task AttributeError versus D-FINE auxiliary losses when denoising is off." + }, + { + "left": "issue:45081", + "right": "issue:45907", + "accept": false, + "reason": "Both touch loading-related code, but one is a Mistral regex patch crash and the other is a hub list_repo_files revision-forwarding issue." + }, + { + "left": "issue:33290", + "right": "issue:45656", + "accept": false, + "reason": "Both mention DeepSpeed/optimizer behavior, but one is OOM with Adafactor and the other is double optimizer stepping; distinct root causes." + }, + { + "left": "issue:43441", + "right": "issue:45834", + "accept": false, + "reason": "Different models and failure modes: FlashAttention regression for Ministral-3 versus long-ocr index error in Kosmos2.5." + }, + { + "left": "issue:42994", + "right": "issue:43317", + "accept": false, + "reason": "Both are quantization/offload related, but saving a quantized model and loading a dequantized model with device_map=auto are different paths." + }, + { + "left": "issue:44623", + "right": "issue:44749", + "accept": false, + "reason": "Processor save_pretrained file-missing bug versus a Chinese report about slower filtering after upgrade; not the same issue." + }, + { + "left": "issue:44261", + "right": "issue:44263", + "accept": false, + "reason": "Same broad GLM family, but one is a missing rms_norm_eps causing precision drift and the other is torch.split return handling in GlmMoeDsaIndexer." + }, + { + "left": "issue:45103", + "right": "issue:45372", + "accept": false, + "reason": "Unrelated: auto_docstring future-annotations crash versus a mistral_common import error breaking Gemma 4 processor loading." + }, + { + "left": "issue:44336", + "right": "issue:46851", + "accept": false, + "reason": "ANSI color handling in loading_report versus Qwen3.5 variable-length position_ids behavior; different code paths." + }, + { + "left": "issue:45600", + "right": "issue:45712", + "accept": false, + "reason": "Both are cleanup/CI-style issues, but auto_mappings stale config references and leftover dummy classes are separate problems." + }, + { + "left": "issue:45859", + "right": "issue:46851", + "accept": false, + "reason": "Same model family, but CPU offload incompatibility and variable-length position_ids are different bugs." + }, + { + "left": "issue:42175", + "right": "issue:43976", + "accept": false, + "reason": "Different packaging/version problems: missing TensorFlow backend in torch extra versus Python version compatibility for Transformers 5.1.0." + }, + { + "left": "issue:44623", + "right": "issue:45397", + "accept": false, + "reason": "Processor.save_pretrained missing files versus gemma-4 zero3 from_pretrained failure; different components and symptoms." + }, + { + "left": "issue:44671", + "right": "issue:44779", + "accept": false, + "reason": "Both are model-specific v5 regressions, but CamemBERT masked LM predictions and DeepSeek tokenizer outputs are separate code paths." + }, + { + "left": "issue:43901", + "right": "issue:44246", + "accept": false, + "reason": "Docs mismatch for TextClassificationPipeline versus intermittent import slowness; unrelated." + }, + { + "left": "issue:43723", + "right": "issue:44162", + "accept": false, + "reason": "AutoTokenizer loading regression versus ESM2 broken workflows; different models and failure mechanisms." + }, + { + "left": "issue:43673", + "right": "issue:43761", + "accept": false, + "reason": "Chunked_prefill cache regression in GenerationMixin versus CLIPVisionModel hidden_states behavior; unrelated model paths." + }, + { + "left": "issue:43824", + "right": "issue:43976", + "accept": false, + "reason": "Missing Qwen2_5_VL export versus Python version support issue; no shared bug." + }, + { + "left": "issue:45542", + "right": "issue:45800", + "accept": false, + "reason": "TensorFlow backend packaging issue versus torch/transformers version incompatibility; not the same problem." + }, + { + "left": "issue:42796", + "right": "issue:43441", + "accept": false, + "reason": "Same Ministral family, but tokenizer BPE marker handling and FlashAttention loading failure are distinct issues." + }, + { + "left": "issue:43525", + "right": "issue:43881", + "accept": false, + "reason": "Llama4Config pad_token_id AttributeError versus glm-4v-9b loading failure; unrelated models." + }, + { + "left": "issue:46291", + "right": "issue:46399", + "accept": false, + "reason": "Both involve kernels, but kernels v0.15 revision/version requirements and Qwen3.5 rotary registration 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": 23, + "review_unit_index": 42, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:41628", + "issue:42491", + "issue:42886", + "issue:43299", + "issue:43525", + "issue:43526", + "issue:43531", + "issue:43540", + "issue:43643", + "issue:43784", + "issue:43976", + "issue:44016", + "issue:44188", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44297", + "issue:44336", + "issue:44373", + "issue:44451", + "issue:44485", + "issue:44496", + "issue:44589", + "issue:44871", + "issue:45003", + "issue:45072", + "issue:45083", + "issue:45103", + "issue:45325", + "issue:45362", + "issue:45397", + "issue:45542", + "issue:45561", + "issue:45563", + "issue:45588", + "issue:45593", + "issue:45823", + "issue:46132", + "issue:46399", + "issue:46614", + "issue:46655", + "issue:46665", + "issue:46726", + "issue:46752", + "issue:46821", + "issue:46829", + "issue:46851" + ], + "soft_pairs": [ + "issue:44188|issue:44336", + "issue:44016|issue:45823", + "issue:44589|issue:45362", + "issue:44336|issue:44373", + "issue:43525|issue:43643", + "issue:46655|issue:46726", + "issue:45325|issue:46851", + "issue:41628|issue:43976", + "issue:44297|issue:45397", + "issue:43784|issue:45542", + "issue:44246|issue:44496", + "issue:45083|issue:46851", + "issue:45561|issue:45563", + "issue:46399|issue:46851", + "issue:45103|issue:46399", + "issue:43526|issue:45003", + "issue:46614|issue:46665", + "issue:43540|issue:46752", + "issue:45072|issue:46829", + "issue:43531|issue:45362", + "issue:44295|issue:45397", + "issue:42491|issue:43299", + "issue:44246|issue:44871", + "issue:44242|issue:45593", + "issue:44336|issue:44485", + "issue:45588|issue:46851", + "issue:45003|issue:46821", + "issue:42886|issue:46132", + "issue:44451|issue:46752" + ], + "prepared_review_unit_hash": "e6a61b56e76e7646d01e12fe4b4342e03294878b4357cc37cb8048cae2ebda90", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35845, + "estimated_input_tokens": 8962, + "estimated_eval_tokens": 18180 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These items are mostly unrelated Transformers bugs across different models, configs, processors, and utilities. None of the soft pairs look like true duplicates.", + "confidence": 0.87, + "canonical_issue_reason": "Issue #46851 is the clearest concrete runtime bug in the set; the others are mostly unrelated, documentation, versioning, or different code paths.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46851 is the most actionable representative issue here: it names a specific model family, a specific failure mode, and a concrete input-dependent code path.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44188", + "right": "issue:44336", + "accept": false, + "reason": "Different subsystems: torch.compile attention-kernel branching vs ANSI color output in loading_report." + }, + { + "left": "issue:44016", + "right": "issue:45823", + "accept": false, + "reason": "Notebook syntax error is unrelated to Gemma4 device-map forward dtype/device mismatch." + }, + { + "left": "issue:44589", + "right": "issue:45362", + "accept": false, + "reason": "Float8 storage deserialization error and Qwen3.5 chat crash are different failure modes and code paths." + }, + { + "left": "issue:44336", + "right": "issue:44373", + "accept": false, + "reason": "Terminal ANSI leakage is unrelated to a docstring typo for position_ids." + }, + { + "left": "issue:43525", + "right": "issue:43643", + "accept": false, + "reason": "Both mention missing fields, but one is Llama4Config pad_token_id and the other is trust_remote_code AutoConfig field loss; not the same bug." + }, + { + "left": "issue:46655", + "right": "issue:46726", + "accept": false, + "reason": "Incorrect video padding calculation is different from AutoProcessor\u2019s missing-vision-backend import error." + }, + { + "left": "issue:45325", + "right": "issue:46851", + "accept": false, + "reason": "Both involve position_ids, but they affect different Qwen model families and different logic." + }, + { + "left": "issue:41628", + "right": "issue:43976", + "accept": false, + "reason": "ImportError for AutoImageProcessor is unrelated to the Python version support regression." + }, + { + "left": "issue:44297", + "right": "issue:45397", + "accept": false, + "reason": "Tokenizer save_pretrained metadata mismatch is unrelated to Gemma-4 zero3 loading." + }, + { + "left": "issue:43784", + "right": "issue:45542", + "accept": false, + "reason": "Sentence-transformers import NameError and undefined tf backend from tensorboard-only install are separate dependency/import issues." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import latency and eos_token_id inconsistency are unrelated." + }, + { + "left": "issue:45083", + "right": "issue:46851", + "accept": false, + "reason": "Qwen3-Omni-MoE feature-length helper behavior is different from Qwen3.5 variable-length position_ids handling." + }, + { + "left": "issue:45561", + "right": "issue:45563", + "accept": false, + "reason": "A pytest-xdist race in testing utils is unrelated to a stale generate() warning." + }, + { + "left": "issue:46399", + "right": "issue:46851", + "accept": false, + "reason": "Kernel registration/decorator crash in Qwen3.5 is a different concrete bug than variable-length position_ids." + }, + { + "left": "issue:45103", + "right": "issue:46399", + "accept": false, + "reason": "auto_docstring annotation handling and apply_rotary_pos_emb registration are unrelated code paths." + }, + { + "left": "issue:43526", + "right": "issue:45003", + "accept": false, + "reason": "BeitImageProcessorFast label reduction and unsafe sys.modules access are different bugs." + }, + { + "left": "issue:46614", + "right": "issue:46665", + "accept": false, + "reason": "Stale phonemizer backend state is unrelated to from_pretrained behavior inside Python classes." + }, + { + "left": "issue:43540", + "right": "issue:46752", + "accept": false, + "reason": "Qwen3OmniMoe video-input ValueError and empty apply_chat_template IndexError are different failures." + }, + { + "left": "issue:45072", + "right": "issue:46829", + "accept": false, + "reason": "dtype mismatch in inference and hidden_states/output field confusion are unrelated." + }, + { + "left": "issue:43531", + "right": "issue:45362", + "accept": false, + "reason": "Qwen3-MoE sliding_window behavior and Qwen3.5 chat crashes are different issues." + }, + { + "left": "issue:44295", + "right": "issue:45397", + "accept": false, + "reason": "position_ids buffer access and Gemma-4 zero3 loading are not the same bug." + }, + { + "left": "issue:42491", + "right": "issue:43299", + "accept": false, + "reason": "Both involve Qwen MoE loading, but they are different version/model breakages and not the same concrete change." + }, + { + "left": "issue:44246", + "right": "issue:44871", + "accept": false, + "reason": "Import latency and eos_token_id configuration mismatch are unrelated." + }, + { + "left": "issue:44242", + "right": "issue:45593", + "accept": false, + "reason": "Load-balancing loss gating and D-FINE auxiliary-loss omission are different model-loss bugs." + }, + { + "left": "issue:44336", + "right": "issue:44485", + "accept": false, + "reason": "ANSI output formatting has no relation to GLM-5 RoPE implementation." + }, + { + "left": "issue:45588", + "right": "issue:46851", + "accept": false, + "reason": "flash_attention sink-less-model crash is a different code path than Qwen3.5 variable-length position_ids." + }, + { + "left": "issue:45003", + "right": "issue:46821", + "accept": false, + "reason": "sys.modules access safety and TimesFM window_size attribute handling are unrelated." + }, + { + "left": "issue:42886", + "right": "issue:46132", + "accept": false, + "reason": "Offline tokenizer cache loading and AttentionInterface.register behavior change are unrelated." + }, + { + "left": "issue:44451", + "right": "issue:46752", + "accept": false, + "reason": "ScandiBERT loading failure is unrelated to empty chat template handling." + } + ] + }, + "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": 23, + "review_unit_index": 43, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:36246", + "issue:41950", + "issue:42757", + "issue:43257", + "issue:43262", + "issue:43295", + "issue:43475", + "issue:43526", + "issue:43540", + "issue:43550", + "issue:43596", + "issue:43901", + "issue:44038", + "issue:44162", + "issue:44220", + "issue:44242", + "issue:44246", + "issue:44295", + "issue:44351", + "issue:44466", + "issue:44479", + "issue:44496", + "issue:44623", + "issue:44749", + "issue:44779", + "issue:44857", + "issue:44877", + "issue:44987", + "issue:45003", + "issue:45137", + "issue:45200", + "issue:45292", + "issue:45335", + "issue:45375", + "issue:45459", + "issue:45561", + "issue:45584", + "issue:45600", + "issue:45759", + "issue:45859", + "issue:46369", + "issue:46399", + "issue:46506", + "issue:46655", + "issue:46726", + "issue:46897" + ], + "soft_pairs": [ + "issue:43550|issue:46506", + "issue:44779|issue:45292", + "issue:36246|issue:42757", + "issue:36246|issue:44351", + "issue:41950|issue:44220", + "issue:43901|issue:45003", + "issue:43257|issue:45137", + "issue:45003|issue:45561", + "issue:44496|issue:45003", + "issue:43540|issue:46655", + "issue:44162|issue:45200", + "issue:45375|issue:45759", + "issue:43262|issue:44246", + "issue:44479|issue:46655", + "issue:44295|issue:44749", + "issue:43475|issue:46369", + "issue:45859|issue:46399", + "issue:45137|issue:45600", + "issue:36246|issue:43596", + "issue:44877|issue:45375", + "issue:43295|issue:45292", + "issue:45292|issue:45584", + "issue:44466|issue:45335", + "issue:44623|issue:46726", + "issue:44242|issue:44857", + "issue:43526|issue:46897", + "issue:44038|issue:46506", + "issue:29870|issue:30333", + "issue:44987|issue:45459" + ], + "prepared_review_unit_hash": "005cbf72b70addd76a4062be5de9e3f093ee1f56ba7bf4835d18fc6dc4b848d3", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36554, + "estimated_input_tokens": 9139, + "estimated_eval_tokens": 18534 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous collection of unrelated transformer issues. None of the soft-similarity pairs look like the same underlying bug or change, so there is no true duplicate cluster to merge.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No single issue is a strong cluster representative: the items span unrelated models, processors, configs, and runtime errors, so any choice would be arbitrary.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43550", + "right": "issue:46506", + "accept": false, + "reason": "Different model/runtime failures: torch.compile+SDPA vs CPU/CUDA result mismatch." + }, + { + "left": "issue:44779", + "right": "issue:45292", + "accept": false, + "reason": "Tokenizer correctness regression is unrelated to resize_token_embeddings not updating outputs." + }, + { + "left": "issue:36246", + "right": "issue:42757", + "accept": false, + "reason": "Separate import errors for different symbols and packages." + }, + { + "left": "issue:36246", + "right": "issue:44351", + "accept": false, + "reason": "Different missing-import symbols; not the same bug." + }, + { + "left": "issue:41950", + "right": "issue:44220", + "accept": false, + "reason": "Video pipeline image-processor lookup issue is unrelated to _torch_extract_fbank_features()." + }, + { + "left": "issue:43901", + "right": "issue:45003", + "accept": false, + "reason": "Docs mismatch for a pipeline is unrelated to unsafe sys.modules access." + }, + { + "left": "issue:43257", + "right": "issue:45137", + "accept": false, + "reason": "Both involve Deepspeed-adjacent failures, but the concrete bugs are different." + }, + { + "left": "issue:45003", + "right": "issue:45561", + "accept": false, + "reason": "Unsafe module lookup and xdist file races are unrelated issues." + }, + { + "left": "issue:44496", + "right": "issue:45003", + "accept": false, + "reason": "Unrecognized model/config loading failure is not the same as sys.modules access." + }, + { + "left": "issue:43540", + "right": "issue:46655", + "accept": false, + "reason": "Different video-processor bugs in different model families and code paths." + }, + { + "left": "issue:44162", + "right": "issue:45200", + "accept": false, + "reason": "ESM2 breakage and Gemma 4 mm_token_type_ids behavior are unrelated." + }, + { + "left": "issue:45375", + "right": "issue:45759", + "accept": false, + "reason": "Strict config field handling differs from AutoModel text_config unwrapping." + }, + { + "left": "issue:43262", + "right": "issue:44246", + "accept": false, + "reason": "Audio chat-template sampling-rate bug is unrelated to slow import behavior." + }, + { + "left": "issue:44479", + "right": "issue:46655", + "accept": false, + "reason": "Both are video-related, but they concern different model-specific regressions and fixes." + }, + { + "left": "issue:44295", + "right": "issue:44749", + "accept": false, + "reason": "Position_ids buffer access and slower filtering after upgrade are unrelated." + }, + { + "left": "issue:43475", + "right": "issue:46369", + "accept": false, + "reason": "SAM 3 attribute error and SAM lite text memory overflow are different failures." + }, + { + "left": "issue:45859", + "right": "issue:46399", + "accept": false, + "reason": "CPU-offload incompatibility and missing decorator registration are distinct code-path problems." + }, + { + "left": "issue:45137", + "right": "issue:45600", + "accept": false, + "reason": "DeepSpeed deque underflow is unrelated to stale auto_mappings CI references." + }, + { + "left": "issue:36246", + "right": "issue:43596", + "accept": false, + "reason": "ImportError for a vision processor is unrelated to a ZeRO3 BertModel index error." + }, + { + "left": "issue:44877", + "right": "issue:45375", + "accept": false, + "reason": "Different strict-config problems affecting different model configs and fields." + }, + { + "left": "issue:43295", + "right": "issue:45292", + "accept": false, + "reason": "Regression in processor.tokenizer/image handling is unrelated to embedding resize propagation." + }, + { + "left": "issue:45292", + "right": "issue:45584", + "accept": false, + "reason": "Embedding resize behavior does not match empty-transcription generation failure." + }, + { + "left": "issue:44466", + "right": "issue:45335", + "accept": false, + "reason": "Both mention embedding serialization/resize, but the concrete issues and affected modules differ." + }, + { + "left": "issue:44623", + "right": "issue:46726", + "accept": false, + "reason": "Missing processor save files is not the same as missing vision backend handling." + }, + { + "left": "issue:44242", + "right": "issue:44857", + "accept": false, + "reason": "MoE router-loss behavior and an AMP CUDA image-loss crash are unrelated." + }, + { + "left": "issue:43526", + "right": "issue:46897", + "accept": false, + "reason": "BEiT label reduction bug is unrelated to Florence2 training-loss shifting." + }, + { + "left": "issue:44038", + "right": "issue:46506", + "accept": false, + "reason": "Qwen3-VL-Moe regression and CPU/CUDA output divergence are different problems." + }, + { + "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": "Model-loading failure with physical-intelligence/fast is unrelated to tokenizer protobuf error masking." + } + ] + }, + "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": 23, + "review_unit_index": 44, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:42199", + "issue:42886", + "issue:43065", + "issue:43352", + "issue:43408", + "issue:43475", + "issue:43526", + "issue:43540", + "issue:43653", + "issue:43723", + "issue:43906", + "issue:43931", + "issue:44060", + "issue:44077", + "issue:44162", + "issue:44188", + "issue:44297", + "issue:44368", + "issue:44479", + "issue:44483", + "issue:44623", + "issue:44877", + "issue:44977", + "issue:44995", + "issue:45103", + "issue:45200", + "issue:45229", + "issue:45250", + "issue:45397", + "issue:45412", + "issue:45563", + "issue:45588", + "issue:45656", + "issue:45685", + "issue:45987", + "issue:46082", + "issue:46123", + "issue:46170", + "issue:46202", + "issue:46421", + "issue:46431", + "issue:46710", + "issue:46829", + "issue:46832", + "issue:46851", + "issue:46858", + "issue:46894" + ], + "soft_pairs": [ + "issue:43352|issue:45250", + "issue:42199|issue:46851", + "issue:43526|issue:46894", + "issue:44060|issue:44077", + "issue:44368|issue:44977", + "issue:45412|issue:46123", + "issue:43653|issue:44162", + "issue:44479|issue:46170", + "issue:42886|issue:45229", + "issue:45397|issue:45656", + "issue:43065|issue:43408", + "issue:43475|issue:43540", + "issue:44877|issue:46170", + "issue:46170|issue:46832", + "issue:44368|issue:45563", + "issue:43723|issue:46710", + "issue:44995|issue:45200", + "issue:43931|issue:44368", + "issue:44188|issue:46202", + "issue:44483|issue:46202", + "issue:46829|issue:46858", + "issue:45987|issue:46082", + "issue:43906|issue:46202", + "issue:44995|issue:45588", + "issue:44188|issue:46421", + "issue:45685|issue:46431", + "issue:45103|issue:46858", + "issue:43065|issue:46123", + "issue:44297|issue:44623" + ], + "prepared_review_unit_hash": "18463ae58cec961f8ad88813298466d33c3fea48c14c82405b791a4284a60626", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35968, + "estimated_input_tokens": 8992, + "estimated_eval_tokens": 18240 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "Most pairs are unrelated lookalikes and should be kept separate. The only credible duplicate is the pair about `StaticCache.get_seq_length()` returning a tensor instead of an `int`.", + "confidence": 0.79, + "canonical_issue_reason": "Issue 45987 is the cleanest root-cause report for the `get_seq_length()` return-type bug; it states the contract violation directly and can subsume the backend-specific failure in 46858.", + "canonical_pr_reason": null, + "best_issue_reason": "45987 is the strongest canonical issue candidate overall because it describes the underlying API bug broadly and precisely, rather than just one downstream symptom.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43352", + "right": "issue:45250", + "accept": false, + "reason": "Different scopes: model-specific Flash Attention support error vs a generic FA2 topic/request." + }, + { + "left": "issue:42199", + "right": "issue:46851", + "accept": false, + "reason": "Completely unrelated models and failures: DETR cardinality error vs Qwen3.5 variable-length position_ids." + }, + { + "left": "issue:43526", + "right": "issue:46894", + "accept": false, + "reason": "Different bugs in unrelated pipelines; no shared code-path or symptom." + }, + { + "left": "issue:44060", + "right": "issue:44077", + "accept": false, + "reason": "One is a tied-weights warning in Qwen3-Next; the other is a patchtsmixer post_init policy issue." + }, + { + "left": "issue:44368", + "right": "issue:44977", + "accept": false, + "reason": "Both mention Qwen3.5, but one is a weight-tying warning and the other is a flash-attention generation failure." + }, + { + "left": "issue:45412", + "right": "issue:46123", + "accept": false, + "reason": "RT-DETR memory cleanup and MaskGenerationPipeline batch handling are unrelated." + }, + { + "left": "issue:43653", + "right": "issue:44162", + "accept": false, + "reason": "Tokenizer special-token registration vs ESM2 breakage; no common underlying bug." + }, + { + "left": "issue:44479", + "right": "issue:46170", + "accept": false, + "reason": "Video-input regression in Qwen VL models vs a Windows/ROCm distributed import failure." + }, + { + "left": "issue:42886", + "right": "issue:45229", + "accept": false, + "reason": "Offline tokenizer cache loading and Gemma multi-GPU OOM are different failure modes." + }, + { + "left": "issue:45397", + "right": "issue:45656", + "accept": false, + "reason": "Gemma-4 loading/ZeRO-3 issue is not the same as a DeepSpeed double optimizer-step bug." + }, + { + "left": "issue:43065", + "right": "issue:43408", + "accept": false, + "reason": "Both touch SAM3, but one is a dummy conv implementation detail and the other is a model-type mismatch warning." + }, + { + "left": "issue:43475", + "right": "issue:43540", + "accept": false, + "reason": "SAM3 video attribute error and Qwen3OmniMoe video ValueError are unrelated." + }, + { + "left": "issue:44877", + "right": "issue:46170", + "accept": false, + "reason": "Strict config loading for granite_speech is unrelated to the Windows ROCm dependency error." + }, + { + "left": "issue:46170", + "right": "issue:46832", + "accept": false, + "reason": "Different problems: import/dependency failure vs RT-DETR forward-shape TypeError." + }, + { + "left": "issue:44368", + "right": "issue:45563", + "accept": false, + "reason": "Weight-tying warning and stale `num_return_sequences` warning do not share the same bug." + }, + { + "left": "issue:43723", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer loading regression and model accuracy regression are different issues." + }, + { + "left": "issue:44995", + "right": "issue:45200", + "accept": false, + "reason": "GlmMoeDsa cache crash vs Gemma 4 text-only fine-tuning mask-token defaults." + }, + { + "left": "issue:43931", + "right": "issue:44368", + "accept": false, + "reason": "Model weight-shape mismatch is unrelated to a tied-embeddings warning." + }, + { + "left": "issue:44188", + "right": "issue:46202", + "accept": false, + "reason": "torch.compile attention-kernel divergence and Gemma token roundtrip failure are not the same bug." + }, + { + "left": "issue:44483", + "right": "issue:46202", + "accept": false, + "reason": "API request rejection and tokenizer/pipeline roundtrip behavior are unrelated." + }, + { + "left": "issue:46829", + "right": "issue:46858", + "accept": false, + "reason": "VideoPrism output packaging bug vs cache-length arithmetic bug; different code paths and symptoms." + }, + { + "left": "issue:45987", + "right": "issue:46082", + "accept": false, + "reason": "StaticCache return-type bug and LlamaConfig head_dim validation are unrelated." + }, + { + "left": "issue:43906", + "right": "issue:46202", + "accept": false, + "reason": "An isolated reproduction issue is not the same as the Gemma roundtrip failure reported here." + }, + { + "left": "issue:44995", + "right": "issue:45588", + "accept": false, + "reason": "Second-forward cache corruption in GlmMoeDsa is unrelated to flash_attention.py crashing on `s_aux=None`." + }, + { + "left": "issue:44188", + "right": "issue:46421", + "accept": false, + "reason": "Both involve compile/cache words, but one is kernel divergence and the other is lazy cache init causing recompilation." + }, + { + "left": "issue:45685", + "right": "issue:46431", + "accept": false, + "reason": "MPS histogram/int issue is unrelated to missing chat-template support in a processor." + }, + { + "left": "issue:45103", + "right": "issue:46858", + "accept": false, + "reason": "Docstring annotation handling and static-cache sequence-length arithmetic are unrelated." + }, + { + "left": "issue:43065", + "right": "issue:46123", + "accept": false, + "reason": "SAM3 pixel-decoder implementation details and MaskGenerationPipeline batch termination are unrelated." + }, + { + "left": "issue:44297", + "right": "issue:44623", + "accept": false, + "reason": "Tokenizer save metadata mismatch and processor save missing files are both save_pretrained-adjacent but not the same bug." + }, + { + "left": "issue:45987", + "right": "issue:46858", + "accept": true, + "reason": "Same underlying bug: `get_seq_length()` returns a tensor instead of an `int`, and 46858 is the downstream static-cache/compile manifestation of the same contract violation." + } + ] + }, + "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": 23, + "review_unit_index": 45, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28282", + "issue:30990", + "issue:39692", + "issue:41628", + "issue:42371", + "issue:42994", + "issue:43232", + "issue:43441", + "issue:43475", + "issue:43482", + "issue:43525", + "issue:43643", + "issue:43646", + "issue:43673", + "issue:43701", + "issue:43723", + "issue:43986", + "issue:43994", + "issue:44038", + "issue:44162", + "issue:44188", + "issue:44361", + "issue:44373", + "issue:44623", + "issue:44749", + "issue:44869", + "issue:44871", + "issue:44933", + "issue:44964", + "issue:45092", + "issue:45137", + "issue:45230", + "issue:45310", + "issue:45356", + "issue:45375", + "issue:45397", + "issue:45406", + "issue:45478", + "issue:45656", + "issue:46132", + "issue:46202", + "issue:46326", + "issue:46463", + "issue:46726", + "issue:46752", + "issue:46821", + "issue:46851", + "issue:46877" + ], + "soft_pairs": [ + "issue:44361|issue:44749", + "issue:43994|issue:46726", + "issue:44749|issue:45656", + "issue:43701|issue:44749", + "issue:44871|issue:46132", + "issue:44623|issue:45656", + "issue:43646|issue:44162", + "issue:43441|issue:46752", + "issue:45375|issue:45478", + "issue:42994|issue:43482", + "issue:43475|issue:43643", + "issue:43232|issue:43673", + "issue:44188|issue:46851", + "issue:28282|issue:30990", + "issue:41628|issue:44933", + "issue:43475|issue:46821", + "issue:45310|issue:45375", + "issue:44964|issue:45137", + "issue:45230|issue:45397", + "issue:43986|issue:46726", + "issue:44869|issue:45356", + "issue:44038|issue:44623", + "issue:39692|issue:46726", + "issue:46463|issue:46877", + "issue:44373|issue:46851", + "issue:42994|issue:43525", + "issue:42371|issue:45092", + "issue:43723|issue:46326", + "issue:45406|issue:46202" + ], + "prepared_review_unit_hash": "d8c870de3ab39edd36f51ee753c103fb92c5b7654cdcf687609afb282a083dc9", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36049, + "estimated_input_tokens": 9013, + "estimated_eval_tokens": 18282 + }, + "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 soft duplicate that looks credible is the Qwen3.5-Moe config/from_pretrained regression pair; the rest are different models, different code paths, or different symptoms.", + "confidence": 0.9, + "canonical_issue_reason": "issue:45375 is the better canonical issue for the Qwen3.5-Moe regression because it identifies the likely root cause directly: a missing `deepstack_visual_indexes` field being dropped by `@strict`.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45310 is the best overall representative because it is the clearest user-facing report of the Qwen3.5-Moe `from_pretrained` failure and provides the broadest symptom framing.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44361", + "right": "issue:44749", + "accept": false, + "reason": "Different problems: a tokenizer AttributeError versus a dataset/filtering slowdown regression." + }, + { + "left": "issue:43994", + "right": "issue:46726", + "accept": false, + "reason": "Both involve vision models, but one is wrong output from AutoModel/pipeline while the other is a missing-backend AutoProcessor error; not the same bug." + }, + { + "left": "issue:44749", + "right": "issue:45656", + "accept": false, + "reason": "Unrelated: data filtering performance versus DeepSpeed optimizer stepping twice." + }, + { + "left": "issue:43701", + "right": "issue:44749", + "accept": false, + "reason": "Checkpoint key mismatch is unrelated to the reported slowdown regression." + }, + { + "left": "issue:44871", + "right": "issue:46132", + "accept": false, + "reason": "Gemma eos token config mismatch and AttentionInterface behavior change are different code paths." + }, + { + "left": "issue:44623", + "right": "issue:45656", + "accept": false, + "reason": "Processor save files issue is unrelated to DeepSpeed optimizer stepping." + }, + { + "left": "issue:43646", + "right": "issue:44162", + "accept": false, + "reason": "Different failures: missing fields in AutoConfig with remote code versus ESM2 breakage." + }, + { + "left": "issue:43441", + "right": "issue:46752", + "accept": false, + "reason": "FlashAttention/Ministral regression and empty-chat-template crash are unrelated." + }, + { + "left": "issue:45375", + "right": "issue:45478", + "accept": true, + "reason": "Same underlying Qwen3.5-Moe loading/config regression; the missing field in one issue plausibly explains the from_pretrained error in the other." + }, + { + "left": "issue:42994", + "right": "issue:43482", + "accept": false, + "reason": "Quantized model saving and GGUF loading are different operations with different failure modes." + }, + { + "left": "issue:43475", + "right": "issue:43643", + "accept": false, + "reason": "SAM3 vision encoder attribute error is unrelated to trust_remote_code config field loss." + }, + { + "left": "issue:43232", + "right": "issue:43673", + "accept": false, + "reason": "Both are generation-related, but they describe different internal bugs and scenarios." + }, + { + "left": "issue:44188", + "right": "issue:46851", + "accept": false, + "reason": "Attention-kernel divergence under torch.compile is unrelated to Qwen3.5 variable-length position_ids handling." + }, + { + "left": "issue:28282", + "right": "issue:30990", + "accept": false, + "reason": "ImportError for missing PyTorch and a Sentence Transformers loading hang are not the same issue." + }, + { + "left": "issue:41628", + "right": "issue:44933", + "accept": false, + "reason": "Both mention imports, but one is a missing top-level AutoImageProcessor symbol and the other is a bad/nonexistent image_utils import." + }, + { + "left": "issue:43475", + "right": "issue:46821", + "accept": false, + "reason": "SAM3 encoder attribute error and TimesFM window_size AttributeError are unrelated model bugs." + }, + { + "left": "issue:45310", + "right": "issue:45375", + "accept": true, + "reason": "Same Qwen3.5-Moe regression: one reports the user-facing from_pretrained failure, the other identifies the missing config field root cause." + }, + { + "left": "issue:44964", + "right": "issue:45137", + "accept": false, + "reason": "Phi-4 multimodal loading and DeepSpeed ZeRO3 deque underflow are unrelated." + }, + { + "left": "issue:45230", + "right": "issue:45397", + "accept": false, + "reason": "A generic bug report is not the same as a Gemma-4 ZeRO3 loading failure." + }, + { + "left": "issue:43986", + "right": "issue:46726", + "accept": false, + "reason": "Both are about missing vision dependencies, but one is a video-model AutoProcessor crash and the other is an unrecognized image processor error; too different to merge." + }, + { + "left": "issue:44869", + "right": "issue:45356", + "accept": false, + "reason": "Whisper timestamp decoding crash and Kimi tokenizer regression affect different tokenization paths." + }, + { + "left": "issue:44038", + "right": "issue:44623", + "accept": false, + "reason": "Qwen3-VL-Moe from_pretrained failure and processor.save_pretrained file omission are not the same bug." + }, + { + "left": "issue:39692", + "right": "issue:46726", + "accept": false, + "reason": "Documentation example problems and missing-backend AutoProcessor runtime error are different issues." + }, + { + "left": "issue:46463", + "right": "issue:46877", + "accept": false, + "reason": "T5 decoder-input validation messaging and auto_docstring stdout lint noise are unrelated." + }, + { + "left": "issue:44373", + "right": "issue:46851", + "accept": false, + "reason": "A docstring bug about position_ids is not the same as the Qwen3.5 variable-length input failure." + }, + { + "left": "issue:42994", + "right": "issue:43525", + "accept": false, + "reason": "Quantized saving failure and missing pad_token_id on Llama4Config are different regressions." + }, + { + "left": "issue:42371", + "right": "issue:45092", + "accept": false, + "reason": "TF32 API guidance and InternVL2 meta-init incompatibility are unrelated." + }, + { + "left": "issue:43723", + "right": "issue:46326", + "accept": false, + "reason": "Tokenizer loading failure and continue_final_message behavior change are different tokenizer issues." + }, + { + "left": "issue:45406", + "right": "issue:46202", + "accept": false, + "reason": "Gemma4Processor missing _tokenizer and Gemma pipeline roundtrip on \u2581 are distinct bugs." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The only accepted duplicate cluster is the Qwen3.5-Moe regression, which is supported by matching titles and the explicit root-cause note in issue:45375. The rest of the rejections are appropriately cautious and do not overclaim similarity." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 46, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:30333", + "issue:30990", + "issue:33290", + "issue:35141", + "issue:36010", + "issue:41628", + "issue:42175", + "issue:42673", + "issue:42831", + "issue:42898", + "issue:42994", + "issue:43232", + "issue:43278", + "issue:43479", + "issue:43504", + "issue:43704", + "issue:43784", + "issue:43824", + "issue:43873", + "issue:43976", + "issue:44038", + "issue:44222", + "issue:44351", + "issue:44496", + "issue:44568", + "issue:44623", + "issue:44792", + "issue:44857", + "issue:44933", + "issue:44945", + "issue:45081", + "issue:45103", + "issue:45137", + "issue:45310", + "issue:45405", + "issue:45561", + "issue:45593", + "issue:45600", + "issue:45712", + "issue:45753", + "issue:46077", + "issue:46123", + "issue:46202", + "issue:46399", + "issue:46693", + "issue:46701", + "issue:46829", + "issue:46877" + ], + "soft_pairs": [ + "issue:43873|issue:44945", + "issue:44945|issue:45753", + "issue:43784|issue:46399", + "issue:41628|issue:44351", + "issue:44857|issue:46877", + "issue:45593|issue:46877", + "issue:42673|issue:43704", + "issue:44038|issue:45310", + "issue:36010|issue:41628", + "issue:44222|issue:45310", + "issue:43278|issue:45137", + "issue:46202|issue:46701", + "issue:42994|issue:43824", + "issue:45081|issue:45103", + "issue:42831|issue:44222", + "issue:43232|issue:44792", + "issue:43504|issue:45561", + "issue:42175|issue:44933", + "issue:45561|issue:45600", + "issue:43976|issue:44933", + "issue:44496|issue:45561", + "issue:30333|issue:33290", + "issue:45405|issue:46877", + "issue:30990|issue:35141", + "issue:42898|issue:44568", + "issue:42175|issue:45712", + "issue:43479|issue:46829", + "issue:46077|issue:46123", + "issue:44623|issue:45310", + "issue:42673|issue:46693" + ], + "prepared_review_unit_hash": "3d51abb6f295b79d32931c50ccf270a1a752b10e7747dbafcfae8f7503a4491b", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 30, + "serialized_chars": 36537, + "estimated_input_tokens": 9135, + "estimated_eval_tokens": 18526 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a grab-bag of unrelated Transformers issues; the only clear duplicate is the repeated Qwen3ForCausalLM VRAM leak report. Most other soft matches are merely thematically similar (imports, model loading, tokenizer regressions, FP8, etc.) and should stay separate.", + "confidence": 0.63, + "canonical_issue_reason": "issue:41628 is the best anchor for the cluster\u2019s recurring import/export failures: it is a generic, high-signal missing-symbol bug in `transformers` and sits closest to several other import-related reports.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:41628 is the most representative issue in the import-error subgroup and is broader/more central than the more model-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43873", + "right": "issue:44945", + "accept": false, + "reason": "Different symptoms and code paths: quantization/offloading vs pipeline-parallel generation output correctness." + }, + { + "left": "issue:44945", + "right": "issue:45753", + "accept": false, + "reason": "One is pipeline-parallel output corruption; the other is an image-specific infinite loop in Qwen3_5." + }, + { + "left": "issue:43784", + "right": "issue:46399", + "accept": false, + "reason": "Both are import/runtime errors, but they affect different packages and mechanisms (`nn` import vs rotary registration)." + }, + { + "left": "issue:41628", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different missing symbols and likely different fixes." + }, + { + "left": "issue:44857", + "right": "issue:46877", + "accept": false, + "reason": "Unrelated: AMP/CUDA loss crash versus docstring logging on import." + }, + { + "left": "issue:45593", + "right": "issue:46877", + "accept": false, + "reason": "Different subsystems entirely: model loss behavior vs import-time doc-lint output." + }, + { + "left": "issue:42673", + "right": "issue:43704", + "accept": true, + "reason": "Same exact reported bug: Qwen3ForCausalLM VRAM leak when used in multiple dataloader threads." + }, + { + "left": "issue:44038", + "right": "issue:45310", + "accept": false, + "reason": "Related model family, but different model versions and different failure conditions; not the same concrete bug." + }, + { + "left": "issue:36010", + "right": "issue:41628", + "accept": false, + "reason": "Both are missing-import complaints, but they are different exported symbols and separate failures." + }, + { + "left": "issue:44222", + "right": "issue:45310", + "accept": false, + "reason": "Saving FP8 MoE models vs `from_pretrained` loading failure are distinct code paths." + }, + { + "left": "issue:43278", + "right": "issue:45137", + "accept": false, + "reason": "Training/eval dtype mismatch is unrelated to the DeepSpeed ZeRO3 deque error." + }, + { + "left": "issue:46202", + "right": "issue:46701", + "accept": false, + "reason": "Both involve Gemma-family pipelines, but one is tokenizer roundtrip and the other is an image-text pipeline output issue." + }, + { + "left": "issue:42994", + "right": "issue:43824", + "accept": false, + "reason": "Quantized model saving and a missing `Qwen2_5_VLForConditionalGeneration` import are unrelated." + }, + { + "left": "issue:45081", + "right": "issue:45103", + "accept": false, + "reason": "Both raise AttributeError, but in unrelated utilities and code paths." + }, + { + "left": "issue:42831", + "right": "issue:44222", + "accept": false, + "reason": "FP8 accuracy regression versus FP8 save/load failure are different bugs." + }, + { + "left": "issue:43232", + "right": "issue:44792", + "accept": false, + "reason": "Generation kwargs handling after GPU sync is unrelated to the Janus image-generation test failure." + }, + { + "left": "issue:43504", + "right": "issue:45561", + "accept": false, + "reason": "Model preset loading bug vs pytest-xdist race in testing utils; no shared underlying defect." + }, + { + "left": "issue:42175", + "right": "issue:44933", + "accept": false, + "reason": "Packaging/backend installation issue vs a missing import from `image_utils`; different problems." + }, + { + "left": "issue:45561", + "right": "issue:45600", + "accept": false, + "reason": "Parallel test-file race vs stale auto-mapping references; unrelated." + }, + { + "left": "issue:43976", + "right": "issue:44933", + "accept": false, + "reason": "Python-version compatibility issue vs missing import from `image_utils`; not the same bug." + }, + { + "left": "issue:44496", + "right": "issue:45561", + "accept": false, + "reason": "Model config loading failure vs test utility race; unrelated." + }, + { + "left": "issue:30333", + "right": "issue:33290", + "accept": false, + "reason": "MLflow reporting on failed training vs OOM in Adafactor/DeepSpeed are different failures." + }, + { + "left": "issue:45405", + "right": "issue:46877", + "accept": false, + "reason": "Dependency-version bump issue vs docstring import-time logging; unrelated." + }, + { + "left": "issue:30990", + "right": "issue:35141", + "accept": false, + "reason": "Loading hang vs embedding reinitialization on resize are separate model behaviors." + }, + { + "left": "issue:42898", + "right": "issue:44568", + "accept": false, + "reason": "Both are tokenizer regressions in v5, but they concern different APIs and symptoms." + }, + { + "left": "issue:42175", + "right": "issue:45712", + "accept": false, + "reason": "Extras/backend packaging issue vs dummy class leakage/check-repo failure; different root causes." + }, + { + "left": "issue:43479", + "right": "issue:46829", + "accept": false, + "reason": "Incorrect default multimodal config initialization vs output-field mismatch in video classification are unrelated." + }, + { + "left": "issue:46077", + "right": "issue:46123", + "accept": false, + "reason": "Attention-mask construction bug vs pipeline batching/drop-results bug; not the same change." + }, + { + "left": "issue:44623", + "right": "issue:45310", + "accept": false, + "reason": "Processor save_pretrained file omission vs Qwen3.5 from_pretrained error are different code paths." + }, + { + "left": "issue:42673", + "right": "issue:46693", + "accept": false, + "reason": "VRAM leak in multi-threaded dataloaders vs FA2 long-context performance regression; unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative on the soft edges. The only accepted pair, issue:42673 vs issue:43704, is a clear duplicate by title. The other pair rejections are cautious and consistent with the issue titles; no overstatement of same-bug evidence 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": 23, + "review_unit_index": 47, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28282", + "issue:30333", + "issue:34689", + "issue:35141", + "issue:39692", + "issue:42222", + "issue:42796", + "issue:42831", + "issue:42898", + "issue:43116", + "issue:43295", + "issue:43479", + "issue:43723", + "issue:43784", + "issue:43824", + "issue:43874", + "issue:44016", + "issue:44060", + "issue:44075", + "issue:44079", + "issue:44190", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44568", + "issue:44623", + "issue:44625", + "issue:44734", + "issue:44779", + "issue:44869", + "issue:44933", + "issue:45092", + "issue:45103", + "issue:45125", + "issue:45200", + "issue:45356", + "issue:45405", + "issue:45656", + "issue:45712", + "issue:45812", + "issue:46231", + "issue:46762", + "issue:46877", + "issue:46909" + ], + "soft_pairs": [ + "issue:43479|issue:44869", + "issue:43723|issue:44568", + "issue:44016|issue:46231", + "issue:42898|issue:45812", + "issue:44625|issue:46877", + "issue:42898|issue:44779", + "issue:39692|issue:46909", + "issue:43479|issue:44079", + "issue:43874|issue:45092", + "issue:43784|issue:43824", + "issue:27957|issue:30333", + "issue:44734|issue:45103", + "issue:30333|issue:34689", + "issue:42222|issue:45200", + "issue:42898|issue:43295", + "issue:42898|issue:45356", + "issue:44295|issue:45656", + "issue:43116|issue:44190", + "issue:30333|issue:35141", + "issue:45405|issue:45712", + "issue:44222|issue:44297", + "issue:22355|issue:28282", + "issue:44060|issue:45125", + "issue:44075|issue:46231", + "issue:44933|issue:45712", + "issue:44360|issue:46762", + "issue:42831|issue:44623", + "issue:44297|issue:45656", + "issue:42796|issue:43723", + "issue:44222|issue:44623" + ], + "prepared_review_unit_hash": "52264eef50f917ce4267bcce38fdfe3d054f25fbef1fa658eb408c7a6262c1bc", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 36257, + "estimated_input_tokens": 9065, + "estimated_eval_tokens": 18386 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are unrelated model/import/docs/training bugs. Several pairs are only loosely similar because they mention tokenizers, v5 regressions, or documentation, but they do not describe the same concrete bug or change.", + "confidence": 0.71, + "canonical_issue_reason": "issue:42898 is the best representative of the cluster because it is a core tokenizer regression report in v5 and is one of the few items that recurs across multiple soft-similarity links, even though the cluster is not actually duplicate-heavy.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:42898 is the most central issue to this noisy cluster\u2019s recurring tokenizer-v5 theme; it is the closest thing to a hub, but the surrounding issues are too diverse to treat it as a true duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43479", + "right": "issue:44869", + "accept": false, + "reason": "Different models and failures: multimodal config initialization vs Whisper word timestamp decoding." + }, + { + "left": "issue:43723", + "right": "issue:44568", + "accept": false, + "reason": "Both involve tokenizers in v5, but one is tokenizer loading and the other is add_special_tokens behavior for a specific model." + }, + { + "left": "issue:44016", + "right": "issue:46231", + "accept": false, + "reason": "Both are documentation-related, but one is a notebook syntax error and the other is a parameter-name mismatch in docs." + }, + { + "left": "issue:42898", + "right": "issue:45812", + "accept": false, + "reason": "Both are tokenizer regressions, but they affect different APIs and different model families; not the same bug." + }, + { + "left": "issue:44625", + "right": "issue:46877", + "accept": false, + "reason": "Unrelated: Qwen config propagation vs auto_docstring import-time logging." + }, + { + "left": "issue:42898", + "right": "issue:44779", + "accept": false, + "reason": "Both are tokenizer regressions, but one is a cleanup/decoding behavior change and the other is a Deepseek tokenization correctness bug." + }, + { + "left": "issue:39692", + "right": "issue:46909", + "accept": false, + "reason": "Both are broken docs examples, but they are different examples with different failure modes." + }, + { + "left": "issue:43479", + "right": "issue:44079", + "accept": false, + "reason": "Different subsystems: multimodal config defaults vs ModelOutput key assignment logic." + }, + { + "left": "issue:43874", + "right": "issue:45092", + "accept": false, + "reason": "Both are multimodal model issues, but one is a missing processor method and the other is remote-code/meta-init compatibility." + }, + { + "left": "issue:43784", + "right": "issue:43824", + "accept": false, + "reason": "Different import errors affecting different packages/models." + }, + { + "left": "issue:27957", + "right": "issue:30333", + "accept": false, + "reason": "Completely unrelated: Flash Attention on XLM-R vs MLflow reporting after failed training jobs." + }, + { + "left": "issue:44734", + "right": "issue:45103", + "accept": false, + "reason": "Different crash sites and causes: KV-cache tensor indexing vs annotations-related AttributeError in auto_docstring." + }, + { + "left": "issue:30333", + "right": "issue:34689", + "accept": false, + "reason": "Unrelated training-job reporting bug vs model-loading regression for Llama 3.2 Vision." + }, + { + "left": "issue:42222", + "right": "issue:45200", + "accept": false, + "reason": "Different models and fixes: vitpose breakage vs Gemma 4 mm_token_type_ids defaults." + }, + { + "left": "issue:42898", + "right": "issue:43295", + "accept": false, + "reason": "Both are processor/tokenizer regressions, but the concrete breakage is different: cleanup behavior vs processor.tokenizer access/images passing." + }, + { + "left": "issue:42898", + "right": "issue:45356", + "accept": false, + "reason": "Both mention tokenizer regressions, but they are model-specific issues with different root causes and symptoms." + }, + { + "left": "issue:44295", + "right": "issue:45656", + "accept": false, + "reason": "Unrelated: position_ids buffer handling vs DeepSpeed optimizer stepping twice." + }, + { + "left": "issue:43116", + "right": "issue:44190", + "accept": false, + "reason": "Different example scripts and different failures." + }, + { + "left": "issue:30333", + "right": "issue:35141", + "accept": false, + "reason": "Different bugs: MLflow reporting vs embeddings being reinitialized in post_init." + }, + { + "left": "issue:45405", + "right": "issue:45712", + "accept": false, + "reason": "Unrelated: dependency version bump vs leftover dummy classes in package exports." + }, + { + "left": "issue:44222", + "right": "issue:44297", + "accept": false, + "reason": "Different save_pretrained problems affecting different object types and outputs." + }, + { + "left": "issue:22355", + "right": "issue:28282", + "accept": false, + "reason": "Both are import errors, but one is a missing submodule and the other is a missing PyTorch dependency." + }, + { + "left": "issue:44060", + "right": "issue:45125", + "accept": false, + "reason": "Different Qwen model configuration issues with different consequences." + }, + { + "left": "issue:44075", + "right": "issue:46231", + "accept": false, + "reason": "One is a code bug in SGD argument handling; the other is just a docs parameter mismatch." + }, + { + "left": "issue:44933", + "right": "issue:45712", + "accept": false, + "reason": "Both concern package surface/imports, but they are different namespace problems." + }, + { + "left": "issue:44360", + "right": "issue:46762", + "accept": false, + "reason": "Different technical areas: DSA indexer implementation vs Minimax attention numerical differences." + }, + { + "left": "issue:42831", + "right": "issue:44623", + "accept": false, + "reason": "Different issues: FP8 accuracy vs processor save_pretrained file emission." + }, + { + "left": "issue:42796", + "right": "issue:43723", + "accept": false, + "reason": "Both are tokenizer-related, but one is a BPE marker handling bug and the other is a tokenizer loading issue." + }, + { + "left": "issue:44222", + "right": "issue:44623", + "accept": false, + "reason": "Both mention save_pretrained, but they involve different objects and different missing outputs." + } + ] + }, + "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": 23, + "review_unit_index": 48, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:30990", + "issue:38175", + "issue:41628", + "issue:42757", + "issue:42831", + "issue:42907", + "issue:43122", + "issue:43232", + "issue:43475", + "issue:43479", + "issue:43644", + "issue:43723", + "issue:44190", + "issue:44222", + "issue:44261", + "issue:44263", + "issue:44279", + "issue:44295", + "issue:44297", + "issue:44360", + "issue:44373", + "issue:44448", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44623", + "issue:44977", + "issue:45071", + "issue:45092", + "issue:45245", + "issue:45310", + "issue:45357", + "issue:45381", + "issue:45406", + "issue:45834", + "issue:46050", + "issue:46082", + "issue:46132", + "issue:46209", + "issue:46305", + "issue:46328", + "issue:46369", + "issue:46424", + "issue:46431", + "issue:46655", + "issue:46710", + "issue:46909" + ], + "soft_pairs": [ + "issue:44448|issue:46132", + "issue:45245|issue:46132", + "issue:43479|issue:43644", + "issue:41628|issue:42757", + "issue:44373|issue:44485", + "issue:44297|issue:45310", + "issue:46132|issue:46710", + "issue:45071|issue:46082", + "issue:44222|issue:44484", + "issue:43723|issue:44977", + "issue:43122|issue:46209", + "issue:44263|issue:45092", + "issue:44190|issue:46082", + "issue:43232|issue:46424", + "issue:43232|issue:45834", + "issue:44295|issue:44623", + "issue:43475|issue:46305", + "issue:46209|issue:46431", + "issue:44263|issue:44360", + "issue:46305|issue:46655", + "issue:44263|issue:44373", + "issue:30990|issue:38175", + "issue:42831|issue:45310", + "issue:45092|issue:45357", + "issue:46369|issue:46909", + "issue:42907|issue:46050", + "issue:44279|issue:46328", + "issue:45381|issue:46305", + "issue:44261|issue:45092", + "issue:44483|issue:45406" + ], + "prepared_review_unit_hash": "50d08cc11e0530099f77b06e5367a47946e3c4fece0d481d5db56b2c64adc3b7", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35671, + "estimated_input_tokens": 8918, + "estimated_eval_tokens": 18092 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is broadly about Transformers v5 regressions and compatibility issues, but the soft-edge pairs are mostly only superficially similar. I would not merge any of the suggested pairs as duplicates.", + "confidence": 0.66, + "canonical_issue_reason": "issue:45092 is the clearest, most concrete representative of the cluster\u2019s recurring v5 compatibility/loading regressions: it has a specific failure mode, a defined affected class of checkpoints, and a likely actionable fix path.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45092 is the strongest overall issue to keep as the representative triage item because it is specific, reproducible, and captures a high-impact regression pattern better than the more vague or narrowly scoped alternatives.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44448", + "right": "issue:46132", + "accept": false, + "reason": "Different problems: Pegasus output regression vs AttentionInterface.register changing function behavior. Same broad version/regression theme, but not the same bug." + }, + { + "left": "issue:45245", + "right": "issue:46132", + "accept": false, + "reason": "Unrelated failures: category-cardinality RuntimeError vs attention registration behavior. No shared code-path evidence." + }, + { + "left": "issue:43479", + "right": "issue:43644", + "accept": false, + "reason": "Both involve config/init semantics, but one is default multimodal sub-config initialization and the other is non-persistent buffer serialization junk; too different to treat as duplicates." + }, + { + "left": "issue:41628", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they concern different symbols and different packages; likely separate dependency/version issues." + }, + { + "left": "issue:44373", + "right": "issue:44485", + "accept": false, + "reason": "A docstring issue is not the same as a RoPE implementation discussion; the titles suggest different scope and impact." + }, + { + "left": "issue:44297", + "right": "issue:45310", + "accept": false, + "reason": "Tokenizer save metadata mismatch vs Qwen3.5 MOE from_pretrained error. Overlap is only around model I/O, not the same concrete failure." + }, + { + "left": "issue:46132", + "right": "issue:46710", + "accept": false, + "reason": "Attention registration behavior change vs accuracy regression after upgrading Transformers. Different symptoms and likely different code paths." + }, + { + "left": "issue:45071", + "right": "issue:46082", + "accept": false, + "reason": "Type-checking breakage in PretrainedConfig is not the same as LlamaConfig rejecting explicit head_dim. Related config area, but distinct bugs." + }, + { + "left": "issue:44222", + "right": "issue:44484", + "accept": false, + "reason": "FP8 save_pretrained bug vs max_shard_size question. One is a serialization bug, the other is a size-limit behavior question." + }, + { + "left": "issue:43723", + "right": "issue:44977", + "accept": false, + "reason": "Tokenizer loading issue and Qwen3.5 flash-attention generation issue are unrelated." + }, + { + "left": "issue:43122", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer output differences across versions vs FP8 inference failure on B200. Different subsystems and failure modes." + }, + { + "left": "issue:44263", + "right": "issue:45092", + "accept": false, + "reason": "DSA indexer split semantics vs remote-code checkpoint meta-initialization incompatibility. No concrete shared bug." + }, + { + "left": "issue:44190", + "right": "issue:46082", + "accept": false, + "reason": "Local dataset loading in an example script is unrelated to LlamaConfig head_dim validation." + }, + { + "left": "issue:43232", + "right": "issue:46424", + "accept": false, + "reason": "Generation state update after sync_gpus vs cache length handling for quantized cache. Similar cache/generation area, but different behavior." + }, + { + "left": "issue:43232", + "right": "issue:45834", + "accept": false, + "reason": "Generation kwargs bug and Kosmos2.5 OCR long-input index error are different code paths." + }, + { + "left": "issue:44295", + "right": "issue:44623", + "accept": false, + "reason": "Reading a registered buffer vs processor.save_pretrained missing files. Both serialization-adjacent, but not the same bug." + }, + { + "left": "issue:43475", + "right": "issue:46305", + "accept": false, + "reason": "SAM 3 Vision encoder attribute error and MoonshineStreaming window-mask off-by-one are unrelated model-specific issues." + }, + { + "left": "issue:46209", + "right": "issue:46431", + "accept": false, + "reason": "FP8 inference failure and missing chat template error are entirely different problems." + }, + { + "left": "issue:44263", + "right": "issue:44360", + "accept": false, + "reason": "Both mention the DSA indexer, but one is about torch.split outputs and the other about a missing ReLU; not enough to merge." + }, + { + "left": "issue:46305", + "right": "issue:46655", + "accept": false, + "reason": "Different media processors and different off-by-one/padding bugs; only superficial similarity." + }, + { + "left": "issue:44263", + "right": "issue:44373", + "accept": false, + "reason": "Indexer implementation bug vs docstring issue. No duplicate relationship." + }, + { + "left": "issue:30990", + "right": "issue:38175", + "accept": false, + "reason": "Sentence Transformers loading hang and zero probabilities in SigLIP2 are unrelated." + }, + { + "left": "issue:42831", + "right": "issue:45310", + "accept": false, + "reason": "FP8 accuracy issue and Qwen3.5 MOE from_pretrained error are separate model/runtime bugs." + }, + { + "left": "issue:45092", + "right": "issue:45357", + "accept": false, + "reason": "Different checkpoints and different failure modes: meta-init incompatibility vs incorrect visual encoder keys on save_pretrained." + }, + { + "left": "issue:46369", + "right": "issue:46909", + "accept": false, + "reason": "Model memory overflow and broken documentation example are unrelated." + }, + { + "left": "issue:42907", + "right": "issue:46050", + "accept": false, + "reason": "Both concern save/load of quantized models, but one is dequantized Ministrals/Devstrals serialization and the other is sinq model round-tripping; too different for duplicate triage." + }, + { + "left": "issue:44279", + "right": "issue:46328", + "accept": false, + "reason": "Generic dependency complaint vs a performance problem in SequenceFeatureExtractor.pad; no shared underlying bug." + }, + { + "left": "issue:45381", + "right": "issue:46305", + "accept": false, + "reason": "Qwen2.5-VL vision_position_ids bug and MoonshineStreaming mask off-by-one are unrelated sequence-positioning problems." + }, + { + "left": "issue:44261", + "right": "issue:45092", + "accept": false, + "reason": "MLA q_a_layernorm epsilon precision issue is different from remote-code checkpoint meta initialization incompatibility." + }, + { + "left": "issue:44483", + "right": "issue:45406", + "accept": false, + "reason": "Chat-completions request rejection in serve and Gemma4Processor missing _tokenizer are distinct serving/processor bugs." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 49, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:22355", + "issue:27957", + "issue:28218", + "issue:30990", + "issue:33357", + "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:43540", + "issue:43576", + "issue:43644", + "issue:43646", + "issue:43784", + "issue:43824", + "issue:43937", + "issue:43976", + "issue:44038", + "issue:44188", + "issue:44263", + "issue:44295", + "issue:44297", + "issue:44373", + "issue:44488", + "issue:44568", + "issue:44829", + "issue:44945", + "issue:44964", + "issue:45278", + "issue:45381", + "issue:45593", + "issue:45753", + "issue:46002", + "issue:46172", + "issue:46305", + "issue:46369", + "issue:46402", + "issue:46762", + "issue:46909" + ], + "soft_pairs": [ + "issue:22355|issue:27957", + "issue:44829|issue:46002", + "issue:43644|issue:46305", + "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:42886|issue:43066", + "issue:42886|issue:42994", + "issue:42757|issue:45278", + "issue:27957|issue:30990", + "issue:44188|issue:44373", + "issue:43408|issue:46369", + "issue:43232|issue:45593", + "issue:42796|issue:43122", + "issue:43646|issue:46002", + "issue:45381|issue:46909", + "issue:43540|issue:46909", + "issue:30990|issue:33357", + "issue:42907|issue:44945", + "issue:43576|issue:43784", + "issue:44295|issue:44297", + "issue:42371|issue:44263", + "issue:39401|issue:44488", + "issue:42907|issue:44964", + "issue:43976|issue:44945", + "issue:46172|issue:46762", + "issue:42907|issue:43937", + "issue:41628|issue:45278" + ], + "prepared_review_unit_hash": "c85f9d90a89df609d832dd25afc406684a2c0f42791cadee3596d9b5407a8c50", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36352, + "estimated_input_tokens": 9088, + "estimated_eval_tokens": 18432 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This set is not a true duplicate cluster: it mixes unrelated tokenizer, import, model-loading, training, generation, and numerical-regression issues. None of the soft pairs looks like the same underlying bug/change, so all should be rejected.", + "confidence": 0.95, + "canonical_issue_reason": "issue:45278 is the broadest umbrella-style report in the set ('many import errors after update from 4.57.0 to 5.5.0'), so it is the least-specific anchor if one must pick a canonical issue.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the best triage anchor among these because it is the most general regression report; the rest are narrower and largely unrelated to one another.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:22355", + "right": "issue:27957", + "accept": false, + "reason": "Import error for transformers.onnx vs XLMRoberta Flash Attention 2 behavior; different subsystem and symptom." + }, + { + "left": "issue:44829", + "right": "issue:46002", + "accept": false, + "reason": "Both involve training instability, but one is flash-attention-3 degeneracy and the other is NaNs after checkout/init; not the same bug path." + }, + { + "left": "issue:43644", + "right": "issue:46305", + "accept": false, + "reason": "Non-persistent buffer junk on init is unrelated to MoonshineStreaming sliding-window off-by-one masking." + }, + { + "left": "issue:42831", + "right": "issue:44038", + "accept": false, + "reason": "FineGrainedFP8 accuracy regression and Qwen3-VL-Moe model bug are different problems, not one underlying issue." + }, + { + "left": "issue:41628", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they miss different symbols from different model families; no shared root cause is evident." + }, + { + "left": "issue:43408", + "right": "issue:46402", + "accept": false, + "reason": "SAM3 type warning during instantiation vs SAM3 text-encoder weights not loading; different failure modes." + }, + { + "left": "issue:42831", + "right": "issue:45753", + "accept": false, + "reason": "FP8 accuracy issues are unrelated to Qwen3_5 infinite-loop behavior on a specific image." + }, + { + "left": "issue:22355", + "right": "issue:28218", + "accept": false, + "reason": "Module import failure vs tokenizer adding an extra space after an added token; different code paths." + }, + { + "left": "issue:39401", + "right": "issue:44568", + "accept": false, + "reason": "Wrong offset_mapping and missing BOS/EOS on add_special_tokens are both tokenizer bugs but not the same concrete behavior." + }, + { + "left": "issue:42886", + "right": "issue:43066", + "accept": false, + "reason": "Offline cache load failure is unrelated to wrong tokenizer decoder type serialization in v5." + }, + { + "left": "issue:42886", + "right": "issue:42994", + "accept": false, + "reason": "Tokenizer cache-loading offline and quantized model saving are separate persistence issues." + }, + { + "left": "issue:42757", + "right": "issue:45278", + "accept": false, + "reason": "Missing is_offline_mode from huggingface_hub is a dependency/API mismatch; 45278 is a broader transformers import-regression report." + }, + { + "left": "issue:27957", + "right": "issue:30990", + "accept": false, + "reason": "Flash Attention 2 model-specific bug vs Sentence-Transformers loading hang; different symptoms and paths." + }, + { + "left": "issue:44188", + "right": "issue:44373", + "accept": false, + "reason": "torch.compile attention-kernel divergence is unrelated to a docstring issue for position_ids." + }, + { + "left": "issue:43408", + "right": "issue:46369", + "accept": false, + "reason": "SAM3 model-type warning vs memory overflow in sam3-litetext; different failures." + }, + { + "left": "issue:43232", + "right": "issue:45593", + "accept": false, + "reason": "Generation helper after sync_gpus and D-FINE auxiliary-loss behavior are unrelated subsystems." + }, + { + "left": "issue:42796", + "right": "issue:43122", + "accept": false, + "reason": "Ministral tokenizer BPE-marker handling vs a version-wide tokenization change; similar area but not the same underlying bug." + }, + { + "left": "issue:43646", + "right": "issue:46002", + "accept": false, + "reason": "Custom model initialization breakage vs NaNs in classification heads after checkout/init; not the same code path." + }, + { + "left": "issue:45381", + "right": "issue:46909", + "accept": false, + "reason": "Qwen2.5-VL video position_ids bug vs a broken documentation example for DIA; different artifact types." + }, + { + "left": "issue:43540", + "right": "issue:46909", + "accept": false, + "reason": "Qwen3OmniMoe video-input ValueError is unrelated to a documentation example bug." + }, + { + "left": "issue:30990", + "right": "issue:33357", + "accept": false, + "reason": "Loading hang for Sentence Transformers vs MacOS bus error on a CLIP model; different runtime failures." + }, + { + "left": "issue:42907", + "right": "issue:44945", + "accept": false, + "reason": "Saving dequantized Ministral/Devstral models is unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:43576", + "right": "issue:43784", + "accept": false, + "reason": "Broken transformers env CLI and sentence-transformers import NameError are different issues." + }, + { + "left": "issue:44295", + "right": "issue:44297", + "accept": false, + "reason": "Reading a registered buffer position_ids and tokenizer_class metadata mismatch are unrelated serialization bugs." + }, + { + "left": "issue:42371", + "right": "issue:44263", + "accept": false, + "reason": "TF32 configuration API warning is unrelated to torch.split return-value handling in GlmMoeDsaIndexer." + }, + { + "left": "issue:39401", + "right": "issue:44488", + "accept": false, + "reason": "Qwen3 tokenizer offset mapping bug vs failure to load cjvt/sleng-bert; different tokenizer/model-loading problems." + }, + { + "left": "issue:42907", + "right": "issue:44964", + "accept": false, + "reason": "Dequantized saving failure and loading Phi-4-multimodal-instruct are separate model persistence/load issues." + }, + { + "left": "issue:43976", + "right": "issue:44945", + "accept": false, + "reason": "Python-version compatibility regression and pipeline-parallelism output corruption are unrelated." + }, + { + "left": "issue:46172", + "right": "issue:46762", + "accept": false, + "reason": "DeepSeek-V4 tensor transpose/dtype bug and Minimax MSA numerical differences are different attention/model-path issues." + }, + { + "left": "issue:42907", + "right": "issue:43937", + "accept": false, + "reason": "Saving dequantized models and invalid GenerationConfig are distinct failures." + }, + { + "left": "issue:41628", + "right": "issue:45278", + "accept": false, + "reason": "Single missing AutoImageProcessor import vs broad import errors after upgrade; not the same concrete 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": 23, + "review_unit_index": 50, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:22355", + "issue:29127", + "issue:29159", + "issue:36296", + "issue:39401", + "issue:42175", + "issue:42757", + "issue:42907", + "issue:43116", + "issue:43122", + "issue:43278", + "issue:43408", + "issue:43452", + "issue:43475", + "issue:43493", + "issue:43756", + "issue:43784", + "issue:43824", + "issue:43937", + "issue:44038", + "issue:44060", + "issue:44117", + "issue:44188", + "issue:44222", + "issue:44295", + "issue:44297", + "issue:44351", + "issue:44360", + "issue:44485", + "issue:44521", + "issue:44625", + "issue:44704", + "issue:44779", + "issue:44829", + "issue:44938", + "issue:44945", + "issue:45200", + "issue:45216", + "issue:45290", + "issue:45478", + "issue:45600", + "issue:45753", + "issue:46231", + "issue:46326", + "issue:46328", + "issue:46614", + "issue:46877" + ], + "soft_pairs": [ + "issue:43756|issue:44060", + "issue:43784|issue:45600", + "issue:44117|issue:44704", + "issue:39401|issue:44521", + "issue:44704|issue:45290", + "issue:44625|issue:44829", + "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:44038|issue:44295", + "issue:45216|issue:45478", + "issue:39401|issue:46614", + "issue:22355|issue:29159", + "issue:44038|issue:44222", + "issue:43824|issue:44351", + "issue:43116|issue:45200", + "issue:43122|issue:44779", + "issue:42757|issue:44351", + "issue:36296|issue:43116", + "issue:43278|issue:44945", + "issue:22355|issue:29127", + "issue:42175|issue:44938", + "issue:43493|issue:45753", + "issue:46231|issue:46877", + "issue:44188|issue:46328", + "issue:43937|issue:44485" + ], + "prepared_review_unit_hash": "ed3deab9cc0692ac1bcb15c9fe8142bb1823dd4ed537995335f5e4153cd0c274", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 29, + "serialized_chars": 35672, + "estimated_input_tokens": 8918, + "estimated_eval_tokens": 18092 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These artifacts are mostly unrelated, spanning different models, APIs, and failure modes. None of the soft pairs look like the same underlying bug/change, so all soft edges are rejected.", + "confidence": 0.91, + "canonical_issue_reason": "issue:44521 is the strongest representative issue here: it is open, actively discussed, and describes a concrete, reproducible API bug in a widely used path (`apply_chat_template`) with multimodal inputs.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:44521 is the best overall issue candidate because it is well-scoped, user-facing, and likely to attract/anchor related reports about the same chat-template behavior.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43756", + "right": "issue:44060", + "accept": false, + "reason": "Different bugs in different model families: RoPE layer selection vs a tied-weights warning in Qwen3-Next." + }, + { + "left": "issue:43784", + "right": "issue:45600", + "accept": false, + "reason": "Import-time `nn` NameError in sentence-transformers is unrelated to stale `auto_mappings.py` config references and CI breakage." + }, + { + "left": "issue:44117", + "right": "issue:44704", + "accept": false, + "reason": "Tokenizer mapping `None` handling is a different code path from `AutoProcessor.from_pretrained` kwarg forwarding." + }, + { + "left": "issue:39401", + "right": "issue:44521", + "accept": false, + "reason": "Wrong `offset_mapping` in Qwen3 tokenizer is not the same as zeroed assistant masks in multimodal chat templating." + }, + { + "left": "issue:44704", + "right": "issue:45290", + "accept": false, + "reason": "`cached_file` kwargs forwarding and chat-template crashes with tool-call messages are separate failures." + }, + { + "left": "issue:44625", + "right": "issue:44829", + "accept": false, + "reason": "`num_labels` propagation and flash-attention-3 training degeneration are unrelated regression types." + }, + { + "left": "issue:44360", + "right": "issue:44485", + "accept": false, + "reason": "A discussion about DSA indexer activation is unrelated to GLM-5 RoPE implementation details." + }, + { + "left": "issue:43408", + "right": "issue:43475", + "accept": false, + "reason": "Wrong SAM3 model-type warning and missing `fpn_position_embeddings` attribute are distinct SAM3-video issues." + }, + { + "left": "issue:43122", + "right": "issue:44521", + "accept": false, + "reason": "Tokenizer version drift is a different problem from multimodal assistant-mask generation in chat templates." + }, + { + "left": "issue:44038", + "right": "issue:44297", + "accept": false, + "reason": "Qwen3-VL-Moe/transformers 5.0 bug and tokenizer save metadata mismatch are separate paths." + }, + { + "left": "issue:43122", + "right": "issue:46326", + "accept": false, + "reason": "General tokenization differences across versions are not the same as `continue_final_message` handling for mistral_common tokenizers." + }, + { + "left": "issue:42907", + "right": "issue:43452", + "accept": false, + "reason": "Dequantized save failures and GGUF/from_pretrained loading breakage are different serialization/loading bugs." + }, + { + "left": "issue:44038", + "right": "issue:44295", + "accept": false, + "reason": "Qwen3-VL-Moe model behavior and `position_ids` buffer access are unrelated." + }, + { + "left": "issue:45216", + "right": "issue:45478", + "accept": false, + "reason": "Both involve Qwen3.5, but one is a save-pretrained regression and the other is a from-pretrained error; not the same concrete bug." + }, + { + "left": "issue:39401", + "right": "issue:46614", + "accept": false, + "reason": "Tokenizer offset mapping for Qwen3 is unrelated to stale phonemizer backend state in Wav2Vec2 phoneme tokenization." + }, + { + "left": "issue:22355", + "right": "issue:29159", + "accept": false, + "reason": "Missing `transformers.onnx` import and slow/fast tokenizer inconsistency are different issues." + }, + { + "left": "issue:44038", + "right": "issue:44222", + "accept": false, + "reason": "A Qwen3-VL-Moe issue does not match FP8 `save_pretrained` behavior for MoE models." + }, + { + "left": "issue:43824", + "right": "issue:44351", + "accept": false, + "reason": "Two separate missing-symbol import errors (`Qwen2_5_VLForConditionalGeneration` vs `HybridCache`)." + }, + { + "left": "issue:43116", + "right": "issue:45200", + "accept": false, + "reason": "Empty results in the example classification script and Gemma 4 `mm_token_type_ids` defaults are different bugs." + }, + { + "left": "issue:43122", + "right": "issue:44779", + "accept": false, + "reason": "Tokenizer behavior drift across versions is unrelated to Deepseek v5 incorrect tokenization." + }, + { + "left": "issue:42757", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but for different symbols from different packages and code paths." + }, + { + "left": "issue:36296", + "right": "issue:43116", + "accept": false, + "reason": "Tensor-parallel training failure is not the same as multi-label example script returning empty outputs." + }, + { + "left": "issue:43278", + "right": "issue:44945", + "accept": false, + "reason": "Training/eval dtype mismatch is unrelated to incorrect outputs under pipeline parallelism." + }, + { + "left": "issue:22355", + "right": "issue:29127", + "accept": false, + "reason": "Missing module import and LayoutLMv3 error-message clarity are not the same bug." + }, + { + "left": "issue:42175", + "right": "issue:44938", + "accept": false, + "reason": "Missing TensorFlow in a torch extra install is unrelated to Python 3.14 load failures." + }, + { + "left": "issue:43493", + "right": "issue:45753", + "accept": false, + "reason": "SigLIP2 implementation discrepancy and Qwen3_5 image infinite loop are different model-specific failures." + }, + { + "left": "issue:46231", + "right": "issue:46877", + "accept": false, + "reason": "Documentation parameter mismatch and noisy import-time doc-lint output are not the same underlying issue." + }, + { + "left": "issue:44188", + "right": "issue:46328", + "accept": false, + "reason": "`torch.compile`-dependent attention divergence and `SequenceFeatureExtractor.pad` performance are unrelated." + }, + { + "left": "issue:43937", + "right": "issue:44485", + "accept": false, + "reason": "GLM-5 generation-config validation and GLM-5 RoPE implementation are separate problems, not one duplicate 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": 23, + "review_unit_index": 51, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28348", + "issue:33453", + "issue:38617", + "issue:39401", + "issue:42222", + "issue:42371", + "issue:42673", + "issue:42898", + "issue:43064", + "issue:43066", + "issue:43116", + "issue:43441", + "issue:43493", + "issue:43611", + "issue:43618", + "issue:43673", + "issue:43825", + "issue:43827", + "issue:43874", + "issue:43976", + "issue:43992", + "issue:44038", + "issue:44075", + "issue:44263", + "issue:44355", + "issue:44610", + "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:46002", + "issue:46231", + "issue:46614", + "issue:46640", + "issue:46697", + "issue:46710", + "issue:46752", + "issue:46762", + "issue:46829", + "issue:46909" + ], + "soft_pairs": [ + "issue:38617|issue:45278", + "issue:42898|issue:46614", + "issue:43066|issue:44779", + "issue:43611|issue:46002", + "issue:43976|issue:45446", + "issue:44038|issue:45478", + "issue:44355|issue:46231", + "issue:43116|issue:43493", + "issue:42673|issue:45072", + "issue:45357|issue:45478", + "issue:44625|issue:46002", + "issue:42371|issue:43874", + "issue:45083|issue:45446", + "issue:44610|issue:46909", + "issue:43618|issue:46829", + "issue:43992|issue:44704", + "issue:44075|issue:44936", + "issue:43116|issue:43825", + "issue:43066|issue:46710", + "issue:28348|issue:46640", + "issue:44263|issue:46762", + "issue:46614|issue:46752", + "issue:44995|issue:46697", + "issue:45092|issue:46762", + "issue:42222|issue:43441", + "issue:43827|issue:45446", + "issue:33453|issue:39401", + "issue:43064|issue:43673" + ], + "prepared_review_unit_hash": "39fa5fc3b249e95e1787a212d04ca47483c3121350098f6819bac0fbb32c1be5", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 28, + "serialized_chars": 35744, + "estimated_input_tokens": 8936, + "estimated_eval_tokens": 18128 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most items are separate Transformers v5 regressions or model-specific bugs, not duplicates. No soft pair looks like the same concrete underlying issue/change, so all soft edges should be rejected.", + "confidence": 0.34, + "canonical_issue_reason": "If a representative issue is needed, issue 45278 is the broadest umbrella-style report in the set (\u201cmany import errors after update\u201d), but the cluster overall does not contain a true duplicate group.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45278 is the closest thing to a global representative because it is the most general regression report; however, it still only covers import errors and does not unify the rest of the cluster.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38617", + "right": "issue:45278", + "accept": false, + "reason": "One is a specific missing-symbol import error; the other is a broad report of many import errors after upgrade. Related theme, but not the same bug." + }, + { + "left": "issue:42898", + "right": "issue:46614", + "accept": false, + "reason": "Tokenizer cleanup behavior change vs stale phonemizer backend state. Different code paths and symptoms." + }, + { + "left": "issue:43066", + "right": "issue:44779", + "accept": false, + "reason": "Both involve tokenizers, but one is about decoder type in v5 and the other is DeepSeek incorrect results. Too different to be duplicates." + }, + { + "left": "issue:43611", + "right": "issue:46002", + "accept": false, + "reason": "Model loading via base_model_prefix versus NaNs in classification heads. Not the same underlying problem." + }, + { + "left": "issue:43976", + "right": "issue:45446", + "accept": false, + "reason": "Python-version compatibility issue vs a PyTorch import/version check in flex_attention. Different failure modes." + }, + { + "left": "issue:44038", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen3/Moe-related, but one is Qwen3-VL-Moe and the other is Qwen3.5 Moe from_pretrained. Not the same concrete bug." + }, + { + "left": "issue:44355", + "right": "issue:46231", + "accept": false, + "reason": "Compiled-Python import/runtime errors versus a documentation parameter-name mismatch. Not duplicates." + }, + { + "left": "issue:43116", + "right": "issue:43493", + "accept": false, + "reason": "Example-script multi-label classification failure versus a SigLIP2 implementation discrepancy. Different components and symptoms." + }, + { + "left": "issue:42673", + "right": "issue:45072", + "accept": false, + "reason": "VRAM leak in Qwen3ForCausalLM dataloader threads versus dtype mismatch in inference. Different issues." + }, + { + "left": "issue:45357", + "right": "issue:45478", + "accept": false, + "reason": "Both are Qwen3.5-related, but one is save_pretrained visual-key regression and the other is from_pretrained load failure. Not one bug." + }, + { + "left": "issue:44625", + "right": "issue:46002", + "accept": false, + "reason": "num_labels propagation bug versus NaNs in classification heads. Different root causes." + }, + { + "left": "issue:42371", + "right": "issue:43874", + "accept": false, + "reason": "TF32 API behavior guidance versus a missing image-processor method. No overlap in bug." + }, + { + "left": "issue:45083", + "right": "issue:45446", + "accept": false, + "reason": "Qwen3_omni_moe feature-length helper bug versus flex_attention version check. Unrelated." + }, + { + "left": "issue:44610", + "right": "issue:46909", + "accept": false, + "reason": "OmDet-Turbo processor size mismatch versus broken DIA documentation example. Different problems." + }, + { + "left": "issue:43618", + "right": "issue:46829", + "accept": false, + "reason": "Both are output-field assignment bugs, but they affect different model outputs and code paths; too broad to merge." + }, + { + "left": "issue:43992", + "right": "issue:44704", + "accept": false, + "reason": "Missing weight in UMT5Encoder loading versus AutoProcessor kwargs not forwarded. Different APIs." + }, + { + "left": "issue:44075", + "right": "issue:44936", + "accept": false, + "reason": "SGD optimizer args ignored versus trainer.evaluate failing after train. Not the same bug." + }, + { + "left": "issue:43116", + "right": "issue:43825", + "accept": false, + "reason": "Example script behavior versus a pipeline() error message/doc mismatch. Different symptoms and fixes." + }, + { + "left": "issue:43066", + "right": "issue:46710", + "accept": false, + "reason": "Tokenizer decoder-type regression versus an accuracy regression in a DeepSeek model. Too different to be duplicates." + }, + { + "left": "issue:28348", + "right": "issue:46640", + "accept": false, + "reason": "FlashAttention 2 feature support request for GPT2LMHeadModel versus a separate Flash Attention 2 issue. Not the same concrete code-path problem." + }, + { + "left": "issue:44263", + "right": "issue:46762", + "accept": false, + "reason": "torch.split indexing bug versus numerical differences in Minimax attention. Different issues." + }, + { + "left": "issue:46614", + "right": "issue:46752", + "accept": false, + "reason": "Stale phonemizer language state versus apply_chat_template crash on empty input. No shared bug." + }, + { + "left": "issue:44995", + "right": "issue:46697", + "accept": false, + "reason": "GlmMoeDsa stale cache crash versus Swin2SR over-padding. Unrelated." + }, + { + "left": "issue:45092", + "right": "issue:46762", + "accept": false, + "reason": "Meta-init incompatibility for InternVL2 checkpoints versus Minimax MSA numerical mismatch. Different model paths." + }, + { + "left": "issue:42222", + "right": "issue:43441", + "accept": false, + "reason": "VitPose import/breakage issue versus Ministral-3 FlashAttention failure. Different subsystems." + }, + { + "left": "issue:43827", + "right": "issue:45446", + "accept": false, + "reason": "Documentation still references removed pipeline() versus a PyTorch version check bug. Not duplicates." + }, + { + "left": "issue:33453", + "right": "issue:39401", + "accept": false, + "reason": "Generic tokenizer loading regression versus a specific offset_mapping bug in Qwen3 tokenizer. Related area only." + }, + { + "left": "issue:43064", + "right": "issue:43673", + "accept": false, + "reason": "Trainer/FSDP2/PEFT optimizer-state corruption versus missing GenerationMixin cache during chunked prefill. Different code paths." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 52, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:42617", + "issue:42673", + "issue:42796", + "issue:43122", + "issue:43408", + "issue:43450", + "issue:43673", + "issue:43742", + "issue:43937", + "issue:43986", + "issue:44117", + "issue:44261", + "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:42673|issue:42796", + "issue:43937|issue:45563", + "issue:44380|issue:46077", + "issue:45356|issue:46614", + "issue:43673|issue:46328", + "issue:42673|issue:46655", + "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:39401|issue:44779", + "issue:43937|issue:44977", + "issue:45341|issue:46328", + "issue:43450|issue:43986", + "issue:45341|issue:45561", + "issue:44829|issue:45216", + "issue:45071|issue:46512", + "issue:44117|issue:45081", + "issue:38617|issue:41628", + "issue:43408|issue:43742", + "issue:43122|issue:46298", + "issue:42617|issue:45341", + "issue:45362|issue:46328", + "issue:42222|issue:46178", + "issue:44484|issue:45310", + "issue:44877|issue:46231", + "issue:42371|issue:44261" + ], + "prepared_review_unit_hash": "12cc3301578552ca3613cf0268b3dcf450a672d977b77904b025790bf4f638e8", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 29, + "serialized_chars": 36391, + "estimated_input_tokens": 9098, + "estimated_eval_tokens": 18452 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous set of unrelated transformer bugs and regressions. The soft pairs are mostly same-subsystem but different concrete failures, so none are strong duplicate candidates.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 43122 is the broadest regression-style report in the set and the safest representative issue, but it is not a true duplicate hub for the rest.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43122 has the most generic, version-regression framing and is the closest thing to a cluster representative among otherwise unrelated reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42673", + "right": "issue:42796", + "accept": false, + "reason": "One is a Qwen3ForCausalLM VRAM leak under multithreaded dataloading; the other is a tokenizer BPE-marker handling bug for Ministral. Different subsystem and failure mode." + }, + { + "left": "issue:43937", + "right": "issue:45563", + "accept": false, + "reason": "Both touch generation behavior, but one is an invalid GenerationConfig error and the other is a stale warning about num_return_sequences in paged generate(). Not the same bug." + }, + { + "left": "issue:44380", + "right": "issue:46077", + "accept": false, + "reason": "Different model families and different code paths: GPT2 attention scaling under SDPA/FlashAttention versus Gemma4 bidirectional mask construction." + }, + { + "left": "issue:45356", + "right": "issue:46614", + "accept": false, + "reason": "Both involve tokenizer/processor state, but one is Kimi-K2.5 codec/fix_mistral_regex regression and the other is Wav2Vec2 phonemizer language state retention. Distinct bugs." + }, + { + "left": "issue:43673", + "right": "issue:46328", + "accept": false, + "reason": "Chunked prefill cache handling and SequenceFeatureExtractor.pad list conversion are unrelated; one is a generation correctness issue, the other a performance inefficiency." + }, + { + "left": "issue:42673", + "right": "issue:46655", + "accept": false, + "reason": "A VRAM leak in Qwen3ForCausalLM is unrelated to SmolVLM video padding calculation." + }, + { + "left": "issue:42222", + "right": "issue:46752", + "accept": false, + "reason": "VitPose model breakage and empty-list apply_chat_template IndexError are unrelated failures in different parts of the stack." + }, + { + "left": "issue:44291", + "right": "issue:45071", + "accept": false, + "reason": "Unexpected _is_hf_initialized argument during init_empty_weights is a model-loading bug; PretrainedConfig type checking in v5.4.0 is a separate validation regression." + }, + { + "left": "issue:45405", + "right": "issue:45446", + "accept": false, + "reason": "A dependency version bump to an unreleased PEFT version is unrelated to a PyTorch version check bug in flex_attention." + }, + { + "left": "issue:44117", + "right": "issue:45588", + "accept": false, + "reason": "TOKENIZER_MAPPING_NAMES returning None during from_pretrained is a mapping/registry issue; flash_attention.py s_aux=None crash is an attention integration bug." + }, + { + "left": "issue:44466", + "right": "issue:46002", + "accept": false, + "reason": "Serialization of tied lm_head weights differs by device; NaNs in classification heads after checkout/init is a separate training/initialization problem." + }, + { + "left": "issue:44410", + "right": "issue:45310", + "accept": false, + "reason": "Both mention Qwen3.5, but one is missing projections in qwen3next layer 0 and the other is a from_pretrained load error. Not the same concrete defect." + }, + { + "left": "issue:39401", + "right": "issue:44779", + "accept": false, + "reason": "Qwen3 offset_mapping errors and Deepseek tokenizer regressions are both tokenization issues but clearly distinct models and behaviors." + }, + { + "left": "issue:43937", + "right": "issue:44977", + "accept": false, + "reason": "GenerationConfig validation failure and Qwen3.5 flash-attention generation problems are different code paths and symptoms." + }, + { + "left": "issue:45341", + "right": "issue:46328", + "accept": false, + "reason": "A testing_utils bug and a SequenceFeatureExtractor.pad inefficiency are unrelated; one is correctness, the other performance." + }, + { + "left": "issue:43450", + "right": "issue:43986", + "accept": false, + "reason": "Incorrect batched video processor shapes and AutoProcessor crashing without torchvision are different video-processing issues." + }, + { + "left": "issue:45341", + "right": "issue:45561", + "accept": false, + "reason": "Both involve testing utilities, but one is a generic bug in testing_utils.py and the other is an xdist race on captured_info.txt. Different concrete problems." + }, + { + "left": "issue:44829", + "right": "issue:45216", + "accept": false, + "reason": "FlashAttention-3 training degeneration and Qwen3.5 save_pretrained regression are unrelated model/backends and do not share the same fix." + }, + { + "left": "issue:45071", + "right": "issue:46512", + "accept": false, + "reason": "PretrainedConfig type checking regression and stale model.config.dtype after from_config are separate configuration issues." + }, + { + "left": "issue:44117", + "right": "issue:45081", + "accept": false, + "reason": "Registry None handling during from_pretrained and a Mistral regex patch crash are different failure points in tokenizer loading." + }, + { + "left": "issue:38617", + "right": "issue:41628", + "accept": false, + "reason": "Missing layer_type_validation import and missing AutoImageProcessor import are both import errors but in unrelated modules." + }, + { + "left": "issue:43408", + "right": "issue:43742", + "accept": false, + "reason": "A sam3_video/sam3_tracker type warning is not the same as the facebook/MobileLLM-125M key error when loading." + }, + { + "left": "issue:43122", + "right": "issue:46298", + "accept": false, + "reason": "Different tokenization regression versus a noisy logging message; no shared bug or code path." + }, + { + "left": "issue:42617", + "right": "issue:45341", + "accept": false, + "reason": "Failure to run 3d_parallel.py and a testing_utils bug are unrelated." + }, + { + "left": "issue:45362", + "right": "issue:46328", + "accept": false, + "reason": "Qwen3.5-35B chat crash and SequenceFeatureExtractor.pad inefficiency are unrelated." + }, + { + "left": "issue:42222", + "right": "issue:46178", + "accept": false, + "reason": "VitPose model breakage and a backend mapping ValueError for tensorflow_text are different problems." + }, + { + "left": "issue:44484", + "right": "issue:45310", + "accept": false, + "reason": "max_shard_size documentation/behavior question is unrelated to the Qwen3.5 Moe from_pretrained error." + }, + { + "left": "issue:44877", + "right": "issue:46231", + "accept": false, + "reason": "granite_speech strict config loading and a set_training doc parameter-name mismatch are unrelated." + }, + { + "left": "issue:42371", + "right": "issue:44261", + "accept": false, + "reason": "TF32 behavior API settings and MLA q_a_layernorm rms_norm_eps precision error are unrelated." + }, + { + "left": "issue:45561", + "right": "issue:46752", + "accept": false, + "reason": "xdist captured_info.txt race and empty-list apply_chat_template IndexError are different bugs in different subsystems." + } + ] + }, + "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": 23, + "review_unit_index": 53, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:27957", + "issue:38617", + "issue:39401", + "issue:41628", + "issue:42222", + "issue:42371", + "issue:43012", + "issue:43066", + "issue:43097", + "issue:43122", + "issue:43208", + "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:46172", + "issue:46209", + "issue:46617", + "issue:46762", + "issue:46894" + ], + "soft_pairs": [ + "issue:43937|issue:44496", + "issue:39401|issue:46209", + "issue:44410|issue:45478", + "issue:42371|issue:46617", + "issue:44261|issue:46762", + "issue:43208|issue:46894", + "issue:44933|issue:45278", + "issue:38617|issue:44220", + "issue:44075|issue:45656", + "issue:43066|issue:43122", + "issue:44263|issue:44485", + "issue:43097|issue:44493", + "issue:43931|issue:44410", + "issue:43673|issue:45341", + "issue:41628|issue:45071", + "issue:42222|issue:45542", + "issue:45071|issue:45542", + "issue:44483|issue:45464", + "issue:44484|issue:45478", + "issue:43874|issue:44261", + "issue:44263|issue:46172", + "issue:44360|issue:46172", + "issue:27957|issue:43352", + "issue:43931|issue:44038", + "issue:43012|issue:43577", + "issue:43441|issue:44829", + "issue:43097|issue:43452", + "issue:45003|issue:45341", + "issue:43986|issue:45341", + "issue:44261|issue:44485" + ], + "prepared_review_unit_hash": "a3f61105b700e4cdeb16d2469178d3b99f64a15ddbff08ac91f3d677443a2182", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 30, + "serialized_chars": 35564, + "estimated_input_tokens": 8891, + "estimated_eval_tokens": 18038 + }, + "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 unrelated model-loading, tokenizer, and API-breaking issues. None of the soft-paired artifacts look like the same concrete bug, so all soft edges are rejected. The best representative issue is the broad v5 import/regression umbrella.", + "confidence": 0.68, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report in the set: generic import breakages after upgrading versions, which best matches the cluster\u2019s overall regression-heavy theme.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the most representative global issue because it is broad, version-regression oriented, and closer to the cluster\u2019s common failure mode than the many highly specific model bugs.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43937", + "right": "issue:44496", + "accept": false, + "reason": "Different failures and code paths: GLM-5 generation config validation vs model config recognition for Olmo-Hybrid." + }, + { + "left": "issue:39401", + "right": "issue:46209", + "accept": false, + "reason": "Tokenizer offset_mapping bug is unrelated to FP8 inference failure on B200 GPUs." + }, + { + "left": "issue:44410", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3 variants, but one is missing projections in a layer and the other is a from_pretrained loading error." + }, + { + "left": "issue:42371", + "right": "issue:46617", + "accept": false, + "reason": "TF32 configuration guidance and trust_remote_code cache FileNotFoundError are unrelated issues." + }, + { + "left": "issue:44261", + "right": "issue:46762", + "accept": false, + "reason": "Both are numerical-difference reports, but they involve different models/modules and distinct root causes." + }, + { + "left": "issue:43208", + "right": "issue:46894", + "accept": false, + "reason": "Different model families and different training bugs; the shared \u201ctraining loss\u201d wording is not enough." + }, + { + "left": "issue:44933", + "right": "issue:45278", + "accept": false, + "reason": "One is a specific missing image_utils import, the other is a broad upgrade-wide import regression report." + }, + { + "left": "issue:38617", + "right": "issue:44220", + "accept": false, + "reason": "ImportError for configuration_utils has nothing to do with the _torch_extract_fbank_features issue." + }, + { + "left": "issue:44075", + "right": "issue:45656", + "accept": false, + "reason": "Both are optimizer-related, but one is ignored SGD args and the other is double optimizer steps under deepspeed." + }, + { + "left": "issue:43066", + "right": "issue:43122", + "accept": false, + "reason": "Both involve tokenizer behavior changes, but the symptoms and likely root causes are different." + }, + { + "left": "issue:44263", + "right": "issue:44485", + "accept": false, + "reason": "GlmMoeDsaIndexer split-return bug and GLM-5 RoPE implementation are separate components." + }, + { + "left": "issue:43097", + "right": "issue:44493", + "accept": false, + "reason": "Removed tie_embeddings API and unexpected position_id key errors are different regressions." + }, + { + "left": "issue:43931", + "right": "issue:44410", + "accept": false, + "reason": "Both are Qwen3-family, but one is a weight-shape mismatch and the other is missing attention projections." + }, + { + "left": "issue:43673", + "right": "issue:45341", + "accept": false, + "reason": "Generation cache handling and a testing_utils bug are unrelated." + }, + { + "left": "issue:41628", + "right": "issue:45071", + "accept": false, + "reason": "Importing AutoImageProcessor and PretrainedConfig type-checking are different breakages." + }, + { + "left": "issue:42222", + "right": "issue:45542", + "accept": false, + "reason": "VitPose model breakage and TensorBoard-without-TensorFlow backend errors are unrelated." + }, + { + "left": "issue:45071", + "right": "issue:45542", + "accept": false, + "reason": "PretrainedConfig typing regression does not match the TensorFlow backend import issue." + }, + { + "left": "issue:44483", + "right": "issue:45464", + "accept": false, + "reason": "Chat/completions request rejection and Qwen3.5 streaming failure are distinct API/model-path problems." + }, + { + "left": "issue:44484", + "right": "issue:45478", + "accept": false, + "reason": "save_pretrained shard-size behavior is unrelated to Qwen3.5 MoE loading failure." + }, + { + "left": "issue:43874", + "right": "issue:44261", + "accept": false, + "reason": "Missing image patch count helper and MLA layernorm epsilon are different multimodal model bugs." + }, + { + "left": "issue:44263", + "right": "issue:46172", + "accept": false, + "reason": "GlmMoeDsaIndexer torch.split behavior and deepseek-v4 tensor transpose are separate code paths." + }, + { + "left": "issue:44360", + "right": "issue:46172", + "accept": false, + "reason": "Missing ReLU in the DSA indexer is unrelated to the deepseek-v4 transpose issue." + }, + { + "left": "issue:27957", + "right": "issue:43352", + "accept": false, + "reason": "Both concern Flash Attention 2, but they are different model-specific support limitations." + }, + { + "left": "issue:43931", + "right": "issue:44038", + "accept": false, + "reason": "Qwen3-VL weight-shape mismatch and Qwen3-VL-Moe from_pretrained failure are not the same bug." + }, + { + "left": "issue:43012", + "right": "issue:43577", + "accept": false, + "reason": "Both mention dtype behavior, but one is a warning during compilation and the other is incorrect dtype after loading Blip2." + }, + { + "left": "issue:43441", + "right": "issue:44829", + "accept": false, + "reason": "Ministral FlashAttention failure and degenerate training with flash_attention_3 are different issues and code paths." + }, + { + "left": "issue:43097", + "right": "issue:43452", + "accept": false, + "reason": "An API removal warning and gguf_file loading failures are unrelated." + }, + { + "left": "issue:45003", + "right": "issue:45341", + "accept": false, + "reason": "Unsafe sys.modules access and a testing_utils bug do not describe the same underlying defect." + }, + { + "left": "issue:43986", + "right": "issue:45341", + "accept": false, + "reason": "AutoProcessor crashing without torchvision and a testing_utils bug are unrelated." + }, + { + "left": "issue:44261", + "right": "issue:44485", + "accept": false, + "reason": "MLA q_a_layernorm epsilon handling and GLM-5 RoPE implementation are distinct model 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": 23, + "review_unit_index": 54, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:38617", + "issue:42757", + "issue:42890", + "issue:43012", + "issue:43064", + "issue:43097", + "issue:43208", + "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:44704", + "issue:44936", + "issue:44987", + "issue:45103", + "issue:45341", + "issue:45372", + "issue:45375", + "issue:45440", + "issue:45588", + "issue:45753", + "issue:46050", + "issue:46123", + "issue:46172", + "issue:46298", + "issue:46365", + "issue:46369", + "issue:46421", + "issue:46655", + "issue:46664", + "issue:46701", + "issue:46710", + "issue:46752", + "issue:46821", + "issue:46897" + ], + "soft_pairs": [ + "issue:45753|issue:46369", + "issue:44117|issue:44987", + "issue:43874|issue:44263", + "issue:45440|issue:46172", + "issue:43531|issue:45341", + "issue:43673|issue:46123", + "issue:43097|issue:45588", + "issue:38617|issue:42757", + "issue:44038|issue:44410", + "issue:44410|issue:45375", + "issue:43208|issue:46897", + "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:46172|issue:46710", + "issue:38617|issue:44351", + "issue:46298|issue:46752", + "issue:44062|issue:44117", + "issue:43097|issue:43873", + "issue:43012|issue:46821", + "issue:44117|issue:46701", + "issue:43502|issue:43992", + "issue:44936|issue:46365", + "issue:44704|issue:45103", + "issue:44380|issue:46421" + ], + "prepared_review_unit_hash": "2f918ec9ce39aa1e34912192571527e1fcde5eae6e26f5d7fcd9129202b3eb5d", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 28, + "serialized_chars": 35332, + "estimated_input_tokens": 8833, + "estimated_eval_tokens": 17922 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is broadly heterogeneous: most items are distinct Transformers bugs or regressions, and the soft pairs are mostly same-subsystem lookalikes rather than true duplicates. I would not merge any of the proposed soft edges.", + "confidence": 0.81, + "canonical_issue_reason": "Issue 44117 is the cleanest representative: it describes a concrete upstream loading failure in from_pretrained/tokenizer mapping logic, which is more canonical than the model-specific or one-off reports in this set.", + "canonical_pr_reason": null, + "best_issue_reason": "44117 is the most actionable and broadly representative issue in the cluster; it has a clear root-cause shape and is the closest thing here to a reusable duplicate anchor.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45753", + "right": "issue:46369", + "accept": false, + "reason": "Different models and symptoms: an image-triggered Qwen3_5 loop vs a Trainer import/uninstall-datasets issue." + }, + { + "left": "issue:44117", + "right": "issue:44987", + "accept": false, + "reason": "Both involve loading, but one is a tokenizer mapping None bug and the other is a specific model-loading failure; no clear same root cause." + }, + { + "left": "issue:43874", + "right": "issue:44263", + "accept": false, + "reason": "GLM46V image processor missing method vs a torch.split indexing bug; different code paths." + }, + { + "left": "issue:45440", + "right": "issue:46172", + "accept": false, + "reason": "Different DeepSeek model families and different failures; not the same bug." + }, + { + "left": "issue:43531", + "right": "issue:45341", + "accept": false, + "reason": "Qwen3-MoE sliding-window behavior vs a testing_utils bug; unrelated." + }, + { + "left": "issue:43673", + "right": "issue:46123", + "accept": false, + "reason": "Chunked-prefill cache bug vs MaskGenerationPipeline final-batch handling; different subsystems and failure modes." + }, + { + "left": "issue:43097", + "right": "issue:45588", + "accept": false, + "reason": "API removal in tie_embeddings_and_encoder_decoder vs flash_attention crash on s_aux=None; unrelated." + }, + { + "left": "issue:38617", + "right": "issue:42757", + "accept": false, + "reason": "Both are ImportErrors, but for different symbols and different modules; not the same underlying issue." + }, + { + "left": "issue:44038", + "right": "issue:44410", + "accept": false, + "reason": "Same broad Qwen family, but one is a transformers 5.0 bug and the other is missing layer projections; distinct concrete problems." + }, + { + "left": "issue:44410", + "right": "issue:45375", + "accept": false, + "reason": "Both Qwen3-family config/model issues, but one is missing projections and the other is a missing config field; not one bug." + }, + { + "left": "issue:43208", + "right": "issue:46897", + "accept": false, + "reason": "xLSTM training blockers vs a Florence2 training-loss shift bug; unrelated." + }, + { + "left": "issue:43064", + "right": "issue:44315", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption vs Liger Kernel not applied with model_init; different fixes." + }, + { + "left": "issue:43986", + "right": "issue:45372", + "accept": false, + "reason": "Video AutoProcessor without torchvision vs a mistral_common import crash; unrelated." + }, + { + "left": "issue:43097", + "right": "issue:46050", + "accept": false, + "reason": "Removed tie_embeddings API vs quantized save/load corruption; different code paths." + }, + { + "left": "issue:43986", + "right": "issue:44484", + "accept": false, + "reason": "Video processor import crash vs max_shard_size default question; unrelated." + }, + { + "left": "issue:46655", + "right": "issue:46664", + "accept": false, + "reason": "Both are SmolVLM video-processor reports, but padding math and num_frames semantics are distinct bugs." + }, + { + "left": "issue:42890", + "right": "issue:46298", + "accept": false, + "reason": "Missing set_seed in tests vs a noisy logging message; not duplicates." + }, + { + "left": "issue:46172", + "right": "issue:46710", + "accept": false, + "reason": "DeepSeek-V4 tensor transform bug vs a downstream accuracy regression report; no direct same-root-cause evidence." + }, + { + "left": "issue:38617", + "right": "issue:44351", + "accept": false, + "reason": "Both import-name errors, but for different missing exports and different failure contexts." + }, + { + "left": "issue:46298", + "right": "issue:46752", + "accept": false, + "reason": "Logging noise vs apply_chat_template empty-list crash; unrelated." + }, + { + "left": "issue:44062", + "right": "issue:44117", + "accept": false, + "reason": "Tokenizer AddedToken keyword collision vs tokenizer mapping None; both tokenizer-related but different failures." + }, + { + "left": "issue:43097", + "right": "issue:43873", + "accept": false, + "reason": "API removal in model config vs quantization offloading behavior; different issues." + }, + { + "left": "issue:43012", + "right": "issue:46821", + "accept": false, + "reason": "PyTorch precision warning during bf16 compilation vs TimesFM window_size AttributeError; unrelated." + }, + { + "left": "issue:44117", + "right": "issue:46701", + "accept": false, + "reason": "Tokenizer mapping None bug vs DiffusionGemma pipeline failure; different code paths." + }, + { + "left": "issue:43502", + "right": "issue:43992", + "accept": false, + "reason": "local_files_only still making requests vs missing embed_tokens.weight on load; unrelated loading bugs." + }, + { + "left": "issue:44936", + "right": "issue:46365", + "accept": false, + "reason": "Trainer evaluate-after-train regression vs import failure until datasets is removed; not the same bug." + }, + { + "left": "issue:44704", + "right": "issue:45103", + "accept": false, + "reason": "AutoProcessor kwargs propagation bug vs auto_docstring failure under future annotations; unrelated." + }, + { + "left": "issue:44380", + "right": "issue:46421", + "accept": false, + "reason": "GPT2 attention scaling ignored in backends vs cache lazy-init recompilation; distinct problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 55, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:36683", + "issue:38617", + "issue:42175", + "issue:42490", + "issue:42757", + "issue:42907", + "issue:43012", + "issue:43064", + "issue:43066", + "issue:43097", + "issue:43208", + "issue:43352", + "issue:43502", + "issue:43526", + "issue:43550", + "issue:43582", + "issue:43618", + "issue:43756", + "issue:43824", + "issue:43825", + "issue:43906", + "issue:43976", + "issue:43986", + "issue:43992", + "issue:44038", + "issue:44077", + "issue:44117", + "issue:44162", + "issue:44222", + "issue:44230", + "issue:44368", + "issue:44380", + "issue:44483", + "issue:44484", + "issue:44485", + "issue:44514", + "issue:44655", + "issue:44938", + "issue:45478", + "issue:45479", + "issue:45584", + "issue:45712", + "issue:45753", + "issue:46132", + "issue:46172", + "issue:46178", + "issue:46298", + "issue:46365" + ], + "soft_pairs": [ + "issue:44938|issue:46178", + "issue:43986|issue:44117", + "issue:42907|issue:44655", + "issue:42907|issue:45753", + "issue:43825|issue:44655", + "issue:43012|issue:43550", + "issue:45712|issue:46298", + "issue:43208|issue:43550", + "issue:43012|issue:44368", + "issue:43992|issue:44484", + "issue:43352|issue:43976", + "issue:43352|issue:44380", + "issue:43066|issue:43906", + "issue:44162|issue:46172", + "issue:43097|issue:46298", + "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:42757|issue:43824", + "issue:43064|issue:44380", + "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:43756|issue:44485" + ], + "prepared_review_unit_hash": "c35b10bfd5205f636e10f508fcafb7860444d310c9295990c277a712a7c236f4", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 31, + "serialized_chars": 36743, + "estimated_input_tokens": 9186, + "estimated_eval_tokens": 18628 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is a grab bag of unrelated Transformers issues; none of the soft-edge pairs look like true duplicates. Most are different models, different code paths, or warning-vs-crash variants rather than the same underlying bug.", + "confidence": 0.74, + "canonical_issue_reason": "No strong duplicate center emerges; if one issue must anchor the cluster, issue 36683 is the highest-signal concrete runtime failure and has the most discussion.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 36683 is the best stand-in because it is a clear, user-facing crash with broad relevance and substantial conversation, unlike the many narrowly scoped reports here.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44938", + "right": "issue:46178", + "accept": false, + "reason": "Different failure modes: Python 3.14 import/load failure vs an invalid backend mapping for tensorflow_text." + }, + { + "left": "issue:43986", + "right": "issue:44117", + "accept": false, + "reason": "Both are import/loading-related, but one is missing torchvision handling and the other is a tokenizer mapping None bug; not the same code path." + }, + { + "left": "issue:42907", + "right": "issue:44655", + "accept": false, + "reason": "Saving dequantized models and saving Pipeline objects are different serialization paths." + }, + { + "left": "issue:42907", + "right": "issue:45753", + "accept": false, + "reason": "Model saving failure vs an infinite generation loop on a specific image are unrelated." + }, + { + "left": "issue:43825", + "right": "issue:44655", + "accept": false, + "reason": "Error-message behavior in pipeline() is unrelated to save_pretrained() pipeline object persistence." + }, + { + "left": "issue:43012", + "right": "issue:43550", + "accept": false, + "reason": "A precision warning during compilation is not the same as a torch.compile SDPA failure on a specific model." + }, + { + "left": "issue:45712", + "right": "issue:46298", + "accept": false, + "reason": "Left is about dummy class cleanup/export leaks; right is about a noisy log message." + }, + { + "left": "issue:43208", + "right": "issue:43550", + "accept": false, + "reason": "xLSTM training bugs are unrelated to Bamba torch.compile/SDPA behavior." + }, + { + "left": "issue:43012", + "right": "issue:44368", + "accept": false, + "reason": "Both mention warnings, but one is a bfloat16 compile warning and the other is a tie_word_embeddings config warning." + }, + { + "left": "issue:43992", + "right": "issue:44484", + "accept": false, + "reason": "Missing embed_tokens.weight on load is unrelated to the max_shard_size default in save_pretrained()." + }, + { + "left": "issue:43352", + "right": "issue:43976", + "accept": false, + "reason": "Flash Attention support for a model is unrelated to a Python version compatibility regression." + }, + { + "left": "issue:43352", + "right": "issue:44380", + "accept": false, + "reason": "Different attention/backend issues: model not supporting FA2 vs GPT2 scaling ignored under SDPA/FlashAttention." + }, + { + "left": "issue:43066", + "right": "issue:43906", + "accept": false, + "reason": "Tokenizer decoder type regression is not the same as an isolated reproduction of another issue." + }, + { + "left": "issue:44162", + "right": "issue:46172", + "accept": false, + "reason": "ESM2 model breakage and a deepseek matmul/transpose bug are different model-specific problems." + }, + { + "left": "issue:43097", + "right": "issue:46298", + "accept": false, + "reason": "Removed config flag behavior is unrelated to a noisy logging message." + }, + { + "left": "issue:44380", + "right": "issue:46132", + "accept": false, + "reason": "Attention scaling being ignored is distinct from AttentionInterface.register changing function behavior." + }, + { + "left": "issue:45584", + "right": "issue:46365", + "accept": false, + "reason": "Whisper generation edge case is unrelated to importing Trainer conflicting with datasets." + }, + { + "left": "issue:44077", + "right": "issue:46365", + "accept": false, + "reason": "PatchTSMixer post_init validation is unrelated to a Trainer import conflict." + }, + { + "left": "issue:44222", + "right": "issue:45478", + "accept": false, + "reason": "FP8 save_pretrained behavior and Qwen3.5 MoE from_pretrained errors are different directions and likely different bugs." + }, + { + "left": "issue:36683", + "right": "issue:43582", + "accept": false, + "reason": "Gemma3Config vocab_size crash is unrelated to an Apple Silicon TypeError in caching_allocator_warmup." + }, + { + "left": "issue:44380", + "right": "issue:45479", + "accept": false, + "reason": "GPT2 attention scaling under backend kernels is unrelated to zero-loss behavior for single-label classification." + }, + { + "left": "issue:42175", + "right": "issue:46365", + "accept": false, + "reason": "TensorFlow extra/backend packaging is unrelated to the datasets/Trainer import conflict." + }, + { + "left": "issue:42757", + "right": "issue:43824", + "accept": false, + "reason": "Both are import errors, but they involve different symbols and different packages." + }, + { + "left": "issue:43064", + "right": "issue:44380", + "accept": false, + "reason": "FSDP2/PEFT optimizer-state corruption is unrelated to GPT2 attention scaling behavior." + }, + { + "left": "issue:42490", + "right": "issue:43012", + "accept": false, + "reason": "Local path trailing-slash warning is unrelated to a bfloat16 compile warning." + }, + { + "left": "issue:43208", + "right": "issue:43526", + "accept": false, + "reason": "xLSTM training bugs and BeitImageProcessorFast label reduction are unrelated subsystems." + }, + { + "left": "issue:44483", + "right": "issue:44514", + "accept": false, + "reason": "Chat completions request handling is different from a batched apply_chat_template crash." + }, + { + "left": "issue:43618", + "right": "issue:44380", + "accept": false, + "reason": "CLIPOutput attentions assignment regression is unrelated to GPT2 attention scaling under backend kernels." + }, + { + "left": "issue:38617", + "right": "issue:43502", + "accept": false, + "reason": "Missing configuration_utils symbol import and local_files_only network leakage are different bugs." + }, + { + "left": "issue:44038", + "right": "issue:44230", + "accept": false, + "reason": "Qwen3-VL-Moe support issues here are about different model variants and different failure points." + }, + { + "left": "issue:43756", + "right": "issue:44485", + "accept": false, + "reason": "RoPE-layer dropping in Smollm3 and GLM-5 RoPE implementation questions are model-specific and not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 56, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29870", + "issue:33453", + "issue:36010", + "issue:36296", + "issue:36683", + "issue:42757", + "issue:42890", + "issue:42994", + "issue:43010", + "issue:43064", + "issue:43208", + "issue:43299", + "issue:43352", + "issue:43475", + "issue:43525", + "issue:43611", + "issue:43618", + "issue:43761", + "issue:43867", + "issue:43901", + "issue:43937", + "issue:43976", + "issue:43992", + "issue:44075", + "issue:44077", + "issue:44355", + "issue:44483", + "issue:44492", + "issue:44625", + "issue:44655", + "issue:44855", + "issue:44908", + "issue:44938", + "issue:45071", + "issue:45216", + "issue:45276", + "issue:45278", + "issue:45310", + "issue:45375", + "issue:45561", + "issue:46077", + "issue:46082", + "issue:46202", + "issue:46365", + "issue:46369", + "issue:46829", + "issue:46851", + "issue:46877" + ], + "soft_pairs": [ + "issue:42994|issue:44655", + "issue:43618|issue:43761", + "issue:43299|issue:43867", + "issue:36010|issue:42757", + "issue:44625|issue:46365", + "issue:45276|issue:46077", + "issue:44355|issue:46877", + "issue:43352|issue:43937", + "issue:44483|issue:46369", + "issue:36683|issue:43475", + "issue:43611|issue:44655", + "issue:43901|issue:44655", + "issue:43867|issue:45216", + "issue:43976|issue:44938", + "issue:36683|issue:44077", + "issue:44625|issue:46077", + "issue:46077|issue:46082", + "issue:43618|issue:45216", + "issue:43208|issue:46829", + "issue:36683|issue:43525", + "issue:36683|issue:44625", + "issue:29870|issue:36296", + "issue:43992|issue:45310", + "issue:43064|issue:44492", + "issue:43010|issue:45278", + "issue:44075|issue:46365", + "issue:44908|issue:46851", + "issue:29870|issue:33453", + "issue:44938|issue:45071", + "issue:44355|issue:44855", + "issue:43867|issue:46202", + "issue:36683|issue:43064", + "issue:43867|issue:45375", + "issue:42890|issue:45561" + ], + "prepared_review_unit_hash": "73b75afd55ce0ea8c3a48f22f0fe545d2a4790eace9eb3b74c92d742e486fab4", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 37588, + "estimated_input_tokens": 9397, + "estimated_eval_tokens": 19050 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This is a heterogeneous set of unrelated bugs and regressions. The soft pairs mostly share only broad themes like import/load failures or model-specific regressions, not the same concrete underlying defect, so no pair should be merged.", + "confidence": 0.39, + "canonical_issue_reason": "issue:45278 is the broadest umbrella report here (general import failures after an upgrade), so it is the least narrow representative if one issue must anchor the cluster.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45278 is the closest thing to a cluster anchor because it describes a wide regression class rather than a model-specific symptom, but the cluster overall is not cohesive.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:42994", + "right": "issue:44655", + "accept": false, + "reason": "Both concern saving, but one is quantized model saving and the other is Pipeline.save_pretrained; different failure modes and code paths." + }, + { + "left": "issue:43618", + "right": "issue:43761", + "accept": false, + "reason": "Both touch CLIP v5 regression behavior, but one is missing attentions and the other is missing hidden_states; same subsystem, not clearly the same bug." + }, + { + "left": "issue:43299", + "right": "issue:43867", + "accept": false, + "reason": "Different problems: Qwen3VL MoE loading versus a generic sorted-state_dict load error." + }, + { + "left": "issue:36010", + "right": "issue:42757", + "accept": false, + "reason": "Both are import errors, but they involve different symbols and different dependency breakages." + }, + { + "left": "issue:44625", + "right": "issue:46365", + "accept": false, + "reason": "Unrelated: Qwen3.5 num_labels propagation versus importing Trainer being blocked by datasets." + }, + { + "left": "issue:45276", + "right": "issue:46077", + "accept": false, + "reason": "Same model family, but one is embedding resize propagation and the other is attention-mask behavior; not the same underlying bug." + }, + { + "left": "issue:44355", + "right": "issue:46877", + "accept": false, + "reason": "One is about running compiled Python files, the other about import-time docstring lint output." + }, + { + "left": "issue:43352", + "right": "issue:43937", + "accept": false, + "reason": "Different model-specific errors: Flash Attention support versus invalid GenerationConfig." + }, + { + "left": "issue:44483", + "right": "issue:46369", + "accept": false, + "reason": "Completely different areas: chat/completions request validation versus SAM3 memory overflow." + }, + { + "left": "issue:36683", + "right": "issue:43475", + "accept": false, + "reason": "Different model configs and different missing attributes; only superficial AttributeError similarity." + }, + { + "left": "issue:43611", + "right": "issue:44655", + "accept": false, + "reason": "Model loading with base_model_prefix is unrelated to saving Pipeline objects." + }, + { + "left": "issue:43901", + "right": "issue:44655", + "accept": false, + "reason": "Docs mismatch for TextClassificationPipeline is unrelated to pipeline save_pretrained failure." + }, + { + "left": "issue:43867", + "right": "issue:45216", + "accept": false, + "reason": "Both involve load/save regressions, but one is sorted state_dict loading and the other is checkpoint saving correctness for Qwen3.5." + }, + { + "left": "issue:43976", + "right": "issue:44938", + "accept": false, + "reason": "Both are Python-version compatibility issues, but they are different version bounds and different failures." + }, + { + "left": "issue:36683", + "right": "issue:44077", + "accept": false, + "reason": "Unrelated: Gemma3 vocab_size missing versus SGD optimizer args not being used." + }, + { + "left": "issue:44625", + "right": "issue:46077", + "accept": false, + "reason": "Different Gemma/Qwen model bugs; no shared concrete code-path." + }, + { + "left": "issue:46077", + "right": "issue:46082", + "accept": false, + "reason": "Both are Llama/Gemma-style config issues, but one is attention masking and the other is head_dim validation." + }, + { + "left": "issue:43618", + "right": "issue:45216", + "accept": false, + "reason": "Different regressions in different models: CLIP output fields versus Qwen3.5 checkpoint saving." + }, + { + "left": "issue:43208", + "right": "issue:46829", + "accept": false, + "reason": "xLSTM training bugs are unrelated to VideoPrism output wiring." + }, + { + "left": "issue:36683", + "right": "issue:43525", + "accept": false, + "reason": "Different missing attributes in different configs; same error class only superficially." + }, + { + "left": "issue:36683", + "right": "issue:44625", + "accept": false, + "reason": "Both are config-related, but the missing fields and affected models are different." + }, + { + "left": "issue:29870", + "right": "issue:36296", + "accept": false, + "reason": "Bart generate decoder_attention_mask error is unrelated to tensor-parallel training bugs." + }, + { + "left": "issue:43992", + "right": "issue:45310", + "accept": false, + "reason": "Both involve from_pretrained, but they affect different models and different missing/invalid state handling." + }, + { + "left": "issue:43064", + "right": "issue:44492", + "accept": false, + "reason": "Trainer/FSDP optimizer-state corruption is unrelated to a cache-strategy typo." + }, + { + "left": "issue:43010", + "right": "issue:45278", + "accept": false, + "reason": "Cache update no_grad decoration is unrelated to broad import errors after upgrade." + }, + { + "left": "issue:44075", + "right": "issue:46365", + "accept": false, + "reason": "Optimizer argument handling is unrelated to importing Trainer." + }, + { + "left": "issue:44908", + "right": "issue:46851", + "accept": false, + "reason": "Scheduler kwargs are unrelated to Qwen3.5 variable-length position_ids behavior." + }, + { + "left": "issue:29870", + "right": "issue:33453", + "accept": false, + "reason": "Bart generation error and tokenizer-loading regression are different failures." + }, + { + "left": "issue:44938", + "right": "issue:45071", + "accept": false, + "reason": "Python 3.14 load failure and PretrainedConfig type-checking regression are different compatibility issues." + }, + { + "left": "issue:44355", + "right": "issue:44855", + "accept": false, + "reason": "Both are import-time problems, but one is compiled-file execution and the other is a torch.jit/script parsing issue." + }, + { + "left": "issue:43867", + "right": "issue:46202", + "accept": false, + "reason": "Generic load error and Gemma pipeline roundtrip failure are not the same bug." + }, + { + "left": "issue:36683", + "right": "issue:43064", + "accept": false, + "reason": "Gemma3 config attribute missing is unrelated to FSDP2/PEFT optimizer-state corruption." + }, + { + "left": "issue:43867", + "right": "issue:45375", + "accept": false, + "reason": "Both are config/load regressions, but they concern different models and different missing fields." + }, + { + "left": "issue:42890", + "right": "issue:45561", + "accept": false, + "reason": "One is test flakiness from missing set_seed in a model integration test; the other is an xdist race in patched testing utils." + } + ] + }, + "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": 23, + "review_unit_index": 57, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28282", + "issue:30333", + "issue:33453", + "issue:36296", + "issue:41669", + "issue:42548", + "issue:42890", + "issue:42898", + "issue:43010", + "issue:43065", + "issue:43232", + "issue:43502", + "issue:43575", + "issue:43576", + "issue:43618", + "issue:43723", + "issue:43824", + "issue:43867", + "issue:43874", + "issue:43976", + "issue:43992", + "issue:44062", + "issue:44077", + "issue:44112", + "issue:44222", + "issue:44246", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44387", + "issue:44485", + "issue:44556", + "issue:44734", + "issue:44749", + "issue:44961", + "issue:44964", + "issue:45125", + "issue:45357", + "issue:45375", + "issue:45440", + "issue:45478", + "issue:45600", + "issue:45850", + "issue:46132", + "issue:46285", + "issue:46617", + "issue:46664", + "issue:46726" + ], + "soft_pairs": [ + "issue:43010|issue:45850", + "issue:41669|issue:44246", + "issue:36296|issue:43575", + "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:43723|issue:44556", + "issue:44222|issue:45125", + "issue:44387|issue:44556", + "issue:43824|issue:46617", + "issue:43502|issue:44485", + "issue:43874|issue:46617", + "issue:43576|issue:46132", + "issue:43010|issue:43065", + "issue:45125|issue:45478", + "issue:30333|issue:33453", + "issue:44355|issue:44749", + "issue:42548|issue:44351", + "issue:45125|issue:45375", + "issue:46664|issue:46726", + "issue:43010|issue:43232", + "issue:43502|issue:44222", + "issue:43502|issue:44961" + ], + "prepared_review_unit_hash": "d20a5b5d62049028d8d73c38649b249ce1599735b65559dca9f63885923681f1", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 33, + "serialized_chars": 37183, + "estimated_input_tokens": 9296, + "estimated_eval_tokens": 18848 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is highly heterogeneous: most pairs are false-positive similarity links across unrelated import errors, tokenizer/model regressions, training/test flakiness, and model-specific issues. Only the Qwen3.5 MoE tensor-parallel pair looks plausibly like the same underlying bug.", + "confidence": 0.9, + "canonical_issue_reason": "No single canonical issue is safe here; the items span unrelated code paths and models, with only a narrow Qwen3.5 MoE subset showing real overlap.", + "canonical_pr_reason": "No PRs are present in this cluster.", + "best_issue_reason": "No global best issue for the whole cluster. If anything, issue 45125 is the most concrete root-cause report, but it only fits a small subset of the items.", + "best_pr_reason": "No PRs are present in this cluster.", + "soft_edge_verdicts": [ + { + "left": "issue:43010", + "right": "issue:45850", + "accept": false, + "reason": "Different scopes: a cache/no_grad code fix vs a broad complaint that many models break in v5." + }, + { + "left": "issue:41669", + "right": "issue:44246", + "accept": false, + "reason": "Both mention import performance, but one is a specific import-star cleanup and the other is a generic slow-import report." + }, + { + "left": "issue:36296", + "right": "issue:43575", + "accept": false, + "reason": "Both touch tensor parallelism, but one is a generic training bug and the other is an OOM on loading a specific model." + }, + { + "left": "issue:43010", + "right": "issue:43976", + "accept": false, + "reason": "Unrelated problems: cache update gradients vs Python version compatibility." + }, + { + "left": "issue:43618", + "right": "issue:44077", + "accept": false, + "reason": "Different model/code paths; one is CLIP output fields, the other is a patchtsmixer init restriction." + }, + { + "left": "issue:43992", + "right": "issue:45478", + "accept": false, + "reason": "Both are from_pretrained failures, but they affect different models and appear to have different causes." + }, + { + "left": "issue:43867", + "right": "issue:45357", + "accept": false, + "reason": "Serialization/loading bug vs save_pretrained key mismatch; not the same concrete defect." + }, + { + "left": "issue:28282", + "right": "issue:33453", + "accept": false, + "reason": "Completely unrelated: missing PyTorch import vs tokenizer-loading regression." + }, + { + "left": "issue:43618", + "right": "issue:46285", + "accept": false, + "reason": "Same broad CLIP area, but one is an output attribute issue and the other is a structural model API change." + }, + { + "left": "issue:44360", + "right": "issue:45440", + "accept": false, + "reason": "Different subsystems and symptoms: DSA indexer issue vs DeepSeek MoE implementation divergence." + }, + { + "left": "issue:42890", + "right": "issue:44964", + "accept": false, + "reason": "Test flakiness in SAM HQ is unrelated to a Phi-4 multimodal load failure." + }, + { + "left": "issue:42890", + "right": "issue:44112", + "accept": false, + "reason": "Both are CI/test failures, but they are different tests with different failure mechanisms." + }, + { + "left": "issue:42890", + "right": "issue:45600", + "accept": false, + "reason": "Test seeding issue vs stale auto-mapping references; not the same bug." + }, + { + "left": "issue:42898", + "right": "issue:46664", + "accept": false, + "reason": "Tokenizer behavior change vs video processor frame-count semantics; unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44062", + "accept": false, + "reason": "Network access despite local_files_only is unrelated to AddedToken argument duplication." + }, + { + "left": "issue:44734", + "right": "issue:46617", + "accept": false, + "reason": "KV-cache continuation crash in serve vs cache FileNotFoundError for trust_remote_code; different paths." + }, + { + "left": "issue:43723", + "right": "issue:44556", + "accept": false, + "reason": "Tokenizer-loading regression vs broad checkpoint reload regression after upgrade; too broad to merge." + }, + { + "left": "issue:44222", + "right": "issue:45125", + "accept": false, + "reason": "FP8 save_pretrained issue and missing tensor-parallel plan are different bugs." + }, + { + "left": "issue:44387", + "right": "issue:44556", + "accept": false, + "reason": "Quantization memory regression vs checkpoint reload incompatibility; not the same defect." + }, + { + "left": "issue:43824", + "right": "issue:46617", + "accept": false, + "reason": "Different import/load failures affecting different symbols and code paths." + }, + { + "left": "issue:43502", + "right": "issue:44485", + "accept": false, + "reason": "Network-request bug vs GLM-5 RoPE implementation discussion; unrelated." + }, + { + "left": "issue:43874", + "right": "issue:46617", + "accept": false, + "reason": "Image patch-count helper missing vs trust_remote_code cache loading failure; distinct issues." + }, + { + "left": "issue:43576", + "right": "issue:46132", + "accept": false, + "reason": "Broken env CLI vs AttentionInterface behavior change; different features and failure modes." + }, + { + "left": "issue:43010", + "right": "issue:43065", + "accept": false, + "reason": "Cache update gradient handling vs dummy Conv2d in Sam3PixelDecoder; unrelated." + }, + { + "left": "issue:45125", + "right": "issue:45478", + "accept": true, + "reason": "Both point to the same Qwen3.5 MoE regression around tensor-parallel/from_pretrained behavior, and a single fix could plausibly address both." + }, + { + "left": "issue:30333", + "right": "issue:33453", + "accept": false, + "reason": "MLFlow reporting on failed jobs vs tokenizer loading regression; no shared code path." + }, + { + "left": "issue:44355", + "right": "issue:44749", + "accept": false, + "reason": "Compiled Python file errors vs model-filter slowdown after upgrade; unrelated." + }, + { + "left": "issue:42548", + "right": "issue:44351", + "accept": false, + "reason": "Both are import errors, but they refer to different missing exports and likely different root causes." + }, + { + "left": "issue:45125", + "right": "issue:45375", + "accept": false, + "reason": "Both involve Qwen3.5-related config, but one is tensor-parallel planning and the other is a missing vision config field." + }, + { + "left": "issue:46664", + "right": "issue:46726", + "accept": false, + "reason": "Video processor frame semantics vs missing vision backend image-processor handling; different bugs." + }, + { + "left": "issue:43010", + "right": "issue:43232", + "accept": false, + "reason": "Both touch generation/cache internals, but one is about no_grad decoration and the other about post-sync_gpus kwargs updates." + }, + { + "left": "issue:43502", + "right": "issue:44222", + "accept": false, + "reason": "Local-files-only networking bug vs FP8 save_pretrained issue; unrelated." + }, + { + "left": "issue:43502", + "right": "issue:44961", + "accept": false, + "reason": "No meaningful overlap; one is an API/networking bug, the other is unrelated." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The cluster summary is well-grounded in the packet: the items are broadly heterogeneous, and the only clearly plausible same-bug link is the Qwen3.5 MoE tensor-parallel/from_pretrained pair (45125, 45478). The other soft-edge rejections are conservative and align with the issue titles without overstating evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 58, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:29870", + "issue:30333", + "issue:30990", + "issue:31795", + "issue:33453", + "issue:34689", + "issue:36296", + "issue:38617", + "issue:41628", + "issue:42548", + "issue:42757", + "issue:43299", + "issue:43475", + "issue:43576", + "issue:43673", + "issue:43824", + "issue:43827", + "issue:43901", + "issue:44038", + "issue:44222", + "issue:44230", + "issue:44273", + "issue:44315", + "issue:44351", + "issue:44355", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44623", + "issue:44908", + "issue:44938", + "issue:44961", + "issue:45003", + "issue:45071", + "issue:45125", + "issue:45230", + "issue:45362", + "issue:45468", + "issue:45542", + "issue:45753", + "issue:45850", + "issue:46143", + "issue:46144", + "issue:46209", + "issue:46617", + "issue:46664" + ], + "soft_pairs": [ + "issue:30990|issue:31795", + "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:42757|issue:43576", + "issue:43576|issue:45071", + "issue:44315|issue:44360", + "issue:31795|issue:38617", + "issue:43576|issue:43824", + "issue:44961|issue:45468", + "issue:44485|issue:44961", + "issue:44351|issue:46617", + "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:45071", + "issue:44393|issue:45753", + "issue:43576|issue:44556" + ], + "prepared_review_unit_hash": "ef33d5193e67a47f668e9091a43dc7e4ec81f7dafbc43b3419b4ad8e4057a3ac", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 48, + "item_count": 48, + "soft_pair_count": 34, + "serialized_chars": 36921, + "estimated_input_tokens": 9231, + "estimated_eval_tokens": 18718 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The set is a grab bag of mostly unrelated Transformers issues: loading/import regressions, model-specific bugs, API/export breaks, and documentation problems. None of the soft pairs look safe to collapse into a single duplicate, especially for PR-style strictness.", + "confidence": 0.92, + "canonical_issue_reason": "Issue 33453 is the cleanest concrete regression report in the set and is easier to triage than the very broad or vague titles; if one issue must anchor the group, this is the most representative user-facing bug report.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 33453 is the best single issue to keep as a canonical example because it is specific, clearly describes a regression, and is more actionable than the broad/vague alternatives.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:30990", + "right": "issue:31795", + "accept": false, + "reason": "Stuck-loading behavior and documentation confusion are different problems; no shared failing code path." + }, + { + "left": "issue:43827", + "right": "issue:44961", + "accept": false, + "reason": "Docs about removed pipeline() usage vs an unrelated placeholder title; no duplicate signal." + }, + { + "left": "issue:30333", + "right": "issue:36296", + "accept": false, + "reason": "MLflow reporting on failed jobs and tensor-parallel training bugs are unrelated." + }, + { + "left": "issue:43299", + "right": "issue:44556", + "accept": false, + "reason": "Both mention version upgrades, but one is Qwen3VL-Moe loading and the other is checkpoint reload compatibility; too broad to merge." + }, + { + "left": "issue:29870", + "right": "issue:31795", + "accept": false, + "reason": "Runtime UnboundLocalError in generate() vs documentation confusion; different issue types and code paths." + }, + { + "left": "issue:44355", + "right": "issue:45071", + "accept": false, + "reason": "Compiled Python file errors and PretrainedConfig type-checking are unrelated failures." + }, + { + "left": "issue:33453", + "right": "issue:34689", + "accept": false, + "reason": "Both are loading regressions, but one is tokenizer loading and the other is a model-loading break for a specific Llama VLM; not the same bug." + }, + { + "left": "issue:42757", + "right": "issue:43576", + "accept": false, + "reason": "Missing import from huggingface_hub vs broken transformers env command; no evidence they share the same fix." + }, + { + "left": "issue:43576", + "right": "issue:45071", + "accept": false, + "reason": "CLI env command regression and config type-checking break are separate code paths." + }, + { + "left": "issue:44315", + "right": "issue:44360", + "accept": false, + "reason": "Liger Kernel application in model_init and DSA indexer/ReLU discussion are unrelated." + }, + { + "left": "issue:31795", + "right": "issue:38617", + "accept": false, + "reason": "Docs clarity issue vs ImportError for a missing symbol; different underlying problems." + }, + { + "left": "issue:43576", + "right": "issue:43824", + "accept": false, + "reason": "Broken env command and Qwen2_5_VL import failure are separate API/CLI failures." + }, + { + "left": "issue:44961", + "right": "issue:45468", + "accept": false, + "reason": "Placeholder title and Gemma4 audio positional encoding bug are not duplicates." + }, + { + "left": "issue:44485", + "right": "issue:44961", + "accept": false, + "reason": "GLM-5 RoPE implementation discussion is unrelated to the placeholder issue." + }, + { + "left": "issue:44351", + "right": "issue:46617", + "accept": false, + "reason": "Top-level HybridCache import failure and trust_remote_code cache FileNotFoundError are different loading failures." + }, + { + "left": "issue:44908", + "right": "issue:44961", + "accept": false, + "reason": "Scheduler kwargs handling and placeholder title have no overlap." + }, + { + "left": "issue:43576", + "right": "issue:44938", + "accept": false, + "reason": "Broken env subcommand and Python 3.14 load failure are unrelated." + }, + { + "left": "issue:36296", + "right": "issue:43901", + "accept": false, + "reason": "Tensor-parallel training bug and docs about return_all_scores are not the same issue." + }, + { + "left": "issue:46143", + "right": "issue:46144", + "accept": false, + "reason": "Both RoFormer-related, but one is kwargs propagation and the other is attention interface usage; not one concrete bug." + }, + { + "left": "issue:43475", + "right": "issue:46617", + "accept": false, + "reason": "SAM 3 Video missing attribute and cache FileNotFoundError are unrelated." + }, + { + "left": "issue:44222", + "right": "issue:44230", + "accept": false, + "reason": "Saving FP8 MoE models and inference support for FP8 MoE models are different code paths." + }, + { + "left": "issue:41628", + "right": "issue:42548", + "accept": false, + "reason": "Different missing top-level imports; likely separate export regressions, not one duplicate." + }, + { + "left": "issue:44315", + "right": "issue:46209", + "accept": false, + "reason": "Liger Kernel application bug and FP8 inference on B200 are unrelated." + }, + { + "left": "issue:44485", + "right": "issue:46664", + "accept": false, + "reason": "RoPE implementation discussion and SmolVLM num_frames semantics are different model areas." + }, + { + "left": "issue:43673", + "right": "issue:44492", + "accept": false, + "reason": "Generation cache missing in chunked_prefill and a cache-strategy typo do not describe the same defect." + }, + { + "left": "issue:45362", + "right": "issue:45850", + "accept": false, + "reason": "Specific Qwen3.5 chat crash vs a vague 'breaks too many models' complaint; not the same concrete bug." + }, + { + "left": "issue:44273", + "right": "issue:45542", + "accept": false, + "reason": "Lazy loading regression and TensorBoard-without-TensorFlow backend error are unrelated." + }, + { + "left": "issue:44623", + "right": "issue:45230", + "accept": false, + "reason": "Missing save_pretrained files is specific, while the other is a generic bug report with no concrete overlap." + }, + { + "left": "issue:44360", + "right": "issue:46664", + "accept": false, + "reason": "DSA indexer/ReLU discussion and SmolVLM video-frame semantics are unrelated." + }, + { + "left": "issue:44038", + "right": "issue:45125", + "accept": false, + "reason": "Qwen3-VL-Moe loading breakage and missing _tp_plan for tensor parallelism are separate model-loading vs parallelism bugs." + }, + { + "left": "issue:44273", + "right": "issue:45003", + "accept": false, + "reason": "Lazy loading problems and unsafe sys.modules access are different implementation defects." + }, + { + "left": "issue:44393", + "right": "issue:45071", + "accept": false, + "reason": "Qwen3-VL bounding-box output issue and PretrainedConfig type checking are unrelated." + }, + { + "left": "issue:44393", + "right": "issue:45753", + "accept": false, + "reason": "Bounding-box output errors and infinite-loop behavior on a specific image are different failure modes." + }, + { + "left": "issue:43576", + "right": "issue:44556", + "accept": false, + "reason": "Broken env command and checkpoint reload incompatibility are both version-related but not the same bug." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 59, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:31795", + "issue:33453", + "issue:34689", + "issue:41669", + "issue:42371", + "issue:42890", + "issue:43352", + "issue:43408", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43576", + "issue:43673", + "issue:43704", + "issue:43976", + "issue:44077", + "issue:44230", + "issue:44246", + "issue:44273", + "issue:44295", + "issue:44297", + "issue:44315", + "issue:44351", + "issue:44360", + "issue:44393", + "issue:44485", + "issue:44492", + "issue:44556", + "issue:44655", + "issue:44704", + "issue:44829", + "issue:44861", + "issue:44908", + "issue:45071", + "issue:45092", + "issue:45230", + "issue:45278", + "issue:45310", + "issue:45341", + "issue:45468", + "issue:45478", + "issue:45850", + "issue:46095", + "issue:46218", + "issue:46328", + "issue:46655", + "issue:46894" + ], + "soft_pairs": [ + "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:44230|issue:46095", + "issue:42371|issue:44492", + "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:44908|issue:45071", + "issue:44295|issue:45230", + "issue:44297|issue:45230", + "issue:44704|issue:44908", + "issue:44492|issue:46328", + "issue:41669|issue:44351", + "issue:41669|issue:45341", + "issue:43976|issue:45850", + "issue:44393|issue:45478", + "issue:43976|issue:46218", + "issue:44077|issue:44861", + "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:45230|issue:45310", + "issue:43576|issue:46218", + "issue:43519|issue:44485", + "issue:45468|issue:46894", + "issue:43519|issue:44908", + "issue:41669|issue:43704", + "issue:31795|issue:34689" + ], + "prepared_review_unit_hash": "379b810dcfa87142a013e703a625fcfe2d3c8ae1e70344abb271b13e76ba8a70", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 47, + "item_count": 47, + "soft_pair_count": 40, + "serialized_chars": 38545, + "estimated_input_tokens": 9637, + "estimated_eval_tokens": 19530 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is dominated by v5-era regressions and unrelated model/subsystem-specific bugs that only look similar at a high level; none of the soft edges appear to be the same concrete bug/change.", + "confidence": 0.83, + "canonical_issue_reason": "issue:45850 is the broadest umbrella report in the set and best matches the general v5 regression theme, while the other issues are narrower, model-specific symptoms.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45850 is the most representative issue for this cluster because it summarizes the broad cross-model breakage pattern better than the more specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41669", + "right": "issue:43673", + "accept": false, + "reason": "One is import-star/lazy-loading performance, the other is a generation cache regression; not the same code-path bug." + }, + { + "left": "issue:43525", + "right": "issue:45850", + "accept": false, + "reason": "Specific missing pad_token_id in Llama4Config versus a broad v5 breakage complaint; too generic to be a duplicate." + }, + { + "left": "issue:44556", + "right": "issue:45092", + "accept": false, + "reason": "Both involve loading failures after v5 upgrades, but they affect different checkpoint types and likely different causes." + }, + { + "left": "issue:43572", + "right": "issue:45850", + "accept": false, + "reason": "StableLmConfig pad_token_idx regression is a concrete config bug, not the same underlying issue as the broad v5-breakage report." + }, + { + "left": "issue:44492", + "right": "issue:45850", + "accept": false, + "reason": "A typo in cache strategies is unrelated to a general claim that v5 breaks many models." + }, + { + "left": "issue:45278", + "right": "issue:45850", + "accept": false, + "reason": "General import errors after upgrade are broader, but this is still a separate report from the umbrella complaint." + }, + { + "left": "issue:44230", + "right": "issue:46095", + "accept": false, + "reason": "FP8 inference support for Qwen models is a different feature request from a DeepSeek weight-conversion question." + }, + { + "left": "issue:42371", + "right": "issue:44492", + "accept": false, + "reason": "TF32 API guidance and a cache-strategy typo are unrelated issues." + }, + { + "left": "issue:41669", + "right": "issue:46218", + "accept": false, + "reason": "Import-star cleanup/performance and a missing Qwen3_5 module import error are not the same bug." + }, + { + "left": "issue:43408", + "right": "issue:44315", + "accept": false, + "reason": "Wrong model-type warning versus Liger Kernel not being applied in model_init; different code paths." + }, + { + "left": "issue:42890", + "right": "issue:44908", + "accept": false, + "reason": "Test flakiness from missing set_seed and a scheduler kwarg bug are unrelated." + }, + { + "left": "issue:44246", + "right": "issue:44273", + "accept": false, + "reason": "Both touch import/lazy-loading performance, but one is about slow import and the other claims lazy loading is broken; not clearly the same defect." + }, + { + "left": "issue:43576", + "right": "issue:44908", + "accept": false, + "reason": "Broken `transformers env` command and inverse_sqrt scheduler kwargs are unrelated." + }, + { + "left": "issue:43352", + "right": "issue:44393", + "accept": false, + "reason": "Flash Attention 2.0 support gating and Qwen3-VL bounding-box output errors are different problems." + }, + { + "left": "issue:43976", + "right": "issue:44393", + "accept": false, + "reason": "Python version compatibility and Qwen3-VL output errors are unrelated." + }, + { + "left": "issue:44908", + "right": "issue:45071", + "accept": false, + "reason": "Scheduler kwarg handling and PretrainedConfig type-checking regressions are different code paths." + }, + { + "left": "issue:44295", + "right": "issue:45230", + "accept": false, + "reason": "Position_ids buffer access issue versus a generic bug report; no concrete match." + }, + { + "left": "issue:44297", + "right": "issue:45230", + "accept": false, + "reason": "Tokenizer save_pretrained mismatch is a specific serialization bug, not the same as a generic bug report." + }, + { + "left": "issue:44704", + "right": "issue:44908", + "accept": false, + "reason": "AutoProcessor kwargs forwarding and scheduler kwargs are unrelated subsystems." + }, + { + "left": "issue:44492", + "right": "issue:46328", + "accept": false, + "reason": "Cache strategy typo and SequenceFeatureExtractor.pad performance are unrelated." + }, + { + "left": "issue:41669", + "right": "issue:44351", + "accept": false, + "reason": "Import-star cleanup/performance does not match a missing HybridCache import error." + }, + { + "left": "issue:41669", + "right": "issue:45341", + "accept": false, + "reason": "Model import cleanup and a testing_utils bug are unrelated." + }, + { + "left": "issue:43976", + "right": "issue:45850", + "accept": false, + "reason": "Python compatibility issues are not the same as the broad 'many models break' complaint." + }, + { + "left": "issue:44393", + "right": "issue:45478", + "accept": false, + "reason": "Both mention Qwen3.5/Qwen3-VL, but one is bbox hallucination and the other is from_pretrained failure; different defects." + }, + { + "left": "issue:43976", + "right": "issue:46218", + "accept": false, + "reason": "Python version compatibility and missing Qwen3_5 module import are unrelated." + }, + { + "left": "issue:44077", + "right": "issue:44861", + "accept": false, + "reason": "Optional post_init policy and an AttributeError in tied-weight key handling are different bugs." + }, + { + "left": "issue:31795", + "right": "issue:33453", + "accept": false, + "reason": "Documentation confusion in model.forward versus tokenizer loading regression are unrelated." + }, + { + "left": "issue:43576", + "right": "issue:44273", + "accept": false, + "reason": "Broken env CLI and lazy-loading issues are too different to merge as duplicates." + }, + { + "left": "issue:44315", + "right": "issue:44829", + "accept": false, + "reason": "Liger Kernel not being applied in model_init and flash_attention_3 training degeneration are different failures." + }, + { + "left": "issue:44360", + "right": "issue:45468", + "accept": false, + "reason": "DSA indexer activation discussion and Gemma-4 audio positional encoding are unrelated." + }, + { + "left": "issue:43519", + "right": "issue:46655", + "accept": false, + "reason": "Incorrect timestamp calculation and incorrect padding calculation are both processor bugs but on different modalities and code paths." + }, + { + "left": "issue:44273", + "right": "issue:44908", + "accept": false, + "reason": "Lazy-loading failure and scheduler kwarg handling are unrelated." + }, + { + "left": "issue:44655", + "right": "issue:44861", + "accept": false, + "reason": "Pipeline save_pretrained failure and tied-weight key AttributeError are different bugs." + }, + { + "left": "issue:45230", + "right": "issue:45310", + "accept": false, + "reason": "Generic bug report versus a specific Qwen3.5 MoE from_pretrained regression." + }, + { + "left": "issue:43519", + "right": "issue:44485", + "accept": false, + "reason": "Qwen3VL timestamp handling and GLM-5 RoPE implementation are not the same bug." + }, + { + "left": "issue:45468", + "right": "issue:46894", + "accept": false, + "reason": "Gemma-4 audio positional encoding and CohereASR double-shift training loss are different model-specific problems." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 1, + "component_count": 23, + "review_unit_index": 60, + "review_unit_count": 60, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-10000-500", + "nodes": [ + "issue:28348", + "issue:29870", + "issue:30333", + "issue:31515", + "issue:31795", + "issue:36296", + "issue:37934", + "issue:38617", + "issue:42371", + "issue:42913", + "issue:42936", + "issue:43066", + "issue:43240", + "issue:43298", + "issue:43352", + "issue:43502", + "issue:43519", + "issue:43525", + "issue:43572", + "issue:43618", + "issue:43668", + "issue:43704", + "issue:43717", + "issue:43813", + "issue:44031", + "issue:44230", + "issue:44261", + "issue:44263", + "issue:44351", + "issue:44485", + "issue:44492", + "issue:44623", + "issue:44861", + "issue:44908", + "issue:44938", + "issue:44998", + "issue:45250", + "issue:45290", + "issue:45310", + "issue:45468", + "issue:45850", + "issue:46018", + "issue:46095", + "issue:46144", + "issue:46172", + "issue:46218" + ], + "soft_pairs": [ + "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:44261|issue:45468", + "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:28348|issue:43352", + "issue:44230|issue:44623", + "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": "936027c0d61717483ac8f2016545a69b310a858626fc055eb658f579c2d94da9", + "component_budget": { + "node_count": 500, + "item_count": 500, + "soft_pair_count": 1839, + "serialized_chars": 953879, + "estimated_input_tokens": 238470, + "estimated_eval_tokens": 477196 + }, + "budget": { + "node_count": 46, + "item_count": 46, + "soft_pair_count": 44, + "serialized_chars": 38878, + "estimated_input_tokens": 9720, + "estimated_eval_tokens": 19696 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "The cluster is largely heterogeneous. A recurring theme is Flash Attention 2.0 / version-compatibility regressions, but the items still point to different models or different failure modes, so I would not merge any soft pair as a duplicate.", + "confidence": 0.61, + "canonical_issue_reason": "issue:45250 is the broadest and most representative artifact in the set, since several nearby items relate to Flash Attention 2.0 support or related upgrade breakage. Even so, the cluster is not cohesive enough to treat as a duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:45250 is the best single issue to keep as the cluster representative because it captures the only clear recurring theme and is broader than the model-specific or symptom-specific reports.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46018", + "right": "issue:46172", + "accept": false, + "reason": "Same model family, but different bugs: expert gating logic vs a tensor transpose/cast issue." + }, + { + "left": "issue:43704", + "right": "issue:44485", + "accept": false, + "reason": "Different models and different code paths; VRAM leakage is unrelated to GLM-5 RoPE behavior." + }, + { + "left": "issue:44485", + "right": "issue:45468", + "accept": false, + "reason": "Both are model-specific architecture issues, but they concern different components and symptoms." + }, + { + "left": "issue:43502", + "right": "issue:43519", + "accept": false, + "reason": "One is a local_files_only network request bug; the other is a Qwen3VL timestamp calculation issue." + }, + { + "left": "issue:42371", + "right": "issue:43704", + "accept": false, + "reason": "TF32 API/deprecation guidance is unrelated to a multi-thread VRAM leak." + }, + { + "left": "issue:45850", + "right": "issue:46218", + "accept": false, + "reason": "Both are version-related regressions, but one is broad model breakage and the other is a specific import failure." + }, + { + "left": "issue:43519", + "right": "issue:44623", + "accept": false, + "reason": "Both involve processors, but the failures are in different methods and unrelated behaviors." + }, + { + "left": "issue:44263", + "right": "issue:45468", + "accept": false, + "reason": "Different model internals and different bug classes; no shared code-path evidence." + }, + { + "left": "issue:44261", + "right": "issue:45468", + "accept": false, + "reason": "Different model-specific numerical/configuration issue versus Gemma-4 audio encoding behavior." + }, + { + "left": "issue:43298", + "right": "issue:46095", + "accept": false, + "reason": "Audio continuous batching support and DeepSeek-V4 weight conversion are unrelated problems." + }, + { + "left": "issue:43519", + "right": "issue:43704", + "accept": false, + "reason": "Timestamp computation in a processor is unrelated to VRAM leakage across dataloader threads." + }, + { + "left": "issue:44351", + "right": "issue:46218", + "accept": false, + "reason": "Both are import errors, but they reference different missing symbols/modules and likely different fixes." + }, + { + "left": "issue:43352", + "right": "issue:46218", + "accept": false, + "reason": "Flash Attention support and an import failure for Qwen3_5ForCausalLM are not the same underlying issue." + }, + { + "left": "issue:42913", + "right": "issue:43066", + "accept": false, + "reason": "Both are tokenizer v5-related, but one is a behavior regression and the other is a wrong decoder type bug." + }, + { + "left": "issue:28348", + "right": "issue:45250", + "accept": false, + "reason": "Related on Flash Attention 2.0, but 28348 is GPT2LMHeadModel-specific while 45250 is only a generic request." + }, + { + "left": "issue:44908", + "right": "issue:44998", + "accept": false, + "reason": "A scheduler kwargs bug is unrelated to an unrelated miscellaneous issue." + }, + { + "left": "issue:44492", + "right": "issue:44998", + "accept": false, + "reason": "Cache-strategy typo and the unrelated issue title do not match any underlying bug." + }, + { + "left": "issue:43618", + "right": "issue:44998", + "accept": false, + "reason": "CLIPOutput attentions regression is unrelated to the other issue." + }, + { + "left": "issue:44998", + "right": "issue:45850", + "accept": false, + "reason": "No shared bug or code path; one is a placeholder/misc title and the other is a broad compatibility complaint." + }, + { + "left": "issue:44938", + "right": "issue:44998", + "accept": false, + "reason": "Python 3.14 import failure is unrelated to the other issue." + }, + { + "left": "issue:44861", + "right": "issue:45290", + "accept": false, + "reason": "Different crash sites: tied-weights handling versus chat template/tool-call serialization." + }, + { + "left": "issue:31515", + "right": "issue:38617", + "accept": false, + "reason": "Checkpoint loading performance and a missing import are unrelated." + }, + { + "left": "issue:43572", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_idx after v5 update is unrelated to list handling in tied weights." + }, + { + "left": "issue:43525", + "right": "issue:44861", + "accept": false, + "reason": "Missing pad_token_id on Llama4Config is a different config-attribute bug." + }, + { + "left": "issue:28348", + "right": "issue:43352", + "accept": false, + "reason": "Both mention Flash Attention 2.0, but they target different models and likely different support paths." + }, + { + "left": "issue:44230", + "right": "issue:44623", + "accept": false, + "reason": "FP8 inference support and missing processor.save_pretrained files are unrelated." + }, + { + "left": "issue:29870", + "right": "issue:31515", + "accept": false, + "reason": "BART generate() UnboundLocalError is unrelated to slow checkpoint loading." + }, + { + "left": "issue:30333", + "right": "issue:31515", + "accept": false, + "reason": "MLflow reporting on failed jobs is unrelated to from_pretrained performance." + }, + { + "left": "issue:31515", + "right": "issue:36296", + "accept": false, + "reason": "Checkpoint load slowness and tensor parallel training bugs are different problems." + }, + { + "left": "issue:31515", + "right": "issue:31795", + "accept": false, + "reason": "Performance on loading checkpoints is unrelated to documentation confusion in forward()." + }, + { + "left": "issue:44230", + "right": "issue:45310", + "accept": false, + "reason": "Both touch Qwen3.5 MoE, but one is fp8 support and the other is a from_pretrained error; not the same fix." + }, + { + "left": "issue:43298", + "right": "issue:44230", + "accept": false, + "reason": "Audio continuous batching and Qwen3-VL/Qwen3.5 FP8 inference support are different features." + }, + { + "left": "issue:43717", + "right": "issue:46144", + "accept": false, + "reason": "Mamba init_weights behavior and RoFormer attention interface usage are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:46018", + "accept": false, + "reason": "Mask2Former ignore_value handling is unrelated to DeepSeek-V4 gating." + }, + { + "left": "issue:43668", + "right": "issue:44031", + "accept": false, + "reason": "A config type hint issue and a tokenizer regex warning are not the same bug." + }, + { + "left": "issue:42936", + "right": "issue:43240", + "accept": false, + "reason": "Different subsystems and different symptoms; no shared code path." + }, + { + "left": "issue:43240", + "right": "issue:46144", + "accept": false, + "reason": "Loss kwargs propagation is unrelated to RoFormer attention implementation." + }, + { + "left": "issue:42936", + "right": "issue:44031", + "accept": false, + "reason": "Mask2Former ignore_value behavior and tokenizer regex warnings are unrelated." + }, + { + "left": "issue:42936", + "right": "issue:46144", + "accept": false, + "reason": "Different model families and different bug types." + }, + { + "left": "issue:42936", + "right": "issue:43813", + "accept": false, + "reason": "A Mask2Former bug and a typo in quantization conversion naming are unrelated." + }, + { + "left": "issue:43240", + "right": "issue:43668", + "accept": false, + "reason": "Loss-function kwargs and a ModernBERT type-hint problem are unrelated." + }, + { + "left": "issue:43668", + "right": "issue:43813", + "accept": false, + "reason": "Type-hint correctness and quantization-operation typo are different issues." + }, + { + "left": "issue:37934", + "right": "issue:43668", + "accept": false, + "reason": "Llama4 normalization semantics and ModernBERT type hints do not describe the same change." + }, + { + "left": "issue:42936", + "right": "issue:43668", + "accept": false, + "reason": "Mask2Former ignore_value handling is unrelated to ModernBERT config typing." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 2, + "component_count": 23, + "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": "eb7c71b70f2b1831364dfe97eb6da3ef168ca8e810465b610ee3df3f24f280cc", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 2875, + "estimated_input_tokens": 719, + "estimated_eval_tokens": 1694 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is not a duplicate set: one issue is about tokenizer truncation direction, one about ReduceLROnPlateau-style scheduling, and one about Wav2Vec2 word time offsets. They only share vague ML/NLP similarity, not the same underlying problem.", + "confidence": 0.97, + "canonical_issue_reason": "No single canonical issue: the three issues describe unrelated feature requests in different parts of the library.", + "canonical_pr_reason": null, + "best_issue_reason": "If one issue must anchor the cluster, #11307 is the strongest representative by discussion activity, but it is not a duplicate of the others.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:10519", + "right": "issue:11307", + "accept": false, + "reason": "Tokenizer truncation from the beginning and Wav2Vec2 word offset extraction are unrelated feature requests affecting different code paths." + }, + { + "left": "issue:10519", + "right": "issue:11005", + "accept": false, + "reason": "One request is about truncation behavior in tokenization; the other is about learning-rate scheduler behavior. No shared underlying bug." + }, + { + "left": "issue:11005", + "right": "issue:11307", + "accept": false, + "reason": "ReduceLROnPlateau-like functionality and Wav2Vec2 token timing offsets are separate feature requests with no concrete overlap." + } + ] + }, + "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": 23, + "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": "362a78a1970b535576e8eb50b88b506d0932ca74d5a1c3a88c4d9256231f0264", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 2477, + "estimated_input_tokens": 620, + "estimated_eval_tokens": 1496 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These three closed issues are not duplicates of one another: one is about TAPAS tokenization mismatching TensorFlow, one is about a misleading auth/download error for private models, and one is about failing to load bert-base-uncased. The soft similarities are superficial and do not indicate the same underlying bug.", + "confidence": 0.89, + "canonical_issue_reason": "Issue #16618 is the most representative user-facing failure in the set, but it is still distinct from the other two issues.", + "canonical_pr_reason": null, + "best_issue_reason": "#16618 is the clearest, most concrete model-loading problem and the best single representative if one issue must be chosen.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:13244", + "right": "issue:16618", + "accept": false, + "reason": "Different problem areas: TAPAS tokenization parity vs generic model loading for bert-base-uncased. No shared code-path or same bug." + }, + { + "left": "issue:15980", + "right": "issue:16618", + "accept": false, + "reason": "Both mention loading models, but #15980 is specifically an authentication/error-message issue for private models, while #16618 is a separate load failure for a public model." + } + ] + }, + "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": 23, + "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": "87ddbecc582afcac93f2ffe8cedb1f50ab30f59170e1cb4bd01ccb54c1cc1122", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1590, + "estimated_input_tokens": 398, + "estimated_eval_tokens": 1052 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are both community-contribution feature requests about expanding architecture support, but they target different technologies: Flash Attention 2 vs GGUF. They are related at a high level only and should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": "Issue #26350 is the stronger representative because it is older and has substantially higher discussion activity, indicating broader engagement around that request.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #26350 is the best single issue to represent the cluster because it is the more prominent and more established feature request, even though the two issues are not duplicates.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:26350", + "right": "issue:33260", + "accept": false, + "reason": "Both are feature requests for broader architecture support, but they concern different concrete additions and code paths: Flash Attention 2 support versus GGUF support. 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": "issue", + "component_index": 5, + "component_count": 23, + "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": "4525b0bb5a4776a6532f9c825c997c3ca58e7d1d76dfa846d54f8201badd0dce", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13530, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 24, + "serialized_chars": 13530, + "estimated_input_tokens": 3383, + "estimated_eval_tokens": 7022 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly i18n docs-translation requests, but they split into separate language-specific tasks. Only the exact duplicate placeholder-title issues and the identical Thai-titled pair look like true duplicates; the Turkish/Spanish/Romanian requests are distinct issues, not the same bug/change.", + "confidence": 0.77, + "canonical_issue_reason": "Use issue 43834 as the canonical representative of the exact-duplicate placeholder-template subgroup: it matches the other placeholder-title issues closely and is the cleanest anchor among the true duplicates.", + "canonical_pr_reason": null, + "best_issue_reason": "43834 is the best issue to anchor deduplication for the placeholder-template copies, but it should not be treated as canonical for the language-specific translation requests.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:27088", + "right": "issue:44370", + "accept": false, + "reason": "Different target languages; these are separate translation requests, not the same underlying issue." + }, + { + "left": "issue:27088", + "right": "issue:44365", + "accept": false, + "reason": "Turkish vs. generic placeholder issue; not the same concrete task." + }, + { + "left": "issue:27088", + "right": "issue:43834", + "accept": false, + "reason": "Specific Turkish translation request does not duplicate the generic placeholder-template issue." + }, + { + "left": "issue:27088", + "right": "issue:38435", + "accept": false, + "reason": "Different language translation work; distinct issues." + }, + { + "left": "issue:27088", + "right": "issue:28936", + "accept": false, + "reason": "Turkish and Spanish translation requests are separate issues." + }, + { + "left": "issue:28936", + "right": "issue:43834", + "accept": false, + "reason": "Spanish translation request is not the same as the generic placeholder issue." + }, + { + "left": "issue:28936", + "right": "issue:44365", + "accept": false, + "reason": "Different language target and different issue intent; not duplicates." + }, + { + "left": "issue:28936", + "right": "issue:38435", + "accept": false, + "reason": "Spanish vs. Romanian translation requests are distinct tasks." + }, + { + "left": "issue:28936", + "right": "issue:44370", + "accept": false, + "reason": "Spanish translation request does not match the placeholder-template issue." + }, + { + "left": "issue:38435", + "right": "issue:44370", + "accept": false, + "reason": "Romanian translation request and placeholder-template copy are not the same underlying issue." + }, + { + "left": "issue:38435", + "right": "issue:44365", + "accept": false, + "reason": "Romanian request is distinct from the generic placeholder issue." + }, + { + "left": "issue:38435", + "right": "issue:43834", + "accept": false, + "reason": "Concrete Romanian translation issue is not a duplicate of the placeholder-template issue." + }, + { + "left": "issue:38435", + "right": "issue:46222", + "accept": false, + "reason": "Different translation targets and different issue content; not duplicates." + }, + { + "left": "issue:38435", + "right": "issue:46226", + "accept": false, + "reason": "Romanian translation request is unrelated to the Thai-titled issue pair." + }, + { + "left": "issue:43834", + "right": "issue:44370", + "accept": true, + "reason": "Same placeholder-title translation issue; appears to be an exact duplicate." + }, + { + "left": "issue:43834", + "right": "issue:44365", + "accept": true, + "reason": "Identical placeholder-template translation issue with no substantive difference." + }, + { + "left": "issue:43834", + "right": "issue:46222", + "accept": false, + "reason": "Only superficially similar template text; different title/content, so not the same issue." + }, + { + "left": "issue:43834", + "right": "issue:46226", + "accept": false, + "reason": "Different title/content from the placeholder-template issue; reject as duplicate." + }, + { + "left": "issue:44365", + "right": "issue:44370", + "accept": true, + "reason": "Same exact placeholder-title issue; these look like duplicate copies." + }, + { + "left": "issue:44365", + "right": "issue:46222", + "accept": false, + "reason": "Different issue text and title; not the same underlying request." + }, + { + "left": "issue:44365", + "right": "issue:46226", + "accept": false, + "reason": "Not the same issue; only the placeholder-template subgroup matches." + }, + { + "left": "issue:44370", + "right": "issue:46222", + "accept": false, + "reason": "Different title/content and different underlying translation request." + }, + { + "left": "issue:44370", + "right": "issue:46226", + "accept": false, + "reason": "Not a duplicate; the Thai-titled issue is a separate request." + }, + { + "left": "issue:46222", + "right": "issue:46226", + "accept": true, + "reason": "Exact same Thai title and likely same body; clear duplicate pair." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The assessment is grounded in the packet and stays conservative. The exact-duplicate conclusions are limited to the identical placeholder-title subgroup and the identical Thai-titled pair, while the Turkish/Spanish/Romanian issues are appropriately treated as separate translation requests. No soft-edge verdict appears to overstate the evidence." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 6, + "component_count": 23, + "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-36718-19", + "nodes": [ + "issue:36718", + "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:36718|issue:41115", + "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: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": "745a0fae795c8824cbe0959fc0c0b0fa62d3f68d5db81025b9a9a4960476a025", + "component_budget": { + "node_count": 19, + "item_count": 19, + "soft_pair_count": 40, + "serialized_chars": 24673, + "estimated_input_tokens": 6169, + "estimated_eval_tokens": 12594 + }, + "budget": { + "node_count": 19, + "item_count": 19, + "soft_pair_count": 40, + "serialized_chars": 24673, + "estimated_input_tokens": 6169, + "estimated_eval_tokens": 12594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is a loose topical grouping of model-addition/support requests, not a set of true duplicates. The soft-similarity links are too broad to merge: the issues span different model families and tasks (BERT, DETR, audio ASR/TTS, OCR, vision-language, sparse transformers).", + "confidence": 0.88, + "canonical_issue_reason": "issue:43837 is the most central representative of the cluster: a concrete model-support request with the broadest topical overlap to the other model-addition issues, especially the audio/model-support requests.", + "canonical_pr_reason": null, + "best_issue_reason": "issue:43837 is the best issue-level representative for this cluster because it is specific, current, and sits near the densest area of soft similarity without collapsing distinct requests into one duplicate.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:36718", + "right": "issue:41115", + "accept": false, + "reason": "Different concrete requests in different domains: keypoint matching vs a model architecture addition." + }, + { + "left": "issue:36817", + "right": "issue:41211", + "accept": false, + "reason": "Both are model requests, but for different architectures and code paths; not the same change." + }, + { + "left": "issue:36817", + "right": "issue:43671", + "accept": false, + "reason": "EuroBert config support and Qwen3-TTS support are unrelated model additions." + }, + { + "left": "issue:36817", + "right": "issue:36879", + "accept": false, + "reason": "Different model families (EuroBert vs RF-DETR); no shared underlying fix." + }, + { + "left": "issue:36879", + "right": "issue:41211", + "accept": false, + "reason": "RF-DETR and DEIMv2 are distinct model integrations, not one duplicate request." + }, + { + "left": "issue:36879", + "right": "issue:43909", + "accept": false, + "reason": "Different models and modalities; separate implementation work." + }, + { + "left": "issue:36879", + "right": "issue:44121", + "accept": false, + "reason": "RF-DETR model support is unrelated to a sparse-transformer model request." + }, + { + "left": "issue:36879", + "right": "issue:43837", + "accept": false, + "reason": "Different models and tasks (detection vs ASR); not mergeable as one PR/issue." + }, + { + "left": "issue:36879", + "right": "issue:43671", + "accept": false, + "reason": "RF-DETR and Qwen3-TTS are unrelated model requests." + }, + { + "left": "issue:37015", + "right": "issue:43837", + "accept": false, + "reason": "NeoBERT and Qwen3-ASR are distinct model-support requests." + }, + { + "left": "issue:37015", + "right": "issue:44121", + "accept": false, + "reason": "Different model families with no shared concrete code-path problem." + }, + { + "left": "issue:37015", + "right": "issue:43909", + "accept": false, + "reason": "NeoBERT and LFM2.5 Audio are unrelated model additions." + }, + { + "left": "issue:36879", + "right": "issue:37015", + "accept": false, + "reason": "RF-DETR and NeoBERT are separate model integrations." + }, + { + "left": "issue:37015", + "right": "issue:42503", + "accept": false, + "reason": "NeoBERT and ModernVBERT are different model requests, even if both are BERT-like." + }, + { + "left": "issue:41115", + "right": "issue:43837", + "accept": false, + "reason": "MiniCPM3 architecture work and Qwen3-ASR support are different models and tasks." + }, + { + "left": "issue:41115", + "right": "issue:44121", + "accept": false, + "reason": "MiniCPM3 architecture and sparse-transformer support are unrelated changes." + }, + { + "left": "issue:41115", + "right": "issue:42503", + "accept": false, + "reason": "Different model families; not the same underlying addition." + }, + { + "left": "issue:41115", + "right": "issue:45027", + "accept": false, + "reason": "Model architecture work and Voxtral TTS support are unrelated." + }, + { + "left": "issue:41211", + "right": "issue:43671", + "accept": false, + "reason": "DEIMv2 and Qwen3-TTS are different model requests, not one duplicate." + }, + { + "left": "issue:42503", + "right": "issue:43837", + "accept": false, + "reason": "ModernVBERT and Qwen3-ASR are distinct model-support requests." + }, + { + "left": "issue:42503", + "right": "issue:43671", + "accept": false, + "reason": "ModernVBERT and Qwen3-TTS are unrelated model additions." + }, + { + "left": "issue:42503", + "right": "issue:42738", + "accept": false, + "reason": "ModernVBERT models and BERT-like models with RoPE are related only at a broad theme level, not the same concrete change." + }, + { + "left": "issue:42738", + "right": "issue:45027", + "accept": false, + "reason": "BERT-like RoPE support and Voxtral TTS support are different tasks and code paths." + }, + { + "left": "issue:42738", + "right": "issue:46653", + "accept": false, + "reason": "The named request and the person-only issue title are clearly unrelated." + }, + { + "left": "issue:43671", + "right": "issue:43837", + "accept": false, + "reason": "Related product line, but TTS vs ASR are distinct model-support requests with different implementations." + }, + { + "left": "issue:43671", + "right": "issue:45027", + "accept": false, + "reason": "Qwen3-TTS and Voxtral TTS are different models; same modality is not enough to merge." + }, + { + "left": "issue:43671", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-TTS support and DeepSeek-OCR2 are unrelated model additions." + }, + { + "left": "issue:43671", + "right": "issue:44121", + "accept": false, + "reason": "Audio TTS support and sparse-transformer support are unrelated changes." + }, + { + "left": "issue:43837", + "right": "issue:43909", + "accept": false, + "reason": "Both audio-related, but ASR and audio model support are different concrete requests." + }, + { + "left": "issue:43837", + "right": "issue:45322", + "accept": false, + "reason": "Qwen3-ASR and EUPE are different models and different code paths." + }, + { + "left": "issue:43837", + "right": "issue:44121", + "accept": false, + "reason": "ASR support and sparse-transformer support are unrelated." + }, + { + "left": "issue:43909", + "right": "issue:44205", + "accept": false, + "reason": "LFM2.5 Audio and SAM3-LiteText are different model families/tasks." + }, + { + "left": "issue:43909", + "right": "issue:46645", + "accept": false, + "reason": "Audio model support and a Granite SWA+sink option are not the same change." + }, + { + "left": "issue:44121", + "right": "issue:45322", + "accept": false, + "reason": "Sparse-transformer support and EUPE are distinct model requests." + }, + { + "left": "issue:43837", + "right": "issue:45027", + "accept": false, + "reason": "Different TTS models; similar modality does not make them duplicates." + }, + { + "left": "issue:37015", + "right": "issue:45027", + "accept": false, + "reason": "NeoBERT and Voxtral TTS are unrelated model additions." + }, + { + "left": "issue:43837", + "right": "issue:45177", + "accept": false, + "reason": "Qwen3-ASR and DeepSeek-OCR2 are different models and tasks." + }, + { + "left": "issue:45322", + "right": "issue:46133", + "accept": false, + "reason": "EUPE and TIPSv2 are both vision-language related, but they are distinct model requests." + }, + { + "left": "issue:42503", + "right": "issue:46653", + "accept": false, + "reason": "ModernVBERT and the person-only issue title are clearly unrelated." + }, + { + "left": "issue:45027", + "right": "issue:46653", + "accept": false, + "reason": "Voxtral TTS support and the person-only issue title are unrelated." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 7, + "component_count": 23, + "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": "cf2c65dadb356864ff08fc2d3e4998b4f0fd5aa75e99e07dfee6b32ed7a6f393", + "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": "These issues are unrelated: one is a docs request about adding `pruna` as an optimization framework, while the other is a Vision RoPE refactor. No duplicate relationship is supported.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46443 is the more substantive technical item and is the better representative issue; issue 38740 is a standalone documentation request and not part of the same bug/change.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:38740", + "right": "issue:46443", + "accept": false, + "reason": "Different subjects and scopes: documentation for `pruna` vs a Vision RoPE refactor. No shared files, targets, or concrete code-path overlap." + } + ] + }, + "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": 23, + "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": "c67c095a5a7911926c7488eb42e352a34c375b0dbe1d35bce9675fdf420324c4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1480, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are unrelated: one is about block decoding configuration, the other is about adding DeepSeek v3.2 model support. They do not describe the same underlying bug or feature request.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41084", + "right": "issue:43037", + "accept": false, + "reason": "Different problems and likely different code areas; no shared target, filename, or concrete symptom indicating a duplicate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 9, + "component_count": 23, + "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": "a49ccbd33ff88f5b7b1704092a6a7922040fc475d59099f5ca363fdfbbfe55e8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1585, + "estimated_input_tokens": 397, + "estimated_eval_tokens": 1050 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two issues are about different model components and different requested changes: one is about exposing a TiledFusedLogitsLoss/FLCE flag in `from_pretrained()`, the other is about integrating TiledMLP for lower memory use. They do not look like the same underlying bug or feature request.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No canonical issue in this pair; if forced to pick one, #41306 is more concrete and actionable, but it is not a duplicate of #41826.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:41306", + "right": "issue:41826", + "accept": false, + "reason": "Different requested functionality and code paths: FLCE/from_pretrained configuration versus TiledMLP memory-footprint integration. No strong evidence they describe the same 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": 23, + "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": "facbe6c96bc6c51317ca93f3842260167c1293019191aefd0f3fb11e14e1ae4f", + "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 are both Trainer-related, but they describe different problems: one is about `remove_unused_column` hurting customizability, the other is about FLOPS estimation for custom trainers. They do not look like the same underlying bug.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No single canonical issue stands out; if forced to pick a representative, #43598 is the broader Trainer customizability topic, but it does not cover #43600.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43598", + "right": "issue:43600", + "accept": false, + "reason": "Different concrete issues: `remove_unused_column` behavior vs FLOPS accounting for custom trainers. Shared Trainer context is too broad to treat as a duplicate." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 11, + "component_count": 23, + "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": "d29f07799638e464074169bebbdc292b22c5f2ff81ceb202126ca096b8e7a83b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1612, + "estimated_input_tokens": 403, + "estimated_eval_tokens": 1062 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are two different feature requests for different Qwen model families and different auto-class behaviors, so they should not be deduplicated.", + "confidence": 0.98, + "canonical_issue_reason": "Issue 43684 is the broader and more central request: it asks for model registration in AutoModel and AutoModelForConditionalGeneration for Qwen3-Omni.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 43684 is the better representative issue because it targets core model registration rather than a narrower task-specific auto-class addition.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43684", + "right": "issue:44405", + "accept": false, + "reason": "Different underlying changes: Qwen3-Omni registration for AutoModel/AutoModelForConditionalGeneration versus Qwen3.5 sequence classification support for a different config and auto class." + } + ] + }, + "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": 23, + "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": "f342c57353ea723913f76f5a1080b9fb68fce35083e93bb526fb5b3f2c6239c2", + "component_budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4262, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "budget": { + "node_count": 4, + "item_count": 4, + "soft_pair_count": 6, + "serialized_chars": 4262, + "estimated_input_tokens": 1066, + "estimated_eval_tokens": 2388 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is dominated by three very sparse issues titled \"huggingface \" and one unrelated issue titled \"Fett\". There is not enough concrete description, linkage, or target information to confirm any true duplicate bug, so all soft edges should be rejected.", + "confidence": 0.31, + "canonical_issue_reason": "If one issue must represent the cluster, 43845 is the best placeholder because it is one of the identical \"huggingface \" reports and the earliest of that subgroup, but the evidence is too thin to treat it as a proven duplicate set.", + "canonical_pr_reason": null, + "best_issue_reason": "43845 is the best overall issue representative since it matches the repeated \"huggingface \" pattern and is earliest among the near-identical entries; 43835 appears unrelated by title.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:43835", + "right": "issue:43845", + "accept": false, + "reason": "Different titles and no shared concrete problem details; only weak text similarity." + }, + { + "left": "issue:43835", + "right": "issue:43846", + "accept": false, + "reason": "Different titles and no substantive evidence they describe the same bug." + }, + { + "left": "issue:43835", + "right": "issue:43859", + "accept": false, + "reason": "Different titles with no shared code-path or issue target evidence." + }, + { + "left": "issue:43845", + "right": "issue:43846", + "accept": false, + "reason": "Identical generic title, but the bodies are too sparse to confirm the same underlying issue." + }, + { + "left": "issue:43845", + "right": "issue:43859", + "accept": false, + "reason": "Same vague title, but no concrete details, targets, or links proving duplication." + }, + { + "left": "issue:43846", + "right": "issue:43859", + "accept": false, + "reason": "Same vague title, but insufficient evidence to conclude they are the same issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 13, + "component_count": 23, + "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": "537206ca7e6a754c64abbe20f7e9046cb60cefb461e93a3c1285ededec7aa0d8", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 9, + "serialized_chars": 5794, + "estimated_input_tokens": 1449, + "estimated_eval_tokens": 3154 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster looks like unrelated low-content/spam issues rather than a real duplicate set. The only semi-descriptive item is a documentation suggestion, but there is no evidence that any pair refers to the same underlying bug or request.", + "confidence": 0.16, + "canonical_issue_reason": "Issue 44418 is the least noisy and most descriptive item in the set, so it is the best available representative if one must be chosen.", + "canonical_pr_reason": null, + "best_issue_reason": "No issue here is a strong canonical candidate because the cluster lacks substantive bodies, targets, filenames, or shared problem details; 44418 is only the least-bad representative.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:44371", + "right": "issue:44572", + "accept": false, + "reason": "Both are just '' placeholders with no substantive content; identical titles alone are not enough to prove the same underlying issue." + }, + { + "left": "issue:44371", + "right": "issue:44573", + "accept": false, + "reason": "Both are spam-like placeholders and provide no concrete bug or request details to establish duplication." + }, + { + "left": "issue:44371", + "right": "issue:44418", + "accept": false, + "reason": "One is spam and the other is a generic documentation suggestion; they do not describe the same problem." + }, + { + "left": "issue:44371", + "right": "issue:44937", + "accept": false, + "reason": "A spam placeholder and a random named title do not indicate the same underlying issue." + }, + { + "left": "issue:44418", + "right": "issue:44572", + "accept": false, + "reason": "Generic documentation feedback is unrelated to a '' placeholder." + }, + { + "left": "issue:44418", + "right": "issue:44573", + "accept": false, + "reason": "Generic documentation feedback is unrelated to a '' placeholder." + }, + { + "left": "issue:44572", + "right": "issue:44573", + "accept": false, + "reason": "Although the titles match, there is no body or evidence showing the same concrete bug/request; this looks like duplicate spam, not a confirmed duplicate issue." + }, + { + "left": "issue:44572", + "right": "issue:44937", + "accept": false, + "reason": "A spam placeholder and a named title about 'Google LLC' are not the same underlying issue." + }, + { + "left": "issue:44573", + "right": "issue:44937", + "accept": false, + "reason": "A spam placeholder and a named title about 'Google LLC' 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": 14, + "component_count": 23, + "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": "a6b3bc795eca430edfda6785191100a046569d7af085578aeef65362bf5afe4b", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1646, + "estimated_input_tokens": 412, + "estimated_eval_tokens": 1080 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The pair looks like a strong near-duplicate match: identical titles, very similar body sizes, and no distinguishing links, files, or explicit targets. But the available evidence still doesn\u2019t prove they are the same underlying issue, so this should remain a non-deterministic/low-confidence duplicate candidate rather than a definite merge.", + "confidence": 0.68, + "canonical_issue_reason": "Issue #45145 is the better canonical placeholder because it is slightly earlier and has one inbound reference, making it the more established thread without claiming definitive duplication.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #45145 is the best representative for the cluster because it was created first and has marginally more reference activity; the evidence is otherwise symmetric.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45145", + "right": "issue:45151", + "accept": false, + "reason": "Identical titles and similar body length are suggestive, but there is not enough concrete evidence to conclude these are the same underlying issue. With no links, files, or explicit targets, this should be treated as a possible duplicate rather than accepted as one." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The conclusion is directionally reasonable, but it overstates the evidence. The packet supports identical titles, close body lengths, and no links/files/targets, but not \"identical subject matter\" or that they are definitely duplicates. The soft pair is explicitly non-deterministic, so the accept verdict is too strong for the available evidence." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 15, + "component_count": 23, + "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": 23, + "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": "66d668db9c22dce970cac0f46e4f173f1ba34c4f7eda38e8cd3d64c9e27c0037", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1477, + "estimated_input_tokens": 370, + "estimated_eval_tokens": 996 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: they refer to different people and different dates, with no shared bug report or code change context.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "No single canonical issue here; these are unrelated one-off issues with different titles and no overlapping content.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45762", + "right": "issue:45763", + "accept": false, + "reason": "Different named subjects and dates; no evidence they describe the same underlying problem or request." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 17, + "component_count": 23, + "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": "69c7cc4c5cde221d5be92fd140435795f924b1e326a30abb302226778fc90d27", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1481, + "estimated_input_tokens": 371, + "estimated_eval_tokens": 998 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two closed issues are unrelated personal-name/date entries and do not describe the same bug or request. The soft similarity is superficial only, so the pair should not be merged as duplicates.", + "confidence": 0.98, + "canonical_issue_reason": "No canonical issue: neither item contains actionable problem details, and the titles are different personal-name/date strings rather than duplicate bug reports.", + "canonical_pr_reason": null, + "best_issue_reason": "Neither issue is a good canonical representative for a duplicate cluster because both appear non-technical and unrelated.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45764", + "right": "issue:45765", + "accept": false, + "reason": "Different names and dates with no shared technical subject, target, or problem statement; 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": 18, + "component_count": 23, + "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": "2a19dd98194cfbc824c3fdaf1f95865712819ace28f3822e08a31961d0f50a11", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1542, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1542, + "estimated_input_tokens": 386, + "estimated_eval_tokens": 1028 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are not duplicates: one requests adding a Qwen3_5 token-classification model class for value-model use, while the other asks for FP8 training support. They concern different code paths and different kinds of functionality.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": "issue 45810 is the more concrete and self-contained request, with a specific model-class addition target, so it is the better anchor for this cluster even though it is unrelated to the FP8 training request.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45810", + "right": "issue:46295", + "accept": false, + "reason": "No shared bug or change: model-class support for token classification/value models is unrelated to FP8 training support, and the titles/bodies do not indicate the same underlying work." + } + ] + }, + "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": 23, + "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": "ba3470b9831d79898ba249f4cbe80f78f6b8c5a87e2289c4fed58efdd1761dd7", + "component_budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "budget": { + "node_count": 5, + "item_count": 5, + "soft_pair_count": 8, + "serialized_chars": 5728, + "estimated_input_tokens": 1432, + "estimated_eval_tokens": 3120 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "There are two clear duplicate issue pairs: 45870/45871 and 45872/45873. Issue 45869 is a separate dataset-promo request and does not match the DataForge example/service issues.", + "confidence": 0.96, + "canonical_issue_reason": "Issue 45870 is a clean canonical representative for the exact-duplicate 'Example: Financial time series prediction with DataForge' pair; it is fully redundant with 45871.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 45872 is the strongest standalone issue to anchor because its title is the most specific and it also has an exact duplicate (45873), making it the clearest representative of a distinct topic.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:45869", + "right": "issue:45870", + "accept": false, + "reason": "Different topics: dataset bundle promo vs. a DataForge prediction example." + }, + { + "left": "issue:45869", + "right": "issue:45871", + "accept": false, + "reason": "Different topics: dataset bundle promo vs. a DataForge prediction example." + }, + { + "left": "issue:45870", + "right": "issue:45871", + "accept": true, + "reason": "Exact duplicate titles and matching metadata/body size indicate the same issue." + }, + { + "left": "issue:45870", + "right": "issue:45872", + "accept": false, + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." + }, + { + "left": "issue:45870", + "right": "issue:45873", + "accept": false, + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." + }, + { + "left": "issue:45871", + "right": "issue:45872", + "accept": false, + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." + }, + { + "left": "issue:45871", + "right": "issue:45873", + "accept": false, + "reason": "Related wording, but different underlying request: example notebook vs. DataForge + scraping service." + }, + { + "left": "issue:45872", + "right": "issue:45873", + "accept": true, + "reason": "Exact duplicate titles and matching metadata/body size indicate the same issue." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative. The exact-duplicate calls for 45870/45871 and 45872/45873 are supported by identical titles, identical body lengths, and jaccard 1.0. The rejection of 45869 against the DataForge issues is also appropriately cautious based on the title mismatch and topic difference. No soft-edge verdict appears overstated." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 20, + "component_count": 23, + "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": "4c7033d0843bee84cc965d327fd77442d97a28880dd39a51545637a244f0a5a0", + "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": "Both issues involve static cache behavior around linear-attention/Qwen3.5, but they report different failure modes: one is shape corruption during early initialization, the other is a KeyError for a missing `linear_attention` cache entry. They look related in subsystem, not the same underlying bug.", + "confidence": 0.8, + "canonical_issue_reason": "Issue 46439 is the better representative because it describes the concrete mechanism (`early_initialization`) and the resulting shape corruption, which is a clearer root-cause style report than the model-specific KeyError in 46441.", + "canonical_pr_reason": null, + "best_issue_reason": "46439 is the strongest standalone issue for triage since it is more specific about the failure path and likely applies beyond one model variant.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46439", + "right": "issue:46441", + "accept": false, + "reason": "Related static-cache/linear-attention failures, but the symptoms differ materially: wrong-shape RuntimeError vs `KeyError: 'linear_attention'`. Not enough evidence they are the same bug or that one PR would fix both." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 21, + "component_count": 23, + "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": "9cc0d2e6562aefd3d7d6c86a7d0bebe03224fad474d621606e18b03e5548852c", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1706, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 1706, + "estimated_input_tokens": 427, + "estimated_eval_tokens": 1110 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two issues are very likely duplicates: they have identical titles and identical body length, and both point to the same Wav2Vec2CTCTokenizer v5 special-token regression. The packet does not include the body text, so the duplicate call should be phrased based on metadata rather than claiming the bodies are identical in content. #46552 is the better canonical issue because it is open and already has inbound references.", + "confidence": 0.93, + "canonical_issue_reason": "#46552 is open and has inbound references, so it is the most actionable representative of this duplicate pair.", + "canonical_pr_reason": null, + "best_issue_reason": "#46552 best represents the currently active report for this tokenizer regression.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46551", + "right": "issue:46552", + "accept": true, + "reason": "Same title and identical body length, with matching metadata pointing to the same tokenizer bug; the body text is not available, but these look like duplicate reports of the same issue." + } + ] + }, + "evaluator_result": { + "accept": false, + "feedback": "The pair is well-supported as a duplicate by the identical titles and identical body length, but the summary overstates the evidence by claiming the bodies have the same content. The packet does not include body text, so that part should be stated more conservatively. The canonical choice of #46552 as the open issue with inbound references is grounded." + }, + "evaluator_used": true, + "retried": true, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 22, + "component_count": 23, + "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": "6795f94fe4c483d467eaa2c0e629d8e5c21168695fbe4c95b37d8b65c0d87529", + "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 the same underlying bug: one reports a tensor shape mismatch during object-detection training, while the other reports out-of-memory after several epochs in RT-DETR v2. They overlap only at a broad training/model level.", + "confidence": 0.95, + "canonical_issue_reason": "Issue #46657 is the better canonical issue because it describes a concrete, specific failure mode (tensor size mismatch) rather than a generic training resource problem.", + "canonical_pr_reason": null, + "best_issue_reason": "Issue #46657 is the strongest representative of the cluster since it is more specific and actionable; #46692 appears to be a different symptom class (OOM) and should not be grouped as the same bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46657", + "right": "issue:46692", + "accept": false, + "reason": "Different failure modes: shape mismatch vs. memory exhaustion. They share only the general context of training an object detection model, which is not enough to treat them as the same issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "issue", + "component_index": 23, + "component_count": 23, + "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": "eae3edc28a2a7247f14bae452c047970cc69e164be13ca7a83a7a2054399e755", + "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 two issues are only superficially similar: both involve model loading and memory placement, but one is a ZeRO-3/zero.Init partitioning failure on a composite multimodal submodule, while the other is a 4-bit device_map/max_memory dispatch bug sending a model to CPU unnecessarily. They do not look like the same underlying defect.", + "confidence": 0.97, + "canonical_issue_reason": "No canonical issue for the pair: the reports describe different code paths and different failure modes (ZeRO-3 partitioning vs. 4-bit auto device mapping).", + "canonical_pr_reason": null, + "best_issue_reason": "Issue 46822 is the more concrete, higher-signal report with clearer repro/impact and some discussion activity; 46823 appears to be a separate, unrelated loading bug.", + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "issue:46822", + "right": "issue:46823", + "accept": false, + "reason": "Different underlying bugs: 46822 is about ZeRO-3 zero.Init not partitioning a composite submodule, while 46823 is about 4-bit loading with device_map='auto' incorrectly dispatching to CPU despite available memory." + } + ] + }, + "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": 39, + "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": "10000ca40582744b7e4c57678d139fc0250aef0dc0650a5efba4cce6547e359a", + "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": "Both pull requests are about VibeVoice, but they are not the same change: #40546 implements the core VibeVoice integration, while #42781 adds a separate VibeVoice Realtime path plus acoustic tokenizer work. The overlap is largely shared model-family scaffolding, not a single mergeable fix.", + "confidence": 0.86, + "canonical_issue_reason": null, + "canonical_pr_reason": "#40546 is the better canonical PR because it is the foundational VibeVoice implementation and appears to be the broader, more central change in this cluster.", + "best_issue_reason": null, + "best_pr_reason": "#40546 best represents the cluster as the base VibeVoice implementation; #42781 extends into a distinct realtime/tokenizer feature set rather than duplicating the same PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:40546", + "right": "pull_request:42781", + "accept": false, + "reason": "Reject: the PRs overlap on VibeVoice infrastructure, but #42781 adds a separate realtime feature and acoustic tokenizer work, so they are not the same concrete code-path change and would not plausibly collapse 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": 2, + "component_count": 39, + "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": "f987aca1022b2655246f86c7b323cf4a805e2617d7f2ad630db3f8b8a9f0e2ed", + "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 covariate-based forecasting support for TimesFM, while the other adds DINOv3 image-classification support. They touch different models, different code paths, and solve different product changes, so they are not duplicates.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:40755", + "right": "pull_request:41224", + "accept": false, + "reason": "Different features for different models: TimesFM covariate forecasting vs DINOv3 image classification. No shared bug, no shared code path, and they would not plausibly be merged into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 3, + "component_count": 39, + "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": 39, + "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:43995", + "pull_request:44827", + "pull_request:45170", + "pull_request:45213", + "pull_request:45655", + "pull_request:46048", + "pull_request:46267", + "pull_request:46707", + "pull_request:46911" + ], + "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:43823", + "pull_request:43251|pull_request:44827", + "pull_request:43424|pull_request:46911", + "pull_request:43424|pull_request:46707", + "pull_request:43424|pull_request:44827", + "pull_request:43424|pull_request:43823", + "pull_request:43823|pull_request:45655", + "pull_request:43823|pull_request:43995", + "pull_request:43995|pull_request:45213", + "pull_request:43995|pull_request:44827", + "pull_request:44827|pull_request:46911", + "pull_request:44827|pull_request:46707", + "pull_request:44827|pull_request:45170", + "pull_request:44827|pull_request:46267", + "pull_request:45170|pull_request:46707", + "pull_request:44827|pull_request:45213", + "pull_request:43823|pull_request:45213", + "pull_request:44827|pull_request:45655", + "pull_request:46048|pull_request:46707", + "pull_request:44827|pull_request:46048", + "pull_request:43823|pull_request:46707", + "pull_request:43823|pull_request:46911", + "pull_request:43251|pull_request:43254" + ], + "prepared_review_unit_hash": "0142246c909ff26b86df4b14c309735e6a7457400a18d1f8034450a82c45ce0a", + "component_budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 26, + "serialized_chars": 42249, + "estimated_input_tokens": 10563, + "estimated_eval_tokens": 21382 + }, + "budget": { + "node_count": 15, + "item_count": 15, + "soft_pair_count": 26, + "serialized_chars": 42249, + "estimated_input_tokens": 10563, + "estimated_eval_tokens": 21382 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly a set of unrelated PRs across tokenizers, processors, model additions, docs, and infra. The only clear duplicate is the pair of `fixed_cross_entropy` kwargs fixes in `src/transformers/loss/loss_utils.py`; they address the same concrete bug and should be treated as one change, with PR #43251 as the stronger canonical candidate.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43251 is the better canonical PR for the loss-utils fix: it targets the issue explicitly, has the fuller implementation, and shows more review activity than #43254 while fixing the same `fixed_cross_entropy` code path.", + "best_issue_reason": null, + "best_pr_reason": "PR #43251 is the best PR to represent the duplicate pair because it is the more complete and better-reviewed version of the same kwargs-support fix in `fixed_cross_entropy`.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43251", + "right": "pull_request:43254", + "accept": true, + "reason": "Same file, same function, same issue target, and the same concrete fix: pass supported kwargs through to `nn.functional.cross_entropy`. They are plausible duplicates of the exact same change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet and stays conservative overall. The only minor nuance is that PR #43251 does not literally pass arbitrary kwargs through; both PRs add the same supported `weight`/`label_smoothing` cross-entropy arguments, so describing them as the same bugfix is fine. The duplicate verdict on #43251/#43254 is supported and mergeable into one PR for the same fix." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 5, + "component_count": 39, + "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": "2cbea396c86d80ccced39b5ca94e56fe5470d3731b7bbadcc176c28d420de98e", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5934, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5934, + "estimated_input_tokens": 1484, + "estimated_eval_tokens": 3224 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open PRs both modernize GitHub Actions in workflow files, but they update different action families and address different maintenance goals: one is for Node 24 compatibility of core GitHub actions, the other is for newer Docker build/publish action versions. They are related but not the same change.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 42975 is the stronger representative of the cluster because it is broader and more clearly framed around a specific platform-driven compatibility upgrade across many workflows.", + "best_issue_reason": null, + "best_pr_reason": "PR 42975 is the best standalone PR to keep: it covers a wider set of workflow action upgrades and has the clearest single-purpose maintenance objective, whereas PR 42976 is narrower and focused on Docker action version bumps.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42975", + "right": "pull_request:42976", + "accept": false, + "reason": "Reject: they both touch GitHub Actions workflows, but 42975 upgrades core GitHub actions for Node 24 compatibility while 42976 upgrades docker/build-push-action and related workflow steps. Different action families and different concrete fixes, 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": 6, + "component_count": 39, + "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-42978-6", + "nodes": [ + "pull_request:42978", + "pull_request:43448", + "pull_request:43451", + "pull_request:45490", + "pull_request:46562", + "pull_request:46658" + ], + "soft_pairs": [ + "pull_request:42978|pull_request:43448", + "pull_request:42978|pull_request:43451", + "pull_request:42978|pull_request:45490", + "pull_request:43448|pull_request:43451", + "pull_request:43448|pull_request:45490", + "pull_request:43451|pull_request:45490", + "pull_request:45490|pull_request:46562", + "pull_request:46562|pull_request:46658" + ], + "prepared_review_unit_hash": "30158ac5da763df8f35cb0036a8cd36b18d46fd65377a6149205e427932c80aa", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 21192, + "estimated_input_tokens": 5298, + "estimated_eval_tokens": 10852 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 8, + "serialized_chars": 21192, + "estimated_input_tokens": 5298, + "estimated_eval_tokens": 10852 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is not a duplicate set: it contains several unrelated model-addition PRs (ViT NEPA, Molmo, Molmo2, CTSM, Step 3.7 Flash) plus one WIP Step 3.7 Flash scripting PR. They share some autogenerated framework files and docs plumbing, but not the same underlying bug or change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: the items are separate model implementations or support scripts for different models, not alternate fixes for the same code-path.", + "best_issue_reason": null, + "best_pr_reason": "46562 is the most substantive standalone PR in the set, but it still represents a distinct model addition rather than a duplicate of the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:42978", + "right": "pull_request:43448", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo); overlap is only in shared framework/docs files." + }, + { + "left": "pull_request:42978", + "right": "pull_request:43451", + "accept": false, + "reason": "Different model additions (ViT NEPA vs Molmo2); no shared concrete bug or implementation target." + }, + { + "left": "pull_request:42978", + "right": "pull_request:45490", + "accept": false, + "reason": "Different model additions (ViT NEPA vs CTSM); similar repo plumbing only." + }, + { + "left": "pull_request:43448", + "right": "pull_request:43451", + "accept": false, + "reason": "Molmo and Molmo2 are separate model families, not the same change, despite shared auto-registration/docs edits." + }, + { + "left": "pull_request:43448", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo vs CTSM are unrelated model additions; shared auto/model registry files are not enough." + }, + { + "left": "pull_request:43451", + "right": "pull_request:45490", + "accept": false, + "reason": "Molmo2 and CTSM are unrelated additions with only generic framework overlap." + }, + { + "left": "pull_request:45490", + "right": "pull_request:46562", + "accept": false, + "reason": "CTSM and Step 3.7 Flash are different models; overlap is limited to shared transformer registry files." + }, + { + "left": "pull_request:46562", + "right": "pull_request:46658", + "accept": false, + "reason": "Both are Step 3.7 Flash-related, but 46562 is the core model implementation while 46658 is a WIP script/demo bundle for building or converting a mini model; not the same concrete code-path change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 7, + "component_count": 39, + "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:43665", + "pull_request:44815", + "pull_request:46380", + "pull_request:46539", + "pull_request:46870", + "pull_request:46886", + "pull_request:46922" + ], + "soft_pairs": [ + "pull_request:43291|pull_request:43665", + "pull_request:43291|pull_request:44815", + "pull_request:43291|pull_request:46380", + "pull_request:43291|pull_request:46922", + "pull_request:43665|pull_request:46539", + "pull_request:43665|pull_request:46870", + "pull_request:43665|pull_request:44815", + "pull_request:43665|pull_request:46380", + "pull_request:43665|pull_request:46922", + "pull_request:44815|pull_request:46539", + "pull_request:44815|pull_request:46380", + "pull_request:44815|pull_request:46922", + "pull_request:46380|pull_request:46539", + "pull_request:46380|pull_request:46922", + "pull_request:46539|pull_request:46922", + "pull_request:46886|pull_request:46922" + ], + "prepared_review_unit_hash": "27f1b3e790dca36b6a6f0522f5b7b5afe1b14c50fa6ed5fbde5de3b04c34b19f", + "component_budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 16, + "serialized_chars": 21719, + "estimated_input_tokens": 5430, + "estimated_eval_tokens": 11116 + }, + "budget": { + "node_count": 8, + "item_count": 8, + "soft_pair_count": 16, + "serialized_chars": 21719, + "estimated_input_tokens": 5430, + "estimated_eval_tokens": 11116 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The cluster is mostly heterogeneous, with one clear duplicate pair around YOLOS device-mismatch fixes. Most other PRs are unrelated and should not be merged as duplicates.", + "confidence": 0.87, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46922 is the best canonical PR: it addresses the same YOLOS device_map bug as 46886, is tied to the explicit issue target, and contains the more complete fix (the device move plus the related no-split-module adjustment).", + "best_issue_reason": null, + "best_pr_reason": "PR 46922 is the strongest representative of the cluster\u2019s only clear code-fix duplicate because it directly fixes the YOLOS device mismatch and subsumes the smaller patch in 46886.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43291", + "right": "pull_request:43665", + "accept": false, + "reason": "Whisper tokenizer/test cleanup vs. unrelated CLIP/ViT test edits; no shared bug or concrete change." + }, + { + "left": "pull_request:43291", + "right": "pull_request:44815", + "accept": false, + "reason": "Whisper decoding changes are unrelated to the FP8 dequant/loading fix." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46380", + "accept": false, + "reason": "Different areas and problems: Whisper decode logic vs. generation watermarking validation." + }, + { + "left": "pull_request:43291", + "right": "pull_request:46922", + "accept": false, + "reason": "Whisper tokenizer fix and YOLOS device-mismatch fix are unrelated." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46539", + "accept": false, + "reason": "Bogus model tests vs. CI memory/setup changes for Phimoe; not the same underlying issue." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46870", + "accept": false, + "reason": "Model test placeholders are unrelated to ASR pipeline model-type routing." + }, + { + "left": "pull_request:43665", + "right": "pull_request:44815", + "accept": false, + "reason": "Test-only edits for CLIP/ViT do not match the FP8 dequant/loading bug fix." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46380", + "accept": false, + "reason": "No overlap beyond being PRs; one is test noise, the other is generation/watermarking behavior." + }, + { + "left": "pull_request:43665", + "right": "pull_request:46922", + "accept": false, + "reason": "CLIP/ViT test changes are not the same change as the YOLOS fix." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46539", + "accept": false, + "reason": "FP8 dequant/loading changes and Phimoe CI infrastructure are different bugs." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46380", + "accept": false, + "reason": "Different code paths: FP8 quantization/load conversion vs. generation config validation." + }, + { + "left": "pull_request:44815", + "right": "pull_request:46922", + "accept": false, + "reason": "FP8 dequant fix is unrelated to YOLOS device placement." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46539", + "accept": false, + "reason": "Generation watermarking validation and CI memory patching are unrelated changes." + }, + { + "left": "pull_request:46380", + "right": "pull_request:46922", + "accept": false, + "reason": "Watermarking/generation config work does not match the YOLOS device bug." + }, + { + "left": "pull_request:46539", + "right": "pull_request:46922", + "accept": false, + "reason": "Phimoe integration-test CI fixes are unrelated to YOLOS model runtime behavior." + }, + { + "left": "pull_request:46886", + "right": "pull_request:46922", + "accept": true, + "reason": "Both fix the same YOLOS device-mismatch problem in the same file; 46922 is the broader version of the same change." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: most PRs are clearly unrelated, and the only well-supported duplicate pair is 46886/46922 on the same YOLOS device-mismatch bug. The canonical/representative choice is conservative, and the accepted pair looks mergeable as a single bugfix PR." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 8, + "component_count": 39, + "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": "a940ac07c7d8af4269b9fc5fb4a5f1deb60f267f785bf15ee15e1ad51f9fa444", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6348, + "estimated_input_tokens": 1587, + "estimated_eval_tokens": 3430 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 6348, + "estimated_input_tokens": 1587, + "estimated_eval_tokens": 3430 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "All three pull requests are distinct '[don't merge]' experiment/verification changes. They overlap superficially around CI/testing/workflow files, but they do not fix the same concrete code-path problem.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true canonical duplicate exists in this cluster; the PRs change different mechanisms (setup/dependency/test tweaks vs. workflow trigger rewrites), so any single pick would be arbitrary.", + "best_issue_reason": null, + "best_pr_reason": "If one PR must represent the cluster, 43532 is the broadest and most feature-complete CI/test experiment, but it is still not a duplicate of the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43488", + "right": "pull_request:46747", + "accept": false, + "reason": "43488 changes packaging/dependency/test files, while 46747 only switches a workflow trigger for a security gate. Different subsystems and different fixes." + }, + { + "left": "pull_request:43488", + "right": "pull_request:43532", + "accept": false, + "reason": "43488 is a setup/version/test tweak; 43532 rewrites several workflows and testing utilities. They share a test file, but not the same underlying change or bug." + }, + { + "left": "pull_request:43532", + "right": "pull_request:46747", + "accept": false, + "reason": "Both touch CI workflows, but 43532 is about self-comment CI and related test/util changes, whereas 46747 changes the PR security gate trigger. Not the same concrete fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 9, + "component_count": 39, + "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": "c766b58cbf56e15efe0b0513d6577c768d6090f4217a4da356c5421f1a129e63", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6083, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6083, + "estimated_input_tokens": 1521, + "estimated_eval_tokens": 3298 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related but distinct PRs: both add new promptable segmentation pipelines, but they target different tasks, different model integrations, and different concrete code paths.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: 43612 adds Promptable Concept Segmentation, while 43613 adds Promptable Visual Segmentation with different pipeline implementations, model/test coverage, and task docs.", + "best_issue_reason": null, + "best_pr_reason": "43613 is the broader/more substantial PR, covering more model integrations and tests, so it is the better representative of the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43612", + "right": "pull_request:43613", + "accept": false, + "reason": "Different features and code paths: concept segmentation vs visual segmentation. The overlap is mainly shared docs/registration scaffolding, 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": 10, + "component_count": 39, + "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": "cf52ad3d2a73780fac1fd7c4f137303b2ae4242766d67d7888b0aa6bfbcf82bd", + "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": "Two open PRs touch the same trainer file but address different extension points: gradient-accumulation loss scaling vs. signature-column filtering. They are related in theme but not duplicates.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "pull_request:43651 is the better representative of the cluster because it is a focused trainer-behavior change with a clear user-facing effect; 43663 is a separate internal refactor for column filtering.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43651", + "right": "pull_request:43663", + "accept": false, + "reason": "Both modify src/transformers/trainer.py, but they target different code paths and problems: 43651 changes training_step loss scaling for gradient accumulation, while 43663 extracts signature-column filtering for custom trainer overrideability. Shared file 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": 39, + "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": "ab996854e7ea5d75d2ef167ec8b1a3da156abcffa80b8f9ade2e586d818bdcf0", + "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": "Two open PRs target related TypeAdapter NameError problems in the same CLI file, but they appear to fix different runtime causes rather than the same exact bug.", + "confidence": 0.84, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #43656 is the broader and more direct fix for the CLI TypeAdapter NameError, with a larger runtime-annotation cleanup in serve.py and a title that matches the reported bug closely.", + "best_issue_reason": null, + "best_pr_reason": "PR #43656 is the best representative of the cluster because it addresses the underlying annotation-evaluation problem in the CLI more generally, rather than only a narrow optional-dependency import path.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43656", + "right": "pull_request:43842", + "accept": false, + "reason": "Both touch serve.py and mention TypeAdapter NameError, but #43656 fixes postponed-annotation/runtime typing issues broadly, while #43842 specifically guards the TypeAdapter import when pydantic is absent. They do not look like the same concrete bug or a single mergeable change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 12, + "component_count": 39, + "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": 39, + "review_unit_index": 1, + "review_unit_count": 2, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-43996-28", + "nodes": [ + "pull_request:43996", + "pull_request:43997", + "pull_request:44007", + "pull_request:44010", + "pull_request:44013", + "pull_request:44017", + "pull_request:44018", + "pull_request:44019", + "pull_request:44024", + "pull_request:44044", + "pull_request:44056", + "pull_request:44059", + "pull_request:44066", + "pull_request:44068", + "pull_request:44071", + "pull_request:44072", + "pull_request:44073", + "pull_request:44074", + "pull_request:44076", + "pull_request:44085", + "pull_request:44086", + "pull_request:44098", + "pull_request:44101", + "pull_request:44116", + "pull_request:44129", + "pull_request:44154", + "pull_request:44161", + "pull_request:44722" + ], + "soft_pairs": [ + "pull_request:43996|pull_request:44085", + "pull_request:43996|pull_request:44044", + "pull_request:44066|pull_request:44085", + "pull_request:44007|pull_request:44072", + "pull_request:44072|pull_request:44722", + "pull_request:44066|pull_request:44072", + "pull_request:44013|pull_request:44044", + "pull_request:44066|pull_request:44086", + "pull_request:44018|pull_request:44068", + "pull_request:44066|pull_request:44071", + "pull_request:44066|pull_request:44068", + "pull_request:44018|pull_request:44066", + "pull_request:44019|pull_request:44722", + "pull_request:44086|pull_request:44722", + "pull_request:44019|pull_request:44085", + "pull_request:44018|pull_request:44019", + "pull_request:44019|pull_request:44071", + "pull_request:44068|pull_request: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:44059|pull_request:44722", + "pull_request:44018|pull_request:44071", + "pull_request:44017|pull_request:44161", + "pull_request:44017|pull_request:44722", + "pull_request:44017|pull_request:44073", + "pull_request:44068|pull_request:44071", + "pull_request: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": "dc9ff42294aa7ca9265154f7dd5214676279b5d8faa7979756b4b9e44ef2c126", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84506, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 + }, + "budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 72, + "serialized_chars": 76939, + "estimated_input_tokens": 19235, + "estimated_eval_tokens": 38726 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "This cluster is mostly a broad multi-model migration to standardized output tracing/decorators. Most pairs are only thematically similar; the clear duplicate-like edges are the GPT-Neo pair and the GPT-J/CodeGen pair(s).", + "confidence": 0.91, + "canonical_issue_reason": "No issue artifact is present in the cluster; #43979 is the shared umbrella tracking issue referenced by many of these PRs.", + "canonical_pr_reason": "44722 is the best canonical PR: it directly targets the shared refactor and captures the GPT-J/CodeGen output-tracing change in one coherent implementation.", + "best_issue_reason": null, + "best_pr_reason": "44722 is the most representative PR for the family, with an explicit issue link and the broadest concrete implementation of the same refactor.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43996", + "right": "pull_request:44085", + "accept": false, + "reason": "Different model families (CVT/FNet vs GPT-J); same tracking issue is not enough." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44044", + "accept": false, + "reason": "Unrelated model implementations; both are output-tracing refactors, but not the same code path." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44085", + "accept": true, + "reason": "Both hit the same GPT-J output-tracing code path; 44066 is a broader variant that also includes CodeGen." + }, + { + "left": "pull_request:44007", + "right": "pull_request:44072", + "accept": false, + "reason": "Different backbones and code paths (RegNet/ResNet/RT-DETR vs EfficientNet)." + }, + { + "left": "pull_request:44072", + "right": "pull_request:44722", + "accept": false, + "reason": "Different model families and independent refactors; only the umbrella issue is shared." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44072", + "accept": false, + "reason": "GPT-J/CodeGen vs EfficientNet are unrelated concrete changes." + }, + { + "left": "pull_request:44013", + "right": "pull_request:44044", + "accept": false, + "reason": "MobileNetV2 and DeBERTa are separate model-specific refactors, not one duplicate bug." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-J/CodeGen refactor versus MGP-STR output tracing; same theme, different code path." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44068", + "accept": true, + "reason": "Both are GPT-Neo output-tracing refactors on the same model/file family." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44071", + "accept": false, + "reason": "GPT-J/CodeGen and MPT are unrelated model implementations." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44068", + "accept": false, + "reason": "GPT-J/CodeGen versus GPT-Neo; shared pattern only, not the same fix." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44066", + "accept": false, + "reason": "GPT-Neo and GPT-J are different concrete implementations, despite similar decorator migration." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44722", + "accept": false, + "reason": "ResNet output tracing is unrelated to GPT-J/CodeGen output tracing." + }, + { + "left": "pull_request:44086", + "right": "pull_request:44722", + "accept": false, + "reason": "MGP-STR and GPT-J/CodeGen touch different model code paths." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44085", + "accept": false, + "reason": "ResNet and GPT-J are separate refactors in different files." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44019", + "accept": false, + "reason": "GPT-Neo and ResNet are different model families; same refactor style only." + }, + { + "left": "pull_request:44019", + "right": "pull_request:44071", + "accept": false, + "reason": "ResNet and MPT are unrelated concrete code paths." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-Neo is a different model family from GPT-J/CodeGen." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44076", + "accept": false, + "reason": "GPT-Neo and ImageGPT are separate model-specific changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44116", + "accept": false, + "reason": "GPT-Neo and Flaubert are unrelated implementations." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44722", + "accept": false, + "reason": "MPT and GPT-J/CodeGen do not share the same concrete change." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44085", + "accept": false, + "reason": "GPT-Neo and GPT-J are different model code paths." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are unrelated model implementations." + }, + { + "left": "pull_request:44068", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-Neo and MGP-STR are different concrete refactors." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44086", + "accept": false, + "reason": "GPT-2 and MGP-STR are separate model-specific changes." + }, + { + "left": "pull_request:44098", + "right": "pull_request:44101", + "accept": false, + "reason": "ViLT and Flaubert/XLM are unrelated model families." + }, + { + "left": "pull_request:44059", + "right": "pull_request:44722", + "accept": false, + "reason": "GPT-2 and GPT-J/CodeGen are different concrete code paths." + }, + { + "left": "pull_request:44017", + "right": "pull_request:44161", + "accept": false, + "reason": "SegFormer and LongT5 are unrelated model implementations." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44101", + "accept": false, + "reason": "MPNet and Flaubert/XLM are different model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44017", + "accept": false, + "reason": "RegNet and SegFormer are separate model-specific output-tracing changes." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44056", + "accept": false, + "reason": "RegNet and MPNet are unrelated code paths." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44076", + "accept": false, + "reason": "MPT and ImageGPT are distinct model implementations." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44101", + "accept": false, + "reason": "TextNet and Flaubert/XLM are unrelated changes." + }, + { + "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 separate concrete refactors." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44161", + "accept": false, + "reason": "RegNet and LongT5 do not fix the same underlying change." + }, + { + "left": "pull_request:44074", + "right": "pull_request:44161", + "accept": false, + "reason": "TextNet and LongT5 are unrelated model-specific refactors." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44161", + "accept": false, + "reason": "GPT-Neo and LongT5 are different model code paths." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44071", + "accept": false, + "reason": "SqueezeBert and MPT are separate implementations." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44073", + "accept": false, + "reason": "SqueezeBert and VisualBert are different model-specific changes." + }, + { + "left": "pull_request:44018", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-Neo and SpeechT5 are unrelated model families." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44074", + "accept": false, + "reason": "RegNet and TextNet are different concrete output-tracing changes." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44018", + "accept": false, + "reason": "SqueezeBert and GPT-Neo are unrelated code paths." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44129", + "accept": false, + "reason": "RegNet and SpeechT5 are separate model refactors." + }, + { + "left": "pull_request:44071", + "right": "pull_request:44129", + "accept": false, + "reason": "MPT and SpeechT5 are different model implementations." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44161", + "accept": false, + "reason": "CVT/FNet and LongT5 are not the same underlying change." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44017", + "accept": false, + "reason": "CVT/FNet and SegFormer are distinct model-specific refactors." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44044", + "accept": false, + "reason": "SqueezeBert and DeBERTa are separate model implementations." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44074", + "accept": false, + "reason": "DeBERTa and TextNet are unrelated concrete changes." + }, + { + "left": "pull_request:44066", + "right": "pull_request:44722", + "accept": true, + "reason": "Same GPT-J/CodeGen output-tracing refactor family; 44722 is a later, broader version." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44722", + "accept": true, + "reason": "Both center on the same GPT-J output-tracing change; 44722 also includes CodeGen." + }, + { + "left": "pull_request:44085", + "right": "pull_request:44129", + "accept": false, + "reason": "GPT-J output tracing and SpeechT5 output tracing are different code paths." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44007", + "accept": false, + "reason": "They share the same umbrella theme but cover different model backbones and implementations." + }, + { + "left": "pull_request:43996", + "right": "pull_request:44129", + "accept": false, + "reason": "CVT/FNet and SpeechT5 are unrelated model-specific refactors." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44129", + "accept": false, + "reason": "DeBERTa and SpeechT5 are different concrete changes." + }, + { + "left": "pull_request:44044", + "right": "pull_request:44154", + "accept": false, + "reason": "DeBERTa and ViTS are unrelated model implementations." + }, + { + "left": "pull_request:44024", + "right": "pull_request:44076", + "accept": false, + "reason": "FocalNet and ImageGPT are separate model-specific output-tracing changes." + } + ] + }, + "evaluator_result": { + "accept": true, + "feedback": "The summary is grounded in the packet: it correctly identifies the cluster as a broad standardized-output-tracing migration, and the accepted soft edges stay within the same concrete model/file families (GPT-Neo and GPT-J/CodeGen). The rejected edges are conservative and mostly rely on shared theme or umbrella issue alone, which is not enough to merge them into a single bugfix PR." + }, + "evaluator_used": true, + "retried": false, + "accepted_nontrivial_soft_edge": true, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 13, + "component_count": 39, + "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": "9d1c34d1741458e842c2335bfae1c682cb1f1c8f08dba694d36fc66d7cf59d3d", + "component_budget": { + "node_count": 28, + "item_count": 28, + "soft_pair_count": 77, + "serialized_chars": 84506, + "estimated_input_tokens": 21127, + "estimated_eval_tokens": 42510 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 5, + "serialized_chars": 13067, + "estimated_input_tokens": 3267, + "estimated_eval_tokens": 6790 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": true, + "analyst_result": { + "summary": "These PRs share a common theme\u2014migrating model forward passes to standardized output-capture utilities\u2014but they touch different model code paths, so they are related refactors rather than duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No single canonical PR: each change is model-specific (RegNet, SqueezeBert, MPNet, VisualBert, TextNet, SpeechT5). If a representative is needed, PR 44129 is the strongest candidate because it has an explicit issue target and the most complete output-tracing rewrite.", + "best_issue_reason": null, + "best_pr_reason": "PR 44129 is the best representative of the cluster\u2019s theme: it has an explicit issue link, a substantive refactor, and clear standardized output-capture changes, while the others are similar but isolated to different models.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44073", + "right": "pull_request:44074", + "accept": false, + "reason": "Same refactor pattern, but different model implementations and code paths. This is not a single concrete change that could be merged as one PR." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44073", + "accept": false, + "reason": "Both migrate output tracing, but MPNet and VisualBert are distinct model backends with different implementation details. Not the same underlying bug/change." + }, + { + "left": "pull_request:44056", + "right": "pull_request:44074", + "accept": false, + "reason": "Different models and different refactor shapes; they share a maintenance pattern only, not a duplicate fix." + }, + { + "left": "pull_request:43997", + "right": "pull_request:44010", + "accept": false, + "reason": "RegNet and SqueezeBert are unrelated code paths. Similar standardized-output migration, but not the same concrete change." + }, + { + "left": "pull_request:44010", + "right": "pull_request:44129", + "accept": false, + "reason": "Different model families and different output-tracing implementations; 44129 is tied to SpeechT5-specific layer behavior and an explicit issue, so it is not a duplicate of SqueezeBert." + } + ] + }, + "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": 39, + "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": "058e2b674f763390a24a12bcf78aedbed6e0a3ce76adcc0024179cd08046fb70", + "component_budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40177, + "estimated_input_tokens": 10045, + "estimated_eval_tokens": 20346 + }, + "budget": { + "node_count": 13, + "item_count": 13, + "soft_pair_count": 40, + "serialized_chars": 40177, + "estimated_input_tokens": 10045, + "estimated_eval_tokens": 20346 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs share a broad theme of output-tracing refactors, but they target different model implementations and forward paths, so they are not duplicates of one concrete change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 44003 is the strongest representative of the cluster: it is a broader output-tracing refactor spanning two related Mamba model files and introduces the shared capture/output plumbing seen across the group.", + "best_issue_reason": null, + "best_pr_reason": "PR 44003 is the best single anchor for this cluster because it has the broadest scope and the clearest shared tracing plumbing, making it the most representative refactor among the PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:43998", + "right": "pull_request:44002", + "accept": false, + "reason": "Both are output-tracing refactors, but timm_backbone and upernet are different model code paths and not the same underlying fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44000", + "accept": false, + "reason": "Same theme, but timm_backbone and vision_text_dual_encoder are unrelated implementations with different forward behavior." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44001", + "accept": false, + "reason": "Both touch output handling, but timm_backbone and univnet are separate model paths and cannot be merged as one fix." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44003", + "accept": false, + "reason": "These are different model families; the similarity is only in refactoring style, not a shared bug or change." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44029", + "accept": false, + "reason": "timm_backbone and rwkv have distinct forward paths; this is a thematic similarity, not a duplicate change." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44001", + "accept": false, + "reason": "MobileNetV1 and UnivNet are unrelated model code paths, so this is not the same concrete fix." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44002", + "accept": false, + "reason": "Both are refactors, but MobileNetV1 and UperNet have different output behavior and cannot be treated as one PR." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44025", + "accept": false, + "reason": "MobileNetV1 and DepthAnything are different model implementations; only the refactor pattern matches." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44026", + "accept": false, + "reason": "These modify different model wrappers and forward signatures, so they are not duplicates." + }, + { + "left": "pull_request:43999", + "right": "pull_request:44027", + "accept": false, + "reason": "MobileNetV1 and SpeechEncoderDecoder are unrelated code paths; same broad subsystem is not enough." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44001", + "accept": false, + "reason": "VisionTextDualEncoder and UnivNet are separate architectures; the overlap is only the tracing refactor theme." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44002", + "accept": false, + "reason": "These are different model paths with different output contracts, so they are not the same underlying change." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44003", + "accept": false, + "reason": "VisionTextDualEncoder and the Mamba PR target unrelated models; not mergeable as one concrete fix." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44025", + "accept": false, + "reason": "Same style of cleanup, but different model code paths and forward APIs." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44002", + "accept": false, + "reason": "UnivNet and UperNet are unrelated implementations; the shared label is only output-tracing refactor." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44003", + "accept": false, + "reason": "These touch different model families and do not fix the same concrete bug." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44025", + "accept": false, + "reason": "UnivNet and DepthAnything are distinct model paths; not a single duplicate PR." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44029", + "accept": false, + "reason": "UperNet and RWKV are unrelated model code paths, so this is only thematic similarity." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44003", + "accept": false, + "reason": "These are different model implementations with different forward/output behavior." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44028", + "accept": false, + "reason": "Falcon/Mamba and SuperPoint are unrelated code paths; same refactor style, not same change." + }, + { + "left": "pull_request:44003", + "right": "pull_request:44004", + "accept": false, + "reason": "Mamba/Falcon-Mamba and CodeGen are different model implementations, so they are not duplicates." + }, + { + "left": "pull_request:43998", + "right": "pull_request:44004", + "accept": false, + "reason": "timm_backbone and CodeGen are unrelated model code paths; similar tracing refactor only." + }, + { + "left": "pull_request:44001", + "right": "pull_request:44004", + "accept": false, + "reason": "UnivNet and CodeGen are different architectures and not the same concrete fix." + }, + { + "left": "pull_request:44002", + "right": "pull_request:44004", + "accept": false, + "reason": "UperNet and CodeGen are unrelated outputs/refactors, not one mergeable PR." + }, + { + "left": "pull_request:44000", + "right": "pull_request:44004", + "accept": false, + "reason": "VisionTextDualEncoder and CodeGen modify different forward paths and cannot be merged as one fix." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44026", + "accept": false, + "reason": "DepthAnything and VisionEncoderDecoder are different model wrappers; the overlap is only the refactoring pattern." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44027", + "accept": false, + "reason": "DepthAnything and SpeechEncoderDecoder are unrelated model paths, not a duplicate bug or change." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44028", + "accept": false, + "reason": "DepthAnything and SuperPoint target different architectures and output contracts." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44029", + "accept": false, + "reason": "DepthAnything and RWKV are separate model implementations; no shared concrete fix." + }, + { + "left": "pull_request:44025", + "right": "pull_request:44030", + "accept": false, + "reason": "DepthAnything and DPR are different code paths, so this is not the same underlying change." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44029", + "accept": false, + "reason": "VisionEncoderDecoder and RWKV are unrelated models; similar output-tracing work is not enough." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44027", + "accept": false, + "reason": "VisionEncoderDecoder and SpeechEncoderDecoder are different wrapper implementations and not a single fix." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44028", + "accept": false, + "reason": "These are separate model paths with different output handling, so they are not duplicates." + }, + { + "left": "pull_request:44026", + "right": "pull_request:44030", + "accept": false, + "reason": "VisionEncoderDecoder and DPR are unrelated code paths despite both touching forward/output behavior." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44028", + "accept": false, + "reason": "SpeechEncoderDecoder and SuperPoint are different architectures; only the refactor theme matches." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44029", + "accept": false, + "reason": "SpeechEncoderDecoder and RWKV are not the same underlying bug or code change." + }, + { + "left": "pull_request:44027", + "right": "pull_request:44030", + "accept": false, + "reason": "SpeechEncoderDecoder and DPR modify different model-specific output paths." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44029", + "accept": false, + "reason": "SuperPoint and RWKV are unrelated model implementations, so this pair is not a duplicate." + }, + { + "left": "pull_request:44028", + "right": "pull_request:44030", + "accept": false, + "reason": "SuperPoint and DPR are different code paths; the shared theme is too broad." + }, + { + "left": "pull_request:44029", + "right": "pull_request:44030", + "accept": false, + "reason": "RWKV and DPR are distinct model families with different output contracts, not one mergeable 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": 15, + "component_count": 39, + "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": "aa0a71ceba6b29f1eaab3bc7db2491e97acb16b52f6023f3b5d30eba9ee5df91", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6552, + "estimated_input_tokens": 1638, + "estimated_eval_tokens": 3532 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are unrelated model-addition efforts: one adds PenguinVL, the other adds Voxtral TTS. The shared auto-registration/docs files are expected boilerplate and do not indicate the same underlying fix.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "45401 is the better representative of its topic because it is explicitly tied to issue #45027 and has a narrower, well-scoped Voxtral TTS implementation. 44662 is a separate PenguinVL model addition.", + "best_issue_reason": null, + "best_pr_reason": "45401 is the stronger overall PR in this cluster: it is issue-linked, more focused, and easier to reason about as a standalone change. 44662 is broader and belongs to a different model family.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44662", + "right": "pull_request:45401", + "accept": false, + "reason": "Reject: these implement different models (PenguinVL vs Voxtral TTS) and do not appear to fix the same concrete code-path problem. Shared framework/docs files are just model-integration boilerplate." + } + ] + }, + "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": 39, + "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-44965-2", + "nodes": [ + "pull_request:44965", + "pull_request:46845" + ], + "soft_pairs": [ + "pull_request:44965|pull_request:46845" + ], + "prepared_review_unit_hash": "8b211751f9d6a68b9440d8533a199f3821fd06aef9e7594640de4424fe5e1219", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4220, + "estimated_input_tokens": 1055, + "estimated_eval_tokens": 2366 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4220, + "estimated_input_tokens": 1055, + "estimated_eval_tokens": 2366 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two open PRs are not duplicates. PR 44965 rewrites a docs workflow to convert PRs to draft, while PR 46845 changes CI branch routing and includes an unrelated modeling_utils edit. The soft similarity is superficial only.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true duplicate canonical exists here; if one PR must represent the cluster, 46845 is the more substantive change, but it is not the same underlying change as 44965.", + "best_issue_reason": null, + "best_pr_reason": "46845 is the better representative of the set because it contains the more meaningful CI-related change, whereas 44965 is a separate workflow rewrite with a different purpose.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44965", + "right": "pull_request:46845", + "accept": false, + "reason": "Reject: the PRs address different problems and code paths. 44965 changes a workflow to convert PRs to draft, while 46845 updates CI branch handling and touches modeling_utils; they could not plausibly be merged as one 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": 17, + "component_count": 39, + "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": "b9c3586792d8230f19978f6d41d7fe39b61b2a763912b8f4f7be720643e0a3c4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4407, + "estimated_input_tokens": 1102, + "estimated_eval_tokens": 2460 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "The two PRs are related to module/kernel fusion, but they implement different layers of the feature: one adds a general Module Fusion API, while the other wires n-to-1 kernel fusion through KernelConfig and model loading. They are not the same underlying change and should not be deduplicated.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #44979 is the better canonical representative for the cluster because it introduces the core Module Fusion API and test coverage; PR #45363 is a separate integration of KernelConfig-based kernel fusion built on top of that broader idea.", + "best_issue_reason": null, + "best_pr_reason": "PR #44979 is the best single representative because it is the more foundational, self-contained feature PR. PR #45363 is narrower and more implementation-specific, so it is less representative of the overall fusion topic.", + "soft_edge_verdicts": [ + { + "left": "pull_request:44979", + "right": "pull_request:45363", + "accept": false, + "reason": "Different concrete changes: #44979 adds a new module fusion abstraction/API, while #45363 adds KernelConfig-driven n-to-1 fusion plumbing and an example. Related area, but not the same bug or mergeable PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 18, + "component_count": 39, + "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": "f23bce768d1eecdc60303c90ba6042ff7a649c61ec10e87393b098a7ee1b1923", + "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 fix the same Mistral regex patch bug in `tokenization_utils_tokenizers.py` by switching from `tokenizer.backend_tokenizer.pre_tokenizer` to `tokenizer.pre_tokenizer`; PR 45317 is the more complete version because it also adds tests.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45317 is the better canonical PR because it implements the same fix as 45086 but also includes test coverage and a clearer title for the AttributeError path.", + "best_issue_reason": null, + "best_pr_reason": "PR 45317 is the strongest representative of this cluster: same concrete bug fix, plus tests, making it the most merge-ready and informative.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45086", + "right": "pull_request:45317", + "accept": true, + "reason": "Same underlying code-path bug in `_patch_mistral_regex` with the same one-line fix; 45317 is essentially the tested, slightly expanded version of 45086, so they look like duplicate PRs for the same change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 19, + "component_count": 39, + "review_unit_index": 1, + "review_unit_count": 1, + "status": "reviewed", + "reason": null, + "source": "llm", + "cache_hit": false, + "model": "gpt-5.4-mini?service_tier=flex", + "cluster_id": "cluster-45115-2", + "nodes": [ + "pull_request:45115", + "pull_request:45569" + ], + "soft_pairs": [ + "pull_request:45115|pull_request:45569" + ], + "prepared_review_unit_hash": "677e3b5d98aa1c7786cc7c1b3b65013868eb55a9aa1b7941994e8b2e798120a4", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 6066, + "estimated_input_tokens": 1517, + "estimated_eval_tokens": 3290 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Two open NemotronH-related PRs, but they are not duplicates: one is a refactor/conversion-mapping cleanup, the other is a broader feature split adding NemotronH dense/sparse model variants, docs, and tests.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45569 is the better canonical PR because it is the more complete, user-visible implementation: it introduces the dense/sparse model split, updates auto-mapping/docs, and includes tests.", + "best_issue_reason": null, + "best_pr_reason": "PR #45569 is the strongest representative of the cluster since it covers the main architecture change and its surrounding API/docs support, whereas #45115 is a narrower refactor/conversion-mapping update.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45115", + "right": "pull_request:45569", + "accept": false, + "reason": "Both concern NemotronH, but they do not appear to fix the same concrete code-path problem. #45115 focuses on conversion-mapping and inheritance/refactor work, while #45569 introduces a broader dense/sparse model split with docs and tests. They 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": 20, + "component_count": 39, + "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": "f0e3aef2b3a89697a70686e666fcee532485f165f510bd12597ead200d2ea9f6", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5953, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5953, + "estimated_input_tokens": 1489, + "estimated_eval_tokens": 3234 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are closely related but not duplicates: #45333 adds per-layer heterogeneous configuration support, while #45332 extends that work into full heterogeneous model/runtime support. They overlap on the config/doc/test scaffolding, but they target different scopes and code paths.", + "confidence": 0.9, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "#45332 is the better cluster representative because it is the broader, end-to-end heterogeneous modeling PR and subsumes the config-layer groundwork rather than only the configuration piece.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45332", + "right": "pull_request:45333", + "accept": false, + "reason": "Related scope, but not the same change: #45332 covers heterogeneous modeling/runtime integration across cache, masking, model utils, and model support, while #45333 is limited to per-layer config plumbing. They could be sequential PRs, not 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": 21, + "component_count": 39, + "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": "a91af0794df3a69dd19919667b2a0e36349d231dfe89488b0940c4494557c823", + "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": "These are both typing-focused PRs, but they are not the same change: one is a broad repository-wide type-checking sweep, while the other is a narrower typing support update centered on `_typing.py` and `modeling_utils`-related protocols.", + "confidence": 0.89, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45425 is the more coherent concrete typing change: it introduces specific helper protocols and typing aliases used by `modeling_utils` and related code. PR 45415 is a much broader type-checking sweep across many files.", + "best_issue_reason": null, + "best_pr_reason": "45425 is the best representative PR for the cluster because it is focused on one typing problem domain and appears more self-contained; 45415 is too broad and mostly a repo-wide annotation pass.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45415", + "right": "pull_request:45425", + "accept": false, + "reason": "Both are typing-related and share some touched files, but they address different scopes: 45415 is a broad annotation/type-checking sweep, while 45425 adds specific typing aliases/protocols for modeling utilities. They do not look like the same underlying change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 22, + "component_count": 39, + "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": "fcc16111e712199093ec244257c70540cc898355d82270f159fca746a4b8dfa4", + "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 pull requests concern expert-parallelism loading behavior, but they fix different concrete failure modes: 45548 targets EP + DeepSpeed ZeRO-3 loading via accelerate launch and config serialization, while 45662 targets EP + FSDP2 where experts can be overwritten by rank-0 broadcast/post-shard handling. They are related by subsystem, not the same underlying change.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "45662 is the better representative PR because it addresses a concrete EP loading overwrite bug with a broader, more explicit fix across the loading path and EP integrations.", + "best_issue_reason": null, + "best_pr_reason": "45662 is the strongest standalone PR in this cluster: the title and diff point to a specific bug (experts being silently overwritten) and the code changes align with that failure mode.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45548", + "right": "pull_request:45662", + "accept": false, + "reason": "Related EP-loading area, but different bugs and code paths: 45548 is about serialize/config handling for accelerate+ZeRO-3, while 45662 is about post-shard wrapping and rank-0 overwrite in FSDP2." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 23, + "component_count": 39, + "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": "e5006ba79292d7e33f6f3a01301321f894b2378d0379bd78e7c2e592efc7ef69", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4673, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4673, + "estimated_input_tokens": 1169, + "estimated_eval_tokens": 2594 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both PRs address the same underlying xdist/captured_info logging collision in testing_utils, with PR 45645 expanding the fix to CI artifact collection as well. PR 45645 is the broader, more complete change set.", + "confidence": 0.88, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 45645 is the better canonical PR because it covers the core xdist-safe file handling plus the workflow-side log aggregation needed to preserve multiple captured_info artifacts.", + "best_issue_reason": null, + "best_pr_reason": "PR 45645 is the strongest representative of the cluster: it fixes the collision in the code path and also updates CI handling, making it the more complete end-to-end solution.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45639", + "right": "pull_request:45645", + "accept": true, + "reason": "Both PRs target the same captured_info/xdist collision problem in the testing debug-log path, and PR 45645 looks like a superset that could plausibly absorb PR 45639's core 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": 39, + "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": "0e2717cffb963ceaa8b8606f2ac72780343e88e0f9e771f96d3881e86274dace", + "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 unrelated linter/chore PRs: one implements TRF003 across model forwards, the other adds TRF018 initialization-rule changes. They touch overlapping model files, but the underlying code paths and fixes are different, so they should not be merged as duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #45720 is the more focused representative: it addresses TRF003 by adding `can_return_tuple` and adjusting `forward` signatures/return handling across multiple model implementations.", + "best_issue_reason": null, + "best_pr_reason": "PR #45720 is the best representative for the cluster because its scope is narrower and more directly tied to a single concrete rule change, making it easier to summarize as the canonical PR.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45720", + "right": "pull_request:46259", + "accept": false, + "reason": "Different lint rules and different fix patterns: TRF003 changes model `forward` return behavior, while TRF018 refactors weight initialization. Shared model files are incidental and not enough to indicate 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": 25, + "component_count": 39, + "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": 26, + "component_count": 39, + "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-45932-3", + "nodes": [ + "pull_request:45932", + "pull_request:46100", + "pull_request:46891" + ], + "soft_pairs": [ + "pull_request:45932|pull_request:46100", + "pull_request:45932|pull_request:46891", + "pull_request:46100|pull_request:46891" + ], + "prepared_review_unit_hash": "5dcacce9dccc81055f460d5437752d45829d1572675b789b932580830d8cf682", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7881, + "estimated_input_tokens": 1971, + "estimated_eval_tokens": 4198 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7881, + "estimated_input_tokens": 1971, + "estimated_eval_tokens": 4198 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These three PRs are only superficially similar because they all touch tests/CI, but they address different changes: grafana/telemetry plumbing, test migration/kernel coverage, and PEFT integration CI. I would not deduplicate any pair.", + "confidence": 0.93, + "canonical_issue_reason": null, + "canonical_pr_reason": "No issue artifact is present; among the PRs, #46891 is the most coherent representative because it has the clearest concrete change set centered on PEFT integration and related CI/test plumbing.", + "best_issue_reason": null, + "best_pr_reason": "#46891 is the best single representative: it has the most specific, self-contained purpose and a clearer code-path focus than the mostly infrastructure/test-migration PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:45932", + "right": "pull_request:46100", + "accept": false, + "reason": "Both are test-related, but they change different things: #45932 is grafana/telemetry CI plus model config tweaks, while #46100 adds AltCLIP/ViT kernel and DDP tests. No shared concrete bug or mergeable single change." + }, + { + "left": "pull_request:45932", + "right": "pull_request:46891", + "accept": false, + "reason": "They only overlap on a workflow file. #45932 is testing grafana/otel plumbing with unrelated model edits, while #46891 is PEFT integration CI and peft.py/test-fetcher changes. Different underlying changes." + }, + { + "left": "pull_request:46100", + "right": "pull_request:46891", + "accept": false, + "reason": "Both involve tests, but one is a kernel-test migration for AltCLIP/ViT and the other is PEFT integration CI with peft integration code. They do not fix the same code-path problem and would not plausibly merge 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": 27, + "component_count": 39, + "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": "6a0d4528e46620f371a11239f3aef067211851a9e8b39031b7a5fed07e83f67a", + "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 duplicates: PR 46126 is a focused EP/Trainer integration feature, while PR 46269 is a broad distributed branch base touching many unrelated areas. The overlap in a few distributed files is not enough to merge them as the same change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46126 is the more canonical change because it targets one concrete feature area (EP + Trainer integration) with a narrower, coherent scope. PR 46269 is a broad base branch update spanning many subsystems, so it is less suitable as the representative PR.", + "best_issue_reason": null, + "best_pr_reason": "PR 46126 is the best representative of a specific code-path change. PR 46269 is too wide and mixed to serve as a clean duplicate-triage anchor for this cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46126", + "right": "pull_request:46269", + "accept": false, + "reason": "Reject: despite shared distributed filenames, the PRs do not appear to fix the same concrete bug or implement the same change. 46126 is a targeted EP/Trainer integration PR, while 46269 is a broad distributed branch base that bundles many unrelated updates." + } + ] + }, + "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": 39, + "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": "e6a31b6b72a229c62c8b6c75abe8e43081cc6f631adf05fd77549696f2cd2f70", + "component_budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 29, + "serialized_chars": 25924, + "estimated_input_tokens": 6481, + "estimated_eval_tokens": 13218 + }, + "budget": { + "node_count": 10, + "item_count": 10, + "soft_pair_count": 29, + "serialized_chars": 25924, + "estimated_input_tokens": 6481, + "estimated_eval_tokens": 13218 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is mostly unrelated workflow-maintenance PRs. The pr-ci-security-gate items are thematically similar, but they change different security controls/steps, so they are not true duplicates under a strict mergeability test.", + "confidence": 0.91, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46554 is the most concrete and self-contained workflow-security hardening change in the set: it pins the downloaded monitor script, verifies its checksum, and clears HF_TOKEN before SSH.", + "best_issue_reason": null, + "best_pr_reason": "PR 46554 is the best representative PR for this cluster because it has the clearest behavioral impact and strongest security rationale, unlike the other mostly generic workflow tweak PRs.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46138", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflow files and different behavior; only the generic chore style is shared." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46558", + "accept": false, + "reason": "Unrelated workflows: one changes self-comment CI input plumbing, the other hardens the PR CI security gate." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow purposes and code paths; the edits are not the same change." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46716", + "accept": false, + "reason": "Different workflow files and different semantics; no shared concrete bug fix." + }, + { + "left": "pull_request:46138", + "right": "pull_request:46504", + "accept": false, + "reason": "One updates self-comment CI stdin handling, the other tweaks commit-message matching in build image CI." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46503", + "accept": false, + "reason": "SSH runner hardening and benchmark trigger logic are unrelated changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46501", + "accept": false, + "reason": "Different workflows and different fixes; no common underlying bug." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46502", + "accept": false, + "reason": "SSH runner changes do not match reviewer-assignment checkout behavior." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows with unrelated security/automation changes." + }, + { + "left": "pull_request:46432", + "right": "pull_request:46558", + "accept": false, + "reason": "Not the same problem: SSH runner token handling vs PR CI gate permissions/result flow." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46503", + "accept": false, + "reason": "Both touch CI workflows, but one is anti-slop permissions and the other is benchmark dispatch behavior." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46502", + "accept": false, + "reason": "Different workflow files and different automation concerns." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46504", + "accept": false, + "reason": "Anti-slop permissions and build-ci image triggering are unrelated." + }, + { + "left": "pull_request:46501", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different concrete changes; not mergeable as one fix." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46503", + "accept": false, + "reason": "Reviewer checkout ref fix and benchmark trigger expansion are unrelated." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46504", + "accept": false, + "reason": "Assign-reviewers checkout semantics do not match build-ci commit parsing." + }, + { + "left": "pull_request:46502", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows and different code paths; no shared bug/change." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46504", + "accept": false, + "reason": "Benchmark dispatch logic and build-ci message parsing are distinct automation changes." + }, + { + "left": "pull_request:46503", + "right": "pull_request:46554", + "accept": false, + "reason": "Different workflows with no common underlying fix." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46554", + "accept": false, + "reason": "Build-ci commit-message parsing and SSH runner security hardening are unrelated." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46558", + "accept": false, + "reason": "Different workflows and different fixes; only the broad CI theme overlaps." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46558", + "accept": false, + "reason": "Same workflow file, but different hardening edits affecting different steps and semantics." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46563", + "accept": false, + "reason": "Both touch pr-ci-security-gate.yml, but one adds checks/checkout while the other changes reusable-workflow commentary; not the same fix." + }, + { + "left": "pull_request:46554", + "right": "pull_request:46716", + "accept": false, + "reason": "Security-gate hardening vs secrets-passing change in a reusable workflow call; different concrete changes." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46563", + "accept": false, + "reason": "Same file, but one changes permissions and gating, the other adds safety notes to a reusable workflow call." + }, + { + "left": "pull_request:46558", + "right": "pull_request:46716", + "accept": false, + "reason": "Permissions/approval-flow tweak is not the same change as replacing inherited secrets with an empty secrets map." + }, + { + "left": "pull_request:46563", + "right": "pull_request:46716", + "accept": false, + "reason": "Both relate to the same reusable workflow, but one only documents safety and the other changes secrets handling; not duplicate." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46563", + "accept": false, + "reason": "Different workflow files and different automation logic." + }, + { + "left": "pull_request:46504", + "right": "pull_request:46716", + "accept": false, + "reason": "Build-ci docker image tagging and PR CI reusable workflow secrets are unrelated." + } + ] + }, + "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": 39, + "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": "d9fcb4695f21a615f1d0ede6ef7b643885a033f71411d7c337bbe9171a737670", + "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": "These PRs share a broad theme (replacing file-open source introspection), but they fix different code paths in different modules, so they are not duplicates.", + "confidence": 0.95, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: PR 46152 and PR 46207 address separate source-reading implementations in different files with different behavior changes.", + "best_issue_reason": null, + "best_pr_reason": "PR 46207 is the strongest standalone PR here because it has a clear, localized fix in modeling_utils.py and includes tests, making it the better representative of the cluster.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46152", + "right": "pull_request:46207", + "accept": false, + "reason": "Both replace file-open source access, but one fixes import_utils.py traversal logic and the other fixes _can_set_attn_implementation in modeling_utils.py; different underlying bugs and not 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": 30, + "component_count": 39, + "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": 31, + "component_count": 39, + "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-46281-2", + "nodes": [ + "pull_request:46281", + "pull_request:46890" + ], + "soft_pairs": [ + "pull_request:46281|pull_request:46890" + ], + "prepared_review_unit_hash": "858b17423563424d455f7837bf2a659640e4a9762ea9e447ace26284964c4e57", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4270, + "estimated_input_tokens": 1068, + "estimated_eval_tokens": 2392 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4270, + "estimated_input_tokens": 1068, + "estimated_eval_tokens": 2392 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests harden local-loading code against path traversal, but they fix different untrusted inputs in different code paths: sharded checkpoint filenames from `weight_map` vs. `transformers_weights` from config. Similar security theme, not the same underlying bug or mergeable change.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "PR 46890 is the cleaner standalone representative of the security-hardening theme: it adds explicit containment checks for one untrusted local config field with a narrowly scoped fix and test coverage.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46281", + "right": "pull_request:46890", + "accept": false, + "reason": "Different attack surfaces and loaders: 46281 validates shard paths from a checkpoint index, while 46890 validates `transformers_weights` from config. Same broad CWE-22 theme, but not the same concrete bug or a single PR candidate." + } + ] + }, + "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": 39, + "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": "802991a032ccebc67464dcddfd5b4c3e0687506ac95b88b994fc7d49f874f1d8", + "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 address the same sliding-window right-boundary off-by-one fix in Moonshine streaming; the later PR expands the same correction to Gemma4 and adds tests.", + "confidence": 0.87, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR #46546 is the better canonical PR because it contains the same Moonshine streaming boundary fix plus a test update and a related Gemma4 application, making it the more complete change set.", + "best_issue_reason": null, + "best_pr_reason": "PR #46546 is the strongest representative of the bug fix because it covers the core Moonshine streaming code path, includes validation, and appears to subsume the smaller Moonshine-only patch in #46310.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46310", + "right": "pull_request:46546", + "accept": true, + "reason": "Same off-by-one right-window boundary bug in the Moonshine streaming mask; the second PR applies the same fix in the same code path and could plausibly subsume the first." + } + ] + }, + "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": 39, + "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-46376-6", + "nodes": [ + "pull_request:46376", + "pull_request:46417", + "pull_request:46425", + "pull_request:46509", + "pull_request:46850", + "pull_request:46917" + ], + "soft_pairs": [ + "pull_request:46376|pull_request:46417", + "pull_request:46376|pull_request:46425", + "pull_request:46425|pull_request:46509", + "pull_request:46425|pull_request:46917", + "pull_request:46425|pull_request:46850", + "pull_request:46850|pull_request:46917" + ], + "prepared_review_unit_hash": "aa876c0dfcac8c902d32330dffae7955c0572bdedef44c3a1462fc1e2dbc536c", + "component_budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 6, + "serialized_chars": 16835, + "estimated_input_tokens": 4209, + "estimated_eval_tokens": 8674 + }, + "budget": { + "node_count": 6, + "item_count": 6, + "soft_pair_count": 6, + "serialized_chars": 16835, + "estimated_input_tokens": 4209, + "estimated_eval_tokens": 8674 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "This cluster is heterogeneous: multimodal continuous batching support, new model additions (HunYuan VL, RADIO, Nemotron Omni), a Gemma masking bug fix, and a deprecation cleanup. They are not duplicates of one another.", + "confidence": 0.96, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": null, + "soft_edge_verdicts": [ + { + "left": "pull_request:46376", + "right": "pull_request:46417", + "accept": false, + "reason": "Different changes: continuous batching multimodal infrastructure vs adding a new HunYuan VL model and docs/tests." + }, + { + "left": "pull_request:46376", + "right": "pull_request:46425", + "accept": false, + "reason": "Different targets and code paths: continuous batching cache/scheduler work vs RADIO model support and auto-mapping/docs." + }, + { + "left": "pull_request:46425", + "right": "pull_request:46509", + "accept": false, + "reason": "Both are model additions, but for different architectures and code paths (RADIO vs Nemotron Omni), so not the same change." + }, + { + "left": "pull_request:46425", + "right": "pull_request:46917", + "accept": false, + "reason": "RADIO model support is unrelated to removing deprecated backend/training-args behavior." + }, + { + "left": "pull_request:46425", + "right": "pull_request:46850", + "accept": false, + "reason": "RADIO model support and Gemma bidirectional/sliding-window masking fixes are unrelated changes." + }, + { + "left": "pull_request:46850", + "right": "pull_request:46917", + "accept": false, + "reason": "Gemma masking logic fix is unrelated to the deprecation cleanup in image processing/training args." + } + ] + }, + "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": 39, + "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": "4c89d7176224392603d7e05e08077f0129adc27dc78e1c0784e2ece1d15ec673", + "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 two PRs are unrelated: one fixes CodeLlama decode behavior around leading spaces, the other fixes DeepSeek checkpoint routing to the correct tokenizer class. They touch different code paths, different models, and would not plausibly merge into one change.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": null, + "best_issue_reason": null, + "best_pr_reason": "No single PR is a good canonical representative because the changes address different bugs in different tokenizer code paths; neither should be treated as the cluster\u2019s sole fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46574", + "right": "pull_request:46580", + "accept": false, + "reason": "Different underlying problems: CodeLlama decode stripping vs. DeepSeek tokenizer-class misloading. Separate models, separate files, separate fixes." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 35, + "component_count": 39, + "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": "299fc604e791f9bc3f4e1e380c871cf9e04185c22ae6a3cbc05e8c1ab85c19cd", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5496, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 5496, + "estimated_input_tokens": 1374, + "estimated_eval_tokens": 3004 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These are related to Mistral internals but they are different PRs: one refactors/moves the existing MistralConverter, the other adds tekken tokenizer detection/conversion/save support. They should not be deduplicated.", + "confidence": 0.97, + "canonical_issue_reason": null, + "canonical_pr_reason": "No canonical PR: 46603 is a package move/refactor for MistralConverter, while 46604 introduces new tekken tokenizer functionality. Different underlying changes.", + "best_issue_reason": null, + "best_pr_reason": "Neither PR is a good global duplicate representative because they solve different problems in the same subsystem.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46603", + "right": "pull_request:46604", + "accept": false, + "reason": "Shared Mistral files do not imply the same change; 46603 is a refactor/move of MistralConverter, whereas 46604 adds tekken tokenizer detection, conversion, and save utilities. They could not plausibly be merged into one PR as the same fix." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 36, + "component_count": 39, + "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-46741-3", + "nodes": [ + "pull_request:46741", + "pull_request:46773", + "pull_request:46781" + ], + "soft_pairs": [ + "pull_request:46741|pull_request:46773", + "pull_request:46741|pull_request:46781" + ], + "prepared_review_unit_hash": "7e7918e6821bece01ec0f5c28bd7eb2c2b45d76020477bfbec6cbcc38081cd55", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8455, + "estimated_input_tokens": 2114, + "estimated_eval_tokens": 4484 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 2, + "serialized_chars": 8455, + "estimated_input_tokens": 2114, + "estimated_eval_tokens": 4484 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These PRs are thematically related because they touch overlapping model families and cache/padding paths, but they fix three different bugs: chunked-prefill/speculative decoding, padding-mask handling for batch size 1, and QuantizedCache construction for hybrid models.", + "confidence": 0.92, + "canonical_issue_reason": null, + "canonical_pr_reason": "No true duplicate PR exists here; if one must be chosen as the representative, 46741 is the broadest and most central model-path fix across the affected family.", + "best_issue_reason": null, + "best_pr_reason": "46741 is the strongest standalone candidate because it targets a concrete runtime bug in the core forward/cache path and spans the same affected model family more directly than the others.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46741", + "right": "pull_request:46773", + "accept": false, + "reason": "Different root causes: 46741 changes precomputed-state handling for chunked prefill/speculative decoding, while 46773 fixes padding-mask application when batch size is 1. Shared files/models are not enough to make them the same bug." + }, + { + "left": "pull_request:46741", + "right": "pull_request:46781", + "accept": false, + "reason": "Different code paths and failures: 46741 is about cache reuse in model forward passes, whereas 46781 fixes QuantizedCache initialization for hybrid layer types. They are related to caching, but not the same underlying issue." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 37, + "component_count": 39, + "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-46895-3", + "nodes": [ + "pull_request:46895", + "pull_request:46898", + "pull_request:46916" + ], + "soft_pairs": [ + "pull_request:46895|pull_request:46898", + "pull_request:46895|pull_request:46916", + "pull_request:46898|pull_request:46916" + ], + "prepared_review_unit_hash": "45350dd8e9b6741bafb86181dbcd404cfd2b8f6d113a17b3ea3347554c11376b", + "component_budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7714, + "estimated_input_tokens": 1929, + "estimated_eval_tokens": 4114 + }, + "budget": { + "node_count": 3, + "item_count": 3, + "soft_pair_count": 3, + "serialized_chars": 7714, + "estimated_input_tokens": 1929, + "estimated_eval_tokens": 4114 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Three open PRs address the same family of training-loss double-shift bugs, but they split across two model code paths: CohereASR and Florence2. The combined PR 46916 is the best representative because it includes the CohereASR fix and also covers Florence2.", + "confidence": 0.79, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46916 is the broadest and most central change: it includes the CohereASR loss fix from 46895 and the Florence2 fix from 46898, so it best represents the shared double-shift correction work in this cluster.", + "best_issue_reason": null, + "best_pr_reason": "PR 46916 is the strongest cluster anchor because it subsumes one concrete fix from each of the two smaller PRs and directly reflects the common training-loss double-shift pattern.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46895", + "right": "pull_request:46898", + "accept": false, + "reason": "They fix the same kind of bug pattern, but in different model implementations and different concrete code paths (CohereASR vs Florence2), so they are not the same PR-level change." + }, + { + "left": "pull_request:46895", + "right": "pull_request:46916", + "accept": true, + "reason": "46916 includes the same CohereASR loss-fix code path as 46895, so this is the same underlying CohereASR change even though 46916 also adds Florence2." + }, + { + "left": "pull_request:46898", + "right": "pull_request:46916", + "accept": true, + "reason": "46916 includes the same Florence2 double-shift fix as 46898, so they overlap on the same underlying Florence2 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": 38, + "component_count": 39, + "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-46907-2", + "nodes": [ + "pull_request:46907", + "pull_request:46908" + ], + "soft_pairs": [ + "pull_request:46907|pull_request:46908" + ], + "prepared_review_unit_hash": "0b308c8fd0da32e7663075c1b69aa955e780fa85e3cfb54111be129f74c47547", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4384, + "estimated_input_tokens": 1096, + "estimated_eval_tokens": 2448 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4384, + "estimated_input_tokens": 1096, + "estimated_eval_tokens": 2448 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "These two PRs are unrelated: one fixes EdgeTam\u2019s forward pass by dropping an invalid `original_sizes` kwarg, while the other adjusts DeepSeekV32 integration tests to supply an offload folder during `from_pretrained` loading.", + "confidence": 0.98, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46907 is the better canonical representative because it fixes a concrete runtime/code-path bug in the model implementation itself, whereas PR 46908 is a narrower test-only workaround for loading setup.", + "best_issue_reason": null, + "best_pr_reason": "PR 46907 is the strongest standalone PR in the cluster: it addresses a specific integration failure at the source code boundary and is more broadly meaningful than the test harness change in PR 46908.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46907", + "right": "pull_request:46908", + "accept": false, + "reason": "Different models, different failure modes, and different fix types. EdgeTam removes an unsupported kwarg before calling the backbone; DeepSeekV32 only adds `offload_folder` in tests for loading. They do not look like the same underlying bug or change." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + }, + { + "label": "pull_request", + "component_index": 39, + "component_count": 39, + "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-46915-2", + "nodes": [ + "pull_request:46915", + "pull_request:46923" + ], + "soft_pairs": [ + "pull_request:46915|pull_request:46923" + ], + "prepared_review_unit_hash": "8df4131cd112becff605df3293d82ab4153e757286944b903cecc76445bebdf8", + "component_budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4252, + "estimated_input_tokens": 1063, + "estimated_eval_tokens": 2382 + }, + "budget": { + "node_count": 2, + "item_count": 2, + "soft_pair_count": 1, + "serialized_chars": 4252, + "estimated_input_tokens": 1063, + "estimated_eval_tokens": 2382 + }, + "overflow_policy": "truncate_then_skip", + "trimmed": false, + "aggressively_trimmed": false, + "split": false, + "analyst_result": { + "summary": "Both pull requests target the same DIA docs page and the same tracking issue, but they make different fixes to different code examples, so they look related rather than true duplicates.", + "confidence": 0.76, + "canonical_issue_reason": null, + "canonical_pr_reason": "PR 46923 is the better canonical candidate because it appears to provide the more complete docs correction, covering multiple example blocks on the DIA page with a consistent fix.", + "best_issue_reason": null, + "best_pr_reason": "PR 46923 is the strongest standalone choice for the cluster: it addresses the broken DIA documentation examples more broadly and looks like the fuller, cleaner fix.", + "soft_edge_verdicts": [ + { + "left": "pull_request:46915", + "right": "pull_request:46923", + "accept": false, + "reason": "Same docs file and same issue target, but the diffs change different example snippets and use different remediation approaches, so they are not clearly the same underlying change and would not safely merge into one PR." + } + ] + }, + "evaluator_result": null, + "evaluator_used": false, + "retried": false, + "accepted_nontrivial_soft_edge": false, + "error_kind": null, + "error_message": null + } + ] +} diff --git a/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/manifest.json b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..4885be8e58a934425cc17ce9d861c4d84a2ee279 --- /dev/null +++ b/snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/manifest.json @@ -0,0 +1,14 @@ +{ + "analysis_id": "hybrid-model-20260626t180014z", + "artifacts": { + "hybrid": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "model": null, + "published_at": "2026-06-26T18:10:13Z", + "repo": "huggingface/transformers", + "schema_version": 1, + "snapshot_id": "20260626T180014Z", + "variant": "hybrid" +} diff --git a/snapshots/20260626T180014Z/manifest.json b/snapshots/20260626T180014Z/manifest.json index 75551b4c80fb39c0b9f8133c0fa170dac0ceb819..9f5ad413b93cb6baa314c9f1b14e66f75edba1f5 100644 --- a/snapshots/20260626T180014Z/manifest.json +++ b/snapshots/20260626T180014Z/manifest.json @@ -32,6 +32,24 @@ "timeline_events": 918 }, "extracted_at": "2026-06-26T18:00:14Z", + "published_analysis": { + "canonical_analysis_id": "hybrid-model-20260626t180014z", + "runs": { + "hybrid-model-20260626t180014z": { + "analysis_id": "hybrid-model-20260626t180014z", + "artifacts": { + "hybrid": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.json", + "hybrid_reviews": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/analysis-report-hybrid.llm-reviews.json" + }, + "channel": "canonical", + "manifest_path": "snapshots/20260626T180014Z/analysis-runs/hybrid-model-20260626t180014z/manifest.json", + "model": null, + "published_at": "2026-06-26T18:10:13Z", + "variant": "hybrid" + } + }, + "schema_version": 1 + }, "repo": "huggingface/transformers", "snapshot_id": "20260626T180014Z", "watermark": {